The New England Journal of Medicine Volume 344 Number 14 April 5, 2001

The New England Journal of Medicine
Volume 344
April 5, 2001
Original Articles
Efficacy and Safety of a Specific Inhibitor of the
BCR-ABL Tyrosine Kinase in Chronic Myeloid
Leukemia
Number 14
Editorial
Targeting the BCR-ABL Tyrosine Kinase in Chronic
Myeloid Leukemia
Health Policy 2001
Activity of a Specific Inhibitor of the BCR-ABL
Tyrosine Kinase in the Blast Crisis of Chronic Myeloid
Leukemia and Acute Lymphoblastic Leukemia with the
Philadelphia Chromosome
Long-Term Survival after Ablation of the
Atrioventricular Node and Implantation of a Permanent
Pacemaker in Patients with Atrial Fibrillation
Brief Report: Effect of the Tyrosine Kinase Inhibitor
STI571 in a Patient with a Metastatic Gastrointestinal
Stromal Tumor
Images in Clinical Medicine
Managed Care in Transition
Correspondence
Phenylpropanolamine and Hemorrhagic Stroke
Dietary Supplements Containing Ephedra Alkaloids
The Diagnosis and Treatment of Cough
Syncope
Invasive Pulmonary Aspergillosis Associated with
Infliximab Therapy
A Medical Mystery
Review Articles
Advances in Immunology: Complement (First of Two
Parts)
Medical Progress: Atrial Fibrillation
Clinical Problem-Solving
Less Is More
Book Reviews
Decoding Darkness: The Search for the Genetic Causes
of Alzheimer's Disease
The Hidden Structure: A Scientific Biography of Camillo
Golgi
Dear Mr. Darwin: Letters on the Evolution of Life and
Human Nature
Legal Issues in Medicine
Conjoined Twins -- The Limits of Law at the Limits of
Life
Copyright © 2001 Massachusetts Medical Society. All rights reserved.
The New England
Journal of Medicine
C o py r ig ht © 2 0 0 1 by t he Ma s s ac h u s e t t s Me d ic a l S o c ie t y
V O L U ME 3 4 4
A P R I L 5, 2001
NUMB ER 14
EFFICACY AND SAFETY OF A SPECIFIC INHIBITOR OF THE BCR-ABL TYROSINE
KINASE IN CHRONIC MYELOID LEUKEMIA
BRIAN J. DRUKER, M.D., MOSHE TALPAZ, M.D., DEBRA J. RESTA, R.N., BIN PENG, PH.D., ELISABETH BUCHDUNGER, PH.D.,
JOHN M. FORD, M.D., NICHOLAS B. LYDON, PH.D., HAGOP KANTARJIAN, M.D., RENAUD CAPDEVILLE, M.D.,
SAYURI OHNO-JONES, B.S., AND CHARLES L. SAWYERS, M.D.
ABSTRACT
Background BCR-ABL is a constitutively activated
tyrosine kinase that causes chronic myeloid leukemia (CML). Since tyrosine kinase activity is essential
to the transforming function of BCR-ABL, an inhibitor
of the kinase could be an effective treatment for CML.
Methods We conducted a phase 1, dose-escalating
trial of STI571 (formerly known as CGP 57148B), a specific inhibitor of the BCR-ABL tyrosine kinase. STI571
was administered orally to 83 patients with CML in
the chronic phase in whom treatment with interferon
alfa had failed. Patients were successively assigned
to 1 of 14 doses ranging from 25 to 1000 mg per day.
Results Adverse effects of STI571 were minimal;
the most common were nausea, myalgias, edema, and
diarrhea. A maximal tolerated dose was not identified.
Complete hematologic responses were observed in
53 of 54 patients treated with daily doses of 300 mg
or more and typically occurred in the first four weeks
of therapy. Of the 54 patients treated with doses of
300 mg or more, cytogenetic responses occurred in
29, including 17 (31 percent of the 54 patients who received this dose) with major responses (0 to 35 percent of cells in metaphase positive for the Philadelphia chromosome); 7 of these patients had complete
cytogenetic remissions.
Conclusions STI571 is well tolerated and has significant antileukemic activity in patients with CML in
whom treatment with interferon alfa had failed. Our
results provide evidence of the essential role of BCRABL tyrosine kinase activity in CML and demonstrate
the potential for the development of anticancer drugs
based on the specific molecular abnormality present
in a human cancer. (N Engl J Med 2001;344:1031-7.)
C
HRONIC myeloid leukemia (CML) is a
clonal disorder in which cells of the myeloid lineage undergo massive clonal expansion. The disease progresses through three
distinct phases — chronic phase, accelerated phase,
and blast crisis — during which the leukemic clone
progressively loses its ability to differentiate.1,2 Current therapies include allogeneic bone marrow transplantation and drug regimens including interferon
alfa.3,4 Interferon alfa prolongs overall survival but has
considerable adverse effects. Allogeneic bone marrow
transplantation, the only curative treatment for CML,
is associated with substantial morbidity and mortality and is limited to patients for whom a suitable donor is available.
The characteristic genetic abnormality of CML, the
Philadelphia (Ph) chromosome,5 results from a reciprocal translocation between the long arms of chromosomes 9 and 22.6 The molecular consequence of
this translocation is the generation of the fusion protein BCR-ABL, a constitutively activated tyrosine kinase, which is present in virtually all patients with
CML. In vitro studies and studies in animal models
have established that BCR-ABL alone is sufficient to
cause CML, and mutational analysis has established
that the tyrosine kinase activity of the protein is required for its oncogenic activity.7-10 For these reasons,
an inhibitor of the BCR-ABL tyrosine kinase should
be an effective and selective treatment for CML.
STI571 (4-[(4-methyl-1-piperazinyl)methyl]-N-[4methyl-3-[[4-(3-pyridinyl)-2-pyrimidinyl]amino]phen-
Copyright © 2001 Massachusetts Medical Society.
From the Division of Hematology and Medical Oncology, Oregon
Health Sciences University, Portland (B.J.D., S.O.-J.); the Departments of
Bioimmunotherapy (M.T.) and Leukemia (H.K.), University of Texas M.D.
Anderson Cancer Center, Houston; the Department of Oncology Clinical
Research, Novartis Pharmaceuticals, East Hanover, N.J. (D.J.R.), and Basel,
Switzerland (B.P., E.B., J.M.F., N.B.L., R.C.); and the Division of Hematology and Oncology, University of California at Los Angeles, Los Angeles
(C.L.S.). Address reprint requests to Dr. Druker at Oregon Health Sciences
University, L592, 3181 SW Sam Jackson Park Rd., Portland, OR 97201,
or at [email protected].
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1031
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
yl]benzamide methanesulfonate; Glivec, Novartis,
Basel, Switzerland) was synthesized after a compound
was identified by in vitro screening for tyrosine kinase
inhibitors and its activity was optimized for specific
kinases. STI571 functions through competitive inhibition at the ATP-binding site of the enzyme, which
leads to the inhibition of tyrosine phosphorylation of
proteins involved in BCR-ABL signal transduction. It
shows a high degree of specificity for BCR-ABL, the
receptor for platelet-derived growth factor, and c-kit
tyrosine kinases.11 STI571 causes arrest of growth or
apoptosis in hematopoietic cells that express BCR-ABL
but does not affect normal cells.12-14 On the basis of its
antileukemic activity in preclinical models, we conducted a phase 1 trial of STI571 in patients with CML
in whom treatment with interferon alfa had failed.
METHODS
Characteristics of the Patients
Patients with CML in the chronic phase (defined by the presence
of less than 15 percent blasts or basophils in the peripheral blood
or bone marrow) were eligible if they were 18 years of age or older,
if they tested positive for the Ph chromosome, and if treatment
with interferon alfa had failed. A failure of treatment with interferon
alfa was defined as the lack of a complete hematologic response
despite three months of treatment with a regimen containing interferon alfa (hematologic resistance), the lack of a cytogenetic response despite one year of treatment with a regimen containing
interferon alfa (cytogenetic resistance), or a hematologic or cytogenetic relapse. Patients with severe intolerance to interferon alfa
were included after safety data had been obtained for the first seven
cohorts of patients treated with STI571. The minimal interval between the discontinuation of prior therapies and the initiation of
treatment with STI571 was one week for hydroxyurea, two weeks
for interferon alfa and cytarabine, and six weeks for busulfan. Patients with a platelet count of less than 100,000 per cubic millimeter were excluded; adequate renal, hepatic, and cardiac function
and performance status were required. Written informed consent
was obtained from all patients before they enrolled in the study.
Study Design
The primary end point of this phase 1, dose-escalation trial was
the safety and tolerability of STI571; antileukemic activity was a
secondary end point. Patients were successively assigned to 1 of 14
dose cohorts, which ranged from 25 to 1000 mg per day. Doses of
STI571 were administered orally once daily, except for 800 and
1000 mg, which were administered twice daily as doses of 400 and
500 mg, respectively. Patients received continuous daily therapy
with STI571 unless unacceptable adverse effects or disease progression occurred. There was no intrapatient dose escalation. Dose escalation among cohorts was allowed if after 28 days of therapy,
none of three or one of six patients had grade 3 (severe) or grade
4 (life-threatening) adverse nonhematologic effects. No other cytoreductive agents were allowed during the study. Complete blood
counts were obtained twice a week for the first four weeks, once a
week for the next four weeks, and then once every two weeks. Assessments of bone marrow, including cytogenetic analyses, were
performed after 8 weeks of therapy and then once every 12 weeks.
Assessment of Toxicity and Response
Safety assessments included the evaluation of adverse events,
hematologic assessment, biochemical testing, urinalysis, and physical examination. Toxicity was graded in accordance with the Common Toxicity Criteria of the National Cancer Institute.15
A hematologic response was defined as a 50 percent reduction
in the white-cell count from base line, maintained for at least two
weeks. A complete hematologic response was defined as a reduction in the white-cell count to less than 10,000 per cubic millimeter and in the platelet count to less than 450,000 per cubic
millimeter, maintained for at least four weeks.
Cytogenetic responses were determined by the percentage of
cells in metaphase that were positive for the Ph chromosome in
the bone marrow. Cytogenetic responses, based on analysis of 20
cells in metaphase, were categorized as complete (no cells positive
for the Ph chromosome), partial (1 to 35 percent of cells positive
for the Ph chromosome), minor (36 to 65 percent of cells positive for the Ph chromosome), and absent (over 65 percent of cells
positive for the Ph chromosome). Major responses were defined
as complete or partial responses.
Pharmacokinetics
Samples for pharmacokinetic analysis were collected on day 1 and
day 28 of treatment, and plasma STI571 concentrations were determined with a liquid chromatographic and mass spectrophotometric assay. The concentration–time curves of STI571 in plasma
were evaluated by a noncompartmental analysis (with the use of
WinNonlin Pro, version 2.0, Pharsight, Mountain View, Calif.). The
variables analyzed were the time to the maximal concentration, the
maximal concentration, the terminal half-life, and the area under
the concentration–time curve (AUC) from time zero to infinity.
Assessment of BCR-ABL Tyrosine Kinase Inhibition
BCR-ABL kinase activity was determined from samples of peripheral blood obtained before the first dose of STI571 and two
hours after the second dose of STI571. Samples were collected in
tubes treated with heparin, and white cells were prepared and lysed
as described.16 Cell lysates were separated by electrophoresis on
12.5 percent sodium dodecyl sulfate–polyacrylamide gels, followed
by electrophoretic transfer to nylon membranes.16 The extent of
tyrosine phosphorylation of CRK-oncogene–like protein (CRKL)
in BCR-ABL–positive neutrophils17-19 was assessed by gel electrophoresis and immunoblotting with anti-CRKL antiserum.20
RESULTS
Enrollment of Patients
From June 1998 to May 2000, 83 patients in whom
treatment with interferon alfa had failed, or who could
not tolerate the drug, were enrolled at three participating study centers. The characteristics of the patients are summarized in Table 1. Of the 83 patients,
37 had hematologic resistance or relapse, 33 had cytogenetic resistance or relapse, and 13 could not tolerate interferon alfa. The median duration of disease
was 3.8 years (range, 0.8 to 14), and the median duration of therapy with interferon alfa was 8.5 months
(range, 1 week to 8.5 years). Nineteen patients had
had findings suggestive of accelerated disease (5 to
15 percent blasts or basophils in the bone marrow).
The median duration of treatment with STI571 was
310 days (range, 17 to 607). Half of the patients assigned to receive daily doses of 25, 50, or 85 mg of
STI571 were removed from the study within two
months because of elevated white-cell or platelet
counts requiring therapy prohibited by the protocol.
The study is ongoing and the results presented here
represent an interim analysis of the data.
Pharmacokinetics
STI571 was rapidly absorbed after oral administration, and a mean maximal concentration of 2.3 µg
1032 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
E F F ICACY AND SAF ET Y OF A SP ECIFIC INH IB ITOR OF BC R -A BL IN C H R ONIC MYELOID LEUK EMIA
TABLE 1. CHARACTERISTICS
OF THE
This amount exceeded the concentration required for
the inhibition of cellular phosphorylation by BCRABL (concentration required for 50 percent inhibition, 0.25 µM),12 and this concentration caused the
death of cell lines positive for BCR-ABL in vitro.12-14
The increase in the mean plasma STI571 AUC values was proportional to the administered dose.
83 PATIENTS.
CHARACTERISTIC
VALUE
Sex — no. (%)
Male
Female
History of disease — no. (%)
Hematologically resistant or relapsed CML
Cytogenetically resistant or relapsed CML
Intolerance to treatment with interferon
Age — yr
Median
Range
Duration of disease — yr
Median
Range
White-cell count at base line — cells/mm3
Median
Range
Platelet count at base line — cells/mm3
Median
Range
55 (66)
28 (34)
37 (45)
33 (40)
13 (16)
Safety Profile
STI571 was generally well tolerated, and a maximal tolerated dose was not identified. The frequency
of adverse effects attributable to STI571 is summarized in Table 2. The most common adverse effects
included nausea (in 43 percent of patients), myalgias
(41 percent), edema (39 percent), and diarrhea (25
percent). Most adverse effects, even at the highest doses, were grade 1 (mild) or grade 2 (moderate). Most
patients had a reduction in the hemoglobin level of
1 to 2 g per deciliter; the hemoglobin level typically
increased to base-line values or higher with continued
therapy. Two patients taking the 600-mg dose, one
patient taking the 800-mg dose, and one patient taking the 1000-mg dose had grade 3 anemia. Grade 3
thrombocytopenia and neutropenia occurred in 16
percent and 14 percent of the patients, respectively,
receiving doses of 200 mg or more. Elevations of liver-enzyme levels of grade 2 or higher were reported
in seven patients; in some of these patients, the abnormalities were reversed during treatment with STI571,
whereas in others, persistent elevations required the
temporary interruption of therapy or a reduction in
the dose.
55
19–76
3.8
0.8–14
27,800
9,400–199,000
430,000
102,000–1,814,000
per milliliter (4.6 µM) was reached at steady state by
once-daily administration of 400 mg of STI571. The
half-life of the drug in the circulation ranged from
13 to 16 hours, and the levels of the drug increased
by a factor of 2 or 3 at steady state with once-daily
dosing. The mean plasma trough concentration was
0.72 µg per milliliter (1.46 µM) 24 hours after the
administration of 400 mg of STI571 at steady state.
TABLE 2. DRUG-RELATED ADVERSE EVENTS ACCORDING
ADVERSE EVENT
25–140 mg
(N=14)
GRADE
OR
2
1
200–300 mg
(N=23)
GRADE
OR
4
3
GRADE
OR
2
1
GRADE
OR
4
TO THE
DAILY DOSE
350–500 mg
(N=18)
3
GRADE
OR
2
1
GRADE
OR
4
OF
STI571.*
600–1000 mg
(N=28)
3
GRADE
OR
2
1
GRADE
OR
4
% of patients
Nausea
Myalgias
Edema
Diarrhea
Fatigue
Rash
Dyspepsia
Vomiting
Thrombocytopenia
Neutropenia
Arthralgias
21
21
21
14
14
7
14
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
30
52
22
4
22
17
13
13
4
9
4
0
0
0
0
0
0
0
0
0
4
0
50
33
33
33
11
11
28
11
11
6
6
TOTAL
(N=83)
3
GRADES
1–4
no. (%)
0
6
0
0
0
0
0
0
6
6
0
59
28
55
38
24
28
17
34
7
0
28
0
14
7
3
3
3
0
0
24
24
3
36 (43)
34 (41)
32 (39)
21 (25)
17 (20)
16 (19)
15 (18)
15 (18)
13 (16)
12 (14)
11 (13)
*The adverse events listed here were considered to be related to STI571 and were reported in more than 10 percent
of patients. A grade of 1 indicates a mild adverse effect, a grade of 2 a moderate effect, a grade of 3 a severe effect, and
a grade of 4 a life-threatening effect.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1033
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
Only two patients receiving doses of STI571 of
300 mg or more discontinued therapy prematurely.
One patient with a history of coronary artery disease
discontinued therapy because of the recurrence of angina, and a second patient discontinued therapy because of a persistent and progressive rash. There were
no deaths during the study.
logic response was evident within four weeks after
the initiation of treatment. Complete hematologic responses have been maintained in 51 of 53 patients
with a median follow-up of 265 days (range, 17 to
468). One patient relapsed with chronic-phase disease,
whereas CML progressed to the blast phase in a second patient.
Hematologic Responses
Cytogenetic Responses
Hematologic responses occurred in all patients who
were treated with 140 mg or more of STI571 per
day (Table 3). Of the patients treated with daily doses
of 300 mg or more, 53 of 54 had complete hematologic responses and 1 discontinued therapy prematurely (on day 17) because of the recurrence of angina.
Hematologic responses typically occurred within two
weeks after the initiation of therapy with STI571, as
illustrated in Figure 1. In all but one patient treated
with 300 mg or more per day, a complete hemato-
One patient each in the groups receiving daily doses
of 200 mg and 250 mg had a cytogenetic response. As
shown in Table 4, 29 of the 54 patients treated with
doses of 300 mg or more per day (54 percent) had
major or minor cytogenetic responses. Of the 54 patients, 17 (31 percent of the group receiving 300 mg
or more per day) had major responses (35 percent or
less of cells in metaphase positive for the Ph chromosome); 7 of these were complete cytogenetic remissions (13 percent).
Figure 2 shows data on the 17 patients who had a
major cytogenetic response. Cytogenetic responses occurred as early as 2 months and as late as 10 months
after the initiation of treatment with STI571. The median time to the best cytogenetic response (the lowest percentage of cells in metaphase that were positive for the Ph chromosome) was 148 days (range,
48 to 331). Two of the seven patients who had a complete cytogenetic response tested negative for BCRABL by fluorescence in situ hybridization, and one
patient tested negative for BCR-ABL messenger RNA
(mRNA) by the polymerase chain reaction.
TABLE 3. HEMATOLOGIC RESPONSES.
ALL
DOSE (mg/DAY)
PATIENTS
PATIENTS WITH
RESPONSES
no.
25 or 50
85
140
200 or 250
300–1000
Total
6
4
3
16
54
83
PATIENTS WITH
COMPLETE
RESPONSES
no. (%)
2 (33)
2 (50)
3 (100)
16 (100)
54 (100)
77 (93)
0
1
1
9
53
64
(25)
(33)
(56)
(98)
(77)
Inhibition of BCR-ABL–Induced Tyrosine Phosphorylation
Blood samples from treated patients were tested
to determine whether BCR-ABL tyrosine kinase activity was inhibited. A major substrate of the enzyme
is CRKL, which is the most heavily tyrosine-phos-
White-Cell CountE
(cells ¬10¡3/mm3)
100
10
1
0
30
60
90
120
150
Duration of Treatment with STI571 (days)
Figure 1. Hematologic Responses in Six Patients Receiving 500 mg of STI571 per day.
Each line represents the white-cell counts for an individual patient. The dotted line indicates the upper
limit of a normal white-cell count.
1034 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
E F F ICACY AND SAF ET Y OF A SP ECIFIC INH IB ITOR OF BC R -A BL IN C H R ONIC MYELOID LEUK EMIA
tients receiving a daily dose of 140 mg and appeared
to reach a plateau in patients receiving a daily dose
of 250 to 750 mg (Fig. 3).
TABLE 4. CYTOGENETIC RESPONSES.
PATIENTS
COMPLETE
MAJOR
RESPONSES
DOSE (mg/DAY)
OR
PATIENTS
no.
300–350
400
500
600
750
800
1000
Total
DISCUSSION
PATIENTS
MINOR
RESPONSES
WITH
ALL
WITH
The presence of the BCR-ABL fusion protein in
virtually all patients with CML and its required tyrosine kinase activity make CML ideal for testing a
specific inhibitor of this enzyme. In our phase 1 study,
STI571, an oral, specific inhibitor of the BCR-ABL
tyrosine kinase, was well tolerated and had substantial activity against CML. These results were obtained
in patients with late-stage disease, in all of whom
standard therapy with interferon alfa had failed.
The rate of complete hematologic responses increased as the daily dose increased from 85 mg to
250 mg and reached 98 percent in patients treated
with 300 mg or more of STI571. Complete hematologic responses typically occurred within four weeks
after the initiation of therapy. The exception was a patient in the 350-mg group in whom the plasma halflife of STI571 was short (seven hours) and the AUC
was similar to that for patients in the 85-mg group.
STI571 is metabolized primarily by the CYP3A4 enzyme, and the patient in the 350-mg group was being
treated with phenytoin, a known inducer of CYP3A4.
When treatment with phenytoin was discontinued and
the dose of STI571 was increased to 500 mg, the patient had a complete hematologic response associated
with trough levels of STI571 similar to those observed
no. (%)
13
6
6
8
6
8
7
54
5
3
1
4
2
1
1
17
(38)
(50)
(17)
(50)
(33)
(12)
(14)
(31)
2
2
1
4
0
2
1
12
(15)
(33)
(17)
(50)
(0)
(25)
(14)
(22)
phorylated protein in neutrophils from patients with
CML.17-20 CRKL that is phosphorylated by BCR-ABL
migrates more slowly on electrophoresis than the unphosphorylated form.21 Low doses (25 to 50 mg) of
STI571 caused no alteration in the mobility of CRKL.
An increase in the levels of the rapidly migrating unphosphorylated form and a concomitant decrease in
the levels of the slowly migrating phosphorylated form
were seen in patients receiving the 85-mg dose of
STI571; these changes were more prominent in pa-
100
Ph-Chromosome–Positive CellsE
in Metaphase (%)
90
80
70
60
50
40
30
20
10
0
0
50
100
150
200
250
300
350
400
450
Duration of Treatment with STI571 (days)
Figure 2. Patients with a Major Cytogenetic Response.
The percentage of cells in metaphase positive for the Ph chromosome (in bone marrow) and the number of days that the patients received STI571 are shown. Each line represents the cytogenetic response
for an individual patient.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1035
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
Phosphorylated CRKL
250 mg
2
D
ay
1
D
ay
2
1
D
ay
2
1
140 mg
D
ay
D
ay
2
D
ay
85 mg
D
ay
1
2
25 mg
D
ay
D
ay
D
ay
1
Unphosphorylated CRKL
750 mg
Treatment with STI571
Figure 3. Immunoblot Assays Demonstrating the Degree of Phosphorylation of the BCR-ABL Substrate CRKL in Individual Patients
in the Groups Receiving Daily Doses of 25, 85, 140, 250, and 750 mg of STI571.
in other patients in the 500-mg group. This observation suggests that drugs that induce CYP3A4 could
lead to low and ineffective levels of STI571, whereas
drugs that interfere with the activity of the CYP3A4
enzyme could decrease the metabolism of STI571,
leading to increased levels of STI571 and thereby
causing toxic effects.
During treatment with STI571, blood counts gradually returned to normal during the first month, suggesting that the drug does not rapidly induce apoptosis, as would be expected with standard chemotherapy.
Blood counts were maintained within normal limits
regardless of whether a cytogenetic response was observed. This suggests that inhibition of the BCR-ABL
tyrosine kinase restores normal regulatory behavior
to the leukemic clone. Cytogenetic responses might
follow if, over time, the leukemic clone was displaced
either by normal hematopoietic progenitors that had
regained a proliferative advantage or by differentiation of the leukemic stem cell, which leads to its elimination.
Of the 54 patients treated with at least 300 mg of
STI571 per day, 29 (54 percent) had cytogenetic responses, including 7 with complete cytogenetic remissions. As compared with the cytogenetic responses
during therapy with interferon alfa, those during treatment with STI571 occurred relatively rapidly. In one
patient, BCR-ABL mRNA could not be detected by
polymerase chain reaction.
The most frequent adverse effects that seemed to
be related to treatment with STI571 were nausea,
edema, myalgias, and diarrhea; overall, most were
mild. In seven patients, there were elevations of liverenzyme levels of grade 2 or higher, as was also seen
in our study of patients with acute leukemia, reported elsewhere in this issue of the Journal.22 Myelosuppression, which occurred in up to a quarter of the
patients, was not dose limiting and was managed by
temporary interruption of treatment or dose reduction. Myelosuppression may be either a consequence
of a pharmacologic effect of STI571 through inhibi-
tion of c-kit or a reflection of compromised underlying normal hematopoiesis in patients with leukemia.
We did not identify a maximal tolerated dose for
STI571, but other end points could be used to choose
a dose for future trials. One is the pharmacokinetic
profile of STI571. Levels of the drug that killed CML
cells in vitro correlated well with clinical response and
serum drug levels in the 400-mg group. The dose–
response curve clearly demonstrates a relation between
dose and hematologic response. Also, there is substantial in vivo inhibition of the enzymatic activity of
BCR-ABL at the 400-mg dose, as demonstrated by
decreased phosphorylation of CRKL, a substrate of
BCR-ABL. For these reasons, we recommend a daily
dose of at least 400 mg for future studies.
These results show that the BCR-ABL tyrosine kinase is critical to the development of CML and demonstrate the potential for the development of anticancer drugs based on the specific molecular abnormality
in a human cancer.
Supported by grants from the National Cancer Institute (CA65823, to
Dr. Druker, and CA32737, to Dr. Sawyers) and by Novartis Pharmaceuticals. Dr. Druker is the recipient of a Translational Research Award from the
Leukemia and Lymphoma Society, and Dr. Sawyers is a Scholar of the Leukemia and Lymphoma Society.
Drs. Druker, Talpaz, and Sawyers served as consultants to Novartis Pharmaceuticals during the design of this study.
We are indebted to the following people for their assistance with
various aspects of this study: Alex Matter, Juerg Zimmerman, John
Goldman, Gregory Burke, David Parkinson, Michael Hayes, Ulrike
Zoellner, William Palo, Marianne Rosamilia, Carolyn Blasdel, Virginia Naessig, Sheila Broussard, Mary Beth Rios, Ronald Paquette,
Kathryn Kolibaba, Richard Maziarz, Peter Graf, and Hans Michael Buerger.
REFERENCES
1. Faderl S, Talpaz M, Estrov Z, O’Brien S, Kurzrock R, Kantarjian HM.
The biology of chronic myeloid leukemia. N Engl J Med 1999;341:16472.
2. Sawyers CL. Chronic myeloid leukemia. N Engl J Med 1999;340:133040.
3. Kolibaba KS, Druker BJ. Current status of treatment for chronic myelogenous leukemia. Medscape Oncology 2000;3(2). (See http://
www.medscape.com/medscape/oncology/journal/2000/v03.n02/
1036 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
E F F ICACY AND SAF ET Y OF A SP ECIFIC INH IB ITOR OF BC R -A BL IN C H R ONIC MYELOID LEUK EMIA
mo4283.druk/mo4283.druk=01.html.) (See NAPS document no. 05586
for 10 pages, c/o Microfiche Publications, 248 Hempstead Tpke., West
Hempstead, NY 11552.)
4. Silver RT, Woolf SH, Hehlmann R, et al. An evidence-based analysis of
the effect of busulfan, hydroxyurea, interferon, and allogeneic bone marrow transplantation in treating the chronic phase of chronic myeloid leukemias: developed for the American Society of Hematology. Blood 1999;
94:1517-36.
5. Nowell PC, Hungerford DA. A minute chromosome in human granulocytic leukemia. Science 1960;132:1497.
6. Rowley JD. A new consistent chromosomal abnormality in chronic myelogenous leukemia identified by quinacrine fluorescence and Giemsa
staining. Nature 1973;243:290-3.
7. Daley GQ, Van Etten RA, Baltimore D. Induction of chronic myelogenous leukemia in mice by the P210bcr/abl gene of the Philadelphia chromosome. Science 1990;247:824-30.
8. Kelliher MA, McLaughlin J, Witte ON, Rosenberg N. Induction of a
chronic myelogenous leukemia-like syndrome in mice with v-abl and
BCR/ABL. Proc Natl Acad Sci U S A 1990;87:6649-53. [Erratum, Proc
Natl Acad Sci U S A 1990;87:9072.]
9. Heisterkamp N, Jenster G, ten Hoeve J, Zovich D, Pattengale PK,
Groffen J. Acute leukaemia in bcr/abl transgenic mice. Nature 1990;344:
251-3.
10. Lugo TG, Pendergast AM, Muller AJ, Witte ON. Tyrosine kinase activity and transformation potency of bcr-abl oncogene products. Science
1990;247:1079-82.
11. Druker BJ, Lydon NB. Lessons learned from the development of an
abl tyrosine kinase inhibitor for chronic myelogenous leukemia. J Clin Invest 2000;105:3-7.
12. Druker BJ, Tamura S, Buchdunger E, et al. Effects of a selective inhibitor of the Abl tyrosine kinase on growth of Bcr-Abl positive cells. Nat
Med 1996;2:561-6.
13. Deininger MW, Goldman MJ, Lydon N, Melo JV. The tyrosine kinase
inhibitor CGP57148B selectively inhibits the growth of BCR-ABL-positive
cells. Blood 1997;90:3691-8.
14. Gambacorti-Passerini C, le Coutre P, Mologni L, et al. Inhibition of
the ABL kinase activity blocks the proliferation of BCR/ABL+ leukemic
cells and induces apoptosis. Blood Cells Mol Dis 1997;23:380-94.
15. Cancer Therapy Evaluation Program. Common toxicity criteria.
Bethesda, Md.: National Cancer Institute, March 1998.
16. Druker BJ, Neumann M, Okuda K, Franza BR Jr, Griffin JD. rel Is
rapidly tyrosine-phosphorylated following granulocyte-colony stimulating
factor treatment of human neutrophils. J Biol Chem 1994;269:5387-90.
17. Oda T, Heaney C, Hagopian JR, Okuda K, Griffin JD, Druker BJ. Crkl
is the major tyrosine-phosphorylated protein in neutrophils from patients
with chronic myelogenous leukemia. J Biol Chem 1994;269:22925-8.
18. Nichols GL, Raines MA, Vera JC, Lacomis L, Tempst P, Golde DW.
Identification of CRKL as the constitutively phosphorylated 39-kD tyrosine phosphoprotein in chronic myelogenous leukemia cells. Blood 1994;
84:2912-8.
19. ten Hoeve J, Arlinghaus RB, Guo JQ, Heisterkamp N, Groffen J. Tyrosine phosphorylation of CRKL in Philadelphia+ leukemia. Blood 1994;
84:1731-6.
20. Heaney C, Kolibaba K, Bhat A, et al. Direct binding of CRKL to
BCR-ABL is not required for BCR-ABL transformation. Blood 1997;89:
297-306.
21. Senechal K, Heaney C, Druker B, Sawyers CL. Structural requirements for function of the Crk1 adapter protein in fibroblasts and hematopoietic cells. Mol Cell Biol 1998;18:5082-90.
22. Druker BJ, Sawyers CL, Kantarjian H, et al. Activity of a specific inhibitor of the BCR-ABL tyrosine kinase in the blast crisis of chronic myeloid leukemia and acute lymphoblastic leukemia with the Philadelphia
chromosome. N Engl J Med 2001;344:1038-42.
Copyright © 2001 Massachusetts Medical Society.
POSTING PRESENTATIONS AT MEDICAL MEETINGS ON THE INTERNET
Posting an audio recording of an oral presentation at a medical meeting on the
Internet, with selected slides from the presentation, will not be considered prior
publication. This will allow students and physicians who are unable to attend the
meeting to hear the presentation and view the slides. If there are any questions
about this policy, authors should feel free to call the Journal’s Editorial Offices.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1037
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
ACTIVITY OF A SPECIFIC INHIBITOR OF THE BCR-ABL TYROSINE KINASE
IN THE BLAST CRISIS OF CHRONIC MYELOID LEUKEMIA AND ACUTE
LYMPHOBLASTIC LEUKEMIA WITH THE PHILADELPHIA CHROMOSOME
BRIAN J. DRUKER, M.D., CHARLES L. SAWYERS, M.D., HAGOP KANTARJIAN, M.D., DEBRA J. RESTA, R.N.,
SOFIA FERNANDES REESE, M.D., JOHN M. FORD, M.D., RENAUD CAPDEVILLE, M.D., AND MOSHE TALPAZ, M.D.
ABSTRACT
Background BCR-ABL, a constitutively activated
tyrosine kinase, is the product of the Philadelphia (Ph)
chromosome. This enzyme is present in virtually all
cases of chronic myeloid leukemia (CML) throughout
the course of the disease, and in 20 percent of cases
of acute lymphoblastic leukemia (ALL). On the basis
of the substantial activity of the inhibitor in patients in
the chronic phase, we evaluated STI571 (formerly
known as CGP 57148B), a specific inhibitor of the
BCR-ABL tyrosine kinase, in patients who had CML in
blast crisis and in patients with Ph-chromosome–positive ALL.
Methods In this dose-escalating pilot study, 58 patients were treated with STI571; 38 patients had myeloid blast crisis and 20 had ALL or lymphoid blast crisis. Treatment was given orally at daily doses ranging
from 300 to 1000 mg.
Results Responses occurred in 21 of 38 patients
(55 percent) with a myeloid-blast-crisis phenotype;
4 of these 21 patients had a complete hematologic
response. Of 20 patients with lymphoid blast crisis or
ALL, 14 (70 percent) had a response, including 4 who
had complete responses. Seven patients with myeloid blast crisis continue to receive treatment and remain in remission from 101 to 349 days after starting
the treatment. All but one patient with lymphoid blast
crisis or ALL has relapsed. The most frequent adverse effects were nausea, vomiting, edema, thrombocytopenia, and neutropenia.
Conclusions The BCR-ABL tyrosine kinase inhibitor STI571 is well tolerated and has substantial activity in the blast crises of CML and in Ph-chromosome–
positive ALL. (N Engl J Med 2001;344:1038-42.)
as in 50 percent of adults and 80 percent of children
with ALL the BCR-ABL protein is smaller, with a
molecular mass of 185 or 190 kd.1,2
The blast crisis is highly refractory to treatment.
The rate of response to standard induction chemotherapy in patients with myeloid blast crisis is approximately 20 percent, and the rate of complete remission is less than 10 percent. In patients with lymphoid
blast crisis, the rate of response is approximately 50
percent, but remissions are short-lived.3-5 After allogeneic stem-cell transplantation during blast crisis, the
five-year survival rate is only 6 percent6,7; Ph-chromosome–positive ALL also has a poor prognosis.8-10
STI571 (4-[(4-methyl-1-piperazinyl)methyl]-N-[4methyl-3-[[4-(3-pyridinyl)-2-pyrimidinyl]amino]phenyl]benzamide methanesulfonate; Glivec, Novartis, Basel, Switzerland) is a potent and selective inhibitor
of the tyrosine kinase activity of BCR-ABL.11,12 In a
phase 1 trial of STI571 in patients with CML in the
chronic phase, reported elsewhere in this issue of the
Journal, those treated with daily doses of 300 mg or
more had a high rate of response and minimal adverse
effects.13 Disease progression to blast crisis is associated with genetic instability and numerous molecular
abnormalities. Thus, it is possible that other oncogenic abnormalities replace the need for BCR-ABL
tyrosine kinase activity for cellular survival of leukemic
blasts. In this study, we evaluated the effects of STI571
in the treatment of CML in blast crisis and Ph-chromosome–positive ALL.
METHODS
Copyright © 2001 Massachusetts Medical Society.
Patients
T
Patients with CML were eligible if they tested positive for the
Ph chromosome, were at least 18 years of age, and were in blast
crisis (with more than 30 percent blasts in the peripheral blood
or bone marrow), irrespective of prior therapy. Patients with Phchromosome–positive ALL were eligible if they had not had a response to standard induction or consolidation chemotherapy or
had had a relapse after such therapy. Treatment with STI571 was
not initiated until at least 24 hours after treatment with hydroxyurea ended and until at least four weeks after treatment with standard induction or consolidation therapy ended. Adequate renal, hepatic, and cardiac function and performance status were required.
HE BCR-ABL tyrosine kinase, the product
of the chimeric gene produced by the Philadelphia (Ph) chromosome, is the molecular abnormality that causes chronic myeloid
leukemia (CML). During the chronic phase of the disease, there is massive clonal expansion of myeloid cells,
which retain the ability to differentiate. Over time,
however, the leukemic clone loses this ability, and the
disease inevitably progresses to an acute leukemia
known as blast crisis.1,2 In two thirds of patients the
blasts are myeloid, and in one third they are lymphoid. Up to 20 percent of adults and 5 percent of children with acute lymphoblastic leukemia (ALL) have
the BCR-ABL fusion protein.1 In virtually all patients
with CML, including those with blast crisis, the BCRABL protein has a molecular mass of 210 kd, where-
From the Division of Hematology and Medical Oncology, Oregon
Health Sciences University, Portland (B.J.D.); the Division of Hematology
and Oncology, University of California at Los Angeles, Los Angeles
(C.L.S.); the Departments of Leukemia (H.K.) and Bioimmunotherapy
(M.T.), University of Texas M.D. Anderson Cancer Center, Houston; and
the Department of Oncology Clinical Research, Novartis Pharmaceuticals,
East Hanover, N.J. (D.R.J.), and Basel, Switzerland (S.F.R., J.M.F., R.C.).
Address reprint requests to Dr. Druker at Oregon Health Sciences University, L592, 3181 SW Sam Jackson Park Rd., Portland, OR 97201, or at
[email protected].
1038 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
AC T I V I T Y OF AN INHIBITOR OF BCR- ABL IN CH R ONIC MYELOID A ND ACUTE LY MPH OBL ASTIC LEUK EMIA
Written informed consent was obtained from all patients before
they entered the study.
Study Design
This pilot dose-escalation study was designed to assess the antileukemic activity and safety of STI571 in patients with CML in
blast crisis or Ph-chromosome–positive ALL. Patients were assigned to successive dose cohorts of STI571 ranging from 300 to
1000 mg. The starting dose of 300 mg per day was selected on
the basis of its efficacy in a parallel phase 1 study in patients with
CML in the chronic phase.13 To dissociate the confounding role
of underlying illness in acutely ill patients from drug-related toxicity, decisions about dose escalation were based on findings from
the phase 1 study, which was conducted simultaneously. Six to
eight patients were assigned to each dose; STI571 was administered orally once daily, except for the 800-mg and 1000-mg doses, which were administered twice daily in 400-mg and 500-mg
doses, respectively. Patients received continuous therapy unless
unacceptable adverse effects or disease progression occurred.
Therapy with hydroxyurea was permitted after the initiation of
treatment for a maximum of seven days during the first four
weeks as required to maintain acceptable blood counts. No other
anticancer agents were allowed during treatment, and no dose
modifications were allowed for hematologic toxicity during the
first 14 days of therapy. All patients received allopurinol for 48
hours before the initiation of treatment with STI571.
Complete blood counts were obtained three times weekly. Assessments of bone marrow, including cytogenetic assessments, were
performed once every 6 weeks during the first 12 weeks of treatment and then once every 12 weeks. If grade 4 neutropenia, defined as an absolute neutrophil count of less than 500 per cubic
millimeter, occurred on or after 14 days of treatment with STI571,
bone marrow aspiration and biopsy were performed. If marrow
cellularity was less than 10 percent, treatment with STI571 was
interrupted until the absolute neutrophil count rose to more than
1000 per cubic millimeter. If neutropenia recurred, treatment with
STI571 was again interrupted until the absolute neutrophil count
was more than 1000 per cubic millimeter and then resumed at the
dose level of the previous cohort. If marrow cellularity was more
than 10 percent, contained more than 30 percent blasts, or both,
treatment with STI571 was continued. There were no dose modifications for thrombocytopenia.
gether. The phenotype of the blasts in the 58 patients was myeloid in 38 and lymphoid in 20; these
20 included 10 patients with Ph-chromosome–positive ALL. One patient was enrolled as an exception,
on the basis of lymphoid blasts in the breast that were
detected during the course of CML. Sixteen patients
with myeloid blast crisis and seven with lymphoid
blast crisis had received previous therapy for the blast
crises. The study required that all patients with Phchromosome–positive ALL had received prior chemotherapy. The study is ongoing, and the results reported here represent an interim analysis of the data.
TABLE 1. CHARACTERISTICS
OF THE
58 PATIENTS.
CHARACTERISTIC
VALUE
Sex — M/F
Age — yr
Median
Range
History of disease — no. (%)
Myeloid blast crisis
Lymphoid blast crisis
Ph-chromosome–positive ALL
Previous therapy for acute leukemia — no. (%)
Patients with myeloid blast crisis
Patients with lymphoid blast crisis
Additional cytogenetic abnormalities — no. (%)
Patients with myeloid blast crisis
Patients with lymphoid blast crisis or ALL
White-cell count at base line — cells/mm3
Median
Range
Platelet count at base line — cells/mm3
Median
Range
35/23
48
24–76
38 (66)
10 (17)
10 (17)
16 (42)
7 (70)
22 (58)
13 (65)
25,200
100–171,000
92,000
4,000–1,278,000
Assessment of Toxicity and Response
Safety assessments included the evaluation of adverse events
and vital signs, hematologic tests, biochemical tests, urinalysis, and
physical examination. Toxicity was graded in accordance with the
Common Toxicity Criteria of the National Cancer Institute.14
We used standard criteria to define a complete hematologic response 4: a decrease in marrow blasts to 5 percent or less of total
cellularity, a disappearance of blasts from the peripheral blood, an
absolute neutrophil count of more than 1000 per cubic millimeter, and a platelet count of more than 100,000 per cubic millimeter.
In patients who did not have a complete hematologic response, a
marrow response was defined as a decrease in marrow blasts to either no more than 5 percent or between 5 and 15 percent, regardless of the peripheral-blood cell counts. A relapse was defined as
either disease progression (an increase in marrow blasts to more
than 15 percent, in peripheral-blood blasts to more than 5 percent,
or in white cells to more than 20,000 per cubic millimeter) or death.
The time to relapse was calculated from the first dose of STI571.
Cytogenetic responses were classified as previously described.4
RESULTS
Accrual of Patients
From April 1999 through March 2000, 58 patients were enrolled; their characteristics are summarized in Table 1. Patients with lymphoid blast crisis
and Ph-chromosome–positive ALL are grouped to-
TABLE 2. DRUG-RELATED ADVERSE EFFECTS.*
ADVERSE
EFFECT
GROUP RECEIVING GROUP RECEIVING
GROUP RECEIVING
300 mg/DAY
400–500 mg/DAY 600–1000 mg/DAY TOTAL
(N=8)
(N=17)
(N=33)
(N=58)
GRADE
GRADE
GRADE
GRADE
GRADE
GRADE
1
3
1
3
1
3
OR
2
OR
4
OR
2
OR
4
OR
2
OR
percentage of patients
Nausea
Vomiting
Edema
Myalgia
Diarrhea
Rash
Fatigue
Anorexia
25
38
25
25
0
0
12
12
12
0
12
0
0
0
0
0
35
35
18
18
12
18
24
18
6
6
6
0
0
0
0
0
4
GRADES
1–4
no. (%)
55
33
45
21
24
15
3
6
12
9
6
0
0
6
0
0
32 (55)
24 (41)
24 (41)
12 (21)
10 (17)
10 (17)
6 (10)
6 (10)
*The adverse effects listed here were considered to be related to STI571
and were reported in more than 10 percent of patients. A grade of 1 indicates mild adverse effects, a grade of 2 moderate effects, a grade of 3 severe
effects, and a grade of 4 life-threatening effects.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1039
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
Safety Profile
STI571 was generally well tolerated (Table 2). The
most frequent adverse effects were nausea (in 55 percent of patients), vomiting (41 percent), and edema
(41 percent); most of these were grade 1 (mild) or
grade 2 (moderate). Patients treated with higher doses of STI571 were more likely to have grade 1 or 2
nausea, edema, or diarrhea than patients given lower
doses of the drug. Grade 4 neutropenia and thrombocytopenia occurred in 40 percent and 33 percent
of patients, respectively (Table 3). Elevations in liverenzyme levels of grade 3 or 4 were reported in eight
patients (14 percent) a median of 16 days after the
initiation of treatment (range, 7 to 194), without evidence of a dose relation. Treatment with STI571 was
discontinued because of these abnormalities in only
one patient; four of the eight patients had grade 1
elevations in liver-enzyme levels at base line.
There were 16 deaths due to disease progression.
No deaths were considered to be related to treatment
with STI571. The following serious adverse events
in 13 patients were possibly related to STI571: nausea and vomiting in 4 patients, febrile neutropenia
in 3 patients, and elevated liver-enzyme levels, exfoliative dermatitis, gastric hemorrhage, renal failure, pancytopenia, and congestive heart failure in 1 patient
each. These events occurred more frequently in patients treated with 800 or 1000 mg of STI571 per day.
4 and 5). Of the 38 patients with myeloid blast crisis, 4 had a complete hematologic remission and 17
had a decrease in blasts in the marrow to 15 percent
or less (8 of these had a decrease to 5 percent or
less). Of the 20 patients with lymphoid blast crisis
and Ph-chromosome–positive ALL, 4 had a complete hematologic remission and 10 had a marrow
response. In the small groups we studied, there was
no relation between the dose of STI571 and the
proportion of patients with hematologic responses.
In patients who had a response to the drug, the reduction in peripheral blasts typically occurred within one week after the initiation of therapy (Fig. 1).
The median duration of therapy was 74 days (range,
1 to 349). Of the 21 patients with myeloid blast crisis who had a response to STI571, 9 subsequently
relapsed between 42 and 194 days (median, 84) after
the initiation of treatment. Seven of the 21 patients
with myeloid blast crisis continue to receive therapy
TABLE 4. RESPONSES
DOSE OF STI571
(mg/DAY)
PATIENTS
IN
NO. OF
PATIENTS
WITH A
NO. WITH
COMPLETE
HEMATOLOGIC
RESPONSE
«5%
In the intention-to-treat analysis of response rates,
all patients in the study were included whether or
not the response could be properly evaluated. There
was a decrease of 50 percent or more in peripheralblood blasts in 46 of the 58 patients (79 percent).
According to the criteria for responses described in
the Methods section, the overall rates of response were
55 percent and 70 percent among patients with myeloid and lymphoid blast crises, respectively (Tables
TABLE 3. HEMATOLOGIC TOXICITY.*
VARIABLE
GROUP
RECEIVING
400–500
mg/DAY
(N=17)
NO. WITH
MARROW RESPONSE
6%
Hematologic and Bone Marrow Response
GROUP
RECEIVING
300 mg/DAY
(N=8)
MYELOID PHENOTYPE.
GROUP
RECEIVING
600–1000
mg/DAY
(N=33)
TOTAL
(N=58)
300
400
500
600
750
800
1000
6
4
5
8
7
7
1
Total — no. (%)
38
TABLE 5. RESPONSES
IN
DOSE OF STI571
(mg/DAY)
TO
«15%
BLASTS
0
1
1
0
2
0
0
1
0
1
2
1
3
0
1
1
2
1
2
1
1
4 (11)
8 (21)
9 (24)
PATIENTS
NO. OF
PATIENTS
BLASTS
WITH A
NO. WITH
COMPLETE
HEMATOLOGIC
RESPONSE
LYMPHOID PHENOTYPE.
NO. WITH
MARROW RESPONSE
6% TO «15%
«5%
BLASTS
BLASTS
percentage of patients
Neutropenia
Grade 3
Grade 4
Thrombocytopenia
Grade 3
Grade 4
38
38
29
47
21
36
26
40
38
25
35
29
36
36
36
33
*The lowest values reported during the study are listed. A grade of 1
indicates mild adverse effects, a grade of 2 moderate effects, a grade of 3
severe effects, and a grade of 4 life-threatening effects.
300
400
500
600
750
800
1000
Total — no. (%)
1040 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
2
4
4
2
2
1
5
20
0
0
1
1
0
0
2
1
2
0
0
1
1
2
0
1
1
1
0
0
0
4 (20)
7 (35)
3 (15)
AC T I V I T Y OF AN INHIBITOR OF BCR- ABL IN CH R ONIC MYELOID A ND ACUTE LY MPH OBL ASTIC LEUK EMIA
20
responses, five were complete (three in patients with
myeloid and two in patients with lymphoid blast crises) and two were partial, defined as less than 35 percent Ph-chromosome–positive cells (one in a patient
with lymphoid and one in a patient with myeloid
blast crisis).
10
DISCUSSION
50
Peripheral bloodF
Bone marrow
Blasts (%)
40
30
0
0
50
100
150
Day
Figure 1. Kinetics of Complete Response in a Patient with Myeloid Blast Crisis Treated with 400 mg per Day of STI571.
and are in remission, with a follow-up of 101 to 349
days. The other five patients were removed from the
study: one for hematopoietic stem-cell transplantation,
one because of poor compliance with therapy, and
three because of adverse events.
Of the 14 patients with lymphoid blast crisis who
had a response to STI571, 12 relapsed a median of
58 days after the initiation of treatment (range, 42 to
123), 1 underwent hematopoietic stem-cell transplantation, and the data on 1 who was in remission at
day 58 have been censored because of the short length
of follow-up (Fig. 2). The patient with extramedullary
disease had a complete response at the extramedullary site and remains in remission at day 243.
All patients who relapsed remained Ph-chromosome–positive. Major cytogenetic responses were observed in 7 of the 58 patients (12 percent). Of these
This study demonstrates that STI571 as a single
agent is well tolerated and has substantial activity
against acute leukemias characterized by the BCRABL fusion protein. The overall response rate in the
myeloid blast crisis of CML was 55 percent, and the
rate of complete remission was 11 percent. Leukemic
blasts in the marrow were reduced to 5 percent or
less in 12 patients (32 percent). Of these 12 patients
with myeloid blast crisis who initially had a response
to STI571, 7 are still in remission after 101 to 349
days of follow-up.
There was no obvious difference in response rates or
the durability of responses between patients with
lymphoid blast crisis and those with Ph-chromosome–
positive ALL. The overall response rate in patients
with lymphoid blast crisis or Ph-chromosome–positive ALL was 70 percent, and 20 percent had complete
remissions. A decrease in bone marrow blasts to 5 percent or less occurred in 11 patients (55 percent).
However, all but one patient who had only extramedullary disease relapsed. With standard therapy, patients
with lymphoid blast crisis have higher rates and
greater durability of response than those with myeloid blast crisis. However, in this study, there was a
trend toward a more durable response in the group
of patients with myeloid blast crisis.
1.0
Probability of Relapse
0.9
0.8
*
*
0.7
0.6
Myeloid (n=21)
*
0.5
**
*
*
*
0.4
0.3
0.2
Lymphoid (n=14)
0.1
0.0
0
100
200
300
400
Day
Figure 2. Time to Relapse in Patients with Myeloid or Lymphoid Blast Crisis Who Had a Response to
STI571.
Arrows with asterisks indicate patients still enrolled in the study and in remission at the time of the last
follow-up; arrows without asterisks indicate the day on which patients were removed from the study.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1041
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
Since the outcome of hematopoietic stem-cell transplantation is better in patients with blast crisis who are
first returned to the chronic phase of CML than in
patients who undergo transplantation during blast
crisis,15 the reduction in the proportion of blasts in
the marrow of patients with blast crisis suggests that
STI571 may be a useful bridge to transplantation.
Combinations of STI571 with standard antileukemic
agents may also improve the outcome for patients with
blast crisis.16,17 In the patients with blast crisis we treated, STI571 had relatively few adverse effects, the most
frequent of which were nausea, vomiting, and edema. There was some evidence of an increased incidence of toxic effects at the higher doses of STI571,
especially at 800 to 1000 mg per day. Myelosuppression of grade 3 or 4 was more frequent in these patients with blast crisis than in patients with CML in
the chronic phase who were treated with STI571 in a
parallel phase 1 study.13 This difference may reflect the
severely compromised bone marrow function in patients in blast crisis and the fact that severe myelosuppression was allowed in this study because of the lifethreatening nature of the illness, but not in the trial
involving patients with CML in the chronic phase.
Rapid response is also a feature of therapy with
STI571 (Fig. 1); despite this, the tumor lysis syndrome
developed in only one patient. Preliminary data suggest that cells from treated patients undergo rapid apoptosis (data not shown). Although STI571 can be
given to outpatients, careful monitoring during the
initiation of therapy, vigorous hydration, and administration of allopurinol are recommended.
The mechanism of resistance to STI571 or relapse
during treatment with the drug is a subject of intense
interest. Analyses of blast-crisis CML cell lines that
have acquired resistance to STI571 after prolonged
culture in doses below the threshold for inhibition
of growth18-20 have shown amplification of the BCRABL gene, increased expression of the BCR-ABL
protein without amplification of the gene, and increased expression of the multidrug-resistance protein (MDR1).18-20 Preliminary analyses have shown
that leukemic cells in patients who relapse retain the
Ph chromosome and that serum levels of STI571 are
unchanged at the time of relapse. These data are
consistent with the in vitro data that implicate drug
efflux or amplification of the BCR-ABL gene in resistance to STI571, but other mechanisms are also
possible.18-20
This study clearly demonstrates that in the majority of patients with CML in blast crisis and Ph-chromosome–positive ALL, the leukemic clone remains
at least partially dependent on BCR-ABL for survival.
We also show that targeting a critical molecular abnormality, even in advanced stages of disease, is a useful strategy; however, in these cases it is likely that
this agent will need to be combined with other therapies to achieve maximal therapeutic benefits.
Supported by grants from the National Cancer Institute (CA65823, to
Dr. Druker, and CA32737, to Dr. Sawyers) and by Novartis Pharmaceuticals. Dr. Druker is the recipient of a Translational Research Award from the
Leukemia and Lymphoma Society, and Dr. Sawyers is a Scholar of the Leukemia and Lymphoma Society.
Drs. Druker, Sawyers, and Talpaz served as consultants to Novartis Pharmaceuticals during the design of this study.
We are indebted to the following people for their assistance with
various aspects of this study: Alex Matter, Juerg Zimmerman, John
Goldman, Gregory Burke, David Parkinson, Michael Hayes, Ulrike
Zoellner, William Palo, Marianne Rosamilia, Carolyn Blasdel, Virginia Naessig, Sheila Broussard, Mary Beth Rios, Ronald Paquette,
Kathryn Kolibaba, Richard Maziarz, Peter Graf, and Hans Michael Buerger.
REFERENCES
1. Faderl S, Talpaz M, Estrov Z, O’Brien S, Kurzrock R, Kantarjian HM.
The biology of chronic myeloid leukemia. N Engl J Med 1999;341:164-72.
2. Sawyers CL. Chronic myeloid leukemia. N Engl J Med 1999;340:133040.
3. Kantarjian HM, Talpaz M, Keating MJ, et al. Intensive chemotherapy
induction followed by interferon-alpha maintenance in patients with Philadelphia chromosome-positive chronic myelogenous leukemia. Cancer
1991;68:1201-7.
4. Sacchi S, Kantarjian HM, O’Brien S, et al. Chronic myelogenous leukemia in nonlymphoid blastic phase: analysis of the results of first salvage
therapy with three different treatment approaches for 162 patients. Cancer
1999;86:2632-41.
5. Walters RS, Kantarjian HM, Keating MJ, et al. Therapy of lymphoid
and undifferentiated chronic myelogenous leukemia in blast crisis with continuous vincristine and adriamycin infusions plus high-dose decadron. Cancer 1987;60:1708-12.
6. Gratwohl A, Hermans J. Allogeneic bone marrow transplantation for
chronic myeloid leukemia. Bone Marrow Transplant 1996;17:Suppl 3:S7S9.
7. Clift RA, Storb R. Marrow transplantation for CML: the Seattle experience. Bone Marrow Transplant 1996;17:Suppl 3:S1-S3.
8. Copelan EA, McGuire EA. The biology and treatment of acute lymphoblastic leukemia in adults. Blood 1995;85:1151-68.
9. Westbrook CA, Hooberman AL, Spino C, et al. Clinical significance of
the BCR-ABL fusion gene in adult acute lymphoblastic leukemia: a Cancer
and Leukemia Group B Study (8762). Blood 1992;80:2983-90.
10. Aricò M, Valsecchi MG, Camitta B, et al. Outcome of treatment in
children with Philadelphia chromosome–positive acute lymphoblastic leukemia. N Engl J Med 2000;342:998-1006.
11. Buchdunger E, Zimmerman J, Mett H, et al. Inhibition of the Abl
protein-tyrosine kinase in vitro and in vivo by a 2-phenylaminopyrimidine
derivative. Cancer Res 1996;56:100-4.
12. Druker BJ, Tamura S, Buchdunger E, et al. Effects of a selective inhibitor of the Abl tyrosine kinase on the growth of Bcr-Abl positive cells.
Nat Med 1996;2:561-6.
13. Druker BJ, Talpaz M, Resta DJ, et al. Efficacy and safety of a specific
inhibitor of the BCR-ABL tyrosine kinase in chronic myeloid leukemia.
N Engl J Med 2001;344:1031-7.
14. Cancer Therapy Evaluation Program. Common toxicity criteria, version 2.0. Bethesda, Md.: National Cancer Institute, March 1998.
15. Spencer A, O’Brien SG, Goldman JM. Options for therapy in chronic
myeloid leukaemia. Br J Haematol 1995;91:2-7.
16. Thiesing JT, Ohno-Jones S, Kolibaba KS, Druker BJ. Efficacy of
STI571, an abl tyrosine kinase inhibitor, in conjunction with other antileukemic agents against bcr-abl-positive cells. Blood 2000;96:3195-9.
17. Fang G, Kim CN, Perkins CL, et al. CGP57148B (STI-571) induces
differentiation and apoptosis and sensitizes Bcr-Abl–positive human leukemia cells to apoptosis due to antileukemic drugs. Blood 2000;96:2246-53.
18. Mahon FX, Deininger MWN, Schultheis B, et al. Selection and characterization of BCR-ABL positive cell lines with differential sensitivity to
the tyrosine kinase inhibitor STI571: diverse mechanisms of resistance.
Blood 2000;96:1070-9.
19. Weisberg E, Griffin JD. Mechanism of resistance to the ABL tyrosine
kinase inhibitor STI571 in BCR/ABL-transformed hematopoietic cell
lines. Blood 2000;95:3498-505.
20. le Coutre P, Tassi E, Varella-Garcia M, et al. Induction of resistance to
the Abelson inhibitor STI571 in human leukemic cells through gene amplification. Blood 2000;95:1758-66.
1042 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
Copyright © 2001 Massachusetts Medical Society.
S U RV I VA L A F T E R A B L AT I O N O F T H E AT R I OV E N T R I C U L A R N O D E A N D I M P L A N TAT I O N O F A P E R M A N E N T PAC E M A K E R
LONG-TERM SURVIVAL AFTER ABLATION OF THE ATRIOVENTRICULAR NODE
AND IMPLANTATION OF A PERMANENT PACEMAKER IN PATIENTS
WITH ATRIAL FIBRILLATION
CEVHER OZCAN, M.D., ARSHAD JAHANGIR, M.D., PAUL A. FRIEDMAN, M.D., PHILIP J. PATEL, M.D.,
THOMAS M. MUNGER, M.D., ROBERT F. REA, M.D., MARGARET A. LLOYD, M.D., DOUGLAS L. PACKER, M.D.,
DAVID O. HODGE, M.S., BERNARD J. GERSH, M.B., CH.B., D.PHIL., STEPHEN C. HAMMILL, M.D.,
AND WIN-KUANG SHEN, M.D.
ABSTRACT
Background In patients with atrial fibrillation that
is refractory to drug therapy, radio-frequency ablation
of the atrioventricular node and implantation of a
permanent pacemaker are an alternative therapeutic
approach. The effect of this procedure on long-term
survival is unknown.
Methods We studied all patients who underwent
ablation of the atrioventricular node and implantation
of a permanent pacemaker at the Mayo Clinic between
1990 and 1998. Observed survival was compared with
the survival rates in two control populations: age- and
sex-matched members of the Minnesota population
between 1970 and 1990 and consecutive patients with
atrial fibrillation who received drug therapy in 1993.
Results A total of 350 patients (mean [±SD] age,
68±11 years) were studied. During a mean of 36±26
months of follow-up, 78 patients died. The observed
survival rate was significantly lower than the expected survival rate based on the general Minnesota population (P<0.001). Previous myocardial infarction (P<
0.001), a history of congestive heart failure (P=0.02),
and treatment with cardiac drugs after ablation (P=
0.03) were independent predictors of death. Observed
survival among patients without these three risk factors was similar to expected survival (P=0.43). None
of the 26 patients with lone atrial fibrillation died
during follow-up (37±27 months). The observed survival rate among patients who underwent ablation
was similar to that among 229 controls with atrial fibrillation (mean age, 67±12 years) who received drug
therapy (P=0.44).
Conclusions In the absence of underlying heart
disease, survival among patients with atrial fibrillation after ablation of the atrioventricular node is similar to expected survival in the general population.
Long-term survival is similar for patients with atrial
fibrillation, whether they receive ablation or drug therapy. Control of the ventricular rate by ablation of the
atrioventricular node and permanent pacing does not
adversely affect long-term survival. (N Engl J Med
2001;344:1043-51.)
A
TRIAL fibrillation is associated with increased morbidity and mortality1-5 and is an
independent risk factor for stroke.6,7 Although the association between atrial fibrillation and mortality has been debated, a recent report
showed that atrial fibrillation was associated with a
mortality rate that was higher by a factor of 1.5 to
1.9 than the rate expected in the general population,
after adjustment for other cardiovascular conditions.1
The optimal goal in treating atrial fibrillation is to
restore and maintain sinus rhythm — often a formidable task. Despite therapy with antiarrhythmic drugs,
studies have reported recurrence rates of 50 to 60
percent during a mean follow-up of one to two
years.8-12 In patients with severe symptoms in whom
drug therapy fails, ablation of the atrioventricular node
and permanent pacing are effective in controlling the
ventricular rate.13-16 Although ablation of the atrioventricular node does not eliminate atrial fibrillation,
it alleviates symptoms and improves the quality of life,
exercise tolerance, and left ventricular function.17-20
Despite the effectiveness of this treatment in relieving
symptoms, its effect on long-term survival in patients
with severe symptoms in whom drug therapy has
failed is unknown. The potentially deleterious effect
on survival of the creation of permanent atrioventricular block and the resulting lifelong commitment to
the use of a pacemaker is a serious concern.
We assessed long-term survival and predictors of
death after the ablation of the atrioventricular node
and the implantation of a permanent pacemaker in
350 patients with atrial fibrillation. To test the hypothesis that this treatment has an adverse effect on longterm survival, we compared the observed survival with
expected survival calculated on the basis of age- and
sex-specific mortality rates in the Minnesota population and with the observed survival of a group of consecutive patients who received pharmacologic therapy for atrial fibrillation.
Copyright © 2001 Massachusetts Medical Society.
From the Division of Cardiovascular Diseases and Internal Medicine
(C.O., A.J., P.A.F., P.J.P., T.M.M., R.F.R., M.A.L., D.L.P., B.J.G., S.C.H.,
W.-K.S.) and the Section of Biostatistics (D.O.H.), Mayo Clinic, Rochester,
Minn. Address reprint requests to Dr. Shen at the Mayo Clinic, 200 First
St. SW, Rochester, MN 55905.
David L. Hayes, M.D. (Division of Cardiovascular Diseases and Internal
Medicine, Mayo Clinic, Rochester, Minn.), was also an author.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1043
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
METHODS
Study Population
All patients with atrial fibrillation who underwent radio-frequency ablation of the atrioventricular node and implantation of a
permanent pacemaker at the Mayo Clinic between July 1990 and
December 1998 were included in the study. Patients with indications for ablation were those with symptomatic paroxysmal or
chronic atrial fibrillation that was refractory to drug therapy aimed
at controlling the ventricular rate or maintaining sinus rhythm.
None of the 350 patients we studied underwent direct-current
ablation. The potential risks of the procedure were explained, and
oral informed consent was obtained from all patients.
Control Groups
The first control group was constructed on the basis of the age
and sex of all patients who underwent ablation, and the expected
survival rate was calculated on the basis of age- and sex-specific mortality rates in the Minnesota population for the period between
1970 and 1990.21 We assumed that the expected survival rate accounts for the effects of cardiovascular and other medical conditions according to their known prevalence in the reference population.
The second control group was selected from a group of consecutive patients who received drug therapy for atrial fibrillation
at the Mayo Clinic in 1993. These patients were selected from an
existing data base as members of the control group because they
had clinical characteristics and follow-up that were similar, although
not identical, to those of the patients who underwent ablation.
Data Collection
Data were collected from a centralized system that contained
complete records of all patients treated and followed at the Mayo
Clinic and its hospitals. These records provide a detailed history
and diagnosis for all outpatient encounters, including emergency
room visits and home and nursing home visits, as well as data recorded during inpatient care, death certificates, and autopsy reports.
Follow-up
The follow-up period for the patients who underwent ablation
began at the time of the procedure; the follow-up period for the
controls treated with drugs began in 1993. For both groups, follow-up ended in January 1999 or at the time of death. Patients who
underwent ablation had follow-up visits in the pacemaker clinic
every three months for the first year and were surveyed annually
thereafter. Causes of death were determined by a review of hospital
records and death certificates and by telephone interviews of local
physicians or family members. All patients in both groups who entered the study had at least one follow-up visit.
Atrioventricular-Node Ablation and Pacemaker
Implantation
Radio-frequency ablation of the atrioventricular node was performed by standard techniques.22,23 Complete atrioventricular block
was achieved in all patients. Seven patients (2 percent of those enrolled) required a left-sided approach to achieve complete block, and
24 patients (7 percent) required a second or third procedure because of recurrent atrioventricular conduction after the first attempt.
A rate-responsive ventricular pacemaker was implanted if the
patient was in atrial fibrillation at the time of the procedure and
if attempts to restore and maintain sinus rhythm by means of cardioversion were not performed. A dual-chamber, rate-adaptive pacemaker was implanted if the patient was in sinus rhythm at the time
of the procedure.
Statistical Analysis
Survival of the patients who underwent ablation and the controls
treated with drugs was estimated by the Kaplan–Meier method.
For each person who underwent ablation, the expected survival
was calculated on the basis of age- and sex-specific mortality rates
in the Minnesota population during the period between 1970 and
1990.21 The observed and expected survival rates were compared
by means of the one-sample log-rank test.24 All three survival curves
were compared by means of the two-sample log-rank test. Categorical variables were compared between groups with use of the
chi-square test for independence. Continuous variables were compared with the use of the Wilcoxon rank-sum test. Univariate and
multivariate associations between base-line variables and survival
were assessed by means of the log-rank test and a Cox regression
model.25 The following variables were considered as potential prognostic factors: demographic features (age and sex), clinical history
(syncope, angina, and congestive heart failure), and the presence of
heart disease (ischemic heart disease, cardiomyopathy, and valvular heart disease) and associated clinical conditions (diabetes mellitus, chronic obstructive pulmonary disease, cerebrovascular disease,
hypertension, and cancer). Treatment with cardiac medications after ablation was also included as a variable in the analysis. Multivariate models are presented in the form of point estimates of the
risk ratios, with 95 percent confidence intervals.
RESULTS
Demographic Characteristics
A total of 350 patients with atrial fibrillation (185
men and 165 women) who underwent ablation and
had a pacemaker implanted at the Mayo Clinic between 1990 and 1998 were included in the study. A
single-chamber ventricular pacemaker was implanted
in 55 percent of the patients, and a dual-chamber
pacemaker in 45 percent.
The base-line characteristics of the patients who underwent ablation are summarized in Table 1. Drugs
used to control the ventricular rate or to maintain sinus
rhythm before ablation included digoxin (used by 87
percent of patients), a calcium-channel blocker (82
percent), a beta-blocker (56 percent), quinidine (43
percent), procainamide (30 percent), disopyramide
(18 percent), propafenone (47 percent), flecainide (20
percent), encainide (6 percent), sotalol (13 percent),
and amiodarone (41 percent). After ablation, 188 patients (54 percent) continued to take one or more
cardiac drugs because of preexisting cardiovascular
disease. These drugs included digoxin, calcium-channel blockers, beta-blockers, angiotensin-converting–
enzyme inhibitors, nitrates, diuretics, and antiarrhythmic agents (propafenone, sotalol, amiodarone, and
mexiletine). At the time of the ablation, 11 percent of
patients were in New York Heart Association functional class III or IV, and 37 percent had a reduced left
ventricular ejection fraction (a fraction of 40 percent
or lower).
Sixty-eight patients (19 percent) had a history of a
cerebrovascular accident or transient ischemic attack,
and eight patients (2 percent) had peripheral arterial
embolism before ablation. During follow-up, a cerebrovascular accident or transient ischemic attack occurred in 15 patients (of whom 6 [40 percent] had
had a previous such event), and 3 had peripheral arterial embolism. At the time of embolic complications,
all patients except one were receiving warfarin therapy.
The mean (±SD) international normalized ratio was
1044 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
S U RV I VA L A F T E R A B L AT I O N O F T H E AT R I OV E N T R I C U L A R N O D E A N D I M P L A N TAT I O N O F A P E R M A N E N T PAC E M A K E R
TABLE 1. BASE-LINE CHARACTERISTICS OF PATIENTS WITH ATRIAL
FIBRILLATION WHO UNDERWENT ABLATION OF THE
ATRIOVENTRICULAR NODE AND IMPLANTATION OF A PACEMAKER
BETWEEN 1990 AND 1998.*
CHARACTERISTIC
Sex (no.)
Male
Female
Age at time of ablation (yr)
Duration of arrhythmia (yr)
Duration of follow-up (mo)
Type of atrial fibrillation (%)
Chronic
Paroxysmal
Atrial flutter
Lone
No. of antiarrhythmic drugs
Left ventricular ejection fraction (%)
NYHA functional class
Medical history and conditions (%)
Coronary-artery bypass graft
Cardiac-valve surgery
Coronary artery disease
Previous myocardial infarction
Congestive heart failure†
Left ventricular hypertrophy
Nonischemic cardiomyopathy
Syncope
Previous cerebrovascular accident
or transient ischemic attack
Hypertension
Hyperlipidemia
Diabetes mellitus
Chronic obstructive lung disease
Cancer
Renal disease
Liver failure
Smoking
ALL PATIENTS
(N=350)
PATIENTS WHO PATIENTS
SURVIVED WHO DIED
(N=78)
(N=272)
185
165
68±11
7±18
36±26
135
137
68±11
8±8
38±26
50
28
69±10
5±5
27±25
50
45
5
7
4.3±2.0
47±17
1.3±0.7
49
47
4
10
4.3±2.0
49±17
1.2±0.7
54
41
5
0
4.5±2.0
41±17
1.6±0.9
18
17
45
27
21
10
16
17
19
16
15
38
19
17
10
15
8
18
27
24
67
58
36
12
42
28
26
46
36
20
32
21
12
3
53
45
33
16
25
19
8
3
50
50
47
33
56
31
27
4
67
*Plus–minus values are means ±SD. NYHA denotes New York Heart
Association.
†Congestive heart failure was defined as NYHA class II or higher.
2.1±1.0 (range, 1.0 to 4.5) immediately before the
thromboembolic event. The clinical characteristics of
the patients who underwent ablation and the controls
treated with drugs are summarized in Table 2.
Overall Survival
The observed survival among the patients who underwent ablation is shown in Figure 1A, along with
the expected survival for age- and sex-matched members of the Minnesota population. The observed survival was significantly worse than the expected survival (P<0.001). The survival curve of the controls
treated with drugs is also shown in Figure 1A. The
observed survival rates of the patients who underwent
ablation and the controls treated with drugs were not
significantly different (P=0.44; risk ratio for the ab-
TABLE 2. CLINICAL CHARACTERISTICS OF THE PATIENTS WHO
UNDERWENT ABLATION AT THE TIME OF THE PROCEDURE
AND OF THE CONTROLS TREATED WITH DRUGS AT THE TIME
OF IN-HOSPITAL THERAPY.*
CHARACTERISTIC
ABLATION DRUG THERAPY
(N=350)
(N=229)
P VALUE
Age (yr)
Male sex (%)
Duration of follow-up (mo)
Coronary-artery bypass graft (%)
Cardiac-valve surgery (%)
Coronary artery disease (%)
Previous myocardial infarction (%)
Congestive heart failure (%)†
Left ventricular ejection fraction (%)
NYHA functional class
Nonischemic cardiomyopathy (%)
Diabetes mellitus (%)
Previous cerebrovascular accident (%)
Hypertension (%)
No. of antiarrhythmic drugs
68±11
53
36±26
18
17
45
27
21
47±17
1.3±0.7
16
20
19
46
4.3±2.0
67±13
66
48±23
17
13
36
17
16
49±16
1.2±0.6
15
17
17
38
2.0±1.0
0.97
0.003
0.77
0.25
0.07
0.006
0.13
0.17
0.07
0.67
0.49
0.38
0.04
<0.001
*Plus–minus values are means ±SD. NYHA denotes New York Heart
Association.
†Congestive heart failure was defined as NYHA class II or higher.
lation group as compared with the controls, 1.14; 95
percent confidence interval, 0.81 to 1.60).
Overall Survival with Coexisting Heart Disease
In subgroup analyses, the survival rate among 115
patients with atrial fibrillation and congestive heart
failure who underwent ablation was compared with
that among 58 controls with the same indications who
were treated with drugs (Fig. 1B). The difference in
survival between the groups was not significant (P=
0.75; risk ratio, 1.09; 95 percent confidence interval,
0.66 to 1.79). Survival was similar for 156 patients
with coronary artery disease who underwent ablation
and 83 controls with coronary artery disease who
were treated with drugs (P=0.85; risk ratio, 0.96; 95
percent confidence interval, 0.60 to 1.52) (Fig. 1C).
When patients with a history of congestive heart failure and previous myocardial infarction were excluded from the analysis, survival among the 194 patients
who underwent ablation was similar to that among
the 144 controls who were treated with drugs (P=
0.13; risk ratio, 1.57; 95 percent confidence interval,
0.88 to 2.81).
Univariate and multivariate predictors of death,
with associated risk ratios, 95 percent confidence intervals, and P values, are summarized in Table 3. Multivariate analysis showed that previous myocardial
infarction (P<0.001), a history of congestive heart
failure (P=0.02), and use of cardiac drugs after the
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1045
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
All Patients and Controls
A
100
90
Expected
Survival (%)
80
Ablation
70
60
Drug therapy
50
40
30
20
10
0
0
1
2
3
4
6
5
Years of Follow-up
Patients and Controls withI
Congestive Heart Failure
B
100
100
90
90
80
70
60
Drug therapy
80
Ablation
Survival (%)
Survival (%)
Patients and Controls withI
Coronary Artery Disease
C
Drug therapy
50
40
30
70
Ablation
60
50
40
30
20
20
10
10
0
0
0
1
2
3
4
5
0
6
Years of Follow-up
1
2
3
4
5
6
Years of Follow-up
Figure 1. Observed Survival among Patients Who Underwent Ablation of the Atrioventricular Node and among Controls Treated
with Drugs for Atrial Fibrillation, and Expected Survival Rates Based on Mortality in an Age- and Sex-Matched General Population.
As shown in Panel A, observed survival among patients who underwent ablation of the atrioventricular node and implantation of
a permanent pacemaker for atrial fibrillation between 1990 and 1998 was worse than the expected survival based on mortality
among age- and sex-matched members of the Minnesota population (P<0.001); however, it was similar to the survival among controls treated with drugs for atrial fibrillation (P=0.44). In the subgroup of patients with congestive heart failure (Panel B), the survival
among the 115 patients who underwent ablation was similar to that among the 58 controls treated with drugs (P=0.75). In the
subgroup with coronary artery disease (Panel C), the survival rates were not significantly different for the 156 patients who underwent ablation and the 83 controls treated with drugs (P=0.85).
ablation (P=0.03) were independent predictors of
death (Table 3). Cumulative survival rates were significantly worse than expected survival rates for patients who had a history of myocardial infarction or
congestive heart failure or who received cardiac-drug
therapy after ablation (Fig. 2). The observed survival
rates among patients without a history of myocardial
infarction were not significantly different from the expected survival rates (P=0.07); the same was true for
those who did not receive cardiac-drug therapy after
ablation (P=0.32). The observed survival among patients without congestive heart failure was worse than
the expected survival rate (P=0.05), but 17 percent
of the patients with a history of congestive heart failure also had a history of myocardial infarction, and
42 percent of the patients with a history of congestive heart failure were taking cardiac drugs after ablation. For the 121 patients without any of the three
1046 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
S U RV I VA L A F T E R A B L AT I O N O F T H E AT R I OV E N T R I C U L A R N O D E A N D I M P L A N TAT I O N O F A P E R M A N E N T PAC E M A K E R
TABLE 3. PREDICTORS OF DEATH IN PATIENTS WITH ATRIAL
FIBRILLATION AFTER ABLATION OF THE ATRIOVENTRICULAR
NODE AND PERMANENT PACING.*
VARIABLE
RISK RATIO (95% CI)
P VALUE
Univariate predictors
Previous myocardial infarction
History of congestive heart failure
Cardiac-drug therapy after ablation
Diabetes mellitus
Coronary artery disease
NYHA functional class »II
Dilated cardiomyopathy
Longer duration of arrhythmia
Smoking
Left ventricular ejection fraction <40%
Older age (each 10-year increase)
3.67
2.88
2.72
2.52
2.34
1.59
2.00
0.91
1.78
0.59
1.28
(2.34–5.75)
(1.84–4.52)
(1.63–4.53)
(1.56–4.06)
(1.46–3.75)
(1.23–2.05)
(1.27–3.16)
(0.85–0.97)
(1.12–2.85)
(0.37–0.96)
(1.02–1.61)
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
0.003
0.006
0.02
0.03
0.03
Multivariate predictors
Previous myocardial infarction
History of congestive heart failure
Cardiac-drug therapy after ablation
2.70 (1.65–4.28)
1.72 (1.11–2.94)
1.81 (1.05–3.08)
<0.001
0.02
0.03
*CI denotes confidence interval, and NYHA New York Heart Association.
independent risk factors, the observed survival was
similar to the expected survival (P=0.43) (Fig. 3).
Lone Atrial Fibrillation
In our study, patients were considered to have lone
atrial fibrillation if they did not have ischemic heart
disease, hyperthyroidism, congestive heart failure, cardiomyopathy, hypertension, chronic obstructive pulmonary disease, previous cardiac surgery, or potentially life-shortening noncardiac disease (diabetes or
cancer). There was no age restriction. Twenty-six patients met the criteria for lone atrial fibrillation (14
men and 12 women; mean age, 64±13 years; range,
41 to 83 years); none of them died during a mean
follow-up period of 37±27 months.
Mortality and Causes of Death
At the latest assessment, 78 patients had died
(mean follow-up, 27±25 months; range, 3 days to
88 months) (Table 1). Their mean age at the time of
ablation was 69±10 years (range, 39 to 95). The
causes of death are summarized in Table 4.
DISCUSSION
In this long-term follow-up study, we assessed the
survival of patients who presented with symptomatic
atrial fibrillation that was refractory to medical therapy and who then underwent ablation of the atrioventricular node and implantation of a permanent
pacemaker. Among the patients who underwent ablation, the observed overall survival was significantly
worse than the expected survival for age- and sexmatched members of the Minnesota population. The
observed survival among patients who underwent
ablation for atrial fibrillation was similar to that among
controls treated with drugs for atrial fibrillation. In
the absence of previous myocardial infarction, previous congestive heart failure, and treatment with cardiac medications after ablation, the observed survival
among patients who underwent ablation was similar
to the expected survival for age- and sex-matched
members of the Minnesota population. None of the
26 patients with lone atrial fibrillation died during a
mean follow-up period of 37±27 months.
Our observations confirm that the presence of preexisting cardiac disease is the main determinant of
long-term survival in patients with atrial fibrillation
who undergo ablation of the atrioventricular node.
More important, the normal survival rate among patients without clinically significant heart disease, the
excellent survival rate among patients with lone atrial fibrillation, and the similar survival rates for the patients who underwent ablation and the controls with
atrial fibrillation who were treated with drugs suggest
that controlling the ventricular rate and alleviating
symptoms by ablation of the atrioventricular node
and permanent pacing do not have an adverse effect
on long-term survival in this patient population.
Survival data from epidemiologic studies have
demonstrated higher mortality among patients with
atrial fibrillation than among patients in sinus
rhythm.1,2,4,26-28 Data from the Framingham Heart
Study showed that the risk-factor–adjusted odds ratio for mortality was 1.5 for men and 1.9 for women
with atrial fibrillation, as compared with subjects in
sinus rhythm.1 In some patients, atrial fibrillation
may be a marker of atherosclerosis, older age, and loss
of vascular compliance (all of which could be associated with a higher risk of stroke and death); nevertheless, the evidence supports the conclusion that atrial fibrillation is an independent predictor of poor longterm survival.
Results from observational studies17-20 and randomized trials 29,30 have demonstrated that ablation of the
atrioventricular node and permanent pacing are effective in controlling the ventricular rate, alleviating
symptoms, and improving the quality of life, exercise
tolerance, and left ventricular function. However, there
is concern that the creation of complete atrioventricular block that is inherent in this approach and the
requirement for permanent pacing to which it leads
may have an adverse effect on survival. According to
the Framingham Heart Study, overall mortality for
men between 65 and 74 years old is 20.8 percent at
one year and 48.2 percent at five years; for women
in the same age group, overall mortality is 18.2 percent at one year and 38.9 percent at five years.1 The
mean age of our study population was 68±11 years,
and overall mortality was 8 percent at one year and
27 percent at five years; these rates compare favorably
with those in the Framingham Heart Study, in which
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1047
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
Myocardial Infarction
A
No Myocardial Infarction
100
100
Expected
80
80
70
70
60
50
Previous myocardialI
infarction
40
30
P<0.001
20
Expected
90
Survival (%)
Survival (%)
90
No previous myocardialI
infarction
60
50
40
30
P=0.07
20
10
10
0
0
0
1
2
3
4
5
6
0
1
Years after Ablation
100
Expected
80
80
70
70
History of congestiveI
heart failure
60
50
40
30
P<0.001
20
6
5
6
5
6
Expected
No history of congestiveI
heart failure
60
50
40
30
P=0.05
20
10
10
0
0
0
1
2
3
4
5
6
0
1
Years after Ablation
100
Expected
80
80
Survival (%)
90
Cardiac-drug use
60
50
40
30
P<0.001
20
3
4
No Cardiac-Drug Use
90
70
2
Years after Ablation
Cardiac-Drug Use
100
Survival (%)
5
90
Survival (%)
Survival (%)
90
4
No Congestive Heart Failure
100
C
3
Years after Ablation
Congestive Heart Failure
B
2
Expected
No cardiac-drug use
70
60
50
40
30
P=0.32
20
10
10
0
0
0
1
2
3
4
5
6
Years after Ablation
0
1
2
3
4
Years after Ablation
Figure 2. Cumulative Survival for Subgroups of Patients Who Underwent Ablation of the Atrioventricular Node and Implantation of
a Permanent Pacemaker between 1990 and 1998 and Expected Survival Based on Mortality among Age- and Sex-Matched Controls.
Panel A shows the survival curves for those with a history of myocardial infarction (left-hand side) and those without such a history
(right-hand side), Panel B for those with and those without a history of congestive heart failure (CHF), and Panel C for those with
and those without cardiac-drug use after ablation.
1048 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
S U RV I VA L A F T E R A B L AT I O N O F T H E AT R I OV E N T R I C U L A R N O D E A N D I M P L A N TAT I O N O F A P E R M A N E N T PAC E M A K E R
100
Observed
90
Survival (%)
80
Expected
70
60
50
40
30
20
10
0
0
1
2
3
4
5
6
7
Years of Follow-up
Figure 3. Observed Survival among 121 Patients Who Underwent Ablation of the Atrioventricular Node and Implantation of a Permanent Pacemaker for Atrial Fibrillation but Who Had No History of Congestive Heart Failure or Myocardial
Infarction and No Cardiac-Drug Use after Ablation, as Compared with the Expected Survival Based on Mortality in an
Age- and Sex-Matched Control Population.
In this subgroup of patients, the observed survival and the expected survival were not significantly different (P=0.43).
most cases of atrial fibrillation were managed medically. Although direct comparisons cannot be made
between our data and data from the Framingham
Heart Study, because of differences in the study populations, the methods of analysis, and the timing of
the studies, it is encouraging to note that overall
mortality was lower among the patients in our study
who underwent ablation of the atrioventricular node
and permanent pacing than it was among patients in
the Framingham Heart Study.
The safety of ablation for controlling the ventricular rate is confirmed by the similar long-term survival in a group of consecutive patients receiving medical
treatment for atrial fibrillation. Because the controls
who were treated with drugs were identified retrospectively, the clinical characteristics of the two groups
are not identical, but the survival rate in this group
was similar to that among patients who underwent
ablation, despite the fact that the ablation group had
higher proportions of men, of patients with myocardial infarction and hypertension, and of patients in
whom previous drug treatment had failed.
Our study confirmed that preexisting cardiovascular disease and coexisting medical conditions are predictors of a higher risk of death in patients with atrial
fibrillation. The mode of pacing and the type of atrial
fibrillation (chronic or paroxysmal) were not independent predictors of long-term survival. The observation that survival was normal among patients without a history of myocardial infarction or congestive
heart failure who were not taking cardiac medications
after ablation suggests that in addition to having beneficial effects on symptoms, as demonstrated by other
studies, this therapy is unlikely to have a negative ef-
TABLE 4. PRIMARY CAUSES OF DEATH
AMONG PATIENTS WHO UNDERWENT
ABLATION OF THE ATRIOVENTRICULAR
NODE AND PERMANENT PACING FOR
ATRIAL FIBRILLATION.
CAUSE
OF
DEATH
Cardiac causes
Congestive heart failure
Myocardial infarction
Sudden death
Other
Noncardiac causes
Stroke
Respiratory failure
Cancer
Infection
Other
Unknown cause
Total
NO. OF PATIENTS
(%)
49
26
10
5
8
24
2
8
7
3
4
5
78
(63)
(33)
(13)
(6)
(10)
(31)
(3)
(10)
(9)
(4)
(5)
(6)
(100)
fect on long-term survival. The observation of similar
rates of survival in subgroups of patients with coronary artery disease or congestive heart failure whether they were treated medically or with ablation indicates that ablation of the atrioventricular node is as
safe as conventional medical treatment for atrial fibrillation in patients with underlying heart disease.
The clinical features of patients with lone atrial
fibrillation have been highlighted by epidemiologic
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1049
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
studies such as the Framingham Heart Study31 and
a study from Olmsted County, Minnesota,32 which
found a low risk of stroke and a low rate of mortality
overall. In our study, the indications in 26 patients
(7 percent) met the definition of lone atrial fibrillation, and none of them had died by the end of the
study, further supporting the conclusion that ablation of the atrioventricular node and permanent pacing do not negatively affect long-term survival in the
absence of clinically significant heart disease.
During follow-up, 49 of the 78 patients who died
(63 percent) died of cardiac causes. This high proportion is probably the result of the prevalence of cardiovascular diseases in our study population (Table 1).
Five patients (1 percent of all the patients who underwent ablation) had sudden death from cardiac causes,
and in four of them, underlying heart disease with
left ventricular dysfunction had previously been documented. Earlier studies from a registry of patients
who underwent direct-current ablation of the atrioventricular node estimated that the prevalence of
sudden death from cardiac causes after ablation is between 2.04 percent and 3.70 percent.22,33 Most sudden deaths occurred in patients with preexisting heart
disease.
Our observations and conclusions should be interpreted in the light of the limitations imposed by
a retrospective study design. All the information was
obtained from original hospital records of the Mayo
Clinic. Although these records were interpreted and
transferred into a standard data format, most of the
information was qualitative. The multivariate model
was used to minimize the effect of base-line differences. Selection of the study patients and the control
population was not random, but the inclusion of consecutive patients minimized selection bias. Although
the relative benefit of ablation of the atrioventricular
node and permanent pacing, as compared with other methods of treatment, can be determined only by
prospective, randomized trials, it is unlikely that such
studies will be conducted, given the difficulties in
maintaining sinus rhythm and controlling the ventricular rate by other medical and nonmedical methods, the diverse population of patients, and the high
rate of crossover that would be expected.
Although the observed overall survival among the
patients in our study who underwent ablation was
significantly worse than the expected survival among
matched controls from the Minnesota population, the
observed survival among patients without overt heart
disease was similar to that of the general-population
controls, and no deaths occurred during follow-up
among patients with lone atrial fibrillation. Survival
rates were similar in the group receiving medical treatment for atrial fibrillation and the group that underwent ablation of the atrioventricular node. These observations suggest that permanent atrioventricular
block and implantation of a pacemaker after ablation
of the atrioventricular node do not have an important adverse effect on survival, thus reassuring patients
and physicians that ablation of the atrioventricular
node is an acceptable treatment option for symptomatic atrial fibrillation that is refractory to medical
therapy.
REFERENCES
1. Benjamin EJ, Wolf PA, D’Agostino RB, Silbershatz H, Kannel WB,
Levy D. Impact of atrial fibrillation on the risk of death: the Framingham
Heart Study. Circulation 1998;98:946-52.
2. Lake FR, Cullen KJ, de Klerk NH, McCall MG, Rosman DL. Atrial
fibrillation and mortality in an elderly population. Aust N Z J Med 1989;
19:321-6.
3. Kannel WB, Abbott RD, Savage DD, McNamara PM. Epidemiologic
features of chronic atrial fibrillation: the Framingham Study. N Engl J Med
1982;306:1018-22.
4. Gajewski J, Singer RB. Mortality in an insured population with atrial
fibrillation. JAMA 1981;245:1540-4.
5. Phillips SJ, Whisnant JP, O’Fallon WM, Frye RL. Prevalence of cardiovascular disease and diabetes mellitus in residents of Rochester, Minnesota.
Mayo Clin Proc 1990;65:344-59.
6. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent
risk factor for stroke: the Framingham Study. Stroke 1991;22:983-8.
7. 2000 Heart and stroke statistical update. Dallas: American Heart Association, 1999.
8. Nemec J, Shen W-K. Pharmacotherapy of atrial fibrillation. Expert
Opin Pharmacother 1999;1:81-96.
9. Antman EM, Beamer AD, Cantillon C, McGowan N, Friedman PL.
Therapy of refractory symptomatic atrial fibrillation and atrial flutter: a
staged care approach with new antiarrhythmic drugs. J Am Coll Cardiol
1990;15:698-707.
10. Pritchett ELC. Management of atrial fibrillation. N Engl J Med 1992;
326:1264-71.
11. Zehender M, Hohnloser S, Muller B, Meinertz T, Just H. Effects of
amiodarone versus quinidine and verapamil in patients with chronic atrial
fibrillation: results of a comparative study and a 2-year follow-up. J Am
Coll Cardiol 1992;19:1054-9.
12. Crijns HJ, Van Gelder IC, Van Gilst WH, Hillege H, Gosselink AM,
Lie KI. Serial antiarrhythmic drug treatment to maintain sinus rhythm after
electrical cardioversion for chronic atrial fibrillation or atrial flutter. Am J
Cardiol 1991;68:335-41.
13. Prystowsky EN, Benson DW Jr, Fuster V, et al. Management of patients with atrial fibrillation: a statement for healthcare professionals: from
the Subcommittee on Electrocardiography and Electrophysiology, American Heart Association. Circulation 1996;93:1262-77.
14. Touboul P. Atrioventricular nodal ablation and pacemaker implantation
in patients with atrial fibrillation. Am J Cardiol 1999;83:241D-245D.
15. Brignole M, Menozzi C. Control of rapid heart rate in patients with
atrial fibrillation: drugs or ablation? Pacing Clin Electrophysiol 1996;19:
348-56.
16. Brignole M. Ablate and pace: a pragmatic approach to paroxysmal atrial
fibrillation not controlled by antiarrhythmic drugs. Heart 1998;79:531-3.
17. Twidale N, Sutton K, Bartlett L, et al. Effects on cardiac performance
of atrioventricular node catheter ablation using radiofrequency current for
drug-refractory atrial arrhythmias. Pacing Clin Electrophysiol 1993;16:
1275-84.
18. Marshall HJ, Harris ZI, Griffith MJ, Gammage MD. Atrioventricular
nodal ablation and implantation of mode switching dual chamber pacemakers: effective treatment for drug refractory paroxysmal atrial fibrillation.
Heart 1998;79:543-7.
19. Kay GN, Ellenbogen KA, Giudici M, et al. The Ablate and Pace Trial:
a prospective study of catheter ablation of the AV conduction system and
permanent pacemaker implantation for treatment of atrial fibrillation. J Interv Card Electrophysiol 1998;2:121-35.
20. Fitzpatrick AP, Kourouyan HD, Siu A, et al. Quality of life and outcomes after radiofrequency His-bundle catheter ablation and permanent
pacemaker implantation: impact of treatment in paroxysmal and established
atrial fibrillation. Am Heart J 1996;131:499-507.
21. Therneau T, Sicks J, Bergstralh E, Offord J. Expected survival based
on hazard rates. Technical report series. No. 52. Section of biostatistics.
Rochester, Minn.: Mayo Clinic, March 1994.
22. Trohman RG, Simmons TW, Moore SL, Firstenberg MS, Williams D,
Maloney JD. Catheter ablation of the atrioventricular junction using radiofrequency energy and a bilateral cardiac approach. Am J Cardiol 1992;70:
1438-43.
1050 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
S U RV I VA L A F T E R A B L AT I O N O F T H E AT R I OV E N T R I C U L A R N O D E A N D I M P L A N TAT I O N O F A P E R M A N E N T PAC E M A K E R
23. Olgin JE, Scheinman MM. Comparison of high energy direct current
and radiofrequency catheter ablation of the atrioventricular junction. J Am
Coll Cardiol 1993;21:557-64.
24. Peto R, Peto J. Asymptotically efficient rank invariant test procedures.
J R Stat Soc [A] 1972;135:185-206.
25. Cox DR. Regression models and life-tables. J R Stat Soc [B] 1972;34:
187-220.
26. Krahn AD, Manfreda J, Tate RB, Mathewson FA, Cuddy TE. The natural history of atrial fibrillation: incidence, risk factors, and prognosis in the
Manitoba Follow-Up Study. Am J Med 1995;98:476-84.
27. Dries DL, Exner DV, Gersh BJ, Domanski MJ, Waclawiw MA, Stevenson LW. Atrial fibrillation is associated with an increased risk for mortality
and heart failure progression in patients with asymptomatic and symptomatic left ventricular systolic dysfunction: a retrospective analysis of the
SOLVD trials: Studies of Left Ventricular Dysfunction. J Am Coll Cardiol
1998;32:695-703.
28. Wolf PA, Mitchell JB, Baker CS, Kannel WB, D’Agostino RB. Impact
of atrial fibrillation on mortality, stroke, and medical costs. Arch Intern
Med 1998;158:229-34.
29. Brignole M, Menozzi C, Gianfranchi L, et al. Assessment of atrioventricular junction ablation and VVIR pacemaker versus pharmacological
treatment in patients with heart failure and chronic atrial fibrillation: a randomized, controlled study. Circulation 1998;98:953-60.
30. Brignole M, Gianfranchi L, Menozzi C, et al. Assessment of atrioventricular junction ablation and DDDR mode-switching pacemaker versus pharmacological treatment in patients with severely symptomatic paroxysmal atrial fibrillation: a randomized controlled study. Circulation 1997;96:2617-24.
31. Brand FN, Abbott RD, Kannel WB, Wolf PA. Characteristics and
prognosis of lone atrial fibrillation: 30-year follow-up in the Framingham
Study. JAMA 1985;254:3449-53.
32. Kopecky SL, Gersh BJ, McGoon MD, et al. The natural history of lone
atrial fibrillation: a population-based study over three decades. N Engl J
Med 1987;317:669-74.
33. Scheinman MM, Morady F, Hess DS, Gonzalez R. Catheter-induced
ablation of the atrioventricular junction to control refractory supraventricular arrhythmias. JAMA 1982;248:851-5.
Copyright © 2001 Massachusetts Medical Society.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1051
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
Brief Report
E FFECT OF THE T YROSINE
K INASE I NHIBITOR STI571
IN A P ATIENT WITH A M ETASTATIC
G ASTROINTESTINAL S TROMAL T UMOR
HEIKKI JOENSUU, M.D., PETER J. ROBERTS, M.D.,
MAARIT SARLOMO-RIKALA, M.D.,
LEIF C. ANDERSSON, M.D., PEKKA TERVAHARTIALA, M.D.,
DAVID TUVESON, M.D., PH.D.,
SANDRA L. SILBERMAN, M.D., PH.D.,
RENAUD CAPDEVILLE, M.D., SASA DIMITRIJEVIC, PH.D.,
BRIAN DRUKER, M.D., AND GEORGE D. DEMETRI, M.D.
G
ASTROINTESTINAL stromal tumors are a
group of mesenchymal neoplasms that arise
from precursors of the connective-tissue cells
of the gastrointestinal tract.1 They occur predominantly in middle-aged and older persons, and approximately 70 percent of the tumors are found in
the stomach, 20 to 30 percent are found in the small
intestine, and less than 10 percent are found elsewhere in the gastrointestinal tract.1 Recent studies
have shown that cells in gastrointestinal stromal tumors express a growth factor receptor with tyrosine
kinase activity termed c-kit. This receptor, the product of the proto-oncogene c-kit, can be detected by
immunohistochemical staining for CD117, which
appears to be the most specific diagnostic criterion
for the diagnosis of gastrointestinal stromal tumors.2
The ligand for the c-kit receptor is stem-cell factor,
also known as steel factor or c-kit ligand.3 Mutations
of c-kit that cause constitutive activation of the tyrosine kinase function of c-kit are detectable in most
gastrointestinal stromal tumors and appear to play a
central part in the pathogenesis of these tumors.4,5
These mutations result in ligand-independent tyrosine
kinase activity, autophosphorylation of c-kit, uncontrolled cell proliferation, and stimulation of downstream signaling pathways, including those involving
From the Departments of Oncology (H.J.) and Radiology (P.T.), Helsinki University Central Hospital, Helsinki, Finland; the Department of Surgery, Turku University Central Hospital, Turku, Finland (P.J.R.); the Department of Pathology, Haartman Institute, University of Helsinki,
Helsinki (M.S.-R., L.C.A.); the Department of Biology and Howard
Hughes Medical Institute, Massachusetts Institute of Technology, Cambridge (D.T.); the Center for Sarcoma and Bone Oncology, Department of
Adult Oncology, Dana–Farber Cancer Institute and Harvard Medical
School, Boston (D.T., G.D.D.); Novartis Oncology, East Hanover, N.J.
(S.S.), and Basel, Switzerland (R.C., S.D.); and the Department of Medical
Oncology, Oregon Health Sciences University, Portland (B.D.). Address
reprint requests to Dr. Joensuu at the Department of Oncology, Helsinki
University Central Hospital, Haartmaninkatu 4, P.O. Box 180, FIN-00029,
Helsinki, Finland.
Drs. Joensuu and Roberts contributed equally to the article.
phosphatidylinositol 3-kinase and mitogen-activated
protein kinases. Gastrointestinal stromal tumors are
notoriously unresponsive to cancer chemotherapy, and
there is no effective therapy for advanced, metastatic
disease.6
We used STI571 (Glivec, Novartis, Basel, Switzerland),7 an inhibitor of the tyrosine kinase activity of
c-kit, in a patient with a gastrointestinal stromal tumor.
CASE REPORT
In October 1996, a 50-year-old, previously healthy woman presented with mild abdominal discomfort and a large mass in the
upper abdomen. Two tumors, 6.5 and 10 cm in diameter, were removed from the stomach by proximal gastric resection, and the
greater omentum and mesocolic peritoneum were removed because of the presence of multiple metastatic nodules 1 to 2 mm in
diameter. Histologic examination of the specimens revealed more
than 20 cells undergoing mitosis per 10 high-power fields and
identified the masses as a gastrointestinal stromal tumor. The diagnosis was confirmed by immunostaining for CD117, and a c-kit
mutation consisting of a deletion of 15 bp from exon 11 was detected in tumor DNA amplified by the polymerase chain reaction.8
Recurrent tumors in the left upper abdomen, two liver metastases, and multiple small intra-abdominal metastases were excised
in February 1998, and in September 1998 six more liver metastases and an ovarian metastasis were removed. Seven cycles of chemotherapy with mesna, doxorubicin, ifosfamide, and dacarbazine
were given from November 1998 to March 1999 for additional
liver metastases, but there was no clinical response. In March 1999,
progression of the disease prompted removal of a metastasis that
was obstructing the large bowel and 45 smaller metastases by laparotomy. The patient was treated between April 1999 and February 2000 with 400 mg of thalidomide once daily and 900,000 U
of subcutaneous interferon alfa three times a day, but by February
2000 the liver metastases were progressing in size and number,
and several new intra-abdominal and mesenteric metastases were
documented by magnetic resonance imaging (MRI).
The patient then agreed to participate in this study of STI571.
The institutional review board of Helsinki University Central Hospital approved the study, and the patient gave written informed
consent. Treatment with four 100-mg capsules of STI571 once
daily was started in March 2000. This dose was based on evaluations of the safety and tolerability of STI571 in patients with
chronic myeloid leukemia.9 Toxicity was assessed at follow-up visits every two to four weeks, and blood-cell counts and blood chemical values were analyzed every one to two weeks. The response
to treatment was assessed with dynamic MRI, positron-emission
tomography (PET) with [18F]fluorodeoxyglucose as a tracer, and
serial needle biopsies of a liver metastasis.
METHODS
Immunostaining for CD117 was performed with a polyclonal
rabbit antibody (sc-168, Santa Cruz Biotechnology, Santa Cruz,
Calif.) diluted 1:200 and for Ki-67 antigen, a marker of cell proliferation, with another polyclonal rabbit antibody (A0047, Dako,
Glostrup, Denmark) diluted 1:150. Staining was analyzed with a
detection kit (ChemMate Peroxidase/DAB, Dako) designed to
be used with an automated immunostaining system (TechMate
500 Medical Systems, Ventana, Tucson, Ariz.).
RESULTS
Evaluation of the Response by MRI
When measured as the sum of the products of two
perpendicular axes of each of eight large liver metastases, the size of the tumor one day before the start
of treatment with STI571 was 112.5 cm2. On subsequent MRI scans, the size of the tumor was as fol-
1052 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
B R IEF R EPOR T
lows: 67 cm 2 (after 2 weeks of treatment), 54 cm 2
(at 1 month), 42 cm 2 (at 2 months), 36 cm 2 (at 4
months), 33 cm 2 (at 5.5 months), and 28 cm 2 (at 8
months). No new lesions appeared, and 6 of the 28
liver metastases disappeared. At the peripheral rim of
the hepatic metastases, the considerable contrast enhancement that had been seen on the dynamic MRI
scans (a finding consistent with the presence of viable
tumor) before the beginning of STI571 treatment was
dramatically reduced; indeed, no enhancement was
seen on dynamic MRI scans obtained during treatment. In addition, many of the metastases became
hypodense (Fig. 1). As of February 2001, the tumor at
A
C
all sites continued to respond to treatment, and the
patient remained clinically well.
Evaluation by PET Scanning with [18F]Fluorodeoxyglucose
Multiple liver metastases and increased accumulation of [18F]fluorodeoxyglucose in the right renal pelvis and ureter, a finding indicative of hydronephrosis,
were seen on a PET scan obtained four days before
treatment with STI571 was started (Fig. 2A). On a
PET scan obtained one month after STI571 was started, no abnormal uptake of [18F]fluorodeoxyglucose
was seen in the liver or right kidney (Fig. 2B). In a
finding consistent with the changed, hypodense ap-
B
Figure 1. Transaxial Gadolinium-Enhanced T1-Weighted MRI
Studies of the Upper Abdomen.
Before STI571 therapy (Panel A), multiple metastatic lesions
were present in the liver. Contrast enhancement of the metastases was highly heterogeneous, with strong enhancement at
the periphery. Enhancement was less intense in the central parts
of the metastases, suggesting necrosis. After four weeks of treatment with STI571 (Panel B), the metastases had a cyst-like appearance. After eight months of treatment (Panel C), the metastases were smaller, and some had disappeared.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1053
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
pearance of metastases on MRI, “cold” areas, with
less uptake of [18F]fluorodeoxyglucose than in the surrounding liver parenchyma, were seen at the sites of
liver metastases on a PET scan obtained two months
after STI571 was started.
Histologic Findings
Serial needle-biopsy specimens of a ventrally located liver metastasis obtained one and two months
after STI571 treatment was started showed a marked
decrease in the density of the tumor cells, as well as
myxoid degeneration and scarring, with no signs of
an inflammatory reaction or necrosis (Fig. 3). The
few remaining cells in the myxoid background were
probably pyknotic tumor cells and not mast cells, according to their immunohistochemical characteristics (positive for CD117 and negative for CD45 and
Giemsa stain). These cells did not stain for the cellproliferation marker Ki-67, suggesting that they were
not actively dividing. Endothelial cells within the lesion were histologically normal, with no suggestion of
cytotoxic effects.
Side Effects of STI571
STI571 was well tolerated, with only mild, transient
nausea related to the swallowing of the capsules; this
minor symptom improved when the drug was taken
with food. No clinically significant changes were noted
in the peripheral blood-cell counts or blood chemi-
A
cal values. No drug-related adverse effects on the liver,
kidneys, or heart were observed. All of the main subjective adverse effects were mild (grade 1 according to
version 2.0 of the Common Toxicity Criteria of the
National Cancer Institute10) and consisted of an increased frequency of bowel movements (two to four
a day), occasional muscle cramps in the legs, and slight,
transient ankle edema. The World Health Organization performance status improved from 1 (indicating
the presence of cancer-related symptoms) to 0 (normal) during STI571 therapy.
DISCUSSION
There is no effective therapy for unresectable or
metastatic gastrointestinal stromal tumor, which is invariably fatal. STI571, a phenylaminopyrimidine derivative, is a small molecule that selectively inhibits the
enzymatic activity of several tyrosine kinases, including ABL and the BCR-ABL fusion protein of chronic
myeloid leukemia and Philadelphia chromosome–positive acute lymphoblastic leukemia; platelet-derived
growth factor receptor; and the product of the c-kit
gene. This selective activity of STI571 suggests that it
has a relatively narrow spectrum of anticancer activity.
Our results indicate that inhibition by STI571 of the
constitutively active mutant c-kit tyrosine kinase of
gastrointestinal stromal tumors is an effective therapy
for these tumors.
Our patient had a rapidly progressive metastatic gas-
B
Figure 2. PET Studies with [18F]Fluorodeoxyglucose as the Tracer.
Before STI571 therapy (Panel A), there were multiple metastases in the liver and upper abdomen. There was also marked retention of [18F]fluorodeoxyglucose in the right renal pelvis and
ureter, a finding indicative of hydronephrosis. After four weeks of treatment (Panel B), there was
no abnormal uptake of tracer in the liver or right kidney.
1054 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
BR IEF R EPOR T
trointestinal stromal tumor that was resistant to chemotherapy. She had a complete metabolic response
within one month after the start of STI571 treatment,
as shown by negative findings on PET and the 52 percent decrease in tumor volume on MRI. Many of the
liver metastases became hypodense, and the tumor enhancement on dynamic MRI was markedly reduced,
suggesting decreased viability. Histopathological evaluation of serial needle-biopsy specimens of a liver
metastasis confirmed the anticancer activity of this
treatment. With treatment, extensive fibrosis, myxoid
degeneration, and a few scattered, nonproliferating
CD117-positive cells replaced the abundant, frequently mitotic, Ki-67–positive gastrointestinal stromaltumor cells. The absence of visible damage to the
vascular endothelial cells in the biopsy specimens indicated the selective action of STI571 in this patient.
These responses have now continued during more
A
B
C
D
E
F
G
Figure 3. Histologic Appearance of the Primary Gastrointestinal Stromal Tumor (Hematoxylin and
Eosin [Panels A, B, and C] and Immunostaining for Ki-67 [Panels D and E] and CD117 [Panels F and G]).
In 1996, frequent mitotic figures were present (Panel A, ¬400). In 2000, a pretreatment biopsy specimen from a cellular liver metastasis (Panel B, ¬200) had a high frequency of Ki-67–positive nuclei
(Panel D, ¬200) and staining for CD117 (Panel F, ¬200). After three weeks of STI571 treatment, histologic examination of the liver metastasis showed myxoid degeneration and a few pyknotic cells
(Panel C; hematoxylin and eosin, ¬200), no staining for Ki-67 (Panel E, ¬200), and only a few, scattered CD117-positive cells (Panel G, ¬200).
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1055
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
than 11 months of treatment. In addition, the toxicity of STI571 therapy was minimal and consisted
mainly of mild dyspepsia and a slightly increased frequency of bowel movements.
In addition to its activity in BCR-ABL–positive
leukemias, STI571 may be active in solid tumors
that rely on the expression of c-kit, ABL, or plateletderived growth factor receptor. Among the solid tumors, gastrointestinal stromal tumors may be especially responsive to STI571 because they uniformly
express c-kit and because a tumor-specific c-kit mutation appears to be the chief cause of this neoplasm.
Our patient’s favorable response to STI571 supports
the concept that specific inhibition of tyrosine kinase is a clinically useful therapeutic intervention for
tumors in which aberrant tyrosine kinase signaling is
critical.
We are indebted to J. Lasota and M. Miettinen (Department of
Soft Tissue Pathology, Armed Forces Institute of Pathology, Washington, D.C.) for allowing us to refer to the results of c-kit mutation
analysis in this patient; to H. Minn (Turku PET Center, University
of Turku, Turku, Finland) for skillful analyses of PET images; to
Christopher Fletcher, Jonathan Fletcher, and Samuel Singer (Departments of Pathology and Surgical Oncology, Dana–Farber Cancer Institute, Brigham and Women’s Hospital, and Harvard Medical School, Boston); and to Charles D. Blanke and Michael C.
Heinrich (Department of Medical Oncology, Oregon Health Sciences
University, Portland) for helpful discussions.
REFERENCES
1. Miettinen M, Sarlomo-Rikala M, Lasota J. Gastrointestinal stromal tumors: recent advances in understanding of their biology. Hum Pathol
1999;30:1213-20.
2. Sarlomo-Rikala M, Kovatich AJ, Barusevicius A, Miettinen M. CD117:
a sensitive marker for gastrointestinal stromal tumors that is more specific
than CD34. Mod Pathol 1998;11:728-34.
3. Zsebo KM, Williams DA, Geissler EN, et al. Stem cell factor is encoded
at the S1 locus of the mouse and is the ligand for the c-kit tyrosine kinase
receptor. Cell 1990;63:213-24.
4. Hirota S, Isozaki K, Moriyama Y, et al. Gain-of-function mutations
of c-kit in human gastrointestinal stromal tumors. Science 1998;279:57780.
5. Lux ML, Rubin BP, Biase TL, et al. KIT extracellular and kinase domain mutations in gastrointestinal stromal tumors. Am J Pathol 2000;156:
791-5.
6. Plaat BE, Hollema H, Molenaar WM, et al. Soft tissue leiomyosarcomas
and malignant gastrointestinal stromal tumors: differences in clinical outcome and expression of multidrug resistance proteins. J Clin Oncol 2000;
18:3211-20.
7. Druker BJ, Tamura S, Buchdunger E, et al. Effects of a selective inhibitor of the Abl tyrosine kinase on the growth of Bcr-Abl positive cells. Nat
Med 1996;2:561-6.
8. Lasota J, Jasinski M, Sarlomo-Rikala M, Miettinen M. Mutations in
exon 11 of c-kit occur preferentially in malignant versus benign gastrointestinal tumors and do not occur in leiomyomas or leiomyosarcomas. Am
J Pathol 1999;154:53-60.
9. Druker BJ, Talpaz M, Resta DJ, et al. Efficacy and safety of a specific
inhibitor of the BCR-ABL tyrosine kinase in chronic myeloid leukemia.
N Engl J Med 2001;344:1031-7.
10. Cancer Therapy Evaluation Program. Common toxicity criteria, version 2.0. Bethesda, Md.: National Cancer Institute, March 1998.
Copyright © 2001 Massachusetts Medical Society.
RECEIVE THE JOURNAL’ S TABLE OF CONTENTS
EACH WEEK BY E-MAIL
To receive the table of contents of the
New England Journal of Medicine by e-mail
every Wednesday evening,
you can sign up through our Web site at:
http://www.nejm.org
1056 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
IMAGES IN C LINICA L MED IC INE
Images in Clinical Medicine
A
C
B
A Medical Mystery
This 82-year-old woman was referred to a physician. What is the diagnosis?
ARJEN F. NIKKELS, M.D., PH.D.
GÉRALD E. PIÉRARD, M.D., PH.D.
University of Liège
B-4000 Liège, Belgium
Editor’s note: We invite our readers to offer their opinions by e-mail ([email protected]) or fax (617-739-9864). We will not be able
to acknowledge the receipt of responses. We will publish the diagnosis in the Correspondence section of the May 24, 2001, issue. All
replies must be received by April 19, 2001.
Copyright © 2001 Massachusetts Medical Society.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1057
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
Review Articles
Advances in Immunology
I A N R . M A C K A Y , M. D . , A N D F R E D S . R O S E N , M .D .,
Ed i t ors
C OMPLEMENT
First of Two Parts
MARK J. WALPORT, PH.D., F.R.C.P.
C
OMPLEMENT is part of the innate immune
system and underlies one of the main effector mechanisms of antibody-mediated immunity. It has three overarching physiologic activities
(Table 1): defending against pyogenic bacterial infection, bridging innate and adaptive immunity, and
disposing of immune complexes and the products of
inflammatory injury. In this review, each of these activities will be placed in a clinical context.
Complement was first identified as a heat-labile
principle in serum that “complemented” antibodies in
the killing of bacteria. We now know that complement
is a system of more than 30 proteins in plasma and on
cell surfaces. Complement proteins in plasma amount
to more than 3 g per liter and constitute approximately 15 percent of the globulin fraction. The nomenclature of complement follows the historical order of discovery of the proteins and is one of the less
friendly aspects of the complement system.
The first complement pathway that was discovered,
the classical pathway, begins when antibody binds to
a cell surface and ends with lysis of the cell. The proteins of this pathway are designated C1 through C9
(Fig. 1). It was subsequently discovered that the numbering of the proteins did not quite correspond with
the order of the reaction, since C1 is followed in succession by C4, C2, C3, and C5, with numerical sanity
restored from C6 through C9. Proteins of the second pathway to be discovered, the alternative pathway, are called factors, followed by a letter, such as
factor B. Complement proteins on cell membranes can
be receptors for activated complement proteins or
proteins that regulate complement. They often have
From the Rheumatology Section, Division of Medicine, Imperial College of Science, Technology and Medicine, Hammersmith Campus, London. Address reprint requests to Dr. Walport at the Division of Medicine,
Imperial College School of Medicine, Hammersmith Hospital, Du Cane
Rd., London W12 0NN, United Kingdom, or at [email protected].
TABLE 1. THE THREE MAIN PHYSIOLOGIC ACTIVITIES
OF THE COMPLEMENT SYSTEM.
ACTIVITY
COMPLEMENT PROTEIN RESPONSIBLE FOR ACTIVITY
Host defense against
infection
Opsonization
Chemotaxis and activation
of leukocytes
Lysis of bacteria and cells
Covalently bound fragments of C3 and C4
Anaphylatoxins (C5a, C3a, and C4a); anaphylatoxin receptors on leukocytes
Membrane-attack complex (C5b–C9)
Interface between innate
and adaptive immunity
Augmentation of antibody
responses
Enhancement of immunologic memory
Disposal of waste
Clearance of immune complexes from tissues
Clearance of apoptotic cells
C3b and C4b bound to immune complexes
and to antigen; C3 receptors on B cells
and antigen-presenting cells
C3b and C4b bound to immune complexes
and to antigen; C3 receptors on follicular
dendritic cells
C1q; covalently bound fragments of C3
and C4
C1q; covalently bound fragments of C3
and C4
multiple names. Several complement proteins are
cleaved during activation of the system, and the fragments are designated with lowercase suffixes — for
example, C3 is cleaved into two fragments, C3a and
C3b. Normally, the large fragment is designated “b,”
and the small fragment “a.” Anarchy reigns with respect to the fragments of C2: the large fragment is
designated C2a, and the small, C2b, for historical
reasons.
The pathways leading to the cleavage of C3 are
triggered enzyme cascades, analogous to the coagulation, fibrinolysis, and kinin pathways. The terminal
complement pathway, leading to the formation of the
membrane-attack complex, is a unique system that
builds up a lipophilic complex in cell membranes from
several plasma proteins. There are three pathways of activation of the complement system: the classical, mannose-binding lectin, and alternative pathways (Fig. 1
and Table 2).
The regulatory mechanisms of complement are
finely balanced so that, on the one hand, the activation of complement is focused on the surface of invading microorganisms and, on the other hand, the
deposition of complement on normal cells and tissues is limited. When the mechanisms that regulate
this delicate balance go awry, the complement system may cause injury, and some of the resulting dis-
1058 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
ADVA NC ES IN IMMUNOLOGY
Increased susceptibility to pyogenic bacteria such
as Haemophilus influenzae and Streptococcus pneumoniae occurs in patients with defects of antibody production, complement proteins of the classical pathway,
or phagocyte function. Study of these patients shows
that the normal pathway of defense against pyogenic
bacteria is opsonization with antibody, followed by
the activation of complement, phagocytosis, and intracellular killing. The most important complement
opsonins in the defense against bacterial infection are
C3b and iC3b, the covalently-bound cleavage fragments of C3.
plex and infection is with neisserial disease, particularly Neisseria meningitidis.1-3 This complex is necessary
for the complement system to form a lytic channel
in neisseriae. Extracellular lysis is a major mechanism
of killing these organisms, which are capable of intracellular survival.
In parts of the world where meningococcal infections are highly endemic,4,5 there appears to be a high
prevalence of deficiencies of proteins of the membrane-attack complex. It has been argued that the
homozygous deficiency of C6 offers a selective advantage by protecting against the deleterious effects
of complement activation by endotoxin in infantile
gastroenteritis.5,6
There are good data from epidemiologic studies
in Japan on the association between a deficiency of
proteins of the membrane attack complex and neisserial infection. A survey of nearly 150,000 consecutive blood donors7 revealed 154 with the absence
of one of these proteins (138 had a C9 deficiency,
and 16 had a deficiency of C5, C6, C7, or C8), none
of whom had a history of neisserial infection. However, among 17 patients with meningococcal disease
ascertained from a register in Fukuoka, 8 had an inherited complement deficiency (4 of C7 and 4 of C9),
and a 9th patient had systemic lupus erythematosus
with acquired complement deficiency.8 From incidence
data for meningococcal disease,8 the prevalence of
complement deficiency among patients with sporadic
cases of meningococcal disease, and the frequency of
inherited complement deficiencies among Japanese
blood donors,7,8 the risk of meningococcal disease for
a person with a complement deficiency in Japan can
be calculated to be 0.5 percent per year. This is a relative risk of 5000, as compared with the incidence of
meningococcal disease among Japanese persons without a complement deficiency.
Complement Deficiency and Neisserial Infections
Mannose-Binding Lectin Deficiency
The sole clinical association between inherited deficiency of components of the membrane-attack com-
In 19769 a group of children between the ages of
six months and two years was described who had re-
eases will be discussed at the end of this review. Particular attention will be paid to illustrating the ways
in which the meticulous study of patients with abnormalities of the complement system has illuminated
our understanding of the immunobiology of complement.
This review is organized around the three main associations between complement and disease (Table 3):
complement deficiency and susceptibility to infection, the consequences of abnormalities in the regulation of the complement system, and the role of complement deficiency in inflammatory diseases.
COMPLEMENT AND THE DEFENSE
AGAINST INFECTION
Three types of complement deficiency can cause
increased susceptibility to pyogenic infections: a deficiency of the opsonic activities of the complement
system, which causes a general susceptibility to pyogenic organisms; any deficiency that compromises the
lytic activity of complement, which can increase the
susceptibility to neisserial infections; and deficient
function of the mannose-binding lectin pathway.
Pyogenic Infections
Figure 1 (facing page). The Three Activation Pathways of Complement: the Classical, Mannose-Binding Lectin, and Alternative Pathways.
The three pathways converge at the point of cleavage of C3. The classical pathway is initiated by the binding of the C1 complex
(which consists of C1q, two molecules of C1r, and two molecules of C1s) to antibodies bound to an antigen on the surface of a
bacterial cell. C1s first cleaves C4, which binds covalently to the bacterial surface, and then cleaves C2, leading to the formation of
a C4b2a enzyme complex, the C3 convertase of the classical pathway. The mannose-binding lectin pathway is initiated by binding of
the complex of mannose-binding lectin and the serine proteases mannose-binding lectin–associated proteases 1 and 2 (MASP1
and MASP2, respectively) to arrays of mannose groups on the surface of a bacterial cell. MASP2 acts in a fashion similar to that of
C1s to lead to the formation of the C3 convertase enzyme C4b2a. MASP1 may be able to cleave C3 directly. The alternative pathway
is initiated by the covalent binding of a small amount of C3b to hydroxyl groups on cell-surface carbohydrates and proteins and is
activated by low-grade cleavage of C3 in plasma. This C3b binds factor B, a protein homologous to C2, to form a C3bB complex.
Factor D cleaves factor B bound to C3b to form the alternative pathway C3 complex C3bBb. The binding of properdin stabilizes this
enzyme. The C3 convertase enzymes cleave many molecules of C3 to C3b, which bind covalently around the site of complement
activation. Some of this C3b binds to the C4b and C3b in the convertase enzymes of the classical and alternative pathways, respectively, forming C5 convertase enzymes. This C3b acts as an acceptor site for C5, which is cleaved to form the anaphylatoxin C5a
and C5b, which initiates the formation of the membrane-attack complex. The activities of biologically active proteins and protein
fragments of the complement pathway are described in Table 1.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1059
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
Classical3
pathway
Alternative3
pathway
Mannose-binding3
lectin pathway
Bacterial
cell
C3b
Mannose7
7
C4
Bacterial
cell
Factor3
B
C3b
C4b
MASP2
C1s
C4a
Properdin
MASP1
Mannose-7
binding7
lectin
C1r
C1q
C3b
C3b
Factor D
C4b
Ba
C2
Bb
C3 convertase
C4b C3 convertase
C3b
C2a
C2b
C3
C3b
C3b
C3b
Bacterial
cell
C3a
C3b
C5 convertase
C3b
Bb
C5
C5a
C5 convertase
C3b
C3b
C3b
C3b
C5b
C4b
Late steps of complement activation
C2a
C7
C6
C8
C3b
Membrane-7
attack complex
C6
C7
COLOR FIGURE
1
03/14 /01
1060 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
C5b
C6 C5b
C7 C8
C9
ADVA NC ES IN IMMUNOLOGY
TABLE 2. INITIATORS OF THE THREE ACTIVATION
PATHWAYS OF COMPLEMENT.
PATHWAY
INITIATORS
Classical
Immune complexes
Apoptotic cells
Certain viruses and gram-negative bacteria
C-reactive protein bound to ligand
Microbes with terminal mannose groups
Mannosebinding
lectin
Alternative
Many bacteria, fungi, viruses, tumor cells
current pyogenic infections and failure to thrive. Serum from these children failed to opsonize the yeast
Saccharomyces cerevisiae with C3.10 A similar opsonic
defect was present in approximately 5 percent of an
adult population without any obvious immunodeficiency. These results remained a mystery until it was
found that the opsonic defect correlated with reduced
serum levels of mannose-binding lectin.11,12 Mannose-binding lectin is a member of a family of calciumdependent lectins, the collectins (collagenous lectins),
and is homologous in structure to C1q. Mannosebinding lectin, a pattern-recognition molecule of the
innate immune system, binds to arrays of terminal
mannose groups on a variety of bacteria.13,14
A deficiency of mannose-binding lectin is due to
one of three point mutations in the gene for mannose-binding lectin, each of which reduces levels of
the lectin by interfering with the oligomerization of
the protein. A polymorphism in the promoter region
of the gene also influences levels of the protein.15,16
TABLE 3. CLINICAL EFFECTS
OF
The finding that the binding of mannose-binding
lectin to mannose residues can initiate complement activation was followed by the discovery of the mannosebinding lectin–associated serine protease (MASP) enzymes. Mannose-binding lectin activates complement
by interacting with two serine proteases called MASP1
and MASP2. MASP2 cleaves and activates C4 and
C2, and MASP1 may cleave C3 directly.17 These components of the complement system have been named
the mannose-binding lectin pathway (Fig. 1).
The low levels of mannose-binding lectin in young
children with recurrent infections18 suggest that the
mannose-binding lectin pathway is important during
the interval between the loss of passively acquired maternal antibody and the acquisition of a mature immunologic repertoire. Surprisingly, there is a high
frequency of dominantly expressed alleles of the gene
for mannose-binding lectin that result in low levels
of the protein in several ethnic groups. Perhaps this
deficiency during early childhood is counterbalanced
by an advantage later in life. There is epidemiologic
evidence that low levels of mannose-binding lectin
partially protect against mycobacterial infections, although at a population level the effect is not large.19,20
In addition, one study found that Ethiopians with
lepromatous leprosy had higher levels of serum mannose-binding lectin than their counterparts without
the disease.21 It is possible that the opsonization of
intracellular organisms such as mycobacteria by mannose-binding lectin enhances the entry of such pathogens into cells. The other side of this coin is that
pathogenic mycobacteria can synthesize a C4-like
molecule that binds the serine esterase fragment of
C2, C2a, leading to the cleavage of C3 and the deposition of C3b on mycobacterial-cell membranes.22
This mechanism could enhance infection by increas-
HEREDITARY COMPLEMENT DEFICIENCIES.*
COMPLEMENTARY DEFICIENCY
CONSEQUENCE
C3
Loss of major complement opsonin and
failure to activate membrane-attack–complex pathway
Failure to form membrane-attack complex
Pyogenic bacterial infections, may be accompanied by distinctive rash
Membranoproliferative glomerulonephritis
Neisserial infection
Loss of regulation of C1 and failure to
activate kallikrein
Failure to prevent the formation of membrane-attack complex on autologous cells
Failure to activate the classical pathway
Failure to regulate the activation of C3;
severe secondary C3 deficiency
Angioedema
C3, properdin, membraneattack–complex proteins
C1 inhibitor
CD59
C1q, C1r and C1s, C4, C2
Factor H and factor I
OF
COMPLEMENT ACTIVATION
CLINICAL ASSOCIATION
Hemolysis, thrombosis
Systemic lupus erythematosus
Hemolytic–uremic syndrome
Membranoproliferative glomerulonephritis
*C1 inhibitor, CD59, factor H, and factor I are regulatory proteins of the complement system. The other proteins are
members of the activation pathways of complement (as shown in Fig. 1). C3 deficiency is associated with both infectious
and inflammatory diseases.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1061
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
ing the uptake of mycobacteria into macrophages. By
contrast, low levels of mannose-binding lectin, by reducing opsonization, might confer resistance against
mycobacteria.
COMPLEMENT AND THE PATHOGENESIS
OF INFECTIOUS DISEASE
Many organisms take advantage of the complement system to enhance their virulence. Some viruses and intracellular bacteria use cell-bound complement regulatory molecules and receptors as a means
of gaining entry to the cell.23 The Epstein–Barr virus
uses complement receptor type 2 (also called CD21)
as a cellular receptor for its envelope glycoprotein
gp350/220,24,25 which explains the tropism of this
virus for B cells with complement receptor type 2.
Other examples are shown in Table 4. Some organisms activate host complement, thereby causing C3b
to bind to their surface. This process allows the microbe to use C3 receptors to enter the cell. The human immunodeficiency virus (HIV) and pathogenic
mycobacteria exploit this mechanism.22,26
Bacteria can evade complement if they have thick
capsules that form a physical barrier against the membrane-attack complex or express proteins that inhibit
the activation of complement.27 Group A streptococci
use the M protein to bind factor H,28 which increases the catabolism of C3b and reduces the formation
of C3 convertase enzymes. These bacteria also possess a peptidase that inhibits the inflammatory effects
of C5a.29
Viruses have three ways of evading complement.30
Some, such as HIV, incorporate complement regulatory proteins into the viral envelope.31,32 Others
TABLE 4. PROTEINS
MICROORGANISM
Epstein–Barr virus
Measles virus
OF THE
MICROBIAL LIGAND OR
MECHANISM OF ENTRY
INTO HOST CELL
Glycoprotein 350/
220
Hemagglutinin
Picornaviruses, such
as echoviruses and
coxsackieviruses
Mycobacterium tuberculosis
Human immunodeficiency virus
Capsid
Flavivirus, such as
West Nile virus
Deposition of host
C3 fragments
Deposition of host
C3 fragments
Deposition of host
C3 fragments
have proteins that are structural mimics of complement regulatory proteins. For example, the vaccinia
virus complement-control protein acts as a cofactor
to complement factor I, which cleaves C4b and C3b,
thereby inhibiting the activation of complement.33
Still other viruses use proteins that have no structural homology to complement regulatory proteins but
that nevertheless have similar functional properties.
An example is the glycoprotein C of several herpesviruses, which binds C3b and renders the viruses resistant to destruction by complement.34,35
ABNORMALITIES OF COMPLEMENT
REGULATION
Activation of C3
The three activation pathways of complement converge to generate C3 convertase, an enzyme that
cleaves C3. This protein at the heart of the complement system contains an internal thioester bond. The
cleavage of C3 to C3b by C3 convertase activates this
bond and allows, for a very short period, the stable
covalent binding of C3b to hydroxyl groups on carbohydrates and proteins in the immediate vicinity.
Any C3 that does not bind in this way is inactivated
by binding to water molecules. The covalent binding
of C3 to hydroxyl groups is a key feature of the complement system. It tags invading microorganisms as
foreign, and the bound C3 acts as a focus for further
complement activation on and around the microbe.
This leads to the production of anaphylatoxins and
the assembly of the membrane-attack complex on
the membrane of the invading pathogen (Fig. 1 and
Table 1).
Regulation of the cleavage of C3 is critical (Fig.
COMPLEMENT SYSTEM USED
HOST RECEPTOR*
Complement receptor
type 2
CD46 (membrane cofactor protein)
CD55 (decay-accelerating factor)
BY
MICROORGANISMS
OTHER RECEPTORS
CORECEPTORS
AND
Possibly a receptor on
epithelial cells
Coxsackie and adenovirus
receptor, a 2b 1 integrin
(very late antigen-2)
Complement receptor
type 3
Complement receptor
CD4, chemokine receptype 1, complement retor CCR5, chemokine
ceptor type 2, complereceptor CXCR4
ment receptor type 3
Complement receptor
type 3
TO
ENTER HUMAN CELLS.
LOCATION
OF
RECEPTOR
TARGET
OF INFECTION
B cells
B cells, epithelial cells
Many cells
Lymphocytes, macrophages,
dendritic cells, neurons
Cells of alimentary tract and
lymphoid tissues; can disseminate to most tissues
Macrophages
Many cells
Macrophages
Macrophages, some
CD4 T cells, dendritic cells, follicular
dendritic cells
Macrophages
CD4 T cells, dendritic cells,
follicular dendritic cells,
macrophages
Macrophages, neurons
*The cell tropism of most microorganisms cannot be explained by their use of a single receptor to enter cells. Some of the other host receptors used by
these organisms are indicated in the table.
1062 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
ADVA NCES IN IMMUNOLOGY
2). Once C3 is deposited covalently as C3b on a
membrane, it has two possible fates. The first is an
amplification step, in which more C3 is cleaved and
bound to the membrane. This process requires another complement protein, factor B. When factor B binds
to C3b, it is activated by the complement enzyme factor D and forms the C3 convertase enzyme C3bBb.
This enzyme cleaves more C3, causing many more
C3b molecules to be set down on the membrane.
The second possible fate of bound C3b is catabolism to inactive products. This step is mediated by
factor I, a regulatory enzyme of the complement
system, and three other proteins: the plasma protein
factor H and two cell-membrane proteins, complement receptor type 1 (also called CD35) and membrane cofactor protein (also called CD46) (Fig. 2).
Factor H is the dominant complement-control protein, and in its absence the regulation of complement activation breaks down completely.
A key to the regulation of complement activation
is whether factor B or factor H binds to C3b. The
carbohydrate environment of bound C3b influences
the outcome of the competition between factor B
and factor H for binding to C3b on membranes.36
Defective regulation of C3 is typically associated
with glomerulonephritis. The malfunction in these
cases is due to C3 nephritic factor, which increases
the stability of the C3 convertase enzymes, or to reduced function of factor H or factor I.
C3 Nephritic Factor
C3 nephritic factor is an autoantibody that binds
to and stabilizes the C3 convertase enzyme C3bBb.
This autoantibody is not usually connected with other conditions, but it is present in a few patients with
systemic lupus erythematosus.37 What triggers the production of C3 nephritic factor is unknown.38
C3 nephritic factor is associated with type II, densedeposit, membranoproliferative glomerulonephritis
(Fig. 3) and partial lipodystrophy. Membranoproliferative glomerulonephritis39 is characterized by mesangial proliferation, thickening of the capillary wall,
and subendothelial deposits of immunoglobulin and
C3. Electron microscopy of kidneys affected by membranoproliferative glomerulonephritis type II reveals
electron-dense deposits of unknown composition
within the glomerular basement membrane. Partial
lipodystrophy is a disfiguring condition that affects
the body from the waist upward but spares the legs.
The loss of fat in this condition was a mystery until
it was discovered that adipose cells are the main source
of factor D,40 which completes the formation of the
C3 convertase enzyme C3bBb by cleaving factor B
bound to C3b. There is a gradient in the concentration of factor D in the fat cells of the body; more is
present in the upper than the lower half of the body,
which could explain the distribution of the fat loss.41
It is likely that the C3 nephritic antibody in partial
lipodystrophy stabilizes the C3bBb C3 convertase
that forms in the immediate vicinity of adipocytes.
The abnormally stabilized enzyme may then cleave
enough C3 to allow assembly of the membrane-attack
complex, which lyses adipocytes.
Factor H Deficiency
Membranoproliferative glomerulonephritis also occurs with factor H deficiency. The mechanism for
this association is unknown, but in the absence of
factor H, continuous activation and turnover of C3
in the vicinity of the glomerular basement membrane
may cause C3b to bind to glomeruli and incite inflammation.
Some cases of homozygous factor H deficiency
have also been associated with the hemolytic–uremic
syndrome.42-47 Familial cases of the hemolytic–uremic syndrome or membranoproliferative glomerulonephritis have been described in association with low
levels of serum C3 but no obvious deficiency of factor H or factor I. However, in two of the families the
disease was linked by genetic mapping to the gene for
factor H.48 In these families and in a number of sporadic cases of recurrent hemolytic–uremic syndrome,
a heterozygous factor H deficiency has been identified. At present there is no test that will reliably
identify heterozygous carriers of factor H deficiency,
and the gene is not easy to sequence. A high index
of suspicion for factor H deficiency is needed in patients with reduced levels of C3 and recurrent
hemolytic–uremic syndrome or membranoproliferative glomerulonephritis.
C1 Inhibitor Deficiency
The main clinical feature of hereditary angioedema is recurrent angioedema, which may cause severe
illness if it affects the intestinal submucosa or death
by suffocation if it causes obstruction of the upper
airways. In this autosomal dominant disease, the single normal allele of the gene for C1 inhibitor cannot
ensure the production of physiologically adequate
amounts of C1 inhibitor. This serine protease inhibitor inactivates the complement serine esterases C1r
and C1s, kallikrein of the kinin system, and activated
factors XI and XII of the coagulation system. Although C1 inhibitor is not an important inhibitor of
plasmin, it is consumed by plasmin, and plasmin activation is probably the most important trigger of attacks of angioedema. In this disease, treatment by infusion of the deficient protein, C1 inhibitor, relieves
attacks and may be lifesaving.49
The main cause of the increased vascular permeability in hereditary angioedema is the excess bradykinin50 that results from the unregulated cleavage of
high-molecular-weight kininogen by kallikrein. The
angioedema associated with treatment with angiotensin-converting enzyme is also associated with elevated bradykinin levels,50 and angiotensin-convert-
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1063
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
C3
Nonactivating3
surface
Closed 7
thioester7
bond
Factor3
H
C3b
Factor3
H
C3b
Host3
cell
C3b
C3a
Activating3
surface
Factor3
B
Activated 7
thioester7
bond
Factor3
B
C3b
Factor3
H
Factor D
Factor I
3
Bacterial3
cell
Bb
Ba
C3b
iC3b
Factor3
H
C3f
Properdin
Thioester bond
C3b
O
Closed
O
C
C
N
C
C
O
C
O
C
C
C
C
C
N
O
O
Activated
O
N
C
S
C
C
O
C
C
C
O
C
N
C
C
C
C
C
C3
C3b
O
C
N
O
O
C
N
C
S
C
C
C
C
C
C3b
O
O
Figure 2. Regulation of the Cleavage of C3 by Factor H and Factor I.
The first product of the cleavage of C3 by a C3 convertase is C3b, which has an activated internal thioester bond. This bond enables
C3b to bind covalently to hydroxyl groups on nearby carbohydrates and protein-acceptor groups. If the acceptor molecule is on a
host cell surface, then protective regulatory mechanisms come into play. This is illustrated by the binding of factor H to C3b, which
acts as a cofactor to the serine esterase factor I. Factor I cleaves the C3 into an inactive product, iC3b, releasing a small peptide,
C3f. The iC3b can no longer participate in the formation of a C3 convertase enzyme. If C3b binds covalently to a bacterium, then
the enzyme precursor factor B binds to the C3b. Factor B that is bound to C3b is susceptible to cleavage and activation by the
enzyme factor D. This leads to the formation of the C3 convertase enzyme C3bBb, which is stabilized by the binding of properdin.
This enzyme cleaves more C3, leading to the deposition of additional C3b on the bacterium. The carbohydrate environment of the
surface on which the C3b is deposited determines the relative affinity of C3b for factor H or factor B. On host cell surfaces bearing
polyanions such as sialic acid, factor H binds to C3b with a higher affinity than does factor B. On microbial surfaces that lack a
polyanionic coating, factor B binds to C3b with a higher affinity than does factor H, leading to amplified cleavage of C3.
ing–enzyme inhibitors are absolutely contraindicated
in patients with C1 inhibitor deficiency.
In type 1 hereditary angioedema, which accounts
for approximately 85 percent of cases of the disease,
a mutation prevents the transcription of the abnormal allele, and hence, plasma levels of C1 inhibitor
are reduced. In type 2 hereditary angioedema, a point
mutation in the gene for C1 inhibitor alters the amino acid sequence at or near the active center of the
protein and abolishes its activity as a serine protease
inhibitor.51,52 The amount of inert C1 inhibitor in serum is normal or even elevated in patients with type 2
1064 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
ADVA NCES IN IMMUNOLOGY
hereditary angioedema, because the mutant protein
is not consumed by activated serine proteases. It is
easy to miss the diagnosis of this variant of hereditary angioedema if it is not understood that levels of
C1 inhibitor can be normal or high in patients with
the disease.
Another type of angioedema associated with C1
inhibitor deficiency is caused by the presence of autoantibodies against the protein. This syndrome usually occurs in elderly persons and is often associated
with a lymphoproliferative disease.53,54
Paroxysmal Nocturnal Hemoglobinuria
Paroxysmal nocturnal hemoglobinuria is a striking
example of the consequences of the failure to regulate the formation of the membrane-attack complex.55
In this disease, a somatic mutation in a clone of hematopoietic-cell precursors causes a deficiency of phosphatidylinositol glycan class A (PIG-A), a protein required for the synthesis of glycosylphosphatidylinositol
phospholipid. Glycosylphosphatidylinositol is the lipid tail that anchors more than 40 proteins to cell
membranes. As a result of the mutation in the PIG-A
gene, many proteins are deficient on the surface of
cells of patients with paroxysmal nocturnal hemoglobinuria.56,57 The most likely cause of intravascular hemolysis in these patients is the increased susceptibility
of red cells to complement. Two molecules anchored
by glycosylphosphatidylinositol — decay-accelerating
factor (also called CD55), which regulates the formation of C3 convertase, and CD59 (also called membrane inhibitor of reactive lysis), which restricts the
formation of the membrane-attack complex — have
been implicated. The discovery of isolated deficiencies
of decay-accelerating factor and of CD59 showed that
the deficiency of CD59 is responsible for the intravascular hemolysis that characterizes paroxysmal nocturnal hemoglobinuria.58-61
A
REFERENCES
B
Figure 3. Glomerulonephritis in the Presence of C3 Nephritic
Factor.
Panel A shows a glomerulus from a patient with type II membranoproliferative glomerulonephritis (hematoxylin and eosin,
¬200). There is mesangial expansion and hypercellularity, with
thickening of glomerular capillary walls. Panel B shows the
characteristic electron-dense deposits in the glomerular basement membrane (arrows) (¬12,000). Photographs courtesy of
Dr. Terry Cook, Department of Histopathology, Imperial College
School of Medicine, Hammersmith Hospital, London.
1. Ross SC, Densen P. Complement deficiency states and infection: epidemiology, pathogenesis and consequences of neisserial and other infections
in an immune deficiency. Medicine (Baltimore) 1984;63:243-73.
2. Wurzner R, Orren A, Lachmann PJ. Inherited deficiencies of the terminal components of human complement. Immunodefic Rev 1992;3:12347.
3. Lachmann PJ. Heberden oration 1986: complement — friend or foe?
Br J Rheumatol 1987;26:409-15.
4. Schlesinger M, Nave Z, Levy Y, Slater PE, Fishelson Z. Prevalence of
hereditary properdin, C7 and C8 deficiencies in patients with meningococcal infections. Clin Exp Immunol 1990;81:423-7.
5. Orren A, Potter PC, Cooper RC, du Toit E. Deficiency of the sixth
component of complement and susceptibility to Neisseria meningitidis infections: studies in 10 families and five isolated cases. Immunology 1987;
62:249-53.
6. Lehner PJ, Davies KA, Walport MJ, et al. Meningococcal septicaemia
in a C6-deficient patient and effects of plasma transfusion on lipopolysaccharide release. Lancet 1992;340:1379-81.
7. Inai S, Akagaki Y, Moriyama T, et al. Inherited deficiencies of the lateacting complement components other than C9 found among healthy
blood donors. Int Arch Allergy Appl Immunol 1989;90:274-9.
8. Nagata M, Hara T, Aoki T, et al. Inherited deficiency of ninth component of complement: an increased risk of meningococcal meningitis. J Pediatr 1989;114:260-4.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1065
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
9. Soothill JF, Harvey BA. Defective opsonization: a common immunity
deficiency. Arch Dis Child 1976;51:91-9.
10. Idem. A defect of the alternative pathway of complement. Clin Exp
Immunol 1977;27:30-3.
11. Super M, Thiel S, Lu J, Levinsky RJ, Turner MW. Association of low
levels of mannan-binding protein with a common defect of opsonisation.
Lancet 1989;2:1236-9.
12. Sumiya M, Super M, Tabona P, et al. Molecular basis of opsonic defect
in immunodeficient children. Lancet 1991;337:1569-70.
13. Turner MW. Mannose-binding lectin: the pluripotent molecule of the
innate immune system. Immunol Today 1996;17:532-40.
14. Stahl PD, Ezekowitz RA. The mannose receptor is a pattern recognition receptor involved in host defense. Curr Opin Immunol 1998;10:50-5.
15. Madsen HO, Garred P, Thiel S, et al. Interplay between promoter and
structural gene variants control basal serum level of mannan-binding protein. J Immunol 1995;155:3013-20.
16. Madsen HO, Satz ML, Hogh B, Svejgaard A, Garred P. Different molecular events result in low protein levels of mannan-binding lectin in populations from southeast Africa and South America. J Immunol 1998;161:
3169-75.
17. Matsushita M, Thiel S, Jensenius JC, Terai I, Fujita T. Proteolytic activities of two types of mannose-binding lectin-associated serine protease.
J Immunol 2000;165:2637-42.
18. Summerfield JA, Sumiya M, Levin M, Turner MW. Association of mutations in mannose binding protein gene with childhood infection in consecutive hospital series. BMJ 1997;314:1229-32.
19. Hoal-Van Helden EG, Epstein J, Victor TC, et al. Mannose-binding
protein B allele confers protection against tuberculous meningitis. Pediatr
Res 1999;45:459-64.
20. Bellamy R, Ruwende C, McAdam KP, et al. Mannose binding protein
deficiency is not associated with malaria, hepatitis B carriage nor tuberculosis in Africans. QJM 1998;91:13-8.
21. Garred P, Harboe M, Oettinger T, Koch C, Svejgaard A. Dual role of
mannan-binding protein in infections: another case of heterosis? Eur J Immunogenet 1994;21:125-31.
22. Schorey JS, Carroll MC, Brown EJ. A macrophage invasion mechanism
of pathogenic mycobacteria. Science 1997;277:1091-3.
23. Lindahl G, Sjobring U, Johnsson E. Human complement regulators:
a major target for pathogenic microorganisms. Curr Opin Immunol 2000;
12:44-51.
24. Tanner J, Weis J, Fearon D, Whang Y, Kieff E. Epstein-Barr virus
gp350/220 binding to the B lymphocyte C3d receptor mediates adsorption, capping, and endocytosis. Cell 1987;50:203-13.
25. Nemerow GR, Mold C, Schwend VK, Tollefson V, Cooper NR. Identification of gp350 as the viral glycoprotein mediating attachment of Epstein-Barr virus (EBV) to the EBV/C3d receptor of B cells: sequence homology of gp350 and C3 complement fragment C3d. J Virol 1987;61:
1416-20.
26. Speth C, Kacani L, Dierich MP. Complement receptors in HIV infection. Immunol Rev 1997;159:49-67.
27. Fishelson Z. Complement-related proteins in pathogenic organisms.
Springer Semin Immunopathol 1994;15:345-68. [Erratum, Springer Semin Immunopathol 1994;16:129-30.]
28. Horstmann RD, Sievertsen HJ, Knobloch J, Fischetti VA. Antiphagocytic activity of streptococcal M protein: selective binding of complement
control protein factor H. Proc Natl Acad Sci U S A 1988;85:1657-61.
29. Wexler DE, Chenoweth DE, Cleary PP. Mechanism of action of the
group A streptococcal C5a inactivator. Proc Natl Acad Sci U S A 1985;82:
8144-8.
30. Lubinski J, Nagashunmugam T, Friedman HM. Viral interference with
antibody and complement. Semin Cell Dev Biol 1998;9:329-37.
31. Saifuddin M, Parker CJ, Peeples ME, et al. Role of virion-associated
glycosylphosphatidylinositol-linked proteins CD55 and CD59 in complement resistance of cell line-derived and primary isolates of HIV-1. J Exp
Med 1995;182:501-9.
32. Marschang P, Sodroski J, Wurzner R, Dierich MP. Decay-accelerating
factor (CD55) protects human immunodeficiency virus type 1 from inactivation by human complement. Eur J Immunol 1995;25:285-90.
33. Kotwal GJ, Isaacs SN, McKenzie R, Frank MM, Moss B. Inhibition
of the complement cascade by the major secretory protein of vaccinia virus.
Science 1990;250:827-30.
34. Friedman HM, Cohen GH, Eisenberg RJ, Seidel CA, Cines DB. Glycoprotein C of herpes simplex virus 1 acts as a receptor for the C3b complement component on infected cells. Nature 1984;309:633-5.
35. Friedman HM, Wang L, Pangburn MK, Lambris JD, Lubinski J. Novel mechanism of antibody-independent complement neutralization of herpes simplex virus type 1. J Immunol 2000;165:4528-36.
36. Pangburn MK. Host recognition and target differentiation by factor
H, a regulator of the alternative pathway of complement. Immunopharmacology 2000;49:149-57.
37. Walport MJ, Davies KA, Botto M, et al. C3 nephritic factor and SLE:
report of four cases and review of the literature. QJM 1994;87:609-15.
38. Peters DK, Charlesworth JA, Sissons JG, et al. Mesangiocapillary nephritis, partial lipodystrophy, and hypocomplementaemia. Lancet 1973;2:
535-8.
39. Williams DG. Mesangiocapillary glomerulonephritis. In: Davison AM,
Cameron JS, Grünfeld J-P, Kerr DNS, Ritz E, Winearls CG, eds. Oxford
textbook of clinical nephrology. 2nd ed. Vol. 1. Oxford, England: Oxford
University Press, 1998:591-612.
40. White RT, Damm D, Hancock N, et al. Human adipsin is identical to
complement factor D and is expressed at high levels in adipose tissue. J Biol
Chem 1992;267:9210-3.
41. Mathieson PW, Peters DK. Lipodystrophy in MCGN type II: the clue
to links between the adipocyte and the complement system. Nephrol Dial
Transplant 1997;12:1804-6.
42. Levy M, Halbwachs-Mecarelli L, Gubler MC, et al. H deficiency in
two brothers with atypical dense intramembranous deposit disease. Kidney
Int 1986;30:949-56.
43. Ohali M, Shalev H, Schlesinger M, et al. Hypocomplementemic autosomal recessive hemolytic uremic syndrome with decreased factor H. Pediatr Nephrol 1998;12:619-24.
44. Pichette V, Querin S, Schurch W, Brun G, Lehner-Netsch G, Delage
JM. Familial hemolytic-uremic syndrome and homozygous factor H deficiency. Am J Kidney Dis 1994;24:936-41.
45. Noris M, Ruggenenti P, Perna A, et al. Hypocomplementemia discloses genetic predisposition to hemolytic uremic syndrome and thrombotic
thrombocytopenic purpura: role of factor H abnormalities: Italian Registry
of Familial and Recurrent Hemolytic Uremic Syndrome/Thrombotic
Thrombocytopenic Purpura. J Am Soc Nephrol 1999;10:281-93.
46. Wyatt RJ, Julian BA, Weinstein A, Rothfield NF, McLean RH. Partial
H (beta 1H) deficiency and glomerulonephritis in two families. J Clin Immunol 1982;2:110-7.
47. Thompson RA, Winterborn MH. Hypocomplementaemia due to a
genetic deficiency of beta 1H globulin. Clin Exp Immunol 1981;46:110-9.
48. Warwicker P, Donne RL, Goodship JA, et al. Familial relapsing haemolytic uraemic syndrome and complement factor H deficiency. Nephrol
Dial Transplant 1999;14:1229-33.
49. Waytes AT, Rosen FS, Frank MM. Treatment of hereditary angioedema with a vapor-heated C1 inhibitor concentrate. N Engl J Med 1996;334:
1630-4.
50. Nussberger J, Cugno M, Amstutz C, Cicardi M, Pellacani A, Agostoni
A. Plasma bradykinin in angio-oedema. Lancet 1998;351:1693-7.
51. Aulak KS, Pemberton PA, Rosen FS, Carrell RW, Lachmann PJ, Har–
rison RA. Dysfunctional C 1-inhibitor(At), isolated from a type II hereditary-angio-oedema plasma, contains a P1 ‘reactive centre’ (Arg444→His)
mutation. Biochem J 1988;253:615-8.
52. Skriver K, Radziejewska E, Silbermann JA, Donaldson VH, Bock SC.
CpG mutations in the reactive site of human C1 inhibitor. J Biol Chem
1989;264:3066-71.
53. Jackson J, Sim RB, Whelan A, Feighery C. An IgG autoantibody
which inactivates C1-inhibitor. Nature 1986;323:722-4.
54. Cicardi M, Beretta A, Colombo M, Gioffre D, Cugno M, Agostoni
A. Relevance of lymphoproliferative disorders and of anti-C1 inhibitor autoantibodies in acquired angio-oedema. Clin Exp Immunol 1996;106:47580.
55. Hillmen P, Richards SJ. Implications of recent insights into the pathophysiology of paroxysmal nocturnal haemoglobinuria. Br J Haematol
2000;108:470-9.
56. Luzzatto L, Bessler M. The dual pathogenesis of paroxysmal nocturnal
hemoglobinuria. Curr Opin Hematol 1996;3:101-10.
57. Boccuni P, Del Vecchio L, Di Noto R, Rotoli B. Glycosyl phosphatidylinositol (GPI)-anchored molecules and the pathogenesis of paroxysmal
nocturnal hemoglobinuria. Crit Rev Oncol Hematol 2000;33:25-43.
58. Shichishima T, Saitoh Y, Terasawa T, Noji H, Kai T, Maruyama Y.
Complement sensitivity of erythrocytes in a patient with inherited complete deficiency of CD59 or with the Inab phenotype. Br J Haematol 1999;
104:303-6.
59. Yamashina M, Ueda E, Kinoshita T, et al. Inherited complete deficiency of 20-kilodalton homologous restriction factor (CD59) as a cause of
paroxysmal nocturnal hemoglobinuria. N Engl J Med 1990;323:1184-9.
60. Merry AH, Rawlinson VI, Uchikawa M, Daha MR, Sim RB. Studies
on the sensitivity to complement-mediated lysis of erythrocytes (Inab phenotype) with a deficiency of DAF (decay accelerating factor). Br J Haematol 1989;73:248-53.
61. Telen MJ, Green AM. The Inab phenotype: characterization of the
membrane protein and complement regulatory defect. Blood 1989;74:
437-41.
1066 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
Copyright © 2001 Massachusetts Medical Society.
MED ICA L PROGR ES S
Medical Progress
A TRIAL F IBRILLATION
RODNEY H. FALK, M.D.
E
LECTROCARDIOGRAPHICALLY, atrial fibrillation is characterized by the presence of
rapid, irregular, fibrillatory waves that vary in
size, shape, and timing. This set of findings is usually
associated with an irregular ventricular response, although regularization may occur in patients with
complete heart block, an accelerated junctional or idioventricular rhythm, or a ventricular paced rhythm.
In the past decade, we have gained a greater understanding of atrial fibrillation. Experimental studies
have explored the mechanisms of the onset and maintenance of the arrhythmia; drugs have been tailored
to specific cardiac ion channels; nonpharmacologic
therapies have been introduced that are designed to
control or prevent atrial fibrillation; and data have
emerged that demonstrate a genetic predisposition in
some patients.1
EPIDEMIOLOGY
The incidence of atrial fibrillation approximately
doubles with each decade of adult life and ranges
from 2 or 3 new cases per 1000 population per year
between the ages of 55 and 64 years to 35 new cases
per 1000 population per year between the ages of 85
and 94 years. The arrhythmia may be an independent risk factor for death, with a relative risk of about
1.5 for men and 1.9 for women after adjustment for
known risk factors.2 It has been suggested that, in
patients with underlying ventricular dysfunction, this
increased risk of death is due primarily to heart failure.3
The term “lone atrial fibrillation” describes atrial
fibrillation in the absence of demonstrable underlying cardiac disease or a history of hypertension. It
may be due to fibrotic areas in the atrium that predispose patients to arrhythmia, to increased susceptibility to autonomic neural stimuli to the heart,4 or
to localized atrial myocarditis.5 Although the Framingham Heart Study suggested that patients with
lone atrial fibrillation had a risk of stroke that was
four times that of age-matched controls in sinus
rhythm,6 the absence of structural heart disease was
determined in that study without the use of echocardiography, and hypertension was not a criterion
for exclusion. It has been estimated that lone atrial
fibrillation occurs in approximately 3 percent of paFrom the Section of Cardiology, Boston Medical Center, Boston. Address reprint requests to Dr. Falk at the Boston Medical Center, Section of
Cardiology, 88 E. Newton St., Boston, MA 02118, or at [email protected].
tients with atrial fibrillation.7 In patients younger
than 60 years old, lone atrial fibrillation, although uncommon, has a benign prognosis. However, patients
older than 61 years of age who have lone atrial fibrillation have an increased risk of stroke and death.7,8
Whether the treatment of atrial fibrillation reduces
mortality can be evaluated only by prospective, randomized trials. One such study, the Atrial Fibrillation Follow-up Investigation of Rhythm Management
(AFFIRM) trial, is currently being conducted in the
United States.9 Its primary aim is to determine whether allowing atrial fibrillation to persist while controlling the heart rate and administering antithrombotic
therapy is associated with the same rate of mortality
as restoring sinus rhythm with antiarrhythmic drugs.
Several secondary analyses will define the optimal therapy and the risks for specific subgroups of patients.
HISTOLOGIC AND ELECTROPHYSIOLOGIC
FEATURES
Atrial fibrillation is usually precipitated by underlying cardiac or noncardiac disease. The resultant
atrial abnormality (frequently inflammation or fibrosis) acts as a substrate for the development of the arrhythmia.10 In addition, the onset of atrial fibrillation
usually requires a trigger. Triggers that may initiate
the arrhythmia include alterations in autonomic tone,11
acute or chronic changes in atrial wall tension,12 atrial
ectopic foci, and local factors. Cardiothoracic surgery,
a potent trigger of atrial fibrillation, has been discussed
by Ommen et al.13
In most cases of atrial fibrillation, multiple, small
reentrant circuits are constantly arising in the atria,
colliding, being extinguished, and arising again.14-16
A critical mass of atrial tissue is required to sustain
the minimal number of simultaneous circuits necessary for the perpetuation of the arrhythmia. Drugs
can prevent atrial fibrillation by increasing the circuit
wavelength,17 and invasive techniques can prevent it
by decreasing the size of the atrial segments.18
A second distinct mechanism causing atrial fibrillation has recently been recognized — a rapidly firing focus (or foci), usually located in or near the pulmonary veins. Such foci may mimic the appearance of
atrial fibrillation on the surface electrocardiogram19 or,
more commonly, may degenerate into or trigger classic atrial fibrillation after a brief burst of ectopic activity.20 In animals, repeated episodes of induced atrial fibrillation result in the development of sustained
arrhythmia.21 The electrophysiologic effect of these repeated episodes is a marked shortening of the atrial
refractory period and the loss of the normal lengthening of atrial refractoriness at slower heart rates. This
phenomenon, which may be reversible with the maintenance of sinus rhythm,22 has been termed atrial electrical remodeling.23 Pretreatment with verapamil may
markedly reduce the extent of remodeling,24 suggesting that cytosolic calcium overload is a contributory
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1067
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
factor. This observation may have clinical applications, and the use of verapamil in conjunction with
antiarrhythmic drugs before cardioversion from atrial fibrillation may reduce the risk of recurrence of arrhythmia.25,26
Prolonged episodes of atrial fibrillation frequently
cause mechanical dysfunction of the atrium.27 Restoration of sinus rhythm is generally associated with
the normalization of function over a period of two
to four weeks.28 Possibly because of the delayed recovery of atrial mechanical function, the risk of thromboembolism posed by atrial fibrillation seems to persist for a few weeks after cardioversion.
HEMODYNAMIC EFFECTS
Atrial fibrillation is associated with the loss of the
atrial contribution to ventricular filling. This may result in a decrease in ventricular stroke volume of up
to 20 percent.29 The irregularity of the ventricular
response may also contribute to hemodynamic impairment.30 The ventricular rate in patients with atrial fibrillation frequently has wide swings, with peaks
that exceed those that occur during sinus rhythm.31
In some patients with a poorly controlled ventricular
rate (generally, a mean of more than 100 beats per
minute), persistent tachycardia results in ultrastructural changes that cause ventricular dysfunction.24 This
tachycardia-mediated cardiomyopathy is often reversible after sinus rhythm has been restored or when the
heart rate during atrial fibrillation is controlled.32
SYMPTOMS
Some patients with atrial fibrillation have minimal
symptoms or none, whereas others may have severe
symptoms, particularly at the onset of the arrhythmia. Symptoms may range from palpitations to acute
pulmonary edema, but fatigue and other nonspecific
symptoms are probably the most common.33 The cognitive function of elderly patients with persistent arrhythmia may be impaired, as compared with that of
age-matched controls in sinus rhythm.34 Whether this
impairment is due to recurrent cerebral embolism or
cerebral hypoperfusion is unclear.
Not all episodes of arrhythmia are symptomatic,
and monitoring studies in patients with paroxysmal
atrial fibrillation demonstrate that asymptomatic episodes occur more frequently than do symptomatic
ones.35 Preliminary data suggest that the quality of life
is significantly impaired during atrial fibrillation, as
compared with the quality of life after the restoration
of sinus rhythm.36 However, in a recent small trial, control of the heart rate with the use of diltiazem during
atrial fibrillation produced as much relief of symptoms
as did attempts at the maintenance of sinus rhythm
with amiodarone.37 The impairment in the quality of
life in patients with paroxysmal atrial fibrillation is
equivalent to that seen in patients with more severe
cardiac disease, such as those who have undergone angioplasty.38 The ablation of the atrioventricular node
along with the implantation of a pacemaker significantly improves quality-of-life scores.39
APPROACH TO THE PATIENT
WITH ATRIAL FIBRILLATION
In the assessment of a patient with atrial fibrillation, it is important to determine the clinical significance of the arrhythmia and to identify any associated
conditions. The physician must be aware of any potentially negative effect that the therapy for the arrhythmia may have on the underlying heart disease
(for example, the negative inotropic effects of some
antiarrhythmic drugs). A careful history taking and
physical examination are mandatory. In particular,
physicians should seek evidence of a new onset or exacerbation of angina or congestive heart failure, since
such evidence may suggest a need for early cardioversion. Echocardiography is invaluable for evaluating
cardiac abnormalities, and thyroid-function tests occasionally reveal unexpected hyperthyroidism.
A three-part approach to treatment should be considered: physicians should assess the need for, the proper timing of, and the appropriate method for the restoration of sinus rhythm; they should evaluate the need
for anticoagulation to prevent embolic stroke; and they
should ensure appropriate control of the ventricular
rate while the patient is in atrial fibrillation.
Figure 1 (facing page). An Approach to the Management of Newly Diagnosed Atrial Fibrillation or Atrial Fibrillation of Recent Onset.
The pharmacologic therapies suggested for the termination of atrial fibrillation of less than 48 hours’ duration are not presented in
order of preference; to avoid drug interactions, no more than one should be used. Direct-current cardioversion can be attempted
as the initial strategy or used if drug therapy fails. This figure does not detail the investigation into the cause of atrial fibrillation.
Strong consideration should be given to the performance of echocardiography and thyroid-function tests, at minimum, for the evaluation of the cause and evaluation of ventricular function. Although low-molecular-weight heparin has not been compared in a
clinical trial with unfractionated heparin in patients with atrial fibrillation of recent onset, it has been found to be at least as effective
as unfractionated heparin in other situations when used for the prevention of arterial thromboembolism. It should also be noted
that intravenous unfractionated heparin has never been formally evaluated as a therapy for atrial fibrillation of recent onset. Although the Food and Drug Administration has not approved low-molecular-weight heparin for use in atrial fibrillation, it is a logical
alternative to intravenous unfractionated heparin. Both intravenous ibutilide and oral quinidine may provoke torsade de pointes.
Although oral quinidine is quite effective for the termination of an episode of acute atrial fibrillation, long-term oral quinidine is not
recommended for the maintenance of sinus rhythm (see Fig. 2). IV denotes intravenous, LV left ventricular, and TEE transesophageal echocardiography.
1068 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
MED ICA L PROGR ES S
Atrial fibrillation4
of recent onset
Hemodynamic instability,4
angina, or preexcited4
atrial fibrillation
Urgent cardioversion
Patient’s condition4
stable
Heart-rate control with4
IV diltiazem,4
IV beta-blocker, digoxin,4
or some combination
Spontaneous conversion
Remains in4
atrial fibrillation
Home, with follow-up4
to assess cause and4
likelihood of recurrence
IV unfractionated4
or subcutaneous4
low-molecular-weight4
heparin
Duration of atrial4
fibrillation «48 hr and4
no clinically significant LV4
dysfunction, mitral-valve4
disease, or previous embolism
Duration of atrial4
fibrillation >48 hr,4
unknown duration,4
or high risk of embolism
IV ibutilide;4
or oral propafenone (600 mg)4
or flecainide (300 mg);4
or oral quinidine (400–600 mg);4
or direct-current shock
TEE-guided cardioversion;4
or adequate anticoagulation4
for 3 wk, followed by direct-4
current cardioversion, with4
or without concomitant4
antiarrhythmic drugs
Sinus rhythm restored4
and maintained
Failed cardioversion or early4
recurrence of atrial fibrillation
Warfarin for 6–12 wk,4
followed by assessment4
of need for long-term4
antithrombotic therapy
Long-term antithrombotic4
therapy and rate control4
or repeated direct-current4
cardioversion with new4
antiarrhythmic drug
Recurrent or sustained4
atrial fibrillation with poor4
rate control or symptoms4
related to the irregular rhythm
Consider atrioventricular nodal4
ablation or other4
nonpharmacologic therapy
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1069
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
NEWLY DIAGNOSED ATRIAL
FIBRILLATION
An approach to newly diagnosed atrial fibrillation
is outlined in Figure 1. Hospitalization is not required
for all patients and can be limited to those with hemodynamic compromise or severely symptomatic arrhythmia, those at high risk for embolism (such as patients with heart failure), and patients in whom early
cardioversion is considered. In the absence of angina,
electrocardiographic evidence of myocardial ischemia,
or a recent infarction, there is no need for admission
to a coronary care unit in order to rule out myocardial
infarction, since ischemic heart disease rarely presents
as atrial fibrillation with no other signs or symptoms.40
In some patients, such as those with pulmonary edema, acute myocardial infarction, or unstable angina,
urgent cardioversion may be necessary as the initial
treatment. Even if their condition is clinically stable,
patients with atrial fibrillation and a rapid, wide-complex ventricular response related to the preexcitation
syndrome should also be considered for early electrical cardioversion, since the response to antiarrhythmic
agents is unpredictable in such patients and most
agents used for rate control are contraindicated.41
In the absence of an urgent need for cardioversion,
consideration should be given to pharmacologic rate
control. Although the atrial rate usually exceeds 350
beats per minute, the mean resting ventricular rate in
a patient with atrial fibrillation of new onset is between
110 and 130 beats per minute.42 In patients with the
Wolff–Parkinson–White syndrome and a short refrac-
tory period of the accessory pathway, the ventricular
response may exceed 250 beats per minute. In such
cases, the electrocardiogram demonstrates a widecomplex tachycardia, due to predominant accessorypathway conduction. A resting ventricular rate higher
than 150 beats per minute in the absence of preexcitation should raise the suspicion of a hyperadrenergic state, such as occurs in thyrotoxicosis, fever, or
acute gastrointestinal bleeding. A slow ventricular response in the absence of medication may occur with
high vagal tone in young athletes43 or in patients with
conduction-system disease.
Digoxin is somewhat effective for slowing the ventricular rate in a patient at rest, but its maximal action is achieved only after several hours,42,44 and it is
of little value in patients who are in a hyperadrenergic
state. Intravenous beta-blocking or calcium-channel–
blocking drugs produce more rapid rate control, regardless of the level of sympathetic tone. The appropriate doses of these drugs are given in Table 1.
Spontaneous conversion to sinus rhythm within 24
hours after the onset of atrial fibrillation is common,
occurring in up to two thirds of patients.45 Once the
duration of atrial fibrillation exceeds 24 hours, the likelihood of conversion decreases. After one week of persistent arrhythmia, spontaneous conversion is rare.45,46
Self-terminating episodes of atrial fibrillation often
recur. However, the arrhythmia-free period is unpredictable, and it may not be necessary to prescribe either long-term antiarrhythmic therapy or anticoagulation for all patients after the first documented episode.
TABLE 1. PHARMACOLOGIC HEART-RATE CONTROL
DRUG
CONTROL
OF
ACUTE EPISODE
Calcium-channel
blockers
Diltiazem
20-mg bolus followed, if necessary, by 25 mg
given 15 min later. Maintenance infusion
of 5–15 mg/hr.
Verapamil
5–10 mg IV over 2–3 min, repeated once,
30 min later. Maintenance infusion rate is
not reliably documented.
Beta-blockers†
Esmolol
0.5 mg/kg of body weight IV, repeated if
necessary. Follow with infusion at 0.05
mg/kg/min, increasing as needed to 0.2
mg/kg/min.
Metoprolol
5-mg bolus IV, repeated twice at intervals of
2 min. No data on maintenance infusion.
Propranolol
1–5 mg IV, given over 10 min.
Digoxin
1.0–1.5 mg IV or orally over 24 hr in doses
of 0.25 to 0.5 mg.
IN
CONTROL OF SUSTAINED
ATRIAL FIBRILLATION
Oral controlled-release formulation, 180–300 mg
daily.
Slow-release formulation,
120–240 mg once or
twice daily.
ATRIAL FIBRILLATION.*
COMMENTS
Long-term control may be better with the addition
of digoxin.
Causes elevation in digoxin level. May be more negatively
inotropic than diltiazem.
Not available in oral forms. Hypotension may be troublesome but responds to drug
discontinuation.
50–400 mg daily in divided Useful if there is concomitant coronary disease.
doses.
30–360 mg in divided dos- Noncardioselective: use cautiously in patients with a history
es or in long-acting form.
of bronchospasm.
0.125–0.5 mg daily.
Renally excreted. Slow onset even if given IV, with less
effective control than other agents, although may be
synergistic with them. Poor efficacy for exertional heartrate control.
*IV denotes intravenously.
†The beta-blockers listed are representative of agents in this category. Other intravenous or oral beta-blockers may be equally acceptable.
1070 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
MED ICA L PROGR ES S
Exceptions include highly symptomatic patients, patients who have had an embolic event, and those who
are at high risk for thromboembolism. In such patients, it is prudent to administer antiarrhythmic therapy, anticoagulant therapy, or both after the initial
episode.
rhythm. The drugs whose efficacy has been demonstrated are listed in Table 2.
Anticoagulation
In many patients, the precise time of onset of atrial
fibrillation cannot be determined accurately. Under
these circumstances, it is highly advisable to administer
anticoagulant therapy to the patient before attempting
cardioversion. There are two alternative approaches:
outpatient systemic anticoagulation with warfarin to
achieve an international normalized ratio of 2.0 to 3.0
for at least three weeks, followed by cardioversion; and
cardioversion guided by transesophageal echocardiography. In the latter approach, multiplane transesophageal echocardiography that indicates the absence of
thrombus is associated with an extremely low rate of
thromboembolism after cardioversion,49 provided that
short-term anticoagulant therapy is used before and
during the procedure and that warfarin is prescribed
after the procedure. Regardless of which of these
approaches is taken, anticoagulant therapy is mandatory for a minimum of three to four weeks after cardioversion. Since the greatest likelihood of reversion
to atrial fibrillation occurs in the first three months after the restoration of sinus rhythm,50 it is prudent to
continue anticoagulation for this period unless there is
a contraindication.
Antiarrhythmic-Drug Therapy
Early drug therapy to restore sinus rhythm can be
considered in patients in whom the arrhythmia has
lasted less than 48 hours or who are receiving longterm warfarin therapy. Digoxin is not effective in
converting atrial fibrillation to sinus rhythm,42,47 but
antiarrhythmic therapy increases the likelihood of
conversion to as much as 90 percent, if the drugs are
administered early and in adequate doses.48 If pharmacologic therapy is contemplated, continuous electrocardiographic monitoring during the first 48 to
72 hours after the initiation of antiarrhythmic therapy should be considered. It is not necessary during
the administration of amiodarone, unless sinus-node
dysfunction is suspected. Although a limited number of drugs are approved by the Food and Drug
Administration for the treatment of atrial fibrillation,
there is considerable information about the use of
oral and intravenous antiarrhythmic drugs for the
conversion of recent-onset atrial fibrillation to sinus
TABLE 2. DOSES
DRUG
Flecainide
Propafenone
Procainamide
Quinidine
Disopyramide
DOSE
FOR
OF
MEDICATIONS USED FOR CONVERSION OF RECENT-ONSET ATRIAL FIBRILLATION
AND MAINTENANCE OF SINUS RHYTHM.*
CONVERSION
300 mg orally (2 mg/kg of body
weight IV)
600 mg orally (2 mg/kg IV)
100 mg IV every 5 min to maximum
of 1000 mg
200 mg sulfate orally, followed 1–2
hr later by 400 mg
200 mg orally every 4 hr to maximum of 800 mg
DOSE
FOR
MAINTENANCE
50–150 mg twice daily
150–300 mg twice daily
Slow-release formulation, 1000–2000 mg
twice daily
200–400 mg sulfate 4 times daily, or
324–648 mg gluconate 3 times daily
100–150 mg 4 times daily or 200–300
mg controlled-release formulation
twice daily
120–160 mg twice daily
Sotalol
Not recommended (conversion rate
is low)
Dofetilide
0.5 mg twice daily orally (adjust dose
downward for patients with renal
disease)
1200 mg IV in 24 hr
0.5 mg twice daily (adjust dose downward
for patients with renal disease)
1 mg IV over 10 min in patients
weighing »60 kg, or 0.01 mg/kg
over 10 min in patients weighing
<60 kg; may be repeated once if
arrhythmia does not end within 10
min after end of initial infusion
Not available for maintenance (IV formulation only)
Amiodarone
Ibutilide
600 mg/day for 2 wk, then 200–400 mg
daily (lower dose is preferable)
COMMENTS
IV formulation not available in U.S. Approved only for
paroxysmal AF with structurally normal heart.
Same limitations as flecainide.
Long-term use associated with lupus. Not FDAapproved for AF.
Approved for AF but risk of death increased during
long-term therapy.
Not FDA-approved for AF. Strong negative inotropic
effect.
Poor conversion efficacy. Approved for maintenance of
sinus rhythm. Hospitalization for initiation is mandatory.
FDA-approved for conversion and maintenance. Hospitalization for initiation is mandatory.
IV amiodarone moderately effective for conversion,
but onset is slow. Good rate slowing in AF. Not
FDA-approved for this indication.
Do not use in patients with hypokalemia, a prolonged
QT interval, or torsade de pointes.
*Intravenous (IV) flecainide and propafenone (the doses of which are given in parentheses) are not available in the United States. AF denotes atrial
fibrillation, and FDA Food and Drug Administration.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1071
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
RECURRENT PAROXYSMAL ATRIAL
FIBRILLATION
In trials of anticoagulant therapy, patients with paroxysmal atrial fibrillation had the same risk of stroke
as subjects with persistent atrial fibrillation.51,52 Thus,
unless a patient with paroxysmal arrhythmia is younger than 65 years old and has no hypertension or underlying heart disease, long-term warfarin therapy
should be instituted.
Antiarrhythmic-Drug Therapy
Several drugs have been shown to be effective in the
treatment of paroxysmal atrial fibrillation. These include propafenone,53 flecainide,54 and sotalol.55 These
agents often do not totally abolish the arrhythmia, but
they increase the length of the interval between the
paroxysms. Although this decrease in the frequency of
paroxysm is often satisfactory for a reduction of symptoms, there are no data supporting the possibility that
having fewer episodes of atrial fibrillation decreases the
risk of thromboembolism. Furthermore, patients who
have symptomatic episodes of paroxysmal arrhythmia
may also have multiple episodes of asymptomatic atrial fibrillation.35 Although asymptomatic episodes tend
to be shorter than the symptomatic episodes, they may
still pose a risk of thromboembolism.
The treatment of paroxysmal atrial fibrillation with
antiarrhythmic drugs that also slow the heart rate
(such as sotalol) may convert symptomatic episodes to
asymptomatic ones.56 Decisions regarding the discontinuation of anticoagulation in such patients remain
clinically challenging, requiring physicians to weigh
the risk of bleeding against the risk of a stroke from
asymptomatic arrhythmia. Holter monitoring may be
valuable for assessing the presence of brief, asymptomatic runs of atrial fibrillation in such cases.
PERSISTENT ATRIAL FIBRILLATION
Once an episode of atrial fibrillation has lasted more
than seven days, spontaneous conversion is rare and
the condition can be defined as persistent. The decision to attempt to restore sinus rhythm in a patient
with persistent atrial fibrillation is not always clearcut. Restoration of sinus rhythm will generally improve
the patient’s symptoms, but not all patients have symptoms. There is thus a need to strike a balance between
the need for antiarrhythmic therapy and the likelihood
of side effects, particularly proarrhythmia.41
Cardioversion
Unless it is deemed urgent, the restoration of sinus
rhythm should be attempted only after adequate anticoagulant therapy, as described above. Synchronized,
direct-current cardioversion is usually required in order to restore sinus rhythm, although pharmacologic
conversion is successful in 10 to 30 percent of cases,
depending on the drug used and on the duration of
the arrhythmia.46,48,57 Restoration of sinus rhythm in
patients with atrial fibrillation frequently requires at
least 300 J of energy with most defibrillators currently
in use. However, the recent introduction of defibrillators with a biphasic wave form, rather than the traditional monophasic damped-sine wave form, is associated with a marked decrease in the energy required
for atrial defibrillation and with fewer failures.58
Failure to terminate an arrhythmia with a specific
antiarrhythmic agent does not mean that the same
drug will be ineffective in maintaining sinus rhythm
after electrical cardioversion. For the patient with minimal symptoms, it may be sufficient to perform directcurrent cardioversion without the administration of
an antiarrhythmic drug. If the arrhythmia recurs, the
cardioversion can be repeated in combination with the
use of an antiarrhythmic agent.
In some cases, sinus rhythm is not restored or is
restored only very briefly by electrical cardioversion.
In such a case, the use of intravenous ibutilide followed by another shock increases the likelihood of
restoration and the maintenance of sinus rhythm.59
However, ibutilide should be used with caution in
patients with impaired ventricular function, since it
may cause torsade de pointes,60 and the safety of this
approach in patients already receiving another antiarrhythmic drug has not been established. If a patient
fails to return to sinus rhythm even for one or two
beats despite these measures, transvenous internal cardioversion may be successful.61
The decision regarding which antiarrhythmic agent
to use for the maintenance of sinus rhythm should be
based on the known properties of the drugs, their side
effects, and their safety in the presence of structural
heart disease. There are few studies of comparative efficacy, but amiodarone has been shown to be superior
to both sotalol and propafenone for the maintenance
of sinus rhythm.62 To date, only dofetilide and amiodarone have been shown not to increase mortality
when prescribed to patients with heart failure.63,64 A
proposed outline for drug therapy is given in Figure 2.
Most recurrences of atrial fibrillation occur within three months after cardioversion of a first episode
of atrial fibrillation, regardless of the antiarrhythmic
agent used.50 Recurrence during this three-month period usually indicates a failure or an inadequate dose
of the drug and suggests the need to change the drug
or increase the dose if repeated cardioversion is contemplated. If the period after cardioversion during
which the patient is free of atrial fibrillation is longer
than three months, and particularly if it is longer than
six months, it may be reasonable to repeat direct-current cardioversion with the use of the same regimen
(after readministration of anticoagulant therapy if necessary), particularly if there was evidence of clinical
improvement when the patient was in sinus rhythm.
Acceptance of the atrial fibrillation along with the institution of rate control and adequate anticoagulation
1072 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
MED IC A L PROGR ES S
is an alternative, especially if symptoms related to arrhythmia are minimal.
Long-Term Anticoagulation
Several large trials have demonstrated the efficacy
of warfarin for the prevention of stroke in patients
with atrial fibrillation. Both the risk of stroke and the
Lone atrial fibrillation4
or hypertension
efficacy of warfarin among patients with persistent
arrhythmia were equivalent to those among patients
with paroxysmal arrhythmia. A high rate of intracranial hemorrhage complicated the use of warfarin in
elderly patients in one trial (the Stroke Prevention in
Atrial Fibrillation II study),65 but analysis of the degree of anticoagulation at the time of bleeding indicat-
Coronary artery disease4
without congestive heart failure
Propafenone4
or flecainide
Use with caution if there is4
hepatic impairment.4
Use with caution if atrial flutter4
has intermittently occurred.
Congestive heart failure4
or ejection fraction <35%
Sotalol
Dofetilide
Dofetilide
Amiodarone
Amiodarone
Sotalol
Disopyramide
Reduce dose (or avoid) if there4
is renal impairment.4
Use with caution if there is a history4
of bradycardia.4
Correct hypokalemia before using.
Disopyramide4
or quinidine (?)
Avoid disopyramide if prostatic4
symptoms are present, and reduce dose4
(or avoid) if there is renal impairment.4
Avoid quinidine if left ventricular4
hypertrophy is present.
Dofetilide
Reduce dose (or avoid) if there4
is renal impairment.4
Correct hypokalemia before using.
Amiodarone
Carefully consider long-term toxicity.4
Use with caution if there is previous4
bradycardia or serious pulmonary disease.
Figure 2. A Suggested Approach for the Selection of Antiarrhythmic Therapy to Maintain Sinus Rhythm after Cardioversion.
The order of the listed drugs is merely a guideline. Which drug to use first in each group will depend, to some extent, on the preference of the physician as well as the clinical characteristics of the specific patient. In patients with coronary disease, especially
active ischemia, it is prudent to avoid the class IC agents flecainide and propafenone, even though propafenone was not evaluated
in the Cardiac Arrhythmia Suppression Trial. Quinidine is included in this figure despite concern about increased mortality with its
use, since it is still used widely in some countries and no comparative mortality studies have been performed.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1073
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
ed that virtually all episodes occurred at an international normalized ratio greater than 3.0.66
Subsequent analyses suggest that the optimal international normalized ratio for the prevention of
stroke in patients with atrial fibrillation lies between
2.0 and 3.0.67,68 Pooled data from the anticoagulation trials offer insights into risk stratification with
respect to stroke. Clinical risk factors included a previous stroke or transient ischemic attack, hypertension (current or past), an age of more than 70 years,
diabetes, and congestive heart failure.51,69
The role of aspirin in the prevention of stroke in
patients with atrial fibrillation remains controversial.
In one trial, aspirin, in a prescribed dose of 325 mg
daily, reduced the annual rate of stroke by 42 percent, as compared with placebo (absolute reduction,
from 6.3 percent to 3.6 percent).70 A statistically insignificant reduction was found in two other trials,
one of which included only high-risk patients who
had had a previous stroke or transient ischemic attack.71 The other trial used a lower dose of aspirin
(82 mg).72 Aspirin (325 mg) prescribed in conjunction with “mini-dose” warfarin (1 to 3 mg daily) was
also found to be ineffective for the prevention of stroke
in patients with clinical risk factors.69 The Stroke Prevention in Atrial Fibrillation III investigators studied
the effects of 325 mg of aspirin alone in a group of patients initially believed to be at low risk for stroke.69,73
The stroke rate was 2.2 percent per year among the
patients for whom aspirin was prescribed but was
higher in the subgroup with a history of hypertension (3.6 percent per year, as compared with 1.1 percent among the patients who had never had hypertension).73 Although this study demonstrated that it
is possible to identify a cohort of patients with atrial
fibrillation and a low risk of stroke, it did not have a
placebo group and thus did not prove that aspirin is
superior to no therapy.
Recommendations for antithrombotic therapy in
patients with atrial fibrillation are summarized in Table 3. As a rule of thumb, all patients with atrial fibrillation should receive long-term anticoagulant therapy
with warfarin unless they are young (younger than 65
years old) and have none of the risk factors described
above, or unless there is a major contraindication to
the use of warfarin. In the absence of risk factors, aspirin alone (or no antithrombotic therapy) may be adequate.74 Advanced age is a risk factor for both stroke
and bleeding in patients receiving anticoagulation
therapy. However, the relative risk of stroke exceeds
that of bleeding, and whenever possible, elderly patients with atrial fibrillation should receive warfarin
therapy.67
Heart-Rate Control
The aims of pharmacologic control of the heart rate
in patients with persistent atrial fibrillation are to minimize symptoms related to swings in heart rate and
TABLE 3. ANTITHROMBOTIC THERAPY IN PATIENTS WITH
PERSISTENT OR PAROXYSMAL ATRIAL FIBRILLATION.
AGE
«65 yr
RISK FACTORS
FOR STROKE*
THERAPY
None
Aspirin or none (no proof that aspirin is superior in this group)
>65–75
None
Aspirin or warfarin (assess risk of warfarin as
yr
compared with the small risk of stroke)
Any
One or more (in- Warfarin (strongly advised unless very clear
cluding age >75 yr)
contraindications are present)
*Risk factors for stroke in patients with atrial fibrillation include mitral
stenosis, hypertension (including treated hypertension), previous transient
ischemic attack or stroke, congestive heart failure or left ventricular dysfunction, and an age of more than 75 years.
prevent excessive tachycardia during normal daily activities. Digoxin may be acceptable as the sole therapy
in an elderly, sedentary patient, but it is not very effective for preventing excessive tachycardia during moderate exertion. Beta-blocking drugs, verapamil, and
diltiazem are much more effective, and there is synergism between these drugs and digoxin.31,75 Beta-blocking agents are probably the drugs of choice in patients
with both atrial fibrillation and coronary artery disease,
and they may also be valuable when systolic dysfunction is present. Verapamil may elevate serum digoxin
levels into the toxic range, so the dose of digoxin
should be reduced if it is used with verapamil.76
DRUG-REFRACTORY ATRIAL
FIBRILLATION
Ablation of the Atrioventricular Node and Implantation
of a Pacemaker
The combination of persistent atrial fibrillation
and systolic dysfunction poses a challenge for ventricular rate control, since digoxin is often ineffective
and other agents may have a negative inotropic effect. Radio-frequency energy applied to the atrioventricular junction is highly effective in treating patients with these conditions. It produces complete
heart block, usually with a slow junctional escape
rhythm. Implantation of a permanent pacemaker (either single- or dual-chamber, depending on whether
the patient has paroxysmal or persistent atrial fibrillation) is required in order to maintain an adequate
heart rate after ablation. Patients with paroxysmal
atrial fibrillation often have symptoms caused by a
rapid, irregular ventricular response. After ablation
of the atrioventricular node and initiation of permanent pacing, there is usually a marked improvement
in the patient’s sense of well-being.77
Both paroxysmal atrial fibrillation and persistent
atrial fibrillation with an uncontrolled and rapid ventricular response have been associated with the development of tachycardia-mediated cardiomyopathy.78
1074 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
MED IC A L PROGR ES S
Uncontrolled studies have demonstrated that ablation of the atrioventricular node and implantation of
a pacemaker are associated with improvement in ventricular function in a substantial proportion of selected patients.78,79
Focal Ablation
Recently, a group of patients has been described
in whom atrial fibrillation is triggered by a rapidly
firing atrial focus located in the pulmonary veins or
(far less commonly) in the right atrium. In a substantial number of patients, the application of radiofrequency energy in the pulmonary veins at the site
of the ectopic foci or the electrical isolation of the
pulmonary veins from the atrium results in a marked
reduction in spontaneous atrial ectopy and the abolition of atrial fibrillation.20,80,81 The prevalence of
this mechanism of arrhythmia is unknown, but it may
be relatively common in young patients who have paroxysmal atrial fibrillation associated with a structurally normal heart and frequent atrial ectopic beats.
A novel approach to the treatment of drug-resistant atrial fibrillation is a hybrid of pharmacologic
and nonpharmacologic therapy.82 This approach is
suitable for patients in whom atrial fibrillation is transformed to atrial flutter after the initiation of drug
therapy, most commonly with a class IC antiarrhythmic agent or amiodarone. After it has been demonstrated that atrial flutter has become the sole rhythm,
radio-frequency ablation of the flutter frequently results in the maintenance of sinus rhythm, although
antiarrhythmic therapy must be continued in order to
prevent reemergence of the fibrillation.
The Maze Procedure
In 1987, Cox and colleagues introduced a surgical
procedure that they called the maze procedure,18 in
which the atrial appendages are excised and the pulmonary veins isolated. With the use of additional,
carefully placed incisions, a narrow, tortuous path of
atrial tissue is created that directs the sinus-node impulses across the atria to the atrioventricular node.
The incisions are placed so that no area is wide
enough to sustain multiple reentry circuits, and thus
atrial fibrillation cannot occur. Several dead-end “alleyways” create a maze-like pathway and permit the
depolarization of all the atrial tissue. As an isolated
technique for the treatment of atrial fibrillation, the
maze procedure has the limitation of requiring cardiopulmonary bypass. However, it has been used successfully in conjunction with other cardiac operations,
particularly mitral-valve surgery.83 The most recent
modification of the maze procedure uses minimally
invasive surgery and cryoablation, resulting in fewer
atriotomy procedures,84 and a preliminary report of
an experimental study suggests the possibility that
the procedure can be performed in the beating heart
without the need for cardiopulmonary bypass.85
Attempts have been made to duplicate the effects
of the surgical maze procedure with the creation by
radio-frequency energy of lesions in the atria — the
so-called “catheter maze.”86 This procedure is time
consuming and is associated with a risk of serious
complications.87 Attempts to modify and shorten the
procedure by limiting lesions to the right atrium have
been reported. However, initial results suggest a high
recurrence rate,88 and it is unlikely that right-atrial
lesions alone will prevent the recurrence of the arrhythmia. At present, the catheter maze procedure
should be considered experimental.
Pacemaker Therapy
Several novel pacing techniques are being investigated for the prevention of paroxysmal atrial fibrillation. These are based on the concept that inhomogeneous or delayed interatrial or intraatrial conduction
times predispose persons to the development of arrhythmia. Both dual-site atrial pacing (high in the
right atrium and at the coronary-sinus ostium) and
biatrial pacing (high in the right atrium and in the
mid- or proximal coronary sinus) reduce the duration
of the P wave and result in a more homogeneous atrial
depolarization. Data from uncontrolled trials in patients with atrial fibrillation that is refractory to drugs
suggest a benefit of dual-site pacing over single-site
pacing for the prevention of paroxysmal atrial fibrillation.89,90 Among such patients, pharmacologic therapy usually has to be combined with pacing for optimal results, and pacing at 80 to 90 beats per minute
is usually required.
Implantable Atrial Defibrillators
The success of the implantable ventricular defibrillator led to the concept of a device that would terminate
atrial fibrillation by means of an internal shock.91-93
The first such devices were designed for atrial defibrillation only, but they are no longer being manufactured. A new model of implantable atrial defibrillator
combines the option of atrial defibrillation with the
capacity for ventricular defibrillation.94 This device
permits the termination of atrial arrhythmias in patients with coexisting paroxysmal atrial fibrillation and
ventricular tachycardia, and it can act as a safety device
in the very rare event that an atrial shock precipitates
ventricular fibrillation. In addition to the capability of
delivering an atrial shock, the new device offers the
option of applying antitachycardia pacing and burst
atrial pacing in a tiered fashion, as programmed by
the physician. Since atrial fibrillation is rarely associated with sudden hemodynamic instability, the atrial
defibrillator can be activated by the patient, rather
than automatically delivering a shock to the patient.
This feature allows the patient to avoid unexpected,
painful shocks and permits patients who prefer to have
sedation before a shock to seek medical help.95
Despite the vast number of patients with atrial fi-
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1075
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
brillation, there is still uncertainty about the appropriate role of implantable atrial defibrillators. Clinical
experience remains limited, but the device appears
to be safe,96,97 and it is probable that the indications
for its use will expand as experience with it grows.
However, unless a ventricular defibrillator is also required for coexisting ventricular arrhythmias, it is unlikely that the atrial defibrillator will be used in more
than a very small proportion of patients — those with
highly symptomatic paroxysmal atrial fibrillation that
has proved resistant to other therapies.
CONCLUSIONS
The past decade has witnessed extraordinary growth
in all fields of knowledge regarding atrial fibrillation.
There is little doubt that pharmacologic therapy will
remain a mainstay of treatment, although it is likely
that the results of the AFFIRM trial,9 when they become available, will modify practice in some fashion.
There is also little doubt that nonpharmacologic therapy will have an increasing role in the treatment of
highly symptomatic patients with atrial fibrillation that
is refractory to drug therapy. The precise direction that
these therapies will take is intimately connected to
ongoing investigations of the electrophysiologic and
molecular changes that cause, and are produced by,
this arrhythmia.98 The next decade promises to be an
exciting one, in which we may finally overcome the
challenge of atrial fibrillation — so aptly termed
“the last big hurdle in treating supraventricular tachycardia.”99
REFERENCES
1. Brugada R, Tapscott T, Czernuszewicz GZ, et al. Identification of a genetic locus for familial atrial fibrillation. N Engl J Med 1997;336:905-11.
2. Benjamin EJ, Wolf PA, D’Agostino RB, Silbershatz H, Kannel WB,
Levy D. Impact of atrial fibrillation on the risk of death: the Framingham
Heart Study. Circulation 1998;98:946-52.
3. Dries DL, Exner DV, Gersh BJ, Domanski MJ, Waclawiw MA, Stevenson LW. Atrial fibrillation is associated with an increased risk for mortality
and heart failure progression in patients with asymptomatic and symptomatic left ventricular systolic dysfunction: a retrospective analysis of the
SOLVD trials. J Am Coll Cardiol 1998;32:695-703.
4. Chen YJ, Chen SA, Tai CT, et al. Role of atrial electrophysiology and
autonomic nervous system in patients with supraventricular tachycardia and
paroxysmal atrial fibrillation. J Am Coll Cardiol 1998;32:732-8.
5. Frustaci A, Chimenti C, Bellocci F, Morgante E, Russo MA, Maseri A.
Histological substrate of atrial biopsies in patients with lone atrial fibrillation. Circulation 1997;96:1180-4.
6. Brand FN, Abbott RD, Kannel WB, Wolf PA. Characteristics and prognosis of lone atrial fibrillation: 30-year follow-up in the Framingham Study.
JAMA 1985;254:3449-53.
7. Kopecky SL, Gersh BJ, McGoon MD, et al. Lone atrial fibrillation in
elderly persons: a marker for cardiovascular risk. Arch Intern Med 1999;
159:1118-22.
8. Jouven X, Desnos M, Guerot C, Ducimetiere P. Idiopathic atrial fibrillation as a risk factor for mortality: the Paris Prospective Study I. Eur Heart
J 1999;20:896-9.
9. Wyse DG, Anderson JL, Antman EM, et al. Atrial Fibrillation Followup Investigation of Rhythm Management — the AFFIRM study design.
Am J Cardiol 1997;79:1198-202.
10. Falk RH. Etiology and complications of atrial fibrillation: insights
from pathology studies. Am J Cardiol 1998;82:10N-17N.
11. Coumel P. Autonomic influences in atrial tachyarrhythmias. J Cardiovasc Electrophysiol 1996;7:999-1007.
12. Satoh T, Zipes DP. Unequal atrial stretch in dogs increases dispersion
of refractoriness conducive to developing atrial fibrillation. J Cardiovasc
Electrophysiol 1996;7:833-42.
13. Ommen SR, Odell JA, Stanton MS. Atrial arrhythmias after cardiothoracic surgery. N Engl J Med 1997;336:1429-34. [Erratum, N Engl J Med
1997;337:209.]
14. Moe GK. On the multiple wavelet hypothesis of atrial fibrillation.
Arch Int Pharmacodyn 1962;140:183-8.
15. Allessie MA, Lammers WJEP, Bonke FIM, Hollen J. Experimental
evaluation of Moe’s multiple wavelet hypothesis of atrial fibrillation. In:
Zipes DP, Jalife J, eds. Cardiac electrophysiology and arrhythmias. Orlando,
Fla.: Grune & Stratton, 1985:265-75.
16. Konings KTS, Kirchhof CJ, Smeets JR, Wellens HJJ, Penn OC, Allessie MA. High-density mapping of electrically induced atrial fibrillation
in humans. Circulation 1994;89:1665-80.
17. Rensma PL, Allessie MA, Lammers WJ, Bonke FI, Schalij MJ. Length
of excitation wave and susceptibility to reentrant atrial arrhythmias in normal conscious dogs. Circ Res 1988;62:395-410.
18. Cox JL, Schuessler RB, D’Agostino HJ Jr, et al. The surgical treatment
of atrial fibrillation. III. Development of a definitive surgical procedure.
J Thorac Cardiovasc Surg 1991;101:569-83.
19. Jais P, Haissaguerre M, Shah DC, et al. A focal source of atrial fibrillation treated by discrete radiofrequency ablation. Circulation 1997;95:
572-6.
20. Haïssaguerre M, Jaïs P, Shah DC, et al. Spontaneous initiation of atrial
fibrillation by ectopic beats originating in the pulmonary veins. N Engl J
Med 1998;339:659-66.
21. Wijffels MC, Kirchhof CJ, Dorland R, Allessie MA. Atrial fibrillation
begets atrial fibrillation: a study in awake chronically instrumented goats.
Circulation 1995;92:1954-68.
22. Hobbs WJC, Fynn S, Todd DM, Wolfson P, Galloway M, Garratt CJ.
Reversal of atrial electrical remodeling after cardioversion of persistent atrial fibrillation in humans. Circulation 2000;101:1145-51.
23. Zipes DP. Electrophysiological remodeling of the heart owing to rate.
Circulation 1997;95:1745-8.
24. Goette A, Honeycutt C, Langberg JJ. Electrical remodeling in atrial
fibrillation: time course and mechanisms. Circulation 1996;94:2968-74.
25. De Simone A, Stabile G, Vitale DF, et al. Pretreatment with verapamil
in patients with persistent or chronic atrial fibrillation who underwent electrical cardioversion. J Am Coll Cardiol 1999;34:810-4.
26. Tieleman RG, Van Gelder IC, Crijns HJ, et al. Early recurrences
of atrial fibrillation after electrical cardioversion: a result of fibrillationinduced electrical remodeling of the atria? J Am Coll Cardiol 1998;
31:167-73.
27. Manning WJ, Silverman DI, Katz SE, et al. Impaired left atrial mechanical function after cardioversion: relation to the duration of atrial fibrillation. J Am Coll Cardiol 1994;23:1535-40.
28. Manning WJ, Leeman DE, Gotch PJ, Come PC. Pulsed Doppler evaluation of atrial mechanical function after electrical cardioversion of atrial
fibrillation. J Am Coll Cardiol 1989;13:617-23.
29. Upshaw CB Jr. Hemodynamic changes after cardioversion of chronic
atrial fibrillation. Arch Intern Med 1997;157:1070-6.
30. Daoud EG, Weiss R, Bahu M, et al. Effect of an irregular ventricular
rhythm on cardiac output. Am J Cardiol 1996;78:1433-6.
31. Farshi R, Kistner D, Sarma JSM, Longmate JA, Singh BN. Ventricular
rate control in chronic atrial fibrillation during daily activity and programmed exercise: a crossover open-label study of five drug regimens.
J Am Coll Cardiol 1999;33:304-10.
32. Shinbane JS, Wood MA, Jensen DN, Ellenbogen KA, Fitzpatrick AP,
Scheinman MM. Tachycardia-induced cardiomyopathy: a review of animal
models and clinical studies. J Am Coll Cardiol 1997;29:709-15.
33. Kerr C, Boone J, Connolly S, et al. Follow-up of atrial fibrillation —
the initial experience of the Canadian Registry of Atrial Fibrillation. Eur
Heart J 1996;17:Suppl C:48-51.
34. Sabatini T, Frisoni GB, Barbisoni P, Bellelli G, Rozzini R, Trabucchi
M. Atrial fibrillation and cognitive disorders in older people. J Am Geriatr
Soc 2000;48:387-90.
35. Page RL, Wilkinson WE, Clair WK, McCarthy EA, Pritchett ELC.
Asymptomatic arrhythmias in patients with symptomatic paroxysmal atrial
fibrillation and paroxysmal supraventricular tachycardia. Circulation 1994;
89:224-7.
36. Jung W, Luderitz B. Quality of life in patients with atrial fibrillation.
J Cardiovasc Electrophysiol 1998;9:Suppl:S177-S186.
37. Hohnloser SH, Kuck KH, Lilienthal J. Rhythm or rate control in atrial
fibrillation — Pharmacological Intervention in Atrial Fibrillation (PIAF): a
randomised trial. Lancet 2000;356:1789-94.
38. Dorian P, Jung W, Newman D, et al. The impairment of health-related
quality of life in patients with intermittent atrial fibrillation: implications
for the assessment of investigational therapy. J Am Coll Cardiol 2000;36:
1303-9.
39. Wood MA, Brown-Mahoney C, Kay GN, Ellenbogen KA. Clinical
outcomes after ablation and pacing therapy for atrial fibrillation: a metaanalysis. Circulation 2000;101:1138-44.
1076 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
MED IC A L PROGR ES S
40. Friedman HZ, Weber-Bornstein N, Deboe SF, Mancini GB. Cardiac
care unit admission criteria for suspected acute myocardial infarction in
new-onset atrial fibrillation. Am J Cardiol 1987;59:866-9.
41. Falk RH. Proarrhythmia in patients treated for atrial fibrillation or flutter. Ann Intern Med 1992;117:141-50. [Erratum, Ann Intern Med 1992;
117:446.]
42. Intravenous digoxin in acute atrial fibrillation: results of a randomized,
placebo-controlled multicentre trial in 239 patients. Eur Heart J 1997;18:
649-54.
43. Furlanello F, Bertoldi A, Dallago M, et al. Atrial fibrillation in elite
athletes. J Cardiovasc Electrophysiol 1998;9:Suppl:S63-S68.
44. Roberts SA, Diaz C, Nolan PE, et al. Effectiveness and costs of digoxin
treatment for atrial fibrillation and flutter. Am J Cardiol 1993;72:567-73.
45. Danias PG, Caulfield TA, Weigner MJ, Silverman DI, Manning WJ.
Likelihood of spontaneous conversion of atrial fibrillation to sinus rhythm.
J Am Coll Cardiol 1998;31:588-92.
46. Reisinger J, Gatterer E, Heinze G, et al. Prospective comparison of flecainide versus sotalol for immediate cardioversion of atrial fibrillation. Am
J Cardiol 1998;81:1450-4.
47. Falk RH, Knowlton AA, Bernard SA, Gotlieb NE, Battinelli NJ. Digoxin for converting recent-onset atrial fibrillation to sinus rhythm: a randomized, double-blind trial. Ann Intern Med 1987;106:503-6.
48. Boriani G, Biffi M, Capucci A, et al. Conversion of recent-onset atrial
fibrillation to sinus rhythm: effects of different drug protocols. Pacing Clin
Electrophysiol 1998;21:2470-4.
49. Manning WJ, Silverman DI, Keighley CS, Oettgen P, Douglas PS.
Transesophageal echocardiographically facilitated early cardioversion from
atrial fibrillation using short-term anticoagulation: final results of a prospective 4.5-year study. J Am Coll Cardiol 1995;25:1354-61.
50. Van Gelder IC, Crijns HJ, Van Gilst WH, Verwer R, Lie KI. Prediction of uneventful cardioversion and maintenance of sinus rhythm from direct-current electrical cardioversion of chronic atrial fibrillation and flutter.
Am J Cardiol 1991;68:41-6.
51. Laupacis A, Boysen G, Connolly S, et al. Risk factors for stroke and
efficacy of antithrombotic therapy in atrial fibrillation: analysis of pooled
data from five randomized controlled trials. Arch Intern Med 1994;154:
1449-57. [Erratum, Arch Intern Med 1994;154:2254.]
52. Hart RG, Pearce LA, Rothbart RM, McAnulty JH, Asinger RW, Halperin JL. Stroke with intermittent atrial fibrillation: incidence and predictors during aspirin therapy. J Am Coll Cardiol 2000;35:183-7.
53. Cobbe SM, Rae AP, Poloniecki JD, et al. A randomized, placebo-controlled trial of propafenone in the prophylaxis of paroxysmal supraventricular tachycardia and paroxysmal atrial fibrillation. Circulation 1995;92:
2550-7.
54. Anderson JL, Gilbert EM, Alpert BL, et al. Prevention of symptomatic recurrences of paroxysmal atrial fibrillation in patients initially tolerating
antiarrhythmic therapy: a multicenter, double-blind, crossover study of flecainide and placebo with transtelephonic monitoring. Circulation 1989;80:
1557-70.
55. Lee SH, Chen SA, Tai CT, et al. Comparisons of oral propafenone
and sotalol as an initial treatment in patients with symptomatic paroxysmal
atrial fibrillation. Am J Cardiol 1997;79:905-8.
56. Juul-Möller S, Edvardsson N, Rehnqvist-Ahlberg N. Sotalol versus
quinidine for the maintenance of sinus rhythm after direct current conversion of atrial fibrillation. Circulation 1990;82:1932-9.
57. Singh S, Zoble RG, Yellen L, et al. Efficacy and safety of oral dofetilide
in converting to and maintaining sinus rhythm in patients with chronic
atrial fibrillation or atrial flutter: the Symptomatic Atrial Fibrillation Investigative Research on Dofetilide (SAFIRE-D) study. Circulation 2000;102:
2385-90.
58. Mittal S, Ayati S, Stein KM, et al. Transthoracic cardioversion of atrial
fibrillation: comparison of rectilinear biphasic versus damped sine wave
monophasic shocks. Circulation 2000;101:1282-7.
59. Oral H, Souza JJ, Michaud GF, et al. Facilitating transthoracic cardioversion of atrial fibrillation with ibutilide pretreatment. N Engl J Med
1999;340:1849-54.
60. Stambler BS, Wood MA, Ellenbogen KA, Perry KT, Wakefield LK,
VanderLugt JT. Efficacy and safety of repeated intravenous doses of ibutilide for rapid conversion of atrial flutter or fibrillation. Circulation 1996;
94:1613-21.
61. Schmit C, Alt E, Plewan A, et al. Low energy intracardiac cardioversion after failed conventional external cardioversion of atrial fibrillation.
J Am Coll Cardiol 1996;28:994-9.
62. Roy D, Talajic M, Dorian P, et al. Amiodarone to prevent recurrence
of atrial fibrillation. N Engl J Med 2000;342:913-20.
63. Singh SN, Fletcher RD, Fisher SG, et al. Amiodarone in patients with
congestive heart failure and asymptomatic ventricular arrhythmia. N Engl
J Med 1995;333:77-82.
64. Torp-Pedersen C, Møller M, Bloch-Thomsen PE, et al. Dofetilide in
patients with congestive heart failure and left ventricular dysfunction.
N Engl J Med 1999;341:857-65.
65. McBride R, Chesebro JH, Wiebers DO, et al. Warfarin versus aspirin
for prevention of thromboembolism in atrial fibrillation: Stroke Prevention
in Atrial Fibrillation II Study. Lancet 1994;343:687-91.
66. Chesebro JH, Wiebers DO, Holland AE, et al. Bleeding during antithrombotic therapy in patients with atrial fibrillation. Arch Intern Med
1996;156:409-16.
67. Gullov AL, Koefoed BG, Petersen P. Bleeding during warfarin and aspirin therapy in patients with atrial fibrillation: the AFASAK 2 study. Arch
Intern Med 1999;159:1322-8.
68. Hylek EM, Skates SJ, Sheehan MA, Singer DE. An analysis of the lowest effective intensity of prophylactic anticoagulation for patients with nonrheumatic atrial fibrillation. N Engl J Med 1996;335:540-6.
69. Blackshear JL, Baker VS, Rubino F, et al. Adjusted-dose warfarin versus low-intensity, fixed-dose warfarin plus aspirin for high-risk patients
with atrial fibrillation: Stroke Prevention in Atrial Fibrillation III randomised clinical trial. Lancet 1996;348:633-8.
70. The Stroke Prevention in Atrial Fibrillation Investigators. Stroke Prevention in Atrial Fibrillation Study: final results. Circulation 1991;84:527-39.
71. EAFT (European Atrial Fibrillation Trial) Study Group. Secondary
prevention in non-rheumatic atrial fibrillation after transient ischaemic attack or minor stroke. Lancet 1993;342:1255-62.
72. Petersen P, Boysen G, Godtfredsen J, Andersen ED, Andersen B. Placebo-controlled, randomised trial of warfarin and aspirin for prevention of
thromboembolic complications in chronic atrial fibrillation: the Copenhagen AFASAK study. Lancet 1989;1:175-9.
73. Anderson DC, Halperin JL, Hart RG, et al. Patients with nonvalvular
atrial fibrillation at low risk of stroke during treatment with aspirin: Stroke
Prevention in Atrial Fibrillation III Study. JAMA 1998;279:1273-7.
74. Laupacis A, Albers G, Dalen JE, Dunn MI, Jacobson AK, Singer DE.
Antithrombotic therapy in atrial fibrillation. Chest 1998;114:Suppl:579S589S.
75. Roth A, Harrison E, Mitani G, Cohen J, Rahimtoola SH, Elkayam U.
Efficacy and safety of medium- and high-dose diltiazem alone and in combination with digoxin for control of heart rate at rest and during exercise
in patients with chronic atrial fibrillation. Circulation 1986;73:316-24.
76. Panidis IP, Morganroth J, Baessler C. Effectiveness and safety of oral
verapamil to control exercise-induced tachycardia in patients with atrial fibrillation receiving digitalis. Am J Cardiol 1983;52:1197-201.
77. Fitzpatrick AP, Kourouyan HD, Siu A, et al. Quality of life and outcomes after radiofrequency His-bundle catheter ablation and permanent
pacemaker implantation: impact of treatment in paroxysmal and established
atrial fibrillation. Am Heart J 1996;131:499-507.
78. Rodriguez LM, Smeets JL, Xie BY, et al. Improvement in left ventricular function by ablation of atrioventricular nodal conduction in selected
patients with lone atrial fibrillation. Am J Cardiol 1993;72:1137-41.
79. Redfield MM, Kay GN, Jenkins LS, Mianulli M, Jensen DN, Ellenbogen KA. Tachycardia-related cardiomyopathy: a common cause of ventricular dysfunction in patients with atrial fibrillation referred for atrioventricular ablation. Mayo Clin Proc 2000;75:790-5.
80. Haissaguerre M, Jais P, Shah DC, et al. Electrophysiological end point
for catheter ablation of atrial fibrillation initiated from multiple pulmonary
venous foci. Circulation 2000;101:1409-17.
81. Pappone C, Rosanio S, Oreto G, et al. Circumferential radiofrequency
ablation of pulmonary vein ostia: a new anatomic approach for curing atrial
fibrillation. Circulation 2000;102:2619-28.
82. Huang DT, Monahan KM, Zimetbaum P, Papageorgiou P, Epstein
LM, Josephson ME. Hybrid pharmacologic and ablative therapy: a novel
and effective approach for the management of atrial fibrillation. J Cardiovasc Electrophysiol 1998;9:462-9.
83. Izumoto H, Kawazoe K, Eishi K, Kamata J. Medium-term results after the modified Cox/Maze procedure combined with other cardiac surgery. Eur J Cardiothorac Surg 2000;17:25-9.
84. Cox JL. Minimally-invasive Maze procedure. Circulation 1999;100:
Suppl I:I-778. abstract.
85. Lee R, Nitta T, Schuessler RB, Johnson DC, Boineau JP, Cox JL. The
closed heart MAZE: a nonbypass surgical technique. Ann Thorac Surg
1999;67:1696-702.
86. Pappone C, Oreto G, Lamberti F, et al. Catheter ablation of paroxysmal atrial fibrillation using a 3D mapping system. Circulation 1999;100:
1203-8.
87. Zhou L, Keane D, Reed G, Ruskin J. Thromboembolic complications
of cardiac radiofrequency catheter ablation: a review of the reported incidence, pathogenesis and current research directions. J Cardiovasc Electrophysiol 1999;10:611-20.
88. Natale A, Leonelli F, Beheiry S, et al. Catheter ablation approach on
the right side only for paroxysmal atrial fibrillation therapy: long-term results. Pacing Clin Electrophysiol 2000;23:224-33.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1077
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
89. Saksena S, Delfaut P, Prakash A, Kaushik RR, Krol RB. Multisite electrode pacing for prevention of atrial fibrillation. J Cardiovasc Electrophysiol 1998;9:Suppl:S155-S162.
90. Saksena S, Prakash A, Hill M, et al. Prevention of recurrent atrial fibrillation with chronic dual-site right atrial pacing. J Am Coll Cardiol
1996;28:687-94.
91. Tse HF, Lau CP, Yomtov BM, Ayers GM. Implantable atrial defibrillator with a single-pass dual-electrode lead. J Am Coll Cardiol 1999;33:
1974-80.
92. Wellens HJJ, Lau CP, Luderitz B, et al. Atrioverter: an implantable device for the treatment of atrial fibrillation. Circulation 1998;98:1651-6.
93. Tse HF, Lau CP, Yu CM, et al. Effect of the implantable atrial defibrillator on the natural history of atrial fibrillation. J Cardiovasc Electrophysiol 1999;10:1200-9.
94. Swerdlow CD, Sch≤ls W, Dijkman B, et al. Detection of atrial fibrilla-
tion and flutter by a dual-chamber implantable cardioverter-defibrillator.
Circulation 2000;101:878-85.
95. Timmermans C, Nabar A, Rodriguez LM, Ayers G, Wellens HJJ. Use
of sedation during cardioversion with the implantable atrial defibrillator.
Circulation 1999;100:1499-501.
96. Jung W, Wolpert C, Esmailzadeh B, et al. Clinical experience with implantable atrial and combined atrioventricular defibrillators. J Interv Card
Electrophysiol 2000;4:Suppl 1:185-95.
97. Timmermans C, Levy S, Ayers GM, et al. Spontaneous episodes of
atrial fibrillation after implantation of the Metrix Atrioverter: observations
on treated and nontreated episodes. J Am Coll Cardiol 2000;35:1428-33.
98. Nattel S. Electrophysiologic remodeling: are ion channels static players
or dynamic movers? J Cardiovasc Electrophysiol 1999;10:1553-6.
99. Wellens HJJ. Atrial fibrillation — the last big hurdle in treating supraventricular tachycardia. N Engl J Med 1994;331:944-5.
Copyright © 2001 Massachusetts Medical Society.
1078 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
CLINIC A L PROBLEM-SOLV ING
Clinical Problem-Solving
L ESS I S M ORE
C. CHRISTOPHER SMITH, M.D., JESS MANDEL, M.D.,
AND BOOKER BUSH, M.D.
A 63-year-old woman presented to her physician with fatigue, weakness, anorexia, and an unintentional weight loss of 14.5 kg (32 lb) during
the previous four months. She reported increased
constipation in the past year, but not melena,
hematochezia, or a change in stool caliber. She
had intermittent nausea, but not vomiting, dysphagia, distention, or postprandial bloating. She
reported no fevers, chills, chest pain, shortness of
breath, cough, abdominal pain, dysuria, hematuria, tremor, heat intolerance, or risk factors for infection with the human immunodeficiency virus
(HIV). She had a history of hypertension and
type 2 diabetes mellitus (her glycosylated hemoglobin value had been 11 percent at a recent visit).
She had had a positive skin test for tuberculosis
with purified protein derivative (PPD) 17 years
earlier, for which she had not received preventive
therapy. She had emigrated from Trinidad 30 years
earlier and was a retired nurse’s aide. Her medications were aspirin, furosemide, losartan, and
isophane insulin suspension.
On examination, the patient had a blood pressure of 170/99 mm Hg, a pulse of 70, and an
oral temperature of 37°C (98.6°F). She appeared
thin. Her sclerae were anicteric. There was no lymphadenopathy. Rectal examination showed brown
stool, and a test for occult blood was negative.
The patient had a slight, right-sided facial droop
and a diminished ability to raise her right eyebrow.
The neurologic findings and two well-healed facial scars were due to an earlier injury and were
unchanged from past examinations. The remainder of the examination was normal.
AN INTERNIST: There is a long list of potential
causes of unintentional weight loss. Often, however,
a specific diagnosis cannot be made.1 Cancer is the
greatest concern. Another consideration is chronic
infection. The patient has a history of a positive PPD
skin test, but she has no apparent history of fever or
From the Beth Israel Deaconess Medical Center, 1 Autumn St., Boston,
MA 02115, where reprint requests should be directed to Dr. Smith.
Based on a presentation at the Beth Israel Deaconess Medical Center
Morbidity and Mortality Conference, Boston, June 6, 2000.
cough, making tuberculosis a less likely diagnosis. The
initial manifestations of HIV infection can be weight
loss and muscle wasting, but she reports no risk factors for the infection.
Endocrine causes of weight loss include hyperthyroidism, hypoadrenalism, and diabetes. The findings
of hyperthyroidism can be subtle in the elderly, and
the thyrotropin level should be checked. It would be
unusual for diabetes to account for such a large
weight loss.
Psychiatric disease may also underlie unintentional
weight loss. The patient should be questioned about
symptoms of depression. In a younger patient with unexplained weight loss, an eating disorder should be
considered. Alcoholism and illicit drug use are also
possibilities.
The patient has vague gastrointestinal symptoms.
In addition to cancer, other gastrointestinal causes of
weight loss include malabsorption and, occasionally,
chronic diverticulitis or peptic ulcer disease. However,
her history is not suggestive of these conditions.
At this point, I would obtain a thorough history
and perform a physical examination. I would also order age-appropriate cancer screening if the patient has
not undergone such tests, a complete blood count,
liver-function tests, measurement of electrolyte and
thyrotropin levels, and chest radiography.
Three tests for fecal occult blood had been
performed within the previous six months, and
all had been negative. The results of a breast examination, mammography, Pap smear, and pelvic
examination had also been normal.
The patient’s hematocrit was 41.3 percent
(mean corpuscular volume, 85 µm3), and the
white-cell count was 6700 per cubic millimeter,
with a normal differential count. The alanine aminotransferase level was 70 U per liter, the aspartate aminotransferase level was 64 U per liter, the
alkaline phosphatase level was 544 U per liter,
and the g-glutamyltransferase level was 133 U per
liter. The serum levels of bilirubin, albumin, thyrotropin, sodium, potassium, calcium, creatinine, and urea nitrogen were normal. The serum
cholesterol level was 303 mg per deciliter. Urinalysis revealed a protein level of more than 300
mg per deciliter and a glucose level of 500 mg
per deciliter.
A HEPATOLOGIST: The patient has a markedly elevated alkaline phosphatase level with a mild elevation
in aminotransferase levels consistent with the presence
of either intrahepatic or extrahepatic cholestasis. Ex-
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1079
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
trahepatic cholestasis could be due to a biliary stricture
or cancer of the pancreas or bile duct. Intrahepatic
causes of cholestasis include primary biliary cirrhosis
and granulomas from either tuberculosis or sarcoidosis. If there is an incomplete obstruction of the extrahepatic biliary system, the bilirubin level will not be
elevated; thus, the normal bilirubin level does not distinguish intrahepatic from extrahepatic cholestasis. In
addition, the magnitude of the increase in serum alkaline phosphatase does not help differentiate intrahepatic from extrahepatic cholestasis. The patient’s
elevated cholesterol level is also consistent with the
presence of cholestasis and is most likely predominantly of the high-density lipoprotein fraction.2
Ultrasonography and computed tomography of
the abdomen showed no abnormalities. A chest
radiograph showed moderate left ventricular enlargement but no infiltrates or lymphadenopathy.
Serologic tests were negative for hepatitis B and
C viruses but positive for hepatitis A virus IgG.
A test for antimitochondrial antibody was negative; serum and urine protein electrophoresis, performed to help rule out autoimmune hepatitis,
showed no abnormalities.
A HEPATOLOGIST: Next I would perform a liver
biopsy to look for hepatic granulomas.
A consultation with a gastroenterologist was obtained. A colonoscopy revealed no tumor or other
clinically significant abnormalities. An esophagogastroduodenoscopy revealed adjacent ulcerated
antral masses of 2 cm and 1 cm. Pathological
analysis showed mild active gastritis and the presence of Helicobacter pylori.
A HEPATOLOGIST: I would still be concerned
about the lesions. There is a dictum in medicine that
states, “the absence of evidence is not evidence of absence.” Thus, if an endoscopist is concerned that these
masses may be neoplastic, further evaluation is indicated even if the initial pathological analysis shows
only H. pylori infection and gastritis. This patient
could still have a gastric cancer, gastric lymphoma, or
gastric mucosa–associated lymphoid tissue lymphoma (MALToma) — all of which are associated with
H. pylori infection. I would still recommend a liver
biopsy, since conditions such as sarcoidosis, tuberculosis, and syphilis can present as an antral mass.
One week later the patient returned to her
physician’s office to initiate therapy for H. pylori
infection. She reported the onset of left-sided facial weakness and drooling. Examination revealed
facial asymmetry, with marked loss of the left nasolabial fold. The patient was unable to raise her
left eyebrow or to close her left eye. The remain-
der of her examination was unchanged. Left-sided
Bell’s palsy was diagnosed. Glucocorticoids were
considered but were not prescribed owing to her
poorly controlled diabetes.
Within a week, the patient returned because of
double vision and redness of her left eye. The leftsided facial weakness was still present. Movement
of her eye did not cause pain, and she had not
had any fever, chills, or headache. Her pupils were
equal and reactive to light. The conjunctiva of the
left eye was injected medially. She was not able
to adduct or move her left eye superiorly.
A NEUROLOGIST: The patient’s drooling and inability to elevate her left eyebrow indicate a lesion of
the lower seventh cranial nerve. Bell’s palsy most commonly is idiopathic. However, given this history,
I would want to be certain there is nothing infiltrating
or compressing the left facial nerve. I would be reluctant to administer glucocorticoids, acyclovir, or
both, because I doubt that this is a case of idiopathic
Bell’s palsy.
The patient’s inability to adduct or move her left eye
superiorly appears to represent a third nerve palsy
with pupillary sparing. Although this problem could
represent a case of mononeuritis multiplex resulting
from poorly controlled diabetes, carcinomatous meningitis or an infiltrative process that affects the cranial
nerves should also be considered. Magnetic resonance
imaging (MRI) with gadolinium enhancement and
a lumbar puncture should be performed.
AN OPHTHALMOLOGIST: The findings of conjunctivitis and iritis are usually diffuse, present in all the
sectors of the eye. Redness in only one sector raises
the possibility of a more focal inflammatory disease
such as scleritis or episcleritis. What appears to be
conjunctival injection might represent an epibulbar
mass that is an extension of an intraorbital process.
I would recommend a full ophthalmologic examination and orbital imaging with MRI.
A neurologist examined the patient. Diabetic
mononeuritis multiplex was considered the most
likely diagnosis. Imaging studies were deferred.
Three weeks later, the patient reported improved appetite and a weight gain of 1 kg (2 lb).
Despite this improvement, she continued to have
profound generalized weakness and to require assistance with most activities. Her left-sided facial-nerve weakness and conjunctival injection had
improved, but she still could not move her left
eye medially or superiorly. She had new swelling
around her left eye, which was thought to be
consistent with the presence of lacrimal-duct obstruction. Two new round, firm, nontender submandibular lymph nodes of 1.5 cm were noted.
MRI showed a soft-tissue lesion of 1.8 by 2.2
cm in the superolateral aspect of the left orbit, in-
1080 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
CLINICA L PROBLEM-SOLV ING
filtrating both the superior rectus muscle and the
surrounding posterior aspect of the globe (Fig. 1).
There was also proptosis of the left globe. The cavernous sinus and leptomeninges were normal.
AN OPHTHALMOLOGIST: The proptosis of the left
globe indicates that there is substantial downward
displacement of the globe by the mass. This mass
seems to be molding itself to the available space, as
is typical of lymphoid tumors in the orbit or of a
granulomatous disorder such as sarcoidosis.
iary cirrhosis, Crohn’s disease, Hodgkin’s disease, and
non-Hodgkin’s lymphoma are just a few of the other
potential causes of noncaseating granulomas on liver
biopsy.3 In general, confidence in the diagnosis of
sarcoidosis requires a compatible clinical and radiographic picture, the finding of noncaseating granulomas in involved tissue, and the exclusion of other
A biopsy of the submandibular lymph node was
performed, and cultures were obtained. Cytologic
analysis revealed numerous granulomas. Smears
for acid-fast bacilli were negative.
A PULMONOLOGIST: The presentation is consistent
with but not typical of sarcoidosis. Patients generally
present with sarcoidosis in their 20s, although a second peak has also been reported among women over
the age of 50 years.3 Approximately half of the patients present with asymptomatic bilateral hilar lymphadenopathy; patients with symptoms most commonly report cough, dyspnea, fatigue, fever, or weight
loss. This patient apparently had no lung disease, although the lung is involved in more than 90 percent
of patients. Hepatic involvement, extrathoracic lymphadenopathy, and a retro-ocular mass of granulomatous inflammatory tissue have all been reported.
Sarcoidosis, however, would not rank high in the differential diagnosis of any of these findings in isolation. Nonetheless, it should be considered more seriously in a patient with multiorgan involvement. Even
when the pattern of involvement is not classic, as was
true in this case, sarcoidosis is a possibility, since atypical presentations are not uncommon.
Figure 1. MRI Scans of the Orbits.
A soft-tissue lesion measuring 1.8 by 2.2 cm infiltrates the superior rectus muscle and the surrounding posterior aspect of
the left globe in the axial projection.
Because of the atypical, multiorgan presentation and the biochemical abnormalities in liver
function, a needle biopsy of the liver was performed. The biopsy revealed that some of the portal tracts were expanded by non-necrotizing granulomas with multinucleated giant cells (Fig. 2).
Special stains for acid-fast bacilli and fungi were
negative. No polarizable material was seen. The
angiotensin-converting enzyme level was 113 U
per liter (normal range, 8 to 52). Given the clinical history, the findings were thought to be most
consistent with a diagnosis of sarcoidosis.
A PULMONOLOGIST: The diagnosis of sarcoidosis
is problematic because, with the possible exception
of the now abandoned Kveim test, there is no goldstandard test. The presence of noncaseating granulomas is consistent with but not specific for sarcoidosis.
Infections (such as those due to mycobacteria, fungi,
toxoplasma, and brucella), drug reactions, primary bil-
Figure 2. Liver-Biopsy Specimen Showing Non-Necrotizing Granulomas Containing Multinucleated Giant Cells (Arrows) (Hematoxylin and Eosin, ¬100).
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1081
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
causes of those granulomas. In an unusual case such
as this, I would consider sarcoidosis the most likely
diagnosis, but I would continue to follow the patient closely for any evidence of another cause of
granulomatous disease.
The value of measuring the angiotensin-converting–enzyme level is questionable. Its sensitivity and
specificity preclude its use as a meaningful diagnostic test.4 Some believe that the finding that the angiotensin-converting–enzyme level is two to three
times the upper limit of normal is helpful; however,
even in a patient with a classic presentation of sarcoidosis, this finding is not always present. This test
is most useful as a marker of disease activity in patients with established sarcoidosis.
Glucocorticoid therapy was recommended because of the neuro-ophthalmologic involvement.
However, the patient reported subjective improvement in her symptoms and declined treatment.
Over the next several weeks her appetite improved,
her vision returned to normal, the palsy of her
seventh cranial nerve resolved, and her alkaline
phosphatase level decreased to 198 U per liter.
Final results of cultures for tuberculosis were
negative.
COMMENTARY
Most patients with sarcoidosis have a spontaneous
remission.3 When sarcoidosis primarily involves the
lungs, the standard criterion for treatment is the development of progressive pulmonary disease; the inflammatory component of sarcoidosis is treated in an
attempt to limit the development of irreversible fibrosis. Other indications for treatment include hypercalcemia and neurologic, cardiac, or ophthalmologic disease. Hepatic sarcoidosis rarely requires treatment.5
Since this patient’s condition was improving without
treatment, it was reasonable to withhold therapy and
follow her closely. Patients with asymptomatic hilar
adenopathy or the acute onset of classic symptoms of
sarcoidosis have a high rate of spontaneous remission.
Other patients, such as this one, should be monitored
serially for evidence of new symptoms or progression
of their existing disease.
We are indebted to Drs. Daniel Sullivan, Sanjiv Chopra, Penny
Greenstein, Mark Kuperwaser, Paula Pinkston, Anthony Harton,
Harvey Goldman, and J.B. McGee for their assistance.
REFERENCES
1. Rabinovitz M, Pitlik SD, Leifer M, Garty M, Rosenfeld JB. Unintentional weight loss: a retrospective analysis of 154 cases. Arch Intern Med
1986;146:186-7.
2. Propst A, Propst T, Lechleitner M, et al. Hypercholesterolemia in primary biliary cirrhosis is no risk factor for atherosclerosis. Dig Dis Sci 1993;
38:379-80.
3. Statement on sarcoidosis: joint statement of the American Thoracic Society (ATS), the European Respiratory Society (ERS) and the World Association of Sarcoidosis and Other Granulomatous Disorders (WASOG)
adopted by the ATS Board of Directors and by the ERS Executive
Committee, February 1999. Am J Respir Crit Care Med 1999;160:73655.
4. Studdy PR, Bird R. Serum angiotensin converting enzyme in sarcoidosis — its value in present clinical practice. Ann Clin Biochem 1989;26:13-8.
5. Newman LS, Rose CS, Maier LA. Sarcoidosis. N Engl J Med 1997;336:
1224-34. [Erratum, N Engl J Med 1997;337:139.]
Copyright © 2001 Massachusetts Medical Society.
FULL TEXT OF ALL JOURNAL ARTICLES ON THE WORLD WIDE WEB
Access to the complete text of the Journal on the Internet is free to all subscribers. To
use this Web site, subscribers should go to the Journal’ s home page (www.nejm.org)
and register by entering their names and subscriber numbers as they appear on their
mailing labels. After this one-time registration, subscribers can use their passwords to
log on for electronic access to the entire Journal from any computer that is connected
to the Internet. Features include a library of all issues since January 1993, a full-text
search capacity, a personal archive for saving articles and search results of interest, and
free software for downloading articles so they can be printed in a format that is
virtually identical to that of the typeset pages.
1082 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
The New England Journal of Medicine -- April 5, 2001 -- Vol. 344, No. 14
Targeting the BCR-ABL Tyrosine Kinase in Chronic
Myeloid Leukemia
Chronic myeloid leukemia (CML) is one of the most remarkable cancers. (1,2) It was
probably the first type of leukemia to be recognized, in the 1840s, as a distinct entity. A
landmark was the discovery of the Philadelphia (Ph) chromosome in 1960. (3) This led to
the identification in CML cells of the BCR-ABL fusion gene and its corresponding protein,
which is now firmly established as the unique cause of the initial, or "chronic," phase of
CML. ABL and BCR, which are located on chromosomes 9 and 22, respectively, are normal
genes whose function is still unknown. The ABL gene encodes a tyrosine kinase whose
activity is tightly regulated. Both genes are truncated in the formation of the t(9;22)
reciprocal translocation that characterizes CML cells, and two fusion genes are generated:
BCR-ABL on the derivative 22q- chromosome (the Ph chromosome) and ABL-BCR on
chromosome 9q+. The BCR-ABL gene encodes a 210-kd protein with deregulated tyrosine
kinase activity. The presence of this protein in the leukemia cells of almost every patient
with CML is strong evidence of its pathogenetic role. This notion is further strengthened by
the report that insertion of the BCR-ABL gene into murine stem cells induces a
leukemia-like disease in mice. (4) The two reports by Druker and colleagues in this issue of
the Journal, (5,6) which document the clinical efficacy of the BCR-ABL tyrosine kinase
inhibitor STI571 in CML, must be accepted as final proof that the BCR-ABL oncoprotein is
the unique cause of chronic-phase CML.
Although this much is clear, numerous questions remain. The only factor known to
predispose people to CML is ionizing radiation. For most patients, no predisposing factors
are identified, and the cause of the chromosomal translocation is obscure. Similarly, we
understand little of the mechanisms of the inevitable progression from the chronic phase to
the ultimately fatal acute phase or blast crisis. The clinical heterogeneity of the disease also
remains unexplained. With standard therapy, the median survival is about 6 years, but some
patients die within a year of diagnosis and others survive for 20 or more years. In some
patients, the disease starts with an aggressive chronic phase, whereas in others it is indolent.
Wherein lies the difference? Not, apparently, in the BCR-ABL oncoprotein, which seems
remarkably consistent from patient to patient. Also, does the BCR-ABL tyrosine kinase
actually cause CML? Logically, it must activate one or more signal-transduction pathways
that lead to the malignant phenotype, but no pathway has been definitely incriminated. (7)
The treatment of CML evolved slowly at first, but the pace has quickened in the past 20
years. Busulfan yielded to hydroxyurea and hydroxyurea to interferon alfa. Until very
recently, all agreed that the best medical treatment for patients with newly diagnosed CML
was interferon alfa, either alone or in conjunction with cytarabine. The relatively few young
patients who have HLA-matched donors can be offered allogeneic stem-cell transplantation,
which is associated with appreciable procedure-related morbidity and mortality but remains
the only approach that can unequivocally eradicate the leukemia. Much, however, has now
changed with the advent of STI571.
In the early 1990s, knowing that the constitutively activated kinase had a major role in
CML, Druker approached scientists at what was then Ciba-Geigy with a proposal to find a
small molecule that would inhibit the BCR-ABL tyrosine kinase activity. This collaboration
yielded a phenylaminopyrimidine molecule, then called CGP 57148B, which occupies the
kinase pocket of the BCR-ABL protein and blocks access to ATP, thereby preventing
phosphorylation of any substrate (8) (Figure 1). Preclinical studies showed that the molecule
was highly effective in blocking the tyrosine kinase activity of ABL; the stem-cell factor
receptor, c-kit; and the platelet-derived growth factor receptor (PDGFR) but had little effect
on other tyrosine kinases. CGP 57148B inhibited proliferation of CML cell lines and
clonogenic cells from patients with CML but did not affect equivalent control cells. (9)
Beginning in 1998, these encouraging results led to clinical studies, from which arose the
trials by Druker et al., involving 83 patients with CML in the chronic phase5 and 58 patients
with CML in blast crisis or Ph-chromosome-positive acute lymphoblastic leukemia (ALL).
(6) The reports of these trials document the impressive capacity of the compound, now
renamed STI571 (for signal-transduction inhibitor), to reverse very rapidly the clinical and
hematologic abnormalities of CML in the chronic phase and, in many cases, to reduce to
zero or to low levels the proportion of Ph-chromosome-positive cells in the bone marrow.
The toxicity profile seems very mild for a drug with this degree of potency.
STI571 has considerable advantages over interferon alfa. It can be given by mouth, whereas
interferon alfa must be injected; hematologic responses are more rapid and probably more
frequent; the rate of cytogenetic response is clearly higher than with interferon alfa; and it
has fewer adverse effects. However, follow-up of patients given the drug is still very short,
and there is no direct evidence that STI571 prolongs life. Important clinical problems could
still emerge. These points provide the rationale for the ongoing prospective comparison of
STI571 with the combination of interferon alfa plus cytarabine. As the story unfolds,
STI571 may well become the single best agent for treating CML. It is expected to be
licensed for use in the United States this fall and in other countries soon thereafter.
The mechanism by which chronic-phase CML cells become resistant to conventional
cytotoxic drugs, such as hydroxyurea, is unknown. Disease progression probably involves
the occurrence of additional molecular changes in the Ph-chromosome-positive clone that
block cellular maturation. If this model is valid, it is surprising that STI571 has efficacy in
the blast crisis of CML. (6) There are two possible explanations: either the BCR-ABL
tyrosine kinase is necessary for blastic transformation, or other signal transduction pathways
that are activated during blastic transformation can also be inhibited by STI571. The former
possibility is supported by the observation that, at least in the laboratory, cell lines resistant
to STI571 overexpress BCR-ABL and may show amplification of the fusion gene.
(10,11,12)
If STI571 inhibits the BCR-ABL kinase so effectively, what is the result of its inhibition of
the ABL kinase in normal cells? So far, the effect appears to be minimal. Presumably, the
ABL protein, like the product of many other so-called housekeeping genes, is not essential
in normal signaling pathways. Does it matter whether STI571 inhibits normal c-kit or
PDGFR? Apparently not. We now know that some gastrointestinal stromal tumors have an
oncogenic mutation in the proto-oncogene c-kit. Thus, the anti-c-kit effect could be the basis
of the clinical efficacy of STI571 in the tumor that responded so impressively, as reported
by Joensuu et al. in a case report in this issue of the Journal. (13) Will STI571 be useful in
the treatment of other cancers? Time will tell, since clinical trials have now started for
patients with lung, prostate, and cerebral tumors.
How does one advise a patient with CML today? For a relatively young patient, for whom a
cure is the chief objective, hematopoietic stem-cell transplantation must remain a serious
option until it becomes clear that STI571 (or the combination of STI571 with other agents)
can cure a high proportion of patients or at least prolong life more than interferon alfa. If a
cure is less important, especially if the estimated transplant-related mortality exceeds 15 to
20 percent, then initial treatment with STI571 seems logical. The time has not yet come to
delay or rule out transplantation in patients with newly diagnosed disease who are
candidates for this procedure. If, however, STI571 is shown to reduce progression to blast
crisis and to obliterate all molecular evidence of residual leukemia, the role of
transplantation in CML will have to be reevaluated.
John M. Goldman, D.M.
Junia V. Melo, M.D., Ph.D.
Imperial College School of Medicine
London W12 0NN, United Kingdom
Managed Care in Transition
Managed care now dominates health care in the United States. By 1999, only 8 percent of
persons with employer-sponsored health insurance coverage had traditional indemnity
insurance. (1) This reflects a sea change in the past two decades -- not just in the financing
of health insurance but also in the way medicine is practiced.
The rapid growth of managed care is not primarily due to enthusiasm for this approach on
the part of patients or providers. Patients have had mixed reactions to managed care; they
like the low copayments and reduced paperwork but view some managed-care practices as
emphasizing cost control over quality. In fact, there is widespread concern among the
public, physicians, and legislators about the effect of managed care on the quality of care. In
this article, we examine how managed care has made such gains, despite the concern about
its effects on quality, and how it may change in the years ahead.
Evolution of Managed Care
The earliest forms of managed-care organizations were group- or staff-model health
maintenance organizations (HMOs), such as Kaiser-Permanente and Group Health
Cooperative of Puget Sound. As costs rose in the 1960s and 1970s, policymakers and
employers alike began to consider prepayment as an alternative to the fee-for-service system
of payment. Under the HMO rubric, prepayment was extended to include not only groupand staff-model organizations but also more loosely organized individual-practice
associations.
Initially, insurance companies could not control costs as well as HMOs could. With
indemnity insurance, the patient chose a physician from among all those in the community,
received care and a bill, and then submitted the bill to the insurer. Insurers had to pay fees
that were "usual and customary" in each community, which meant that the local physicians
set the fees.
In 1982, California passed legislation permitting insurers to establish contracts with selected
providers. With this new option, (2) insurers could exclude physicians who did not accept
their rules and fee schedules. With their extensive experience in negotiating contracts,
insurers began to build networks of individual physicians, called preferred-provider
organizations. These managed-care organizations were gradually expanded into statewide or
even national networks of contracted physicians.
As managed care grew and as more employers provided only managed-care coverage for
their employees, some patients complained about the restricted choice of providers. This led
to broader networks of preferred providers and to the development of point-of-service plans,
which include some coverage for the services of providers who are not part of the network.
Table 1 shows the increase from 1996 to 1999 in enrollment in point-of-service plans and
preferred-provider organizations, at the expense of indemnity plans, among persons with
employer-sponsored insurance. (1)
Insurers have recently begun to offer "multitiered" plans. Enrollment in a multitiered plan
gives the patient three options: full coverage in an HMO with a limited number of providers;
access to a preferred-provider organization, with slightly higher copayments than those for
the HMO; and use of out-of-network providers, with the highest copayments. By combining
features of HMOs, preferred-provider organizations, and point-of-service plans, this
approach offers a choice between lower copayments for restricted access to providers and
higher payments for greater access but allows the patient to make that trade-off at the time
of illness rather than during the annual enrollment period. However, to the extent that
financial barriers prevent low-income persons from choosing out-of-network providers,
multitiered plans may in effect create different classes of coverage for persons with the same
employer but different salaries.
Enrollment in managed-care plans grew because managed care cost less than fee-for-service
care. Managed-care business practices such as preauthorization of hospital and other
services and restricted formularies for medications reduced utilization and cost. As a result
of these strategies, managed-care organizations also had healthier enrollees than did
fee-for-service plans, because patients switching from fee-for-service care to managed care
tended to be healthier than average, and some patients in managed-care plans returned to
fee-for-service care when they got sick. (3) This "risk selection" increased the cost
advantage of managed care.
Some policymakers supported managed care because they believed that prepayment would
improve the quality of care. Fee-for-service payment leads to uncoordinated care, they
argued, whereas prepayment for all inpatient and outpatient services allows physicians to
allocate resources optimally for each patient. Supporters of managed care also argued that
the use of preventive services would increase and that the quality of care for acute and
chronic conditions would improve as plans sought to avoid costly complications. Despite
this rationale for the superior performance of managed-care organizations, most studies have
found little difference in quality between managed care and fee-for-service care.
Deficiencies in quality occur in both types of health plans. (4) In addition, purchasers have
rarely chosen health plans on the basis of the quality of care. Assuming that accreditation of
plans and licensure of providers are sufficient to ensure high quality, they have chosen plans
primarily on the basis of price. (5)
Effects of Managed Care
Effect on Providers
The growth of managed care has profoundly influenced physicians' practices. In 1996, the
American Medical Association began tracking the proportion of revenues physicians
received from managed care. That year, managed care accounted for 38.4 percent of
revenues received by all physicians (6); by 1999, the proportion was 48.9 percent, and
almost all physicians had managed-care contracts (Table 2). (7) About one third of
physicians had capitation contracts. Although capitation accounted for only 7.4 percent of
practice revenues among all physicians in the United States in 1999, it accounted for 21.0
percent of revenues among physicians with capitation contracts. Capitation represented a
higher proportion of practice revenues in primary care specialties (ranging from 12.2 percent
of revenues received by all physicians in general internal medicine to 16.4 percent in
pediatrics) and in certain regions of the country (16.8 percent of revenues received by all
physicians in Pennsylvania and 14.4 percent in California). (7)
Capitation contracts specify that providers will be prepaid on a per capita basis for an
agreed-on list of services (e.g., primary care only, all physicians' services, or hospitalization
only) or for all services and drug costs (global capitation). This approach shifts the financial
risk from the insurer to the provider. Theoretically, capitation could improve the quality of
care if providers focused on prevention and on early diagnosis and treatment and if
prepayment resulted in better coordination of care rather than episodic provision of services.
However, such improvement is based on four assumptions: that the capitation rate covers all
necessary medical care and capital expenses plus enough to cover the occasional case of
catastrophic illness, that providers have control over all aspects of care, that they expect to
keep patients for a long time, and that they have the managerial skills and infrastructure to
respond effectively to the incentives provided through capitation. Although the news
media's coverage of problems with capitation has focused on situations in which the
capitation rate has been too low to cover all services included in the contract (a violation of
the first assumption), there have been cases in which the other assumptions have not held.
Thus, most of the reported problems with capitation reflect problems with contracts and
with implementation and are probably not sufficient to reject capitation as an approach to
the financing of health care.
The relatively low percentage of revenues that physicians receive from capitation and the
slight decline in the percentage from 1996 to 1999 suggest that medical groups have not
been comfortable with the available capitation arrangements. Some plans have also
encountered difficulties in their capitation arrangements with hospitals. In one of the most
publicized retreats from hospital capitation, PacifiCare Health Systems decided to
restructure some of its contracts in California. In 1998, 91 percent of PacifiCare's California
enrollees were covered by plans that included capitation arrangements with hospitals. Under
these arrangements, enrollees chose the hospital where they would receive care, if
necessary, and the hospital received a fixed payment for each enrollee who chose it. In 1999
and 2000, one third of the hospitals withdrew from the capitation arrangements, citing
unbudgeted capital expenditures (e.g., retrofitting of buildings to provide protection against
earthquakes), rising operating expenses (e.g., nurses' salaries), or an inability to manage
utilization (since physicians, not hospitals, admit and discharge patients). In addition,
mergers gave some hospitals enough market power to refuse PacifiCare's capitation rates.
PacifiCare continues to have capitation contracts with most of the hospitals and has
responded to the rest by negotiating shared-risk contracts and by improving its information
and clinical-management systems.
Effect on Patients
Managed care has resulted in major changes for patients and their experience of care.
Measures adopted by managed-care organizations to control costs or improve the quality of
care, or both, include primary care gatekeeping, preauthorization of referrals, utilization
review, profiling of physicians (monitoring of their patterns of utilization or the quality of
their care), pharmaceutical restrictions, practice guidelines, case management, and most
recently, disease management. (There is no consensus on the definition of disease
management, but in general, it involves a multidisciplinary effort to minimize the burden of
disease by educating patients, encouraging them to take an active role in their care, and
establishing a long-term therapeutic plan.) Patients' reactions to these measures depend
primarily on whether they are perceived as attempts to limit expenditures or to ensure proper
care. Thus, gatekeeping is often not well received, because people rarely believe its purpose
is to maintain or improve the quality of care, (8) whereas disease-management programs
may be more acceptable.
The extent to which these measures influence actual care (as opposed to the perception of
care) remains uncertain. Attempts to evaluate managed-care practices are complicated by
variations in other organizational characteristics, the impossibility of conducting blinded,
randomized trials at the organizational level, and inconsistent definitions of each strategy.
Studies of the effect of these measures on cost have similar limitations, but there is some
evidence that certain measures have little effect. For example, a meta-analysis of studies of
physician profiling showed minimal, though statistically significant, reductions in cost. (9)
Experience in the private sector also suggests that some managed-care practices are
ineffective, and some insurers are dropping them, especially those perceived as cost-control
rather than quality-control measures. In response to patients' complaints about gatekeeping
and lack of evidence that it had changed referral patterns, Harvard Pilgrim Health Care, in
New England, discontinued its use of gatekeepers in 1997. UnitedHealthcare, a national
company, recently stopped requiring preauthorization of referrals, after discovering that
requested referrals were denied in less than 1 percent of cases. (10) However, with health
care costs rising again, insurers may reconsider cost-control measures.
Responses to Managed Care
Physicians and Medical Groups
With their purchasing power and negotiating experience, and aided by business and
government support for managed care, insurers obtained substantial concessions from
providers as managed care grew. These included discounts on fee-for-service rates in some
cases, acceptance of capitation arrangements in others, and tolerance of measures to control
utilization in almost every case.
One of the primary strategies physicians used to match the purchasing power of
managed-care organizations was to affiliate with each other. Table 3 shows the declining
percentage of physicians in solo practice from 1983 to 1999 and the rise in the percentage of
employed physicians. The change in practice type has been even more pronounced among
physicians in their first five years of practice: in 1983, approximately one third of such
physicians were employed, as compared with 66 percent in 1994. (11)
Some medical groups also formed larger systems of care. Physician-hospital organizations
were created either to accept global capitation or to ensure hospitals a steady stream of
admissions. However, differences in the cultures and goals of medical groups and hospitals
have limited the growth of these combined organizations. For example, primary care
physicians may join a physician-hospital organization to increase their ability to coordinate
care and prevent complications. This emphasis on primary care may conflict with the
hospital's goals if its reason for forming the organization is to keep the number of
admissions high. The most successful physician-hospital organizations tend to be nonprofit
organizations involving physicians and hospitals that are prominent in the community and
that had established working relationships before the organization was formed. (12)
In the 1990s, some medical groups sold their practices to or signed agreements with
practice-management companies. (12) This trend was initially greeted with enthusiasm by
Wall Street. At one point, two practice-management companies, MedPartners and PhyCor,
proposed a merger that involved 6 percent of all physicians in the United States. (13)
However, the growth of practice-management companies slowed substantially as they
discovered that buying physicians' practices was easier than changing the way they
delivered care. Investors eventually realized that the apparent growth in earnings for
MedPartners was due to acquisitions rather than to substantive practice management. (14)
Recently, whether as a result of desperation or confidence, an increasing number of medical
groups have rejected or terminated capitation contracts. In California, where many medical
groups are on the verge of bankruptcy, Sutter Health, a large network of medical groups and
hospitals based in Sacramento, canceled its contract with Blue Cross. (15) Aetna recently
had to drop its requirement that physicians who join its preferred-provider organization also
accept its HMO contracts. (16)
Employers
Employers have also turned to collective action. Regional coalitions of employers often
attempt to negotiate discounts on health care. In addition, many of these coalitions recognize
the need to address the problems of risk selection and the lack of incentives for health plans
to improve the quality of care.
Some managed-care organizations have used risk selection to reduce their costs, but
purchasers now recognize this strategy. The solution is to adjust the premium paid to
organizations that enroll patients who are sicker (or healthier) than average. Such an
adjustment ensures that managed-care organizations that do not practice risk selection (and
hence enroll sicker patients than plans that do practice risk selection) have the resources
they need to care for their enrollees and that the plans with healthy enrollees do not receive
unwarranted profits. However, risk adjustment requires both data and analytic tools in order
to set appropriate rates; the tools have been under development for two decades. Some
purchasers believe that adjustment of payments on the basis of diagnoses reported on claims
could be an effective approach. Under capitation, however, providers do not submit claims,
and it has taken time for some managed-care organizations to develop systems for collecting
diagnostic data. Since Medicare adopted a diagnosis-based system of adjustment in 2000,
however, the ability of organizations to provide these data has increased. (17) Buyers Health
Care Action Group in Minneapolis, for example, is adjusting payments to delivery systems
on the basis of diagnoses recorded at every inpatient or outpatient encounter.
Risk adjustment provides an indirect incentive to improve the quality of care. Some
employers are also creating direct incentives. The Pacific Business Group on Health
measures the quality of care on several dimensions (18) and uses public recognition
(through annual awards), patient volume, and financial bonuses to give plans and providers
incentives to improve quality. The Central Florida Health Care Coalition plans to rate the
quality of inpatient and outpatient care provided by physicians. The coalition will allow its
members to select any physician, but copayments will be lower if they select providers with
a high rating for quality of care. In addition, the coalition will make graded fee-for-service
payments, with the top-rated physicians receiving a higher payment for any given service
than lower-rated physicians. The Leapfrog Group is a national coalition of employers that
includes more than 65 Fortune 500 companies and other large employers, with a total of
more than 25 million covered employees. The group is trying to use its leverage to foster
regional improvement in the quality of care. An example is Leapfrog's involvement with the
Michigan Health and Safety Coalition, which includes hospitals, health plans, medical
groups, employers, unions, and government agencies. Leapfrog is collaborating with the
Michigan coalition to improve the safety of hospitalized patients and to inform patients and
providers about issues involving safety.
A parallel initiative by employers is the adoption of a defined-contribution approach, which
specifies the amount the employer will pay for an employee's coverage and requires that the
employee pay the difference for more expensive coverage. This approach is being used in
two different ways. Some employers use it to limit their costs and their involvement in
issues concerning the quality of care. Others view it as part of a strategy (along with
providing data on the quality of care) to make their employees aware of how plans differ
with respect to cost and quality. General Motors makes the largest contributions for the
plans it believes offer the highest-quality care. Currently, however, the growth of this
approach is limited by a tight labor market and the concern that employees will view defined
contributions as a reduction in benefits.
Government
State and federal legislators are responding to public concern about managed care. The
major areas of legislative activity are the benefits package (what should be covered), use of
emergency room services, the physician-patient relationship, resolution of disputes, and
liability (Table 4). (19,20)
Congress has also considered legislation. (20) Passage of the proposed Patient Protection
Act has been held up primarily by disagreements over issues involving the liability of health
plans -- that is, how easy it should be for patients to sue health plans when disputes arise,
and whether such suits should be brought in state or federal courts. There is strong public
support for a patients' bill of rights, but the prospects for meaningful legislation are unclear.
Managed-care organizations claim they should not be sued for malpractice, even when they
deny care. The basis of this claim is that they decide what services are covered but do not
make decisions about individual care and that their actions are, in many cases, exempt from
state regulation under the federal Employee Retirement Income Security Act. Recent
judicial decisions suggest that managed-care organizations, although still protected by the
act from claims that they must provide specific services (i.e., they cannot be challenged with
respect to the scope of the services they provide), can be held responsible under state laws
for the quality of the services that are delivered with their authorization. (21) In addition,
managed-care organizations in several states are being sued on the grounds of false
advertising. The contention is that although their advertisements state that decisions are
made by physicians, in some cases, plans deny coverage for treatments recommended by
physicians or require physicians to adhere to guidelines developed by the plans, not by local
providers.
The Future
Health care is not being transformed in isolation. The Internet allows patients to obtain
information easily, though many sources of data on the Internet are unreliable. Health care
organizations can, however, use the Internet to disseminate data to patients. For example,
the Medicare Web site has data on the quality of care provided by each managed-care
organization participating in the Medicare program
(http://www.medicare.gov/mphcompare/home.asp).
Advances in computer technology have also made the creation of electronic medical records
more feasible. The cost of an Internet-based system of medical records is falling, making it
possible to collect and audit data on quality and risk adjustment less expensively, although
the issue of privacy must be addressed. Thus, the cost of providing incentives for
managed-care plans to improve the quality of care is falling as well.
Conclusions
The health care market and managed care continue to evolve. There has been legitimate
concern about some managed-care practices, and until recently, few attempts had been made
to ensure that patients and the quality of care were protected. However, medical groups
seem to be assuming a stronger stance in their negotiations with managed-care
organizations, and employers, as well as federal and state governments, are becoming more
sophisticated in the use of measures to promote and reward high-quality care. Efforts are
also being made to prevent managed-care organizations from profiting by selecting healthy
patients. Some managed-care organizations are already responding to these changes by
eliminating administrative practices that restrict patients' choices and by establishing
disease-management programs and other measures that increase the coordination of care.
The future of managed care remains uncertain, however. If employers and federal and state
governments continue to emphasize the quality of care, and especially if medical groups and
medical societies support these efforts, physicians may be able to spend more time caring
for their patients and less time arguing with insurers. This might also help patients regain
confidence in a system they have lost trust in. An appropriately designed system of
prepayment -- with rates that are high enough to cover all appropriate care and the costs of
treating catastrophic illness, as well as incentives for providers to plan for the future -- could
facilitate the coordination of care, which has traditionally been fragmented. On the other
hand, medical costs are on the rise again, (1,22) and if the primary focus of the policy debate
returns to financial considerations, efforts to improve the quality of care may be postponed.
R. Adams Dudley, M.D., M.B.A.
Harold S. Luft, Ph.D.
University of California, San Francisco
San Francisco, CA 94118
We are indebted to Eunice Chee, Marchant Wentworth, and Richard Bae for their assistance
in the preparation of the manuscript.
Phenylpropanolamine and Hemorrhagic Stroke
To the Editor:
The request by the Food and Drug Administration (FDA) for the voluntary withdrawal of all
products containing phenylpropanolamine on the basis of the data of Kernan et al. (Dec. 21
issue) (1) is excessive. Although the study was rigorously designed and the data compelling
with respect to the risk of hemorrhagic stroke with this agent, it does not provide
justification for the withdrawal of all products containing phenylpropanolamine, particularly
decongestant preparations that have been used safely in prescription and nonprescription
products for many years. Kernan et al. report an increased risk of hemorrhagic stroke only in
association with the use of appetite suppressants that contained phenylpropanolamine. In our
experience, these products are ingested at higher-than-recommended doses that are often far
in excess of the doses used in decongestants. The study provided no evidence of an
increased risk associated with the amount of phenylpropanolamine contained in
decongestants. The main alternative, pseudoephedrine, is itself not without potential
problems, including its use in the synthesis of illicit methamphetamine and reports of
ischemic colitis. (2) A better approach from the FDA would be to ask for the removal of
appetite suppressants that contain phenylpropanolamine but not of cough and cold
preparations that contain this agent.
Michael E. Ernst, Pharm.D.
Arthur Hartz, M.D., Ph.D.
University of Iowa
Iowa City, IA 52242
References
1. Kernan WN, Viscoli CM, Brass LM, et al. Phenylpropanolamine and the risk of
hemorrhagic stroke. N Engl J Med 2000;343:1826-32.
2. Dowd J, Bailey D, Moussa K, Nair S, Doyle R, Culpepper-Morgan JA. Ischemic colitis
associated with pseudoephedrine: four cases. Am J Gastroenterol 1999;94:2430-4.
To the Editor:
As Paracelsus pointed out in the 16th century, "the right dose differentiates a poison from a
remedy." Would that he could have reviewed the article by Kernan et al. or advised the FDA
on how to respond to the results. Although we do not disagree with the FDA's decision to
remove phenylpropanolamine from the marketplace, several key lines of evidence have been
overlooked.
Surely Kernan et al., Fleming (who wrote the editorial accompanying their report), (1) and
the FDA do not believe that the phenylpropanolamine used in appetite suppressants is a
different molecular entity from the phenylpropanolamine used in decongestants. The
problem may lie not in the molecule, but in its abuse by women who are trying to lose
weight. Kernan et al. hypothesize that there is a sex-based difference in the effect of
phenylpropanolamine on the risk of hemorrhagic stroke; they miss the confounding variable:
difference between the sexes in the pattern of use. A 1983 study (2) of adverse effects after
the ingestion of sympathomimetic weight-control products sold over the counter found that
the average dose of phenylpropanolamine was 575 mg in women. A subsequent review (3)
of 142 instances in which adverse effects were attributed to phenylpropanolamine found that
45 percent of patients had taken an overdose of the drug. Patients who take
phenylpropanolamine as a weight-loss aid commonly abuse the drug by taking it in
higher-than-recommended doses. Therefore, it is the pattern of use and not the drug itself
that poses the danger. Poison-control centers provide pharmacologic vigilance, and both
Kernan et al. and the FDA would have benefited by obtaining the response of this
community to the data.
William R. Wolowich, Pharm.D.
Children's Hospital
Columbus, OH 43205
Marcel J. Casavant, M.D.
Central Ohio Poison Center
Columbus, OH 43205
Brent R. Ekins, Pharm.D.
University of California, San Francisco
San Francisco, CA 94143
References
1. Fleming GA. The FDA, regulation, and the risk of stroke. N Engl J Med
2000;343:1886-7.
2. Ekins BR, Spoerke DG. An estimation of the toxicity of non-prescription diet aids from
seventy exposure cases. Vet Hum Toxicol 1983;25:81-5.
3. Lake CR, Gallant S, Masson E, Miller P. Adverse drug effects attributed to
phenylpropanolamine: a review of 142 case reports. Am J Med 1990;89:195-208.
The authors reply:
To the Editor:
Drs. Ernst and Hartz write that our study reports an increased risk of hemorrhagic stroke
only in association with the use of appetite suppressants containing phenylpropanolamine
and not with the use of decongestants that contain this agent. We disagree. Among women
who were 18 to 49 years of age, the first use of any product containing
phenylpropanolamine was associated with an increased risk of hemorrhagic stroke (odds
ratio, 3.13; P=0.08). All first uses involved decongestants. Although this odds ratio did not
reach conventional criteria for statistical significance (P<0.05), this criterion itself may be
too stringent for evaluating potentially harmful associations.
Drs. Ernst and Hartz further argue that the main alternative to phenylpropanolamine,
pseudoephedrine, is "not without potential problems." We examined the association between
pseudoephedrine and the risk of hemorrhagic stroke in our subjects. Among 702 patients, 9
(1.3 percent) were exposed to pseudoephedrine as a first use (defined as use within 24 hours
before the stroke and no other use during the preceding two weeks), as compared with 18 of
1376 control subjects (1.3 percent), yielding an adjusted odds ratio of 1.07 (95 percent
confidence interval, 0.45 to 2.57; P=0.87). This point estimate for the adjusted odds ratio
suggests that the use of pseudoephedrine is not associated with an increased risk of
hemorrhagic stroke. As indicated by the upper bound of the confidence interval, however,
we cannot exclude the possibility that it has a potentially harmful effect.
Drs. Ernst and Hartz and Dr. Wolowich and colleagues all argue that the stronger
association between the use of appetite suppressants containing phenylpropanolamine and
the risk of stroke than between the use of phenylpropanolamine-containing decongestants
and the risk of stroke may be explained by differences in the dose. We believe that the dose
may provide a partial explanation. The mean amount of phenylpropanolamine in appetite
suppressants consumed over a period of three days was 300 mg (range, 75 to 600), as
compared with 203 mg (range, 13 to 890) of phenylpropanolamine in cough or cold
remedies. As we reported in our paper, higher doses of phenylpropanolamine were
associated with higher odds ratios for stroke.
Walter N. Kernan, M.D.
Catherine M. Viscoli, Ph.D.
Lawrence M. Brass, M.D.
Ralph I. Horwitz, M.D.
Yale University School of Medicine
New Haven, CT 06520-8025
The editorialist replies:
To the Editor:
I accept the hypothesis of Wolowich et al. that women who abuse phenylpropanolamine in
an attempt to lose weight represent a pattern of use that would explain -- but only in part -the risk of stroke reported by Kernan et al. Wolowich et al. cite a 1983 report from a
poison-control center of the adverse effects of phenylpropanolamine that not surprisingly
found that most of the reported cases were in women who took very large doses of
phenylpropanolamine. (1) This report helps characterize acute toxicity in the context of the
abuse of phenylpropanolamine. More important, Kernan et al. establish that stroke is also
associated with the use of phenylpropanolamine in the recommended way. The 1990 report
cited by Wolowich et al. in fact establishes, and the study by Kernan et al. in effect
confirms, that more than half the cases in that series of toxic reactions to
phenylpropanolamine were associated with "non-overdose amounts." (2) No one is surprised
when the abuse of a drug leads to adverse effects. The study by Kernan et al. establishes that
women who apparently took phenylpropanolamine as directed were much more likely to
have a stroke than matched control subjects who had not taken phenylpropanolamine. These
findings, without need for further distinctions, are a sufficient basis for the conclusion that
the risk-benefit ratios for the use of phenylpropanolamine as an antiobesity treatment and
probably as a cough suppressant are unacceptable.
G. Alexander Fleming, M.D.
Ingenix Pharmaceutical Services
Chevy Chase, MD 20815
References
1. Ekins BR, Spoerke DG. An estimation of the toxicity of non-prescription diet aids from
seventy exposure cases. Vet Hum Toxicol 1983;25:81-5.
2. Lake CR, Gallant S, Masson E, Miller P. Adverse drug effects attributed to
phenylpropanolamine: a review of 142 case reports. Am J Med 1990;89:195-208.
Dietary Supplements Containing Ephedra Alkaloids
To the Editor:
As an anatomical pathologist and paid consultant to the Ephedra Education Council, I
reviewed 22 reports of adverse events received by the Food and Drug Administration (FDA)
in which death had occurred and assessed the likelihood that death was related to the use of
ephedrine-type alkaloids. My review, reported August 8, 2000, at the Department of Health
and Human Services's Public Meeting on the Safety of Dietary Supplements Containing
Ephedrine Alkaloids, in Washington, D.C., showed no consistent clinical or pathological
features of the reported adverse events and showed that ephedrine-type alkaloids were not
likely to have been causative or contributing factors in the deaths. (1)
The report by Haller and Benowitz (Dec. 21 issue) (2) included eight of the cases I had
reviewed and interpreted these adverse events as related to the use of ephedrine-type
alkaloids. Table 4 of the report by Haller and Benowitz lists adverse events that were
definitely or probably related to the use of ephedrine-type alkaloids, but in the column
labeled "preexisting conditions or concurrent risks," the authors have omitted the following
data: Patient 4 had chest pain, Patient 5 hypertension, and Patient 7 severe coronary artery
disease. Table 5, which lists events possibly related to the use of ephedrine-type alkaloids,
omits the fact that Patients 2 and 6 collapsed during extreme exercise and fasting for rapid
weight loss. In addition, Patient 7 did not have an adverse event; her premature infant died
from necrotizing enterocolitis. An autopsy in Patient 9 demonstrated the anomalous origin
of the left coronary artery from the pulmonary trunk, a well-known cause of sudden death.
With an adequate explanation of the reported adverse events, the implication of
ephedrine-type alkaloids in deaths from a wide variety of conditions that occur in the
general population is no more than idle speculation.
Grover M. Hutchins, M.D.
1 Stratford Rd.
Baltimore, MD 21218
References
1. The National Women's Health Information Center. Public Meeting on the Safety of
Dietary Supplements Containing Ephedrine Alkaloids, Washington, D.C., August 8 and 9,
2000. (See http://www.4woman.gov/owh/public.)
2. Haller CA, Benowitz NL. Adverse cardiovascular and central nervous system events
associated with dietary supplements containing ephedra alkaloids. N Engl J Med
2000;343:1833-8.
To the Editor:
We wish to report on a previously healthy, 19-year-old male bodybuilder who had a
myocardial infarction after using ephedra.
The patient reported chest pain of 30 minutes' duration that had begun shortly after the use
of Dymetadrine Xtreme. He dissolved the recommended dose (two tablets, each reported to
contain 24 mg of ephedra alkaloids and 100 mg of caffeine (1)) in water and drank the
solution, as he had done in the past. Severe chest pain, radiating to the left arm, developed
15 minutes later. He had no history of chest pain or cardiac disease and reported that he had
no other cardiac risk factors, including cocaine use. Vital signs were as follows: pulse, 116
per minute and regular; blood pressure, 147/84 mm Hg; respirations, 22 per minute;
temperature, 37.3°C (99.1°F). The physical examination was otherwise unremarkable except
for diaphoresis.
The electrocardiogram showed evidence of an inferolateral myocardial infarction. The
patient was given oxygen, aspirin, heparin, and nitroglycerin, and his ST-segment elevation
resolved. Five hours later, inferolateral ST-segment elevation recurred. After treatment with
phentolamine and labetalol, the electrocardiographic findings again returned to normal. The
creatine kinase level was initially 351 U per liter, with an MB fraction of 23 ng per
milliliter; it peaked at 1271 U per liter, with an MB fraction of 104 ng per milliliter. The
value for troponin I peaked at 256.1 ng per milliliter. A toxicologic test of a urine specimen
was negative for cocaine. Echocardiography revealed hypokinesis of the inferior wall.
Cardiac catheterization demonstrated only minimal intimal disease of the distal left anterior
descending artery. The patient recovered and was doing well at follow-up.
The temporal association between the use of the supplement and the infarction, the absence
of clinically significant findings on cardiac catheterization, and the negative test for cocaine
led us to conclude that the myocardial infarction was caused by the use of ephedra.
This case highlights the potential dangers of ephedra use by presumably healthy persons.
Given the growing numbers of reports of ephedra-related adverse events, both the general
public and the medical community should be alerted to the dangers posed by
over-the-counter products containing ephedra. Furthermore, we urge greater regulation of
these potentially lethal products.
Stephen J. Traub, M.D.
New York City Poison Control Center
New York, NY 10016
Wissam Hoyek, M.D.
Staten Island University Hospital
Staten Island, NY 10305
Robert S. Hoffman, M.D.
New York City Poison Control Center
New York, NY 10016
References
1. AST sports science Web site. (See http://www.ast-ss.com/.)
The authors reply:
To the Editor:
Dr. Hutchins argues that because several of the patients who had severe cardiovascular
events while taking ephedra-alkaloid-containing dietary supplements had underlying
disease, it is idle speculation to implicate the dietary supplements as the cause of these
events. Before addressing the individual cases mentioned by Dr. Hutchins, we would like to
restate our point that ephedra-related events are uncommon but are most likely to occur in
vulnerable populations. Persons with underlying cardiovascular disease are an obviously
vulnerable population. The effects of ephedrine and caffeine -- constricting blood vessels,
increasing blood pressure, and releasing catecholamines -- would be most likely to cause
injury in persons with underlying cardiovascular disease. In such persons, ischemia,
infarction, or arrhythmias, or a combination of these events, could well be precipitated by
the sympathomimetic effects of ephedrine and caffeine. Most important, these cases
illustrate that people with unrecognized cardiovascular disease are using products that are
potentially hazardous to them.
In response to Dr. Hutchins's comments on individual cases, it should be noted that in some
cases the information provided to us by the FDA differed from that cited by Dr. Hutchins. In
any case, assuming that the information provided by Dr. Hutchins is correct, one should be
concerned about a 43-year-old man with chest pain (Patient 4 in Table 4 of our article) or a
patient with known hypertension (Patient 5 in Table 4) who is taking a supplement that
contains an ephedra alkaloid. These circumstances raise questions about the adequacy of
warnings about contraindications. The fact that two patients collapsed and died during
extreme exercise and dieting underscores another serious concern about the use of
ephedrine-containing dietary supplements, which are recommended for increased energy
and weight loss. On the basis of the known pharmacologic characteristics of ephedrine and
caffeine, these drugs might be likely to have more injurious effects in the context of intense
exercise. Likewise, a person with a congenital anomaly of a coronary artery might be more
likely to have ischemia in the presence of a sympathomimetic drug. With respect to Patient
7 in Table 5 of our article, the adverse event was fetal death, which was presumed to be due
to premature delivery, which in turn may have been induced by the consumption of
ephedrine-containing dietary supplements.
Thus, it is likely that unrecognized cardiovascular disease confers a predisposition to
adverse events associated with the use of ephedrine-containing dietary supplements. The
fact that several persons had underlying cardiovascular disease does not mean there were no
adverse reactions to these dietary supplements. Even if appropriate warning statements were
listed on the product labels, users with unrecognized risk factors could not be expected to
respond to such warnings. Until there is a way to identify persons who are at risk for adverse
effects, supplements containing ephedra alkaloids should be considered unreasonably
dangerous. Perhaps one solution is to perform coronary angiographic screening of all
patients before they take these products.
Finally, our report describes a series of cases in which the use of ephedrine-containing
dietary supplements was associated with adverse cardiovascular events. Our report does not
prove causation, nor does it provide quantitative information with regard to risk. A
large-scale case-control study similar to the Hemorrhagic Stroke Project for
phenylpropanolamine (1) is needed to determine the risks associated with these dietary
supplements.
Christine A. Haller, M.D.
Neal L. Benowitz, M.D.
University of California, San Francisco
San Francisco, CA 94143-1220
References
1. Kernan WN, Viscoli CM, Brass LM, et al. Phenylpropanolamine and the risk of
hemorrhagic stroke. N Engl J Med 2000;343:1826-32.
The Diagnosis and Treatment of Cough
To the Editor:
We share the concern of Irwin and Madison (Dec. 7 issue) (1) about the unnecessary
treatment of acute cough with antibiotics, which recent meta-analyses have shown has no
clinically significant benefit. (2,3,4) However, we disagree with the authors' terminology.
Irwin and Madison avoid the diagnosis of bronchitis for patients with cough and production
of phlegm, but there are compelling reasons to retain the term. "Acute bronchitis" is
common in the medical literature and is familiar to patients and physicians alike.
Unfortunately, the diagnosis of acute bronchitis has traditionally been used as justification
for the administration of antibiotics and has even been described as "a cough that gets
treated with antibiotics." Instead of jettisoning the term, we favor educating both physicians
and patients by informing them that antibiotics do not alter the course of acute bronchitis.
Several small studies have demonstrated the efficacy of (beta)-agonists for decreasing the
duration of acute cough. In one placebo-controlled trial, (5) 46 patients with acute bronchitis
were randomly assigned, in a two-by-two factorial design, to erythromycin or placebo and
albuterol or placebo. On day seven, 61 percent of the patients in the albuterol group were
still coughing, as compared with 91 percent in the placebo group (P=0.02). The result was
not influenced by the use of erythromycin.
Confirmatory studies are warranted, but the use of (beta)-agonists makes biologic sense.
Patients with acute bronchitis demonstrate obstruction on pulmonary-function testing. (6)
Albuterol allows physicians to prescribe effective therapy, satisfy patients, and avoid
unnecessary antibiotics.
Jeffrey A. Linder, M.D.
Randall S. Stafford, M.D., Ph.D.
Massachusetts General Hospital
Boston, MA 02114
References
1. Irwin RS, Madison JM. The diagnosis and treatment of cough. N Engl J Med
2000;343:1715-21.
2. Bent S, Saint S, Vittinghoff E, Grady D. Antibiotics in acute bronchitis: a meta-analysis.
Am J Med 1999;107:62-7.
3. Smucny JJ, Becker LA, Glazier RH, McIsaac W. Are antibiotics effective treatment for
acute bronchitis? A meta-analysis. J Fam Pract 1998;47:453-60.
4. Fahey T, Stocks N, Thomas T. Quantitative systematic review of randomised controlled
trials comparing antibiotic with placebo for acute cough in adults. BMJ 1998;316:906-10.
5. Hueston WJ. Albuterol delivered by metered-dose inhaler to treat acute bronchitis. J Fam
Pract 1994;39:437-40.
6. Williamson HA. Pulmonary function tests in acute bronchitis: evidence for reversible
airway obstruction. J Fam Pract 1987;25:251-6.
To the Editor:
There is increasing evidence that in many cases of acute cough the cause is not the common
cold or acute bronchitis. Two recent studies showed that acute bronchitis is often (in about
one third of cases) the first manifestation of asthma, which becomes full-blown in the next
several years. (1,2) In nearly half of the patients presenting with a cough of at least two
weeks' duration, there were signs of asthma or features of chronic obstructive pulmonary
disease such as bronchial hyperresponsiveness (as measured with a methacholine challenge).
Most patients could be classified correctly by history taking and physical examination only.
Female sex, prolonged expiration, smoking, reports of wheezing and dyspnea, and
symptoms elicited by allergens helped to predict the risk of asthma or chronic obstructive
pulmonary disease. (3) The physician must decide whether to proceed with further
examination, referral to a respiratory specialist, or initiation of treatment with inhaled
corticosteroids or bronchodilators.
Henk A. Thiadens, M.D., Ph.D.
Machiel P. Springer, M.D., Ph.D.
Leiden University Medical Center
Leiden 2301 CB, the Netherlands
Dirkje S. Postma, M.D., Ph.D.
University Hospital Groningen
Groningen 9700 AD, the Netherlands
References
1. Jonsson JS, Gislason T, Gislason D, Sigurdsson JA. Acute bronchitis and clinical
outcome three years later: prospective cohort study. BMJ 1998;317:1433-4.
2. Thiadens HA, Postma DS, De Bock GH, Huysman DAN, van Houwelingen HC, Springer
MP. Asthma in adult patients presenting with symptoms of acute bronchitis in general
practice. Scand J Prim Health Care 2000;18:188-92.
3. Thiadens HA, de Bock GH, Dekker FW, et al. Identifying asthma and chronic obstructive
pulmonary disease in patients with persistent cough presenting to general practitioners:
descriptive study. BMJ 1998;316:1286-90.
To the Editor:
Irwin and Madison provide excellent guidelines for the management of cough in primary
care but do not address the use of opioid antitussive medications. Among the many opioid
preparations, some contain as little as 2.5 mg of codeine phosphate per dose or as much as 1
mg of hydromorphone per dose. At equivalent doses, the analgesic potency of
hydromorphone is 130 times that of codeine. Preparations containing up to 5 mg of
hydrocodone (which is approximately four times as potent as codeine) per dose are also
available. For codeine, cough suppression has been shown to be dose related, and the
analgesic potency of an opioid medication may also have a bearing on its antitussive
properties.
It is not clear when and how to use opioid antitussive medications. In fact, in a study
involving an office-based population, (1) codeine, dextromethorphan, and guaifenesin were
all found to be equally effective. In addition, opioid agents have the potential for addiction.
Perhaps a case can be made for the use of high-potency opioid antitussive medications in
patients with lung cancer or those needing relief from a cough during radioimaging
procedures.
Ram Kakaiya, M.D.
Jennifer Wamhoff, Pharm.D.
University of Illinois College of Medicine at Rockford
Rockton, IL 61072-0319
References
1. Croughan-Minihane MS, Petitti DB, Rodnick JE, Eliaser G. Clinical trial examining
effectiveness of three cough syrups. J Am Board Fam Pract 1993;6:109-15.
The authors reply:
To the Editor:
We avoid using the term "acute bronchitis" for patients with a syndrome of acute cough and
phlegm, as noted by Linder and Stafford, because it is too often inaccurate. A diagnosis of
acute bronchitis implies that the patient has an acute, noneosinophilic inflammatory
condition of the lower airway, which is often not the case for those with acute cough and
phlegm. For example, the common cold presents as a syndrome of acute cough and phlegm
but is not associated with inflammation of the lower airway. In addition, patients with acute
cough in the setting of unrecognized asthma or chronic obstructive pulmonary disease are
also likely to have a misdiagnosis of acute bronchitis, as stressed by Thiadens et al. We use
the term "acute bronchitis" only when there is objective evidence of noneosinophilic
inflammation of the lower airway. (1) If one splits rather than lumps diseases that present as
a syndrome of acute cough and phlegm, patients with cough due to postnasal drip from an
acute infection of the upper respiratory tract are more likely to be effectively treated, (2) and
one can hope that there will be less overprescription of antibiotics and more appropriate use
of bronchodilators.
Although we agree with Thiadens et al. that patients with asthma and chronic obstructive
pulmonary disease often present to generalists with acute (or subacute) cough, we are
cautious in making a diagnosis of these conditions on the basis of clinical criteria or
physiological testing alone, because these methods can be unreliable. With respect to
asthma, a prospective study showed that only 54 percent of patients considered by
pulmonologists to have symptomatic clinical asthma actually had it. (2) In addition, the
results of a methacholine challenge can falsely predict that cough is due to asthma. (3) For
these reasons, the definitive diagnosis of symptomatic asthma requires an appropriate
clinical context (e.g., cough), compatible results of physiological testing (e.g., positive
results on methacholine challenge), a favorable response to specific therapy, and a clinical
course consistent with asthma during follow-up. (3)
As stressed in both our review and the evidence-based consensus panel report of the
American College of Chest Physicians, (4) there should be only a limited role for
nonspecific therapy, and only with agents shown to be efficacious, such as opioid antitussive
medications. Nonspecific antitussive therapy should be prescribed only when specific
therapy will not work rapidly enough or is unavailable (e.g., for inoperable lung cancer, as
suggested by Kakaiya and Wamhoff).
Richard S. Irwin, M.D.
J. Mark Madison, M.D.
University of Massachusetts Medical School
Worcester, MA 01655
References
1. Pizzichini E, Pizzichini MMM, Efthimiadis A, et al. Indices of airway inflammation in
induced sputum: reproducibility and validity of cell and fluid-phase measurements. Am J
Respir Crit Care Med 1996;154:308-17.
2. Pratter MR, Hingston DM, Irwin RS. Diagnosis of bronchial asthma by clinical
evaluation: an unreliable method. Chest 1983;84:42-7.
3. Irwin RS, French CT, Smyrnios NA, Curley FJ. Interpretation of positive results of a
methacholine inhalation challenge and 1 week of inhaled bronchodilator use in diagnosing
and treating cough-variant asthma. Arch Intern Med 1997;157:1981-7.
4. Irwin RS, Boulet L-P, Cloutier MM, et al. Managing cough as a defense mechanism and
as a symptom: a consensus panel report of the American College of Chest Physicians. Chest
1998;114:Suppl 2:133S-181S.
Syncope
To the Editor:
Kapoor's excellent review of syncope (Dec. 21 issue) (1) failed to mention arrhythmogenic
right ventricular dysplasia as a possible cause of syncope. Although rare, this condition
affects otherwise healthy young persons and can be fatal. Clinicians must have a high degree
of suspicion in order to diagnose this condition, since the physical examination is often
unremarkable. The classic electrocardiogram demonstrates inverted T waves in the right
precordial leads, but it may be read as normal. (2) Since this condition often only grossly
affects the right ventricle, especially early in the disease process, the diagnosis may be
missed on echocardiography. Subtle abnormalities of this chamber may be overlooked
unless the reader of the echocardiogram is alerted to the suspicion of right ventricular
dysplasia.
Since syncope in right ventricular dysplasia (and, for that matter, in patients with anomalous
coronary arteries) often occurs during exertion (3) in patients thought to have no structural
heart disease, we disagree with Kapoor's statement that "episodes associated with exercise in
athletes without heart disease are also examples of neurally mediated syncope." We believe
syncope during exertion in all patients deserves a thorough work-up, including
echocardiography with careful attention to the right ventricle and a provocative test for
arrhythmias such as an exercise stress test or an electrophysiologic study, before the
condition is attributed to neurocardiogenic syncope.
Ralph J. Verdino, M.D.
Francis E. Marchlinski, M.D.
University of Pennsylvania Hospitals
Philadelphia, PA 19104
References
1. Kapoor WN. Syncope. N Engl J Med 2000;343:1856-62.
2. Marcus FI, Fontaine GH, Guiraudon G, et al. Right ventricular dysplasia: a report of 24
adult cases. Circulation 1982;65:384-98.
3. Corrado D, Basso C, Thiene G, et al. Spectrum of clinicopathologic manifestations of
arrhythmogenic right ventricular cardiomyopathy/dysplasia: a multicenter study. J Am Coll
Cardiol 1997;30:1512-20.
To the Editor:
Dr. Kapoor correctly points out that syncope in elderly patients is particularly difficult to
evaluate. In fact, some elderly patients will deny having episodes of syncope and present
with only recurrent unexplained falls. In one study, (1) carotid-sinus massage was
performed on 132 consecutive patients older than 65 years of age who were referred to a
syncope clinic for the investigation of recurrent unexplained dizziness, falls, and syncope. A
total of 59 patients had persistent, reproducible carotid-sinus hypersensitivity, and 17 of
them denied having episodes of syncope and presented instead with dizziness or falls of
unknown origin; in 12 of these patients, loss of consciousness was witnessed during
carotid-sinus massage, and the patients subsequently had retrograde amnesia concerning the
event. Therefore, syncope should be included in the differential diagnosis of older patients
who present with unexplained falls.
Francisco Jose Fernandez-Fernandez, M.D.
Pascual Sesma, M.D.
Hospital Arquitecto Marcide
15405 Ferrol, Spain
References
1. McIntosh SJ, Lawson J, Kenny RA. Clinical characteristics of vasodepressor,
cardioinhibitory, and mixed carotid sinus syndrome in the elderly. Am J Med
1993;95:203-8.
Dr. Kapoor replies:
To the Editor:
The comment made by Drs. Verdino and Marchlinski regarding right ventricular dysplasia
entirely accords with the central point of my article that the most important factor in the
evaluation of syncope is determining whether or not the patient has structural heart disease. I
noted that episodes associated with exercise in athletes without heart disease are examples
of neurally mediated syncope. Right ventricular dysplasia is certainly heart disease and
would require the physician to consider arrhythmias and conduct an additional workup.
Concerning the detection of right ventricular dysplasia, I concur that a high index of
suspicion is necessary. Exertional syncope in any patient merits evaluation with
echocardiography and stress testing to discover any possible cardiac cause for this symptom.
In addition to right ventricular dysplasia, more common causes such as ischemia, valvular
heart disease, hypertrophic cardiomyopathy, and pulmonary hypertension also need to be
considered, as I noted in Table 2 of my article. The careful evaluation of patients with
exertional syncope is likely to uncover right ventricular dysplasia if it is present.
The comments of Drs. Fernandez-Fernandez and Sesma regarding falls in the elderly are
relevant to the evaluation of falls. I focused specifically on syncope. Regarding the
evaluation of falls in the elderly, I concur that brief arrhythmias must be considered, since
some patients may not recall losses of consciousness and may therefore present with falls
rather than with syncope.
Wishwa N. Kapoor, M.D.
University of Pittsburgh School of Medicine
Pittsburgh, PA 15213-2582
Invasive Pulmonary Aspergillosis Associated with
Infliximab Therapy
To the Editor:
A 25-year-old man with fistulizing Crohn's disease began to have high fever, dyspnea, and a
productive cough five days after he received a single intravenous dose of 5 mg of a
monoclonal antibody against tumor necrosis factor (alpha) (anti-TNF-(alpha) antibody;
infliximab, Centocor, Malvern, Pa.) per kilogram of body weight. His current medications
did not include corticosteroids or other immunosuppressive drugs. His chest roentgenogram
showed massive, bilateral infiltrates, and laboratory examination revealed an elevated
C-reactive protein value and leukocytosis. He received treatment with intravenous
broad-spectrum antibiotics. Within 24 hours after admission, respiratory insufficiency
developed. Ventilatory support and extracorporeal membrane oxygenation were needed to
achieve adequate oxygenation. On day 7, Aspergillus fumigatus was grown from a culture
of tracheal secretions but was considered clinically nonsignificant. Treatment with
high-dose corticosteroids was started because of adult respiratory distress syndrome. On day
13, cultures of tracheal secretions and bronchoalveolar-lavage fluid repeatedly yielded A.
fumigatus, and treatment with liposomal amphotericin B (3 mg per kilogram per day) was
started. The results of bacteriologic and virologic examinations remained negative. The
results of analyses for TNF-(alpha) and aspergillus antigen on serial plasma samples were
negative. A hemothorax developed. Samples obtained during a thoracotomy on day 22
showed growth of A. fumigatus, and results of antigen testing were positive. Despite the
intensive treatment, the patient died from multiorgan failure and septic shock on day 24.
Postmortem examination of lung tissue showed invasive growth of septated hyphae, and
cultures revealed A. fumigatus.
Systemic use of anti-TNF-(alpha) antibodies is not a known risk factor for invasive
aspergillosis. As far as we know, this is the first reported case of invasive aspergillosis
associated with this treatment. The patient had had no known risk factors associated with
invasive aspergillosis. He had not had neutropenia; he had not taken corticosteroids or other
immunosuppressive drugs during the three months before he received anti-TNF-(alpha)
treatment; and he had not had a prior influenza infection or any known exposures to
aspergillus. In addition, he had been a nonsmoker and had not used marijuana.
TNF-(alpha) is thought to have a central role in the immunopathology of inflammatory
bowel disease, and several trials have shown a clinical benefit of infliximab. (1,2,3)
Infliximab reduces intestinal inflammation by binding to and neutralizing TNF-(alpha) on
the cell membrane and by destroying TNF-(alpha)-producing cells in the blood. However,
TNF-(alpha) also has a central role in the recruitment of neutrophils into the lungs in
response to pathogens such as A. fumigatus. (4,5) In both normal and neutropenic mice,
antibody-mediated neutralization of TNF-(alpha) resulted in an increase in mortality after
intratracheal challenge with A. fumigatus. (5)
Serious infectious complications with the use of infliximab have been reported rarely,
although the incidence of upper respiratory tract infections with infliximab is higher than
that with placebo. (1)
The systemic use of monoclonal antibodies against TNF-(alpha) in patients with
inflammatory bowel disease increases the risk of serious opportunistic infections by
inhibiting an adequate TNF-(alpha) response. Practitioners should be aware of this risk
when treating patients with anti-TNF-(alpha) antibodies.
Adilia Warris, M.D.
University Medical Center St. Radboud
6500 HB Nijmegen, the Netherlands
Arvid Bjorneklett, M.D., Ph.D.
Peter Gaustad, M.D., Ph.D.
National Hospital of the University of Oslo
0027 Oslo, Norway
References
1. van Deventer SJH. Anti-TNF antibody treatment of Crohn's disease. Ann Rheum Dis
1999;58:Suppl I:I-114-I-120.
2. D'haens G, van Deventer S, van Hogezand R, et al. Endoscopic and histological healing
with infliximab anti-tumor necrosis factor antibodies in Crohn's disease: a European
multicenter trial. Gastroenterology 1999;116:1029-34.
3. Targan SR, Hanauer SB, van Deventer SJH, et al. A short-term study of chimeric
monoclonal antibody cA2 to tumor necrosis factor (alpha) for Crohn's disease. N Engl J
Med 1997;337:1029-35.
4. Schelenz S, Smith DA, Bancroft GJ. Cytokine and chemokine responses following
pulmonary challenge with Aspergillus fumigatus: obligatory role of TNF-alpha and
GM-CSF in neutrophil recruitment. Med Mycol 1999;37:183-94.
5. Mehrad B, Strieter RM, Standiford TJ. Role of TNF-(alpha) in pulmonary host defense in
murine invasive aspergillosis. J Immunol 1999;162:1633-40.
The above letter was referred to Centocor, the manufacturer of infliximab, which
offers the following reply:
To the Editor:
Fungal infections in patients treated with infliximab are rare. As of January 31, 2001, more
than 2000 patients have received infliximab in clinical trials. Fungal infections have
occurred in only two of these patients (<0.1 percent). One patient, described in a previous
report, (1) had disseminated coccidioidomycosis while receiving concurrent corticosteroids
and methotrexate with infliximab, and another patient, in a controlled trial that is currently
blinded, had cryptococcal pneumonia; it is not known whether this patient received
infliximab or placebo.
Since the approval of infliximab by the Food and Drug Administration for Crohn's disease
in 1998 and for rheumatoid arthritis in 1999, nearly 115,000 patients have received
infliximab. Post-marketing reports of documented systemic fungal infections through
January 31, 2001, reveal the following: six cases of aspergillosis, including the case
described by Warris et al. (three of which involved treatment with profound
immunosuppression for graft-versus-host disease before treatment with infliximab for that
disease), five cases of Pneumocystis carinii pneumonia, three cases of histoplasmosis (one
primary and two disseminated), one case of a coccidioidomycosis septic joint, three cases of
systemic candida infections, and five cases of systemic unspecified fungal infections
(several of the patients with candida or unspecified fungal infections had central venous
catheters, with or without total parenteral nutrition). In almost all of these cases, there was
concurrent immunosuppression including therapy with corticosteroids, azathioprine,
methotrexate, or mercaptopurine.
The benefit of infliximab in Crohn's disease and rheumatoid arthritis is well established.
Infliximab alters the course of these illnesses by producing clinical remission and healing of
fistulas in Crohn's disease (2,3) and by reducing signs and symptoms and preventing joint
damage in rheumatoid arthritis. (1) In addition, reports from several tertiary care centers
have shown that infliximab has a corticosteroid-sparing benefit in Crohn's disease, (4) which
may lessen the need for therapeutic immunosuppression with corticosteroids.
Increased risks of infection associated with immunosuppressive agents such as
corticosteroids, azathioprine, methotrexate, and mercaptopurine are recognized. Whether
blockade of TNF-(alpha) significantly increases the risk of opportunistic fungal infections
remains to be established. Finally, invasive pulmonary aspergillosis has been described even
in the absence of an immunocompromised state. (5) Practitioners need to consider the
benefits and risks of all the therapeutic options when developing a treatment plan for
patients with Crohn's disease or rheumatoid arthritis.
Gregory F. Keenan, M.D.
Thomas F. Schaible, Ph.D.
Jerome A. Boscia, M.D.
Centocor
Malvern, PA 19355
References
1. Lipsky PE, van der Heijde DMFM, St Clair EW, et al. Infliximab and methotrexate in the
treatment of rheumatoid arthritis. N Engl J Med 2000;343:1594-1602.
2. Targan SR, Hanauer SB, van Deventer SJH, et al. A short-term study of chimeric
monoclonal antibody cA2 to tumor necrosis factor (alpha) for Crohn's disease. N Engl J
Med 1997;337:1029-35.
3. Present DH, Rutgeerts P, Targan S, et al. Infliximab for the treatment of fistulas in
patients with Crohn's disease. N Engl J Med 1999;340:1398-1405.
4. Cohen RD, Tsang JF, Hanauer SB. Infliximab in Crohn's disease: first anniversary
clinical experience. Am J Gastroenterol 2000;95:3469-77.
5. Karam GH, Griffin FM. Invasive pulmonary aspergillosis in nonimmunocompromised,
nonneutropenic hosts. Rev Infect Dis 1986;8:357-63.
Decoding Darkness: The Search for the Genetic
Causes of Alzheimer's Disease
By Rudolph E. Tanzi and Ann B. Parson. 281 pp., illustrated.
Cambridge, Mass., Perseus, 2000. $26. ISBN 0-7382-0195-2
The writing in medical journals is deliberately devoid of emotion, which may explain why
writers of books about science are flourishing these days. They can add drama and elucidate
complexities. A scientific autobiography written with a journalist can be a winner, as
illustrated by Decoding Darkness, which could have been subtitled "Alzheimer's Research
Goes Molecular." Rudolph Tanzi is director of the Genetics and Aging Unit at
Massachusetts General Hospital. Ann Parson is a professional writer.
As autobiography, Decoding Darkness commands attention. In 1980, after having graduated
from the University of Rochester, Tanzi was playing keyboard in a rock band and writing
best-selling songs. Fifteen years later, he had won two prestigious awards for research on
Alzheimer's disease. The transition was not direct.
When Tanzi returned to Boston in 1980, he answered a bulletin-board advertisement and
became a technician for James Gusella -- hunting for the Huntington's disease gene. Their
success led to a lasting friendship, more than the typical relationship between mentor and
student. Tanzi was there from the beginning -- from the setting up of the laboratory through
the historic mapping of the gene in 1983 that made Huntington's disease the first disease to
be mapped without knowledge of the gene product. At Gusella's suggestion, Tanzi took on a
side project -- developing markers for chromosome 21. That chromosome is home to the
gene for Down's syndrome, and people with that condition are at high risk for Alzheimer's
disease. Tanzi therefore began to work on the laboratory's collection of DNA from families
with Alzheimer's disease.
Gusella advised Tanzi to take graduate courses in neurobiology, not genetics. In 1985, Tanzi
left the laboratory to become a graduate student. He left his Alzheimer's project to Peter St.
George-Hyslop, who was first Tanzi's trainee and then, simultaneously, a partner and
competitor.
In addition to being an autobiography, the book is a primer of molecular genetics. Tanzi
thought the biologic villain in the Alzheimer's story must be the accumulation of insoluble
amyloid in plaques throughout the brain. Before long, he was joined by dozens of
competitors who sought mutations in the gene for the amyloid precursor protein on
chromosome 21. Tanzi describes the tedium and long hours of fruitless searching as well as
the joy when a linkage was found. He explains the methods he and his colleagues used and
provides simple figures to explain how the mutant gene works.
There are no human villains among the investigators, but there is drama aplenty. George
Glenner first characterized the amyloid in the cerebral Alzheimer's plaques; he himself died
of amyloidotic heart disease. Charles Epstein, a leader in recognizing the connection
between Down's syndrome and Alzheimer's disease, was seriously injured by the psychotic
Unabomber. Research partners split up over arguments about patent rights. Friends became
mutually wary. A German scientist publicly accused U.S. investigators of practicing
"murderous science" in their development of a rapid autopsy program to provide brain tissue
for research; one Jewish leader took personal offense. Commercial companies contributed
transgenic mice that carry mutant amyloid genes and develop plaques like those in humans,
but these companies sometimes refused investigators access to the mice and promoted
competitive secrecy. Tanzi himself is more than a consultant to one company; he has "an
equity interest."
Tanzi seems blessed with an amiable disposition. He receives bruises but makes no enemies
and bears no long-term grudges. He describes the anguish of knowing he is in a race with
other investigators without being aware of how many or who they are -- he names 47
investigators who contributed to the progress in this field, but there are more. Tanzi lost one
race when he had almost reached the goal himself, and his marriage was threatened when,
racing, he spent Christmas and New Year's Day in the laboratory.
Tanzi himself is no braggart, and he acknowledges that the genetic discoveries seemed to
come simultaneously in reports from many laboratories. However, he was one of the first to
map the gene for the amyloid precursor protein to chromosome 21. He ruled out mutations
in that gene among the families studied at the Massachusetts General Hospital; he provided
one of the first genetic maps of the chromosome; and he came so close to identifying the
gene for presenilin 2 that he was included as an author in the report that did so. After the
book was published, Tanzi described a locus on chromosome 10. Four other genes had been
implicated: the amyloid precursor protein, two presenilins, and the susceptibility-factor
allele APOE-4. Tanzi and his associates were at the forefront of three of those
achievements.
Funding for research on Alzheimer's disease increased from $13 million in 1980 to more
than $400 million per year in the 1990s. Therapeutic trials based on the amyloid theory are
in progress, investigating a vaccine against amyloid, inhibitors of the proteases that produce
the amyloidogenic peptides, and drugs that lower the levels of zinc and copper in the brain.
The story is unfinished.
By the time you finish the book, you will hope that Tanzi and the others will make further
progress, because Alzheimer's disease is the great white whale of age-related
neurodegenerative diseases. There are 4 million victims now -- a disaster for their loved
ones and their caretakers, at an annual cost of $100 billion for the nation. The situation is
destined to become worse as people live longer. The dominant theory is that genetic
susceptibility interacts with environmental factors to cause Alzheimer's disease. Head injury
increases the risk, whereas the use of estrogens and nonsteroidal antiinflammatory agents
may be protective. By the end of the book, you will also hope that we will see more progress
in therapy, so that Tanzi and Parson can give us an upbeat second edition.
Research on Alzheimer's disease did not begin with molecular genetics, and sporadic cases
account for more than 95 percent of all cases. However, the genetic clues to amyloid
formation apply to the sporadic disease as well as to familial Alzheimer's. Fifty years ago,
not much was known about Alzheimer's disease; modern research began in the 1960s. For
readers who are interested in the days before molecular genetics, oral-history interviews
have been recorded by Robert Katzman and Katherine Bick, themselves pioneering
investigators (Alzheimer Disease: The Changing View. San Diego, Calif.: Academic Press,
2000). You do not have to be a geneticist or a neurologist to enjoy Decoding Darkness;
Alzheimer's disease is a commanding problem for all of us. The story is invigorating, the
progress is fantastic, and the writing is lively.
Lewis P. Rowland, M.D.
Columbia-Presbyterian Medical Center
New York, NY 10032
The Hidden Structure: A Scientific Biography of
Camillo Golgi
By Paolo Mazzarello. Translated and edited by Henry A. Buchtel and
Aldo Badiani. 407 pp., illustrated. New York, Oxford University Press,
2000. $90. ISBN 0-19-852444-7
Great conceptual advances in science are often based on great technical advances. Either
type of discovery can bring scientific fame. In 1896 Riva-Rocci devised the mercury
sphygmomanometer for measuring blood pressure, but it took several decades for
hypertension as a cause of disease to be unmasked. Conversely, Watson and Crick in 1953
discovered the structure of DNA but not the technique of x-ray crystallography that limited
the number of possibilities for their model. There are countless other examples, but it is rare
for a scientist to develop both a new instrument and new ideas. Camillo Golgi (1843-1925)
did his utmost to be more than the inventor of a revolutionary staining technique for nerve
tissue. The "reazione nera," or black reaction, which he discovered in 1873 after systematic
experiments, consisted of immersing specimens in silver nitrate after fixation with
potassium dichromate. It allowed visualization of nerve cells and their ramifications in
unprecedented detail. In The Hidden Structure, the excitement of this discovery is conveyed
to the reader mainly through words, because the histologic illustrations are sparse. Once
Golgi's technique was used in other countries (he published almost exclusively in Italian), it
stirred a flurry of new hypotheses about the anatomical organization and eventually the
function of the nervous system.
Golgi was less successful in staking his claims with regard to these wider implications. His
theory that nerve-cell processes formed a giant anastomotic network was initially attractive,
because it fit with the emerging notions about electricity in the nervous system. But with the
ascendancy of the cell theory in the second half of the 19th century, fewer and fewer
scientists were prepared to make an exception for the nervous system. The doctrine of
separate nerve cells, first proposed by His (in 1886) and Forel (in 1887), was anathema to
Golgi until the end of his life. The ultimate insults were the term "neuron" (coined by
Waldeyer in 1891) and the successful campaigning for the individual nerve cell by Ramon y
Cajal. The Spaniard made a rather sudden appearance on the international stage (he too
published only in his own language) and swiftly convinced the scientific community with
his superb preparations, made with modified Golgi stains. He later added the concept of
"dynamic polarization" (i.e., one-way traffic in nerve cells). Cajal and Golgi would never
get on well together. Irony dictated that in 1906 they were to share the Nobel Prize for
Medicine. Sadly enough, Golgi chose to use his official lecture for another desperate attack
on the neuron doctrine.
Though in his main theme of research Golgi's work exemplified a mixture of success and
error, he was undoubtedly a great scientist in many other respects. That Mazzarello's book
makes this abundantly clear is its greatest merit. The book follows Golgi from the cradle to
the grave, in 24 chronologically arranged chapters. It describes his lifelong attachment to the
University of Pavia, which was interrupted only by a stint as the director of a psychiatric
hospital in the country. It was there, in his spare time and in the kitchen of his private
apartment, that he discovered the black reaction. His name is also linked with the discovery
of several microscopic cellular structures (tendon organ, muscle spindle) and subcellular
structures (the Golgi apparatus). What will be new to most readers outside Italy are his
contributions to general medicine with regard to intestinal-worm infections, Bright's disease
of the kidney, and especially malaria. Golgi and his pupils not only accepted and defended
Laveran's theory of a parasitic origin for malaria, but they also provided many new pieces of
evidence and wholeheartedly threw their weight into a prolonged and complicated battle
with proponents of a "bacillus malariae."
The book has some weak points, in addition to the elaboration of histologic details without
corresponding illustrations. The style does not always run smoothly, and in places it is too
encyclopedic. Also, Golgi does not really come to life as a human being, apart from his
stubborn defense of "reticularism"; however, he was probably reserved in his private life,
leaving few emotional traces for his biographers.
Where the book does succeed is in depicting a man who, until his last moments, was totally
dedicated to medicine and to his university, at which he served as rector for a long period
late in his life. Inevitably, the reader is given many a glimpse of the intricacies of Italian
politics. Only one year before his death, when he was 80 years old and had received all
imaginable honors, Golgi had to endure the transfer of Pavia's medical school to Milan, an
event he had tried to avoid for decades. All in all, the book is a good read, especially for
aficionados of histology and neuroscience.
J. van Gijn, M.D.
University Medical Center Utrecht
3584 CX Utrecht, the Netherlands
Dear Mr. Darwin: Letters on the Evolution of Life
and Human Nature
By Gabriel Dover. 268 pp., illustrated. Berkeley, University of
California Press, 2000. $27.50. ISBN 0-520-22790-5
What would happen if a modern biologist could get in touch with Charles Darwin and enter
into a lively and stimulating discussion about recent developments in evolutionary theory?
Here, we have the answer. Molecular biologist Gabriel Dover uses imaginary
correspondence as a literary device for explaining how our ideas about evolution have
evolved since Darwin's day.
Dover argues that evolution involves more than just natural selection and cites such
phenomena of sampling error as genetic drift. The basic idea behind sampling error is easily
recalled. Gene frequencies inevitably fluctuate at random. Some alleles within a population
may happen not to be present in any of the zygotes that ultimately become the next
generation of adult organisms. This is more apt to happen when the alleles are rare and in
small populations. So pure accident may result in the elimination or fixation of an allele.
The result is change in gene frequencies, but not adaptation.
According to Darwin's theory of natural selection, adaptation results when one organism has
properties that allow it to out-reproduce another of the same species. Dover does not
mention another of Darwin's mechanisms -- correlated variability, or pleiotropy. Variation
sometimes involves traits that always go together, so that they both increase in frequency
even if only one of them is selectively advantageous. Dover's additional mechanisms are
something different. Because of the way in which chromosomes behave in the course of
reproduction, the genome is constantly being reorganized. Sexual reproduction generates
change, and Dover sees in it a cause of evolution (a "molecular drive") that interacts with
selection and sampling error.
Although Dover explains all this very well, his real goal is to rebut the metaphysics of
Richard Dawkins, an Oxford behaviorist who studied chickens before branching out and
writing popular books. Dover laments the influence of Dawkins's reasoning on persons who
are not equipped to see through it, especially textbook writers and social scientists. Let me
clarify what Dover is complaining about. Dawkins decided to call genes, and other things of
which copies are made, "replicators." The problem with that term is that in ordinary English,
the suffix "-or" refers to the doer of the action. Thus, a replicator should be that which does
the replication, not that which is replicated. (Likewise, a photocopier is not the copy that is
produced by the machine.) Dawkins's term is apt to dupe the unwary into thinking that a
passive participant is an active agent. This is what Dover has to grapple with: the
metaphysics of agency. He makes it abundantly clear that genes are not replicators, in the
sense of things that carry out replication. Rather, they are replicated by the cells that contain
them. More important, he argues that organisms are active agents in evolution by virtue of
their roles in restructuring the genome in producing compatibility among genes,
chromosomes, and other components of organisms and species.
This suggests an important role for sex. Dover maintains that molecular drive produces
compatibility within reproductive populations. The compatibility within species, together
with the lack of it between them, is fundamental to the modern "biological species concept."
The point that species are not just abstractions but, rather, higher-level units that play an
important part in evolution is very much in line with Dover's antireductionist metaphysics.
However true it may be that species and other populations are not likely to have adaptations
over and above those of their component organisms, there is no legitimate reason to
extrapolate and treat species and organisms as mere epiphenomena of molecules.
Species are important because they are historical units -- things that evolve and give rise to
the branches of the phylogenetic tree. Dover is off the mark when he suggests that biology is
history, pure and simple, and that we seek in vain for its laws of nature. Rather, biologists
have been seeking laws of nature in the wrong place. Although it is true that there are no
laws of nature for organisms and species, this is because organisms and species are concrete,
particular things, or "individuals" in the broad, metaphysical sense. All laws of nature are
about kinds, or things in general, and not about instances of such kinds. In evolutionary
biology, laws apply to kinds of populations, such as large populations and small
populations. For example, as the effective population size goes down, the frequency with
which alleles are fixed by sampling error goes up. This is a perfectly legitimate law of
nature: it is necessarily true of everything to which it applies, irrespective of time and place.
However, it is a statistical law, since it does not predict which particular version of a gene
will be eliminated from the population. The fact that so much of the lawfulness (such as it
is) of biology must be conceptualized in such statistical terms gives scant comfort to those
who would have us treat human behavior in Laplacian, deterministic style.
One might wonder whether debunking the metaphysical pretensions of one's colleagues is
perhaps beneath the dignity of a good scientist like Dover. Isn't it enough to joke about
selfish chromosomal deletions and then get on with one's research? The trouble is that
selfish genes are becoming part of popular culture. The medical community should brace
itself for an onslaught of belief systems, alternative therapies, and nostrums -- all justified
on the basis of what purports to be legitimate science.
Michael T. Ghiselin, Ph.D.
California Academy of Sciences
San Francisco, CA 94118
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
Legal Issues in Medicine
C ONJOINED T WINS — T HE L IMITS
OF L AW AT THE L IMITS OF L IFE
GEORGE J. ANNAS, J.D., M.P.H.
C
ONJOINED twins have been the subject of
scientific exhibits, medical study, human curiosity, and even entertainment, but until the
year 2000, conjoined twins had never been the subject of a courtroom battle. A unique case that was
the subject of two British court decisions deserves
study.1 The case illustrates the difficulty of applying
legal principles to unprecedented life-and-death decisions involving proposed medical interventions for
children — particularly when parents and physicians
disagree about what should be done.
The conjoined twins who were the subject of the
court decisions are identified by the judges only as
Jodie and Mary. They are the children of Michaelangelo and Rina Attard of the Maltese island of Gozo.
The couple, who are Roman Catholic, came to England for medical care at about five months’ gestation.
The children, who were joined at the pelvis, their
spinal columns on the same axis, with each having two
arms and two legs, were born on August 8, 2000.
The physicians saw no hope that the twins would survive for more than a year if they remained joined.
They believed that if Mary (the weaker of the two and
whose continued survival depended on sharing Jodie’s
circulatory system) was separated from Jodie, Mary
would die, but Jodie would survive and do well. The
parents refused to authorize the separation on the
basis that it was wrong to choose between the lives
of their two innocent children and that it was contrary to their religious beliefs. Physicians have historically honored the wishes of parents in such cases.2
In this case, however, the physicians decided to go to
court for authorization to proceed with the separation over the objections of the parents.
In the United States, the decision of the parents
would have been final unless the physicians or the state
could have persuaded a judge that this was a case of
child neglect.3 In Britain, the law is different: once
a case is placed before a judge, the judge must decide
what the welfare or best interests of the child require
by exercising “an independent and objective judgment.”1 The parents’ wishes are just one piece of evidence to be considered in making this decision. The
trial-court judge concluded that separation was in the
best interests of both children and that separation
was not a case of killing Mary but one of passive euthanasia in which her food and hydration would be
withdrawn (by clamping off her blood supply from
Jodie).1 The parents and the official solicitor, whose
task was to represent Mary, appealed.
Each of the three judges on the appeals panel issued a separate opinion, as is customary in British
courts. Although all the judges agreed with the trialcourt judge that the separation should be performed,
none agreed with the legal reasoning of that judge,
and none of the three judges on the appeals panel fully agreed with one another’s legal reasoning. There
are many explanations for these conflicting opinions,
including the unprecedented nature of the dispute itself, a reliance on analogies that did not quite work,
and a strong desire to authorize physicians to do what
they think is best for their newborn patients.
THE OPINION OF LORD JUSTICE
ALAN WARD
Lord Justice Alan Ward begins his analysis by noting that this “truly is a unique case” that “in a nutshell” involves killing the weaker twin, Mary (who
would not have been viable had she been a singleton), to “give Jodie a life which will be worthwhile.”
Ward describes the physical condition of the twins
in detail, quoting from medical reports that document,
among other things, that Jodie has “an anatomically
normal brain, heart, lungs and liver” and that she “appears to be a bright little girl,” who is expected to be
of “normal intelligence.” Mary, on the other hand, is
described by physicians as “severely abnormal,” having
a “primitive” brain, a very poorly functioning heart,
and the absence of lung tissue. Ward concludes that
Mary is incapable of surviving separately: “She lives on
borrowed time, all of which is borrowed from Jodie.
It is a debt she can never repay.” He notes that separation will cause Mary’s death (which will be quick
and painless) and that a heart–lung transplant is not
an option for Mary.1
Lord Justice Ward then turns to the question of
why the court is involved at all, noting that although
“every instinct of the medical team has been to save
life where it can be saved,” it would have been “perfectly acceptable” for the medical team and hospital
to have respected the parents’ wishes, even though
this would have resulted in the death of both twins.
But seeking the court’s authorization of surgery is
also acceptable in Ward’s view, because “here sincere
professionals could not allay a collective medical conscience and see children in their care die when they
know one was capable of being saved. They could not
proceed in the absence of parental consent. The only
arbiter of that sincerely held difference of opinion is
the court. Deciding disputed matters of life and death
is surely and pre-eminently a matter for a court of
law to judge.”1
In analyzing the existing law, Ward strongly disagrees with the trial court. He states that it is “utterly fanciful” to classify the operation as “an omis-
1104 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
L EGA L ISS UES IN MED ICINE
sion” of treatment (the failure to continue to provide
nutrition) rather than as an active surgical intervention that will end Mary’s life and that there is no way
that killing Mary can be “in Mary’s best interests.”
Instead, he concludes that the only proper legal path
when there is a conflict of interest between conjoined
twins is “to choose the lesser of two evils.”
Ward condemns the parents’ refusal to choose life
for Jodie in dramatic terms: “In my judgment, parents who are placed on the horns of such a terrible
dilemma simply have to choose the lesser of their inevitable loss. If a family at the gates of a concentration camp were told they might free one of their
children but if no choice were made both would die,
compassionate parents with equal love for their twins
would elect to save the stronger and see the weak one
destined for death pass through the gates.” He goes
on to say, “My heart bleeds for them. But . . . it is
I who must now make the decision.”
The decision, of course, seems to have been made
once the condition of the two twins was described.
But it still must be legally justified. Ward does this
by accusing Mary of killing Jodie and thus making a
decision to kill Mary justifiable homicide, a case of
“quasi self defence”: “Mary may have a right to life,
but she has little right to be alive. She is alive because . . . she [parasitically] sucks the lifeblood out
of Jodie. If Jodie could speak, she would surely protest, ‘Stop it, Mary, you’re killing me.’” Ward concludes that the physicians have a legal duty to Jodie,
which gives them an obligation to act, and that “doctors cannot be denied a right of choice if they are
under a duty to choose.”1
THE OPINION OF LORD JUSTICE
ROBERT BROOKE
Lord Justice Robert Brooke horrified the parents
when, in open court, he looked at pictures of the
twins and asked, “What is this creature in the eyes of
the law?”4,5 His opinion, however, is more analytical.
He agrees with Ward’s analysis of family law but believes more is required to conclude persuasively that
the operation that will kill Mary is lawful. The official solicitor, who opposed the separation, suggested
nonetheless that the court might wish to develop new
law that permitted such an operation if it was “proportionate and necessary” and “approved in advance
by the court.” Brooke essentially adopts this approach,
and much of his opinion explores the legal doctrine
of necessity.
The chief case that he examines is that of Regina
v. Dudley and Stephens, a famous 1884 case that involved shipwreck and survival on the high seas by
means of murder and cannibalism.6 In that case, a
crew of four was sailing the yacht Mignonette from
England to Australia when the ship came apart in a
storm in the South Atlantic Ocean 2000 miles from
land. The crew escaped in a lifeboat with only two
cans of turnips. After 19 days the three senior members of the crew killed 17-year-old Richard Parker,
the youngest and weakest member of the crew, and
ate him in order to survive. They later explained that
the point of killing him before he died naturally was
to be able to drink his blood.7 After being rescued and
returned to England, they were arrested and tried
for murder — a charge they did not deny. Their defense was “necessity.” The British courts rejected this
defense, noting among other things that the boy did
not threaten the rest of the crew and that the law
could not justify the killing of “the weakest, the
youngest, the most unresisting. . . . Was it more
necessary to kill him than one of the grown men?
The answer must be ‘No.’”6
Although so far rejected by British law, as this case
illustrates, Brooke suggests that there may be circumstances in which the necessity defense should be
allowed. He gives several examples. The first is the
case of a mountain climber who must cut the rope
holding him to another climber who has fallen, otherwise both will perish. The next is the 1987 sinking
of the passenger ferry Herald of Free Enterprise near
Zeebrugge, Belgium, in which almost 200 passengers
drowned. An army corporal said that he and dozens
of other people were in the water near the foot of a
rope ladder and all were in danger of drowning. Their
route to safety was blocked by a young man on the
ladder who was paralyzed with fear. Eventually, the
corporal ordered the man to be pushed off the ladder so that the others could climb to safety.
Two other examples had been used in the United
States in 1977 by a rabbinical scholar who counseled
a Jewish couple considering a similar operation on
their conjoined twins who shared a heart.8 The rabbi
reportedly said that if two men jump from a burning
plane and the parachute of the second one fails to
open, and he grabs the legs of the first man, the man
whose parachute did open is morally justified in kicking the second man away to save himself because the
man whose parachute did not open was “designated
for death.” Likewise, if a caravan is surrounded by
bandits, and the bandits demand that a particular person be turned over to them or they will kill everyone,
it is permissible to turn that person over because he
has been “designated for death.”1,8
Many more legal authorities are quoted at length,
but ultimately Lord Justice Brooke concludes that the
objections to the necessity defense presented in the
Dudley case of cannibalism on the high seas — who
can judge this sort of necessity, and how can the comparative value of lives be measured? — are not applicable to the case at hand, because “Mary is, sadly,
self-designated for a very early death.” He also thinks
there is no danger of the misuse of the necessity defense by physicians in other cases of conjoined twins
because “there will be in practically every case the
opportunity for the doctors to place the relevant facts
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1105
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
before a court for approval (or otherwise) before the
operation is attempted.”1
THE OPINION OF LORD JUSTICE
ROBERT WALKER
Lord Justice Walker opens his opinion by describing this case as “tragic” and “unprecedented anywhere
in the world.” Although conjoined twins are unique,
Walker insists that “there is no longer any place in the
legal textbooks, any more than there is in the medical
textbooks, for expressions (such as ‘monster’) which
are redolent of superstitious horror. Such disparagingly emotive language should never be used to describe a human being, however disabled or dysmorphic.” He nonetheless concludes that continued life
joined to Jodie would “confer no benefit [on Mary]
but would be to her disadvantage.” Walker agrees with
Brooke that the question of whether Mary can be
lawfully killed for Jodie’s sake rests on the issues of
intention and necessity. Like Brooke he uses a series
of analogies, but unlike him, Walker concludes that
“there is no helpful analogy or parallel to . . . this
case.”1
Ultimately, Walker determines that the doctors’
duties to the twins are in conflict. Nonetheless, he
believes the dilemma does not involve choosing “the
relative worth of two human beings” but rather “undertaking surgery without which neither life will have
the bodily integrity (or wholeness) which is its due.”
He believes that having her “bodily integrity,” if only
for a few seconds, is a benefit to Mary. He ultimately
concludes that physicians would separate the twins
not with the intent of killing Mary, but with the intent of making each twin whole and acting in the
best interests of both. What seems to persuade Walker the most, however, is the testimony of the physicians: “Highly skilled and conscientious doctors believe that the best course, in the interests of both
twins, is to undertake elective surgery in order to separate them and save Jodie.”
AFTERMATH
The twins were separated six weeks after the opinion of the Court of Appeal was issued.9 Before the
surgery was performed, there was further debate
over which surgical team — the one with more experience or the one that brought the case to court
— should perform the surgery.10 It was ultimately
performed by the less experienced team. The physicians involved later told the press that they sought
the court’s approval because they were worried about
being prosecuted for the murder of Mary.11 They continued to believe that separation was in the best interests of both twins (although it caused Mary’s
death), and when the final blood vessels connecting
the twins were cut, an act that would result in the
death of Mary, the two lead surgeons said they cut
the blood vessels together, in silence and with “great
respect.”5 The coroner’s verdict stated simply that
Mary died “following surgery separating her from her
conjoined twin, which surgery was permitted by an
order of the High Court, confirmed by the Court of
Appeal.”12
The opinion of the appeals court has been praised.13
Jodie is doing well and may soon go home to Gozo
with her parents.14 She will reportedly require extensive surgery over the next five years, most of which
will be performed in Britain.9 Mary was buried on
Gozo in January.14
PROBLEMS WITH THE LEGAL ANALYSIS
It is easy to see why all the judges involved characterized this case as unique and hoped that it would
not set a precedent. The case seems to have been decided not on the basis of the law (which most of the
judges found of little help) but on an intuitive judgment that the state of being a conjoined twin is a
disease and that separation is the indicated treatment
for it, at least if such treatment affords one of the twins
a chance to live. The judges identified strongly with
the physicians and had little empathy with the parents or their religious beliefs. I think all these factors
led each judge to make problematic legal statements.
Failure to Identify with the Parents
Lord Justice Ward is the hardest on the parents,
using the Sophie’s Choice analogy of a parent at the
gates of a concentration camp. The Nazi physician
in charge of determining who is to go straight to the
gas chambers and who can work or be used in medical experiments tells Sophie that both her children
will be killed if she does not choose one to save.15
Ward insists that a parent in this situation must choose.
Sophie, of course, did choose, although she ultimately lost both children to the Nazis and killed herself
because she was unable to live with her decision.15
Ward’s reasoning, at the heart of his analysis, is troubling in at least two respects. The first is his conclusion that parents must choose which child will die
when only one can be saved. We would not condemn
a parent for making this terrible choice, but neither
should we condemn a parent for refusing to make it.
For example, if a father jumps from a burning plane
holding his two children, one in each arm, and then
begins to lose his grip on both and realizes that he
will drop them both if he does not drop one to save
the other, we would not fault him for dropping one.
Neither, I believe, should we fault a parent for refusing to choose and trying to hang on to both children
for as long as possible. Second, and more disturbing
with respect to the concentration-camp example, is
the question of who the judge thinks is in the role
of the Nazi physician. Ultimately, Ward concludes that
it is the British physicians who “should be given the
right of choice,” but he also seems to place himself
in that role, saying, “it is I who must now make the
1106 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
L EGA L ISS UES IN MED ICINE
decision.” Of course, Ward is not choosing to kill
both twins, and perhaps he sees nature as the Nazis.
Nonetheless, it is unsettling to have a British judge
rely on what might be termed “concentration-camp
ethics” to reach a decision.
Lord Justice Ward insists that the law requires him
to do what is in the “best interests” of the children
and that British law prohibits the use of the doctrine
of “substituted judgment” (determining what an incompetent person would decide if he or she were capable of making a decision). Nonetheless, his primary
argument turns out to be based on this doctrine: in
colorful language he likens Mary to a parasite who
is “poisoning” Jodie and sucking out her “lifeblood.”
He knows what Jodie would decide if she could decide: “If Jodie could speak, she would surely protest,
‘Stop it, Mary, you’re killing me.’” But the problem
with using substituted judgment in the case of very
young children is that we have no way of knowing
what they would say and tend to speculate on the basis of our own adult values. For example, Jodie could
equally well say to her identical and attached twin
that “I love you as myself and will do everything, including sacrificing my life, to keep you alive as long
as possible.” Likewise, Mary might reasonably say to
Jodie, “You are my identical twin, and because I love
you, I’m willing to die so that you can live, since this
is the only chance for my genes to be transmitted to
the next generation.” Each twin might also, of course,
consider the other twin to be an integral part of herself, a view that would preclude separation. Any of
these hypotheses is plausible, but made-up monologues cannot take the place of legal analysis.
Problems with Analogies
Justice Brooke’s opinion is problematic because in
my view he has not properly interpreted the analogies he uses. His reliance on the necessity defense, for
example, is ultimately based almost exclusively on the
two analogies that were reportedly used by a rabbi
in counseling a Philadelphia couple in a similar situation in 1977: the men jumping from the burning
plane and the caravan surrounded by bandits. In each
case, the necessity defense is said to be appropriate
because the person killed was “designated for death,”
a phrase Brooke adopts as his primary justification
for killing Mary to save Jodie. In fact, he goes further, concluding, “Mary is, sadly, self-designated for
a very early death.”
There are two problems with this conclusion.
First, Mary did not designate herself for anything,
she was simply born and survived. But even the simple conclusion that one can be “designated for death”
may not be a proper interpretation of the two analogies. The description of the cases was drawn from
an article I wrote in 1987, and I used a 1977 newspaper report as my own source.8,16,17 More important
than what might have been lost in the retelling, how-
ever, is that expert commentary on these examples
has since been published, and the court seems unaware of it.
A leading U.S. rabbinical authority, Rabbi J. David Bleich, has written that these two stories are not
examples of instances in which a person is “designated for death.”18 Instead, the example involving the
parachutists is more correctly thought of as a case of
pursuit, in which the first man’s kicking off of the
man who is clinging to him is justified by the fact
that that man’s intentional actions would otherwise
kill him. With respect to the second example, Rabbi
Bleich argues that the caravan is justified in turning
over the named person only if that person is guilty
of some crime; if the person is innocent, he may not
be given up to face certain death. Others have suggested that in a similar situation the group could
lawfully agree to use a random device, such as drawing straws, to decide who would be sacrificed for the
good of the group.19,20
Rabbi Bleich does, nonetheless, offer Lord Justice
Brooke another justification for his conclusions. Bleich
believes that there may be exceptional circumstances
in which one conjoined twin can be judged a pursuer of the other: “If the heart can be shown to belong to one twin exclusively, the second is, in effect,
a parasite . . . [and having] no claim to the heart,
is then quite literally a pursuer.”18 Pursuers must be
stopped before they kill, and self-defense would have
provided Brooke with a much sounder rationale than
the “designated for death” approach.
Problems with Conjoined Twins Themselves
Lord Justice Walker’s opinion is, I think, most notable in attempting to consider the conjoined twins
as both a single entity and two persons. Walker wants
to discourage the use of terms such as “monster” (and
probably “creature” as well) to describe conjoined
twins. Nonetheless, he speaks of them not as one entity, but as two separate “innocent children” and believes the “court must consider the welfare of each.”
The problem is that once the twins are separated verbally, it is only a matter of time before they will be
separated surgically. Walker sees these conjoined twins
as a serious, lethal anomaly that must be medically
corrected so that at least one of the twins can appear
normal. In this regard Walker seems correct in concluding, “in truth there is no helpful analogy or parallel to the current situation.” He thus seems to find
the condition of being a conjoined twin, at least when
one could live if the other were killed, itself adequate
justification for separation. That is why he can conclude, with the physicians, the trial-court judge, and
Lord Justice Brooke, that separation would be in the
best interests of both children. Stated another way,
three of the four judges believed that Mary was better off dead than continuing to live for a few months
as a conjoined twin.
N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org · 1107
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
LESSONS
Perhaps the most important lesson of the case of
Jodie and Mary is that there are severe limits to the
law in making unprecedented, complex, life-and-death
decisions. The most important shortcoming of the
decision of the judges is that it did not rest on any
legal principle. That is why if the circumstances of
this case were to be duplicated tomorrow at the Great
Ormond Street Hospital for Children in London, the
physicians involved could, on the basis of the reasoning of this case (and contrary to its conclusion), decide to follow the wishes of the parents and let both
twins die. The conclusion of Lord Justice Ward that
it would have been “perfectly acceptable” for physicians to decide either way must be wrong: if Mary is
a pursuer who is killing Jodie, saving Jodie’s life (and
that of others in her situation) by ending Mary’s life
must be mandatory. The court’s ruling that physicians
can do whatever they think is best (with the court’s
prior approval) is no legal rule at all. Nor is it true
that there will almost always be ample time to seek
court review in cases such as this.10,21
Closely related is the question of the court’s role
in similar cases. Is it to determine whether a particular course of action, chosen by both parent and
physician, is legally permissible, or is it to determine
whether a particular medical intervention is required
by law? The first role seems reasonable; the second
seems justified only in cases in which the failure to
act (on the part of either parent or physician) is child
neglect. In this regard, had Jodie been a singleton,
her parents might well have been justified in refusing
to consent to three or four years of complicated surgical procedures with an uncertain outcome on the
basis that they did not believe the burdens of these
interventions on Jodie could be justified by the expected outcome even if the physicians believed the operation was in her best interest.22
My own view is that in this case, it would have
been better had the physicians not sought court intervention, or if they had, for the trial court to have
refused to hear the case and to have instructed the
physicians that they must obtain the parents’ consent
before separating the twins. I would have liked to have
had the parents agree to the separation (since giving
Jodie a chance to live at the cost of cutting Mary’s
life short does seem the lesser of two evils), but I do
not believe the case for separation is so strong that it
demands that the authority to make the decision about
the medical care of their children be taken away from
the parents.
REFERENCES
1. Re A (children), (2000) 4 All ER 961.
2. O’Neill JA, ed. Pediatric surgery. 5th ed. St. Louis: Mosby–Year Book,
1998:1925-38.
3. Rosato JL. Using bioethics discourse to determine when parents should
make health care decisions for their children: is deference justified? Temple
L Rev 2000;73:1
4. Cullen K. In London, an agonizing decision. Boston Globe. September
11, 2000:A1, A8.
5. Wells M. Surgeons speak of life-death op on twins. Guardian (London).
December 8, 2000:13.
6. Regina v. Dudley & Stephens, (1884) 14 QBD 273.
7. Simpson AWB. Cannibalism and the common law: the story of the tragic last voyage of the Mignonette and the strange legal proceedings to which
it gave rise. Chicago: University of Chicago Press, 1984.
8. Annas GJ. Siamese twins: killing one to save the other. Hastings Cent
Rep 1987;17:27-9.
9. Laville S. Mary was freed by death, says father. Daily Telegraph (London). December 7, 2000:1.
10. Spitz L, Kiely E. Success rate for surgery of conjoined twins. Lancet
2000;356:1765.
11. British surgeon reflects on decision to let Siamese twin die. Toronto
Star. January 7, 2001:1.
12. Bunyan N. Doctors ‘had no option but to separate twins.’ Daily Telegraph (London). December 16, 2000:9.
13. Smith AM. The separating of conjoined twins. BMJ 2000;321:782.
14. Cramb A. Village children mourn as Siamese twin is laid to rest. Daily
Telegraph (London). January 20, 2001:3.
15. Styron W. Sophie’s choice. New York: Random House, 1979.
16. Drake DC. The twins decision: one must die so one can live. Philadelphia Inquirer. October 16, 1977.
17. Thomasma DC, Muraskas J, Marshall PA, Myers T, Tomich P, O’Neill
JA. The ethics of caring for conjoined twins: the Lakeberg twins. Hastings
Cent Rep 1996;26:4-12.
18. Bleich JD. Conjoined twins. Tradition 1996;31:92-125.
19. U.S. v. Holmes, 26 F. 360 (E.D. Pa. 1842).
20. Fuller L. The case of the Speluncean explorers. Harv L Rev 1949;62:
616.
21. Norwitz ER, Hoyte LP, Jenkins KJ, et al. Separation of conjoined
twins with the twin reversed: arterial-perfusion sequence after prenatal
planning with three-dimensional modeling. N Engl J Med 2000;343:399402.
22. Re T, (1997) 1 All ER 906.
1108 · N Engl J Med, Vol. 344, No. 14 · April 5, 2001 · www.nejm.org
Copyright © 2001 Massachusetts Medical Society.