Simplifying the TNM System for Clinical Use in Differentiated Thyroid Cancer

VOLUME
27
䡠
NUMBER
11
䡠
APRIL
10
2009
JOURNAL OF CLINICAL ONCOLOGY
O R I G I N A L
R E P O R T
Simplifying the TNM System for Clinical Use in
Differentiated Thyroid Cancer
Adedayo A. Onitilo, Jessica M. Engel, Catharina Ihre Lundgren, Per Hall, Lukman Thalib, and Suhail A.R. Doi
From the Marshfield Clinic Weston
Center, Weston, WI; Department of
Molecular Medicine and Surgery, Karolinska University Hospital; Department
of Medical Epidemiology and Biostatistics, Karolinska Institute, Stockholm,
Sweden; and the Department of
Community Medicine (Biostatistics),
Kuwait University; Division of Endocrinology, Mubarak Al-Kabeer Teaching
Hospital; and the Department of Medicine, Kuwait University, Kuwait.
Submitted October 6, 2008; accepted
November 25, 2008; published online
ahead of print at www.jco.org on
March 9, 2009.
Authors’ disclosures of potential conflicts of interest and author contributions are found at the end of this
article.
Corresponding author: Suhail A.R. Doi,
PhD, Department of Medicine, Kuwait
University, PO Box 24923 Safat, 13110
Kuwait; e-mail: [email protected]
The Appendix is included in the
full-text version of this article,
available online at www.jco.org.
It is not included in the PDF version
(via Adobe® Reader®).
A
B
S
T
R
A
C
T
Purpose
The TNM stratification has been found useful at stratifying patients with differentiated thyroid
carcinoma (DTC) into prognostic risk groups. However, it is cumbersome to implement clinically
given the large number of bins within this system and the complicated system of arriving at
stage information.
Patients and Methods
We decided to quantify each variable in this system to arrive at a simplified quantitative alternative
to the TNM system (QTNM) and compare this with the conventional system. We used our
electronic record system to identify 614 cases of DTC managed at our institution from 1987 to
2006. Cancer-specific survival (CSS) and disease-free survival (DFS) were calculated by the
Kaplan-Meier method, and a simplified QTNM score was devised using a Cox proportional
hazards model.
Results
We were able to quantify the TNM system as follows: 4 points each for age older than 45 years
and presence of neck nodal metastases while 6 points for tumor size larger than 4 cm or
extrathyroidal extension and 1 point for nonpapillary DTC. A sum of 0 to 5 points was low risk, 6
to 10 points intermediate, and 11 to 15 points high risk. Comparison with the conventional TNM
system and two other systems revealed similar or better discrimination with the QTNM and this
discrimination was maintained when this risk stratification was applied to a unique validation set.
Conclusion
The QTNM system as opposed to the conventional TNM system seems to be a simple and
effective method for risk stratification for both recurrence and cancer-specific mortality.
J Clin Oncol 27:1872-1878. © 2009 by American Society of Clinical Oncology
© 2009 by American Society of Clinical
Oncology
0732-183X/09/2711-1872/$20.00
INTRODUCTION
DOI: 10.1200/JCO.2008.20.2382
Nonmedullary differentiated thyroid carcinomas
(DTC; papillary, follicular, follicular type of papillary, and Hurthle) have a favorable prognosis, but a
proportion of patients will develop recurrences and
eventually die of the disease indicating a lack of reliable prognosticators. For this reason, systems have
been devised that can order patients by a decreasing
probability of survival which can be used for selecting patients for therapy and for providing patients
with an estimate of their prognosis. The TNM system has been widely used for this purpose and is in
effect a “bin model” wherein the TNM prognostic
factors are used to create a mutually exclusive and
exhaustive partitioning of patients, so that every patient is in one and only one bin.1 This system uses the
mean survival of the patients already in the bin to
predict what will happen to a new patient placed in
that bin. For example, if a new patient is placed in the
1872
(T1, N0, M0) bin, then that patient’s 5-year diseasespecific survival is predicted to be the same as the
mean survival of all the patients who were placed in
that bin 5 years ago.2
Although the TNM is based on the anatomic
extent of disease, there are several problems. Distant
metastases at diagnosis is in itself a sufficient criterion for aggressive management and effectively bypasses the selection process for therapy. It was
realized as early as 1993 that there were sufficient
data available to indicate the independent prognostic significance of additional nonanatomic variables.3 As the TNM system is updated, there has
been an increase in the number of subgroups within
the TNM criteria themselves. The problem faced
with adding all these to the TNM system is that
stratifying by additional prognostic variables or
TNM subgroups (eg, splitting T4 into T4a and T4b
or N1 into N1a and N1b) leads to a proliferation of
bins.2 For example, if one adds to the TNM stages for
© 2009 by American Society of Clinical Oncology
Downloaded from jco.ascopubs.org on June 9, 2014. For personal use only. No other uses without permission.
Copyright © 2009 American Society of Clinical Oncology. All rights reserved.
Quantitative TNM in Thyroid Cancer
Table 1. Results of Model Building Using Cox Proportional Hazard Regression
Parameter
A†
Age, ⬎ 45 years
Tumor, ⬎ 4 cm
Node present
Not papillary
Valid observations, total No.
Uncensored
Censored
Log likelihood
Final solution
Null model (all ß’s ⫽ 0)
␹2 (null model final solution)
df
P
B‡
Age ⬎ 45 years
Tumor, ⬎ 4 cm
Node present
Valid observations, total No.
Uncensored
Censored
Log likelihood
Final solution
Null model (all ß’s ⫽ 0)
␹2 (null model final solution)
df
P
C§
Age, ⬎ 45 years
Tumor, ⬎ 4 cm
Node present
Metastasis present
Valid observations, total No.
Uncensored
Censored
Log likelihood
Final solution
Null model (all ß’s ⫽ 0)
␹2 (null model final solution)
df
P
␤
SE
t Value
ORⴱ
Wald Statistic
P
1.11
1.79
1.18
0.31
0.32
0.31
0.32
0.33
3.44
5.71
3.73
0.95
3.03
5.97
3.24
1.37
559
49; 8.77%
510; 91.23%
11.9
32.6
13.9
0.9
⬍ .001
⬍ .001
⬍ .001
.34
12.98
35.02
13.36
⬍ .001
⬍ .001
⬍ .001
7.2
28.6
9.3
1.9
.007
⬍ .001
.002
.16
⫺248.57
⫺283.54
69.9
4
⬍ .001
1.14
1.83
1.07
0.32
0.31
0.29
3.15
6.25
2.92
559
49; 8.8%
510; 91.2%
⫺249
⫺283.5
69.1
3
⬍ .001
0.88
1.77
0.96
0.76
0.33
0.33
0.32
0.55
2.4
5.9
2.6
2.1
403
43; 10.7%
360; 89.3%
208.7
⫺236.2
54.9
4
⬍ .001
Abbreviations: OR, odds ratio; DTC, differentiated thyroid carcinoma.
ⴱ
Indicates the odds that an individual in the group with that risk factor reaches the end point first (end point is DTC related recurrence or death).
†Results of model building using Cox proportional hazard regression.
‡Results of model building using Cox proportional hazard regression but not forcing histopathology into the model.
§Results of model building using Cox proportional hazard regression but not forcing histopathology into the model and including distant metastases.
thyroid cancer age information (two groups: age ⬎ 45 years v others)
and histologic type information (two groups: follicular v others), one
would increase the number of bins from 36 bins (6 T ⫻ 3 N ⫻ 2
M ⫽ 36) to a number that will be 144 bins (36 ⫻ 2 age groups ⫻ 2
histologic types ⫽ 144). Because the primary utility of the TNM
staging system is its simplicity as a “look-up” table in which one looks
up the stage of disease, the organization and use of a system with 144
bins becomes an impractical task.2 This problem may be compounded
by our future need to add newer nonanatomic prognostic markers to
the current TNM system.
Eventually, clinicians will make use of the bins by grouping them
together into larger bins called stages for clinical use.2 The TNM
www.jco.org
system for thyroid cancer groups the bins into four stages, each associated with a specific prognosis. The problem with this system of risk
stratification is that prognostic risk stratification is not an exact science, and combining 144 bins into four stages may not necessarily
improve prognostic risk stratification over and above quantitative
evaluation of risk factors defined at say two levels of each TNM factor
and other important nonanatomic factors, such as age and histologic type for DTC. We therefore decided to evaluate a quantitative
simplified TNM system for thyroid cancer against the standard TNM
classification using information on disease-free survival (DFS) and
cancer-specific survival (CSS) obtained via our comprehensive patient database.
© 2009 by American Society of Clinical Oncology
Downloaded from jco.ascopubs.org on June 9, 2014. For personal use only. No other uses without permission.
Copyright © 2009 American Society of Clinical Oncology. All rights reserved.
1873
Onitilo et al
PATIENTS AND METHODS
Study Target Populations
Patients diagnosed with well DTC were electronically extracted from
the Marshfield Clinic (MC) combined medical record using ICD-9 diagnostic code 193, with a rule of two (same diagnosis on two separate service
dates). The resulting cases were then run against the MC/Saint Joseph’s
Hospital (SJH) Cancer Registry, which identified a total of 658 patients
during the time-frame from January 1, 1987, through December 31, 2006.
Further review and application of exclusions reduced the number of study
patients from 658 to 614. Patients were excluded due to diagnoses of
non-DTC (anaplastic or medullary thyroid cancer) and if there were
missing data verified through manual abstraction.
A validation cohort (separate from the cohort in the United States described earlier) was selected from the Swedish Cancer Registry (SCR), which
was established in 1958. The SCR does not register tumors for which the sole
source of information is the death certificate. Tumors detected incidentally at
autopsy were excluded from the validation study. From 1958 to 1987, when
Sweden had 8.4 million inhabitants, 7,906 thyroid cancers were reported to the
SCR. After excluding 1,405 patients (17.8%) with anaplastic and medullary
thyroid cancers and 947 with follicular thyroid adenomas (12.0%), 5,554
individuals with DTC remained, of whom 5,123 had survived for at least 1 year
after diagnosis.4 Matching of these 5,123 patients with the Swedish Causes of
Death Register for 1959 to 1999 identified 693 patients (potential cases) for
whom thyroid cancer was reported as the cause of death.
Treatment protocols for DTC are standard and as such were similar in
both Sweden and the United States. However, this does not exclude possible
minor differences between the practices of institutions this being a limitation
of this study.
Data Collection
The MC/SJH Cancer Registry database was queried to provide the following data on each patient: age, sex, histologic type, tumor size, presence of
distant metastases, nodal disease, pathologic diagnosis, recurrence, location of
recurrence, cause of death, and length of survival. The majority of data was
available electronically. Supplemental manual data abstraction as needed was
performed by two of the investigators (A.A.O., J.M.E.). This study protocol
was approved by the MC institutional review board. The setting was a large
multispecialty group practice located in central Wisconsin. Data sources included the combined medical record of MC and affiliated hospitals and the
MC/SJH Cancer Registry.5,6
For the validation study, data on histopathologic findings, growth patterns, tumor differentiation, surgical procedures, and findings at follow-up
were abstracted from the case records of these 693 patients. A group of three
specialists (a cardiologist, an oncologist, and an endocrine surgeon) independently evaluated the medical records to confirm DTC as the cause of death.
Ninety-eight patients had to be excluded for various reasons,4 and 595 sets of
patients and controls were generated by randomly sampling one control for
each patient, matched for age at diagnosis (5-year age groups), sex, and 10-year
calendar periods of diagnosis via incidence density sampling.
Statistical Analysis
Variables used in the TNM system, as well as sex and histopathologic
type, were dichotomized and entered into a forward stepwise Cox proportional hazards model, the outcome being DFS, defined as time to first recurrence or to a cancer-specific death. These factors had all been documented
when the diagnosis was made. Only age, tumor size/extension, and node status
were selected. Sex, metastases at diagnosis, and histopathologic status were
excluded from the model. The results of this multivariate analysis were then
used to develop a clinical prediction model, after forcing histopathology into
the model instead of distant metastases even though the latter and not the
former is within the conventional TNM system (Table 1). This was done for
two reasons: histopathologic status is widely recognized as a prognostic variable and was associated with CSS in our cohort and we decided not to force
distant metastases into the final model given the fact that distant metastases at
diagnosis must be treated aggressively regardless of ultimate prognosis. Nev1874
© 2009 by American Society of Clinical Oncology
Table 2. QTNM Staging for Differentiated Thyroid Cancer
Stage
Risk
Criteria by
Score
10-Year DFS (%)
10-Year CSS (%)
1
2
3
Low
Intermediate
High
0-5
6-10
11-15
95
70
45
100
90
65
NOTE. The TNM factor distant metastases was substituted instead with
histopathologic status. Quantitative alternative to the TNM system (QTNM)
score is the sum of the following: histopathology ⫽ 1 (not papillary thyroid
cancer) otherwise 0; age ⫽ 4 (age ⱖ 45 years) otherwise 0; nodes ⫽ 4
(regional lymph node metastasis) otherwise 0; tumor ⫽ 6 (tumor ⬎ 4 cm in
greatest dimension limited to the thyroid or any tumor with extrathyroid
extension) otherwise 0.
Abbreviations: DFS, disease-free survival; CSS, cancer-specific survival.
ertheless, for comparison, we analyzed models without forcing histopathology
and after forcing distant metastases (Table 1).
Each ␤ coefficient was multiplied by 3.3 and rounded to the nearest
integer. The risk score for an individual patient was determined by assigning
points for each factor present and summing. The resulting continuous distribution of total risk scores across all patients in the model (range, 0 to 15) was
then stratified into three equal categories of points that grouped patients
according to the level of risk (low, intermediate, and high risk) as presented in
Table 2. The quantitative alternative to the TNM system (QTNM) score is the
sum of the following: histopathology ⫽ 1 (not papillary thyroid cancer) otherwise 0; age ⫽ 4 (age ⱖ 45 years) otherwise 0; nodes ⫽ 4 (regional lymph node
metastasis) otherwise 0; tumor ⫽ 6 (tumor ⬎ 4 cm in greatest dimension
limited to the thyroid or any tumor with extrathyroid extension) otherwise 0.
Table 3. Characteristics of the Population With Differentiated Thyroid
Carcinoma in Terms of Difference in Survival Between Subgroups
Characteristic
Sex
Female
Male
Age, years
ⱕ 45
⬎ 45
Histopathology
Papillary
Other
Tumor size
ⱕ 4 cm and no extrathyroid extension
⬎ 4 cm or having extrathyroid
extension
Nodal metastases
N⫹
N⫺
Distant metastases
M1
M0
Surgery
Total thyroidectomy
Less than total surgery
Post surgery radioiodine ablation therapy
No
Yes
No. of
Patients
Gehan’s Wilcoxon
P
DFS
CSS
.09
.2
⬍ .001
⬍ .001
.3
.05
⬍ .0001
⬍ .0001
⬍ .0001
.03
⬍ .001
⬍ .0001
.1
⬍ .0001
.5
⬍ .001
457
154
305
306
419
192
425
145
168
414
12
398
503
102
158
449
Abbreviations: DFS, disease-free survival; CSS, cancer-specific survival.
JOURNAL OF CLINICAL ONCOLOGY
Downloaded from jco.ascopubs.org on June 9, 2014. For personal use only. No other uses without permission.
Copyright © 2009 American Society of Clinical Oncology. All rights reserved.
Quantitative TNM in Thyroid Cancer
classification seems to be the currently most predictive system.10,11 We also did
a comparison with the University of Alabama and M. D. Anderson Cancer
Care (Houston, TX) system12 because its highest risk category is made up of
patients with distant metastases at diagnosis, a variable that was dropped from
the multivariate analysis in this study.
Cumulative Survival (proportion)
1.0
0.9
0.8
0.7
RESULTS
Group 1
Group 2
Group 3
Complete
Censored
0.6
0.5
0.4
0.3
0
5
10
15
20
25
Time (years)
Fig 1. Disease-free survival by quantitative alternative to the TNM system risk
group. Group 1 is low risk, group 2 is intermediate risk, and group 3 is high risk.
Although this stratification method resulted in relatively few patients in the
highest risk group, it allowed for discrimination of this small subset of patients
at high risk. The univariate relationships between survival and the staging
groups were analyzed via the Kaplan-Meier method. The difference between
two risk groups was compared with Gehan’s Wilcoxon test since this test gives
more weight to recurrence or deaths at early time points. Statistical analyses
were performed using the Statistica software (Statsoft, Tulsa, OK) for Windows 6.0 computer software ( StatSoft Inc).
For the validation study, a conditional logistic regression model for the
data set matched for 10-year calendar period, sex, and survival time in completed years was used to estimate odds ratios and their respective 95% CIs
(PROC PHREG in SAS, version 9; SAS Institute Inc, Cary, NC). The odds ratio
was used as an estimate of the relative risk.
Comparisons With Other Systems
Comparison with MACIS (Mayo Clinic’s metastasis, age, completeness,
invasive, size score)7 was made since, after the TNM8,9 system, this staging
We examined 614 DTC patients retrospectively to validate known
prognostic factors that enable them to be recognized as having
either a low or a high risk of death related to a recurrence of DTC,
by reference to our new QTNM stage or the conventional TNM
stage or the two other staging classifications (MACIS and M. D.
Anderson classifications). Patient characteristics are presented in
Table 3. Fifty-nine patients (9.6%) had local or distant recurrence,
or died of a recurrence, with a 10-year recurrence free survival of
85%. The 10-year CSS was 93%. Sex, age, larger tumor size, extrathyroidal invasion, nodal and distant metastases were all related to
a higher incidence of tumor recurrence. Cancer-specific death,
however, was associated with all of these factors except sex and in
addition with nonpapillary tumor type, extent of surgery, and
postsurgical remnant ablation (Table 3). Survival by QTNM risk
group is shown in Figure 1 and Appendix Fig A1 (online only) for
DFS and CSS, respectively. Cox proportional hazards regression of
each staging system revealed that the TNM stages I and II as well as
MACIS groups 2 and 3 were not very discriminatory since the odds
indicate that an individual in the baseline group is as likely to reach
the end point first as an individual in the next group (Table 4). The
validation results confirmed that the hazard ratios for DTC related
mortality were significantly greater in the intermediate and high
risk categories than in the low-risk group in patients from a different center (Table 5), and again the standard TNM staging groups
were not very discriminatory even though there was a trend (Table 6).
Finally a cross-tabulation of the 16 bins possible (from these four
Table 4. Multivariate Cox Regression Analysis
Classification
MACIS score
0-5.99
6-6.99
7-7.99
8⫹
TNM stage
I
II
III
IV
University of Alabama and M. D. Anderson stage
1
2
3
QTNM score (this study)
0-5
6-10
11-15
No. of Patients
Odds of Recurrence
95% CI
P
317
35
18
23
1 (ref)
2.1
5.9
11.7
0.7 to 6.3
2.2 to 16.1
5.7 to 23.9
.17
⬍ .001
⬍ .001
271
38
36
27
1 (ref)
1.4
3.5
16.9
0.4 to 4.8
1.3 to 9.1
8.1 to 35.3
.6
.01
⬍ .001
258
140
12
1 (ref)
3.3
7.3
1.7 to 6.3
2.4 to 22
⬍ .001
⬍ .001
386
117
56
1 (ref)
5.1
14.3
2.5 to 10.6
6.9 to 29.7
⬍ .001
⬍ .001
NOTE. Odds that an individual in the group with the higher risk will reach the end point (recurrence of disease) first according to the various classification systems.
Abbreviations: MACIS, Mayo Clinic’s metastasis, age, completeness, invasive, size score; QTNM, quantitative alternative to the TNM system.
www.jco.org
© 2009 by American Society of Clinical Oncology
Downloaded from jco.ascopubs.org on June 9, 2014. For personal use only. No other uses without permission.
Copyright © 2009 American Society of Clinical Oncology. All rights reserved.
1875
Onitilo et al
Table 5. Validation of QTNM Staging for Differentiated Thyroid Cancer Mortality Using Conditional Logistic Regression on Matched Cases (died) and Controls
(survived) Sampled Using Incidence Density Sampling
Matching Variables Onlyⴱ
No.
Stage
Risk
Score
Patient
Control
Odds Ratio
1
2
3
Low
Intermediate
High
0-5
6-10
11-15
297
176
122
441
114
37
Ref
1.64
2.02
Other Variables†
95% CI
Odds Ratio
95% CI
1.14 to 2.37
1.22 to 3.32
Ref
1.58
1.96
1.09 to 2.28
1.19 to 3.22
NOTE. The hazard is the odds of being a patient. Quantitative alternative to the TNM system (QTNM) score is the sum of the following: histopathology ⫽ 1 (not
papillary thyroid cancer) otherwise 0; age ⫽ 4 (age ⱖ 45 years) otherwise 0; nodes ⫽ 4 (regional lymph node metastasis) otherwise 0; tumor ⫽ 6 (tumor ⬎ 4 cm
in greatest dimension limited to the thyroid or any tumor with extrathyroid extension) otherwise 0.
ⴱ
Odds ratio adjusted for matching variables (sex, survival time in completed years, and 10-year calendar periods of diagnosis).
†Odds ratio adjusted for extent of surgery (total v other procedures) in addition to matching variables.
dichotomous risk parameters) versus quantitative risk categories reveals that the three categories are fairly homogenous (Appendix Table
A1, online only).
DISCUSSION
The QTNM results suggest that despite being a simple and easily
applied quantitative score, it retains a similar prognostic significance
to the TNM, MACIS, and M. D. Anderson classifications. At the same
time it obviates the necessity to make use of the bins in the complicated
fashion of grouping them together into larger bins called stages for
clinical use as dictated by the TNM classification.
We only pursued the four variables in the TNM system (age, neck
lymph node involvement, tumor size/extrathyroidal invasion, and
distant metastases at diagnosis) and two other commonly employed
variables (histopathology and sex). It has been suggested that of these
six, the most important prognostic variables are age (cutoff value, 45 to
50 years) and extracapsular invasion of the thyroid gland.13 These two
factors have also been found previously to be associated with a poor
clinical outcome in several commonly used staging systems.7,12,14-19
Similarly, we found that tumor size/extension was most significant,
followed by age and node status, the latter two being equally important. Other risk factors have been proposed over the years such as
tumor grading20, DNA ploidy,21 nuclear atypia,22 gross lymph node
involvement,23 microscopic and macroscopic tumor multifocality,24
histologic papillary thyroid cancer variants,25 size of lymph node
Table 6. Comparative Validation of TNM Staging for Differentiated Thyroid
Cancer Mortality Using Conditional Logistic Regression on Matched Cases
(died) and Controls (survived) Sampled Using Incidence Density Sampling
Matching Variables
Onlyⴱ
No.
Stage Patient Control Odds Ratio
I
II
III
IV
32
101
288
161
78
218
231
48
Ref
0.66
1.21
1.89
Other Variables†
95% CI
Odds Ratio
95% CI
0.33 to 1.28
0.63 to 2.32
0.9 to 4.0
Ref
0.67
1.21
1.88
0.34 to 1.31
0.63 to 2.32
0.9 to 4.0
NOTE. The hazard is the odds of being a patient.
ⴱ
Odds ratio adjusted for only matching variables (sex, survival time in
completed years, and 10-year calendar periods of diagnosis).
†Odds ratio adjusted for extent of surgery (total v other procedures) in
addition to matching variables.
1876
© 2009 by American Society of Clinical Oncology
metastases,26 and tumor angioinvasion.27 However, due to lack of
widespread acceptance they have not been utilized routinely in clinical practice.
Sex and histopathologic status were found to be the weakest
prognostic indicators. Indeed, in univariate analyses, sex retained only
borderline association with recurrence while histopathologic status
retained borderline association with death. In the multivariate model,
histopathologic status carried the lowest weight and had to be forced
into the DFS model. This could be because patients with follicular
thyroid cancer present at a higher stage, and thus in the univariate
analysis there was a significant difference but not in the adjusted
model. Histopathologic status was retained nevertheless as a risk factor
in the model because there is reasonable evidence for its use as a
prognostic variable,28 and while confounding by presentation at a
higher stage is an issue, the fact that this happens suggests that the two
tumor types behave differently in terms of aggressiveness.29 Indeed,
others suggest that after adjusting for TNM stage, patients with follicular thyroid cancer have a 40% higher risk of dying from DTC
compared with those patients diagnosed with papillary thyroid
cancer.28 Sex, however, was not included in our model. Even
though thyroid disease is more common in women, sex has not
been demonstrated to have direct implication in the overall outcome in patients with thyroid cancer.30,31
In this study, nodal metastases had a significant association with
DFS and CSS. The prognostic impact of regional lymph node metastases is still a controversial issue,32,33 and according to the recent TNM
classification, patients with a minimal extrathyroidal tumor extension
and/or LN metastases belong to the same stage I group when age
younger than 45 years and stage III or IV otherwise.9 However, the
respective impact on prognosis of lymph node metastases and minimal extrathyroidal tumor extension has not been properly documented. A recent nested case-control study suggests that patients
with locoregional spread are more likely to die of DTC while the
number of lymph node metastases do not appear to influence survival.28 We found a stronger relationship with recurrence than with
death, suggesting that while it is associated with both, the former is
more significant.
Distant metastases did have a significant association with both
recurrence and death in this study, but was not included in the multivariate model. It is well known that thyroid cancer with distant metastases at the time of diagnosis has an adverse effect on survival.28,34
However, this may itself be related to other factors, such as age, which
not only influence long-term survival in patients with existing distant
JOURNAL OF CLINICAL ONCOLOGY
Downloaded from jco.ascopubs.org on June 9, 2014. For personal use only. No other uses without permission.
Copyright © 2009 American Society of Clinical Oncology. All rights reserved.
Quantitative TNM in Thyroid Cancer
metastases but also predict the development of such metastases.35
Indeed, when distant metastases was forced into the model, the significance of age decreased confirming what we already know from
TNM—age influences the impact of distant metastases in DTC. Also,
patients with distant metastases at diagnosis must be treated aggressively and as such a staging system may not be relevant if a patient has,
for example, pulmonary metastases identified on the radioiodine scan
after total thyroidectomy. Finally, as distant metastases at diagnosis is
relatively uncommon and routinely treated aggressively, systems with
or without this factor did not differ in terms of risk stratification.
The MACIS staging classification has previously been reported to
result in a risk category for the two intermediate groups to be too
broad, and in the case of the TNM classification, stages I and II
overlapped considerably in terms of risk definition.13 We also found a
similar trend for DFS (Table 4) with both MACIS and the TNM
system, again suggesting that three risk groups are possibly the optimum number when stratifying DFS risk for DTC. In the QTNM,
quantitatively measured risk was divided into three equal groups
based on the assumption that quantitative risk is a continuum and
thus does not require binning where risk groups need to be chosen. In
this respect, the QTNM, with three risk categories, seems more appropriate than these systems. When we compare DFS with CSS with our
QTNM risk stratification system, low- and intermediate-risk groups
come closer for CSS as compared with DFS. This is probably a reflection of the fact that prolonged survival is possible even after disease
recurrence and hence does not confer a major survival disadvantage.
Therefore, while three risk categories seem optimal for DFS, two
might be more optimal for CSS. With the QTNM, predominantly
high- and low-risk groups emerge for CSS (Fig A1), and this may be a
REFERENCES
1. Burke HB, Henson DE: The American Joint
Committee on Cancer: Criteria for prognostic factors
and for an enhanced prognostic system. Cancer
72:3131-3135, 1993
2. Burke HB: Outcome prediction and the future
of the TNM staging system. J Natl Cancer Inst
96:1408-1409, 2004
3. Hermanek P, Sobin LH, Wittekind C: How to
improve the present TNM staging system. Cancer
86:2189-2191, 1999
4. Lundgren CI, Hall P, Dickman PW, Zedenius
J: Influence of surgical and postoperative treatment
on survival in differentiated thyroid cancer. Br J Surg
94:571-577, 2007
5. Greenlee RT: Measuring disease frequency in
the Marshfield Epidemiologic Study Area (MESA).
Clin Med Res 1:273-280, 2003
6. Wilke RA, Berg RL, Peissig P, et al: Use of an
electronic medical record for the identification of
research subjects with diabetes mellitus. Clin Med
Res 5:1-7, 2007
7. Hay ID, Bergstralh EJ, Goellner JR, et al:
Predicting outcome in papillary thyroid carcinoma:
Development of a reliable prognostic scoring system in a cohort of 1779 patients surgically treated at
one institution during 1940 through 1989. Surgery
114:1050-1057, 1993; discussion 1057-1058, 1993
8. Shaha AR: TNM classification of thyroid carcinoma. World J Surg 31:879-887, 2007
9. Wada N, Nakayama H, Suganuma N, et al:
Prognostic value of the sixth edition AJCC/UICC
TNM classification for differentiated thyroid carciwww.jco.org
reflection of more complete surgery and postsurgery radioiodine therapy improving survival in the intermediate-risk groups, as suggested
by univariate analysis (Table 3). This interaction of treatment parameters with mortality as opposed to recurrence needs to be elucidated in
future research.
In conclusion, the QTNM is a simple and effective method of
prognostic stratification for both recurrence and cancer-specific mortality. There is no loss of discrimination when compared with current
systems, including its parent TNM staging system, despite being simpler and less complicated. Substantiation of this system in other settings is recommended.
AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS
OF INTEREST
The author(s) indicated no potential conflicts of interest.
AUTHOR CONTRIBUTIONS
Conception and design: Adedayo Onitilo, Suhail Doi
Administrative support: Adedayo Onitilo, Jessica Engel
Provision of study materials or patients: Adedayo Onitilo, Jessica Engel,
Catharina Ihre Lundgren, Per Hall
Collection and assembly of data: Adedayo Onitilo, Jessica Engel,
Catharina Ihre Lundgren, Per Hall
Data analysis and interpretation: Lukman Thalib, Suhail Doi
Manuscript writing: Adedayo Onitilo, Catharina Ihre Lundgren, Per
Hall, Lukman Thalib, Suhail Doi
Final approval of manuscript: Adedayo Onitilo, Jessica Engel, Catharina
Ihre Lundgren, Per Hall, Lukman Thalib, Suhail Doi
noma with extrathyroid extension. J Clin Endocrinol
Metab 92:215-218, 2007
10. Passler C, Prager G, Scheuba C, et al: Application of staging systems for differentiated thyroid
carcinoma in an endemic goiter region with iodine
substitution. Ann Surg 237:227-234, 2003
11. Lang BH, Lo CY, Chan WF, et al: Staging
systems for papillary thyroid carcinoma: A review
and comparison. Ann Surg 245:366-378, 2007
12. Beenken S, Roye D, Weiss H, et al: Extent of
surgery for intermediate-risk well-differentiated thyroid cancer. Am J Surg 179:51-56, 2000
13. Jukkola A, Bloigu R, Ebeling T, et al: Prognostic factors in differentiated thyroid carcinomas and
their implications for current staging classifications.
Endocr Relat Cancer 11:571-579, 2004
14. Byar DP, Green SB, Dor P, et al: A prognostic
index for thyroid carcinoma. A study of the
E.O.R.T.C. Thyroid Cancer Cooperative Group. Eur J
Cancer 15:1033-1041, 1979
15. Cady B, Rossi R: An expanded view of riskgroup definition in differentiated thyroid carcinoma.
Surgery 104:947-953, 1988
16. DeGroot LJ, Kaplan EL, McCormick M, et al:
Natural history, treatment, and course of papillary
thyroid carcinoma. J Clin Endocrinol Metab 71:414424, 1990
17. Mazzaferri EL, Jhiang SM: Long-term impact
of initial surgical and medical therapy on papillary
and follicular thyroid cancer. Am J Med 97:418-428,
1994
18. Shaha AR, Loree TR, Shah JP: Intermediaterisk group for differentiated carcinoma of thyroid.
Surgery 116:1036-1040, 1994; discussion 10401041, 1994
19. Lerch H, Schober O, Kuwert T, et al: Survival
of differentiated thyroid carcinoma studied in 500
patients. J Clin Oncol 15:2067-2075, 1997
20. Hay ID, Grant CS, Taylor WF, et al: Ipsilateral
lobectomy versus bilateral lobar resection in papillary thyroid carcinoma: A retrospective analysis of
surgical outcome using a novel prognostic scoring
system. Surgery 102:1088-1095, 1987
21. Pasieka JL, Zedenius J, Auer G, et al: Addition
of nuclear DNA content to the AMES risk-group
classification for papillary thyroid cancer. Surgery
112:1154-1159, 1992; discussion 1159-1160, 1992
22. Akslen LA: Prognostic importance of histologic grading in papillary thyroid carcinoma. Cancer
72:2680-2685, 1993
23. Noguchi S, Murakami N, Kawamoto H: Classification of papillary cancer of the thyroid based on
prognosis. World J Surg 18:552-557, 1994; discussion 558, 1994
24. Sherman SI, Brierley JD, Sperling M, et al:
Prospective multicenter study of thyroid carcinoma
treatment: Initial analysis of staging and outcome:
National Thyroid Cancer Treatment Cooperative
Study Registry Group. Cancer 83:1012-1021, 1998
25. Sebastian SO, Gonzalez JM, Paricio PP, et al:
Papillary thyroid carcinoma: Prognostic index for
survival including the histological variety. Arch Surg
135:272-277, 2000
26. Sugitani I, Kasai N, Fujimoto Y, et al: A novel
classification system for patients with PTC: Addition
of the new variables of large (3 cm or greater) nodal
metastases and reclassification during the follow-up
period. Surgery 135:139-148, 2004
27. Yildirim E: A model for predicting outcomes in
patients with differentiated thyroid cancer and model
© 2009 by American Society of Clinical Oncology
Downloaded from jco.ascopubs.org on June 9, 2014. For personal use only. No other uses without permission.
Copyright © 2009 American Society of Clinical Oncology. All rights reserved.
1877
Onitilo et al
performance in comparison with other classification
systems. J Am Coll Surg 200:378-392, 2005
28. Lundgren CI, Hall P, Dickman PW, et al:
Clinically significant prognostic factors for differentiated thyroid carcinoma: A population-based, nested
case-control study. Cancer 106:524-531, 2006
29. Passler C, Scheuba C, Prager G, et al:
Prognostic factors of papillary and follicular thyroid
cancer: Differences in an iodine-replete endemic
goiter region. Endocr Relat Cancer 11:131-139,
2004
30. Shaha A: Treatment of thyroid cancer based
on risk groups. J Surg Oncol 94:683-691, 2006
31. Pelizzo MR, Boschin IM, Toniato A, et al:
Papillary thyroid carcinoma: 35-year outcome and
prognostic factors in 1858 patients. Clin Nucl Med
32:440-444, 2007
32. Sato N, Oyamatsu M, Koyama Y, et al: Do the
level of nodal disease according to the TNM classification and the number of involved cervical nodes reflect
prognosis in patients with differentiated carcinoma of the
thyroid gland? J Surg Oncol 69:151-155, 1998
33. Beasley NJ, Lee J, Eski S, et al: Impact of
nodal metastases on prognosis in patients with
well-differentiated thyroid cancer. Arch Otolaryngol
Head Neck Surg 128:825-828, 2002
34. Haq M, Harmer C: Differentiated thyroid carcinoma with distant metastases at presentation:
Prognostic factors and outcome. Clin Endocrinol
(Oxf) 63:87-93, 2005
35. Clark JR, Lai P, Hall F, et al: Variables predicting distant metastases in thyroid cancer. Laryngoscope 115:661-667, 2005
■ ■ ■
1878
© 2009 by American Society of Clinical Oncology
JOURNAL OF CLINICAL ONCOLOGY
Downloaded from jco.ascopubs.org on June 9, 2014. For personal use only. No other uses without permission.
Copyright © 2009 American Society of Clinical Oncology. All rights reserved.