1. health and fitness
2. disease
3. cholesterol

videdthe limitsofprecision are known
and are acceptable.
It has become common practice to
verify quantitative
accuracy and linearity at a threshold concentration
by
analyzing calibrators
or controls prepared at concentrations
one-half and
twice the cutoff value. It should be
appreciated
that use of these samples
provides neither special information
regarding
accuracy “at or near” the
cutoff nor greater confidence in the
results.
The hypothesis
that the midpoint concentration
on a linear calibration
graph is the most accurate
on
the curve when bracketed
by these
specific lower and higher concentrations is not scientifically
supportable.
Within the limits of precision,
all
points on a linear calibration
graph
are equally reliable and accurate,
as
long as the upper and lower limits of
linearity
are known. The number of
calibrators
necessary
and their concentrations
will depend on the methods used and the intended use of the
analytical results. What is important
in forensic applications
in which a
cutoff is defined administratively
is
the precision of the assay at that concentration.
When a series of calibrators are analyzed by immunoassay,
the calibration curve generated
is curvilinear.
The cutoff concentration
should be on
the linear portion of the curve, the
slopeof which permits discrimination
between positiveand negative samples within established statistical limits. Precision
at this portion
of the
calibration
curve is, therefore,
very
important.
This must be established
and assessed with each batch of urine
specimens by the use of controls. A
control with a validated concentration
of the analyte at the cutoff value, another slightly below, and one slightly
above the cutoff (approximately
plus
and minus 3 SD) should be more than
sufficient
todefine the precision of the
assay at the cutoff concentration.
Currently,many analytical
technologies are used to analyze
biological
fluids
Many
for drugs
clinical
made daily that
ical implications
For employees
work and even
affected, positive
noassays
must
and their metabolites.
determinations
are
have substantial
med-
for patients’
health.
conditions
of
whose
liberty are likely to be
test results by immu-
be confirmedby a second, independent
test, such as GC-MS.
These technologies are absolutely dependent on the concept of calibration
linearity,
accuracy,
and precision
throughout
the concentration
range
required
Accurate
by the purpose of the
and precisely
defined
assay.
quan-
titative or qualitative
results are critical if physicians, employers, and lawyers are to make confident decisions.
To consider
that only one small region
or a single concentration
of the calibration graph defines the method as
“accurate” is misleading
and ignores
the tested and established
concept of
multi-point
calibration.
The finaltest
of the “accuracy”of any analytical
technology
is to actually
apply the
method to a large population of specimens and successfully defend the results obtained against scientific and
legal challenges.
These data are now available from
some of the larger contract laboratories that analyze hundreds
of thousands of urine samples for drugs of
abuse each year, using several different analytical instruments
and procedures, but all within the framework
of NIDA and AACC/CAP accreditation guidelines. The indication is that,
when properly calibrated
by either
multi-point,
full-range
curves or by
arithmetic
means at the threshold
concentration,
the methods are all satisfactory for these purposes.
Bryan
S. Finkle
Center for Human Toxicol.
417 Wakara Way, Room 290
Salt Lake City, UT 84108
David Black
Aegis Analytical Labs., Inc.
624 Grassmere Park Rd., Suite 21
Nashville, TN 37211
Robert V. Blanke
Consultant Toxicologist
4222 Croatan Rd.
Richmond,
VA 23235
Thorne J. Butler
4230 South B urn ham Ave., Suite 250
Las Vegas, NV 89119
Graham
R. Jones
Office of the Chief Medical Examiner
P.O. Box 2257
Edmonton, Alberta, Canada T5J 2P4
R. H.
Barry Sample
Sports Med. Drug ID Lab
Univ. Hospital N440
635 Barrhill Dr.
Indianapolis,
IN 46223
Furosemide as a Displacing Agent
In Assay of Total Trilodothyronine
To the Editor:
An
assay
for the
measurement
of
total triiodothyronine
(T3) has been
developed for the Abbott IMx#{174}
Analyzer (1). Reagents
consist of human
serum calibrators, a T3-alkaline
phosphatase conjugate, anti-T3-coated microparticles, and substrate, 4-methylumbelliferyl phosphate.
In the assay, calibrators or samples
are incubated
with the microparticle
reagent,which contains an agent to
displace T3 from serum binding proteins. The reaction mixture is transferred to the glass-fiber matrix and
washed, and then the T3 conjugate is
added. After incubation, the matrix is
washed and the substrate
added. The
rate of formation
of 4-methylumbelliferone (4-mu) is then measured. Total T3 concentration
is inversely related to 4-mu formation.
In preliminary
evaluations,
the displacing
agents
furosemide
(2,3),
8-anilino-1-naphthalene
sulfonic acid
(ANS) (4), and fenclofenac
(5) were
compared.
ANS was rejected
because
of potential
quenching
of 4-mu; fenclofenac,
because
of slightly
inferior
displacement
ofT3 in the IMx system:
62% vs 71% for furosemide.
In determining
the optimal amount
of furosemide
to use, we varied its
concentrationin the microparticle
reagent from 0 to 1 gIL. Furosemide
at
125 mg/L providedmaximum T3 displacement without
interferingwith
the bindingof the T3 conjugate to the
antibody-coated
particles.
At this optimal concentration:
#{149}
T3 at 8 and 100 p.g/L was displaced
85% and 98%, respectively.
#{149}
Mean analytical
recovery
of T3
added at four concentrations
to four
separate
samples was 105% (range
95-112%).
#{149}
Mean analytical recovery of three
separate samples diluted with zero
calibrator
to <0.3
ng/L
(range 96-105%);
original
was 100%
T3 content
was 4.0, 2.6,and 2.0 pg/L.
#{149}
Results obtainedby the IMx assay
(y) compared well with those by two
commercially
availableRLAs (x):
No. of samples
Slope
y.Intercept, p.gIL
r
Amersham
166
1.20
0.10
0.97
Dalnabot
84
1.00
-0.18
0.98
Thus, furosemide appears to be an
excellent displacing agent for measuring total T3 in serum. Presumably,
furosemide would be equally effective
in displacing thyroxin and other analytes bound to similar sites in serum
or plasma.
CLINICAL CHEMISTRY, Vol. 37, No. 4, 1991 587
References
1. Groskopf W, Hsu S, Sohn L. A fully
automated assayfortotal T3 utilizing the
AbbottIMx analyzer [Abstract].
ClinChem
1988;34:1210.
2. Stockigt JR, Lim C-F, Barlow JW, et al.
High concentrations of furosemide inhibit
serum binding of thyroxine. J Clin Endocrinol Metab 1984;59:62-6.
3. Stockigt JR. Lim C-F, Barlow JW, et al.
Interaction of furosemide with serum thyroxine-binding sites: in vivo and in vitro studies
and comparison with other inhibitors. J Clin
Endocrinol Metab 1985;60: 1025-31.
4. Chopra LI, Ho RS, Lam R. An improved
radioiminunoassay of triiodothyronine in
serum: its application to clinical and physiological studies. J Lab Clin Med 1972;
80:729-39.
5. Rataliffe WA, Hazelton JA, Thomson
JA, Ratcliffe JG. The effect of fenclofenac
on thyroid function tests in vivo and in
vitro. Clin Endocrinol 1980;13:569-75.
W. Groskopf
B. Green
L. Sohn
S. Hsu
Abbott Laboratories
Abbott Park, IL 60064
Potential Problem wIth
StandardizatIon of Prealbumin
(Transthyretln) Results In
immunoturbldlmetry
To the Editor:
Recently,
several centers have described immunoturbidimetric
(IT) assays for measuring prealbumin
(1, 2)
in serum
with centrifugal
analyzers.
We have encountered
a potential problem in calibrating
one such method
and interpreting
the results.
We used Dakopatts rabbit anti-human prealbuinin antiserum (no. Q362)
and Dakopatts
Calibrator
(no. X908)
to set up an IT method (3) for the
Cobas-Fara’TM (Roche Analytical
Instruments
Inc., Nutley,
NJ 07110)
centrifugal analyzer. Despite obtaining what appeared to be an excellent
calibration
curve, we found a lower
than expected value for Standard Human Serum (no. ORDT 06/07) from
Behringwerke
(Hoechst UK Ltd., Middlesex TW4 6JH, U.K.). All results are
shown in Table 1. Reversing the situ-
ation and using dilutions
bring
product
to calibrate
of the Bethe method
gave a predictably higher valueforthe
Dako calibrator,
demonstrating
that
this was neither a dilution error nor a
problem with the IT methodology; we
repeated the assay, using M-Partigen
immunodiffusion
plates
(Behringwerke) for estimation
of prealbumin,
Table 1. BehavIor of Prealbumln CalIbrators In Immunoturbldlmetry
CalIbration/control materIal
!mmunoturbidimetry
Behnng standard
serum
Dakopatts calibrator
Manufacturer’s
assigned value,
g/L
Spol
SPO1
Behnng standard serum
Dakopatts calibrator
Calibration
Assay result,
g/L
0.26
Dakopatts
0.16
0.25
0.25
0.25
0.26
Behnng
Dakopatts
Behring
SPO1
0.32
0.26
0.34
0.14
0.25
SPO1
0.26
0.25
0.25
Behring
Behring
0.31
0.33
M-Partigen immunodiffusion
Spol
Dakoplatts calibrator
calibrated with Behring Standard
man Serum.
The Dakopatts
Hu-
calibrator
gave a value of 0.33 g/L, not the manufacturer’s
stated value of0.25 g/L.
Hamlin and Pankowsky
(1) describeda similar discrepancy between
calibrationmaterials
from Beckman
and Behring.They assayed the Beckman calibration
material by iT with
the Behring calibrator for standardination, and obtained a result about
16% higher than the given value.
They also found that, unlike the Beckman calibrator, the Behring calibrator
material
apparently
behaved differently in IT and nephelometric
assays,
giving consistently higher results with
the IT method (22% positive bias).
Which calibration material is correct
and which reference limits should be
taken? When we analyzed a third independent control (no. SPO1) obtained
from the Supraregional Protein Service
(Sheffield, U.K.), its assigned value
was compatible with Dakopatts
calibration and not with Behring Standard
Human
Serum. However, there is no
internationally recognized prealbumin
reference material for comparison.
Reference ranges are another problem. Dakopatts
recommends
the
ranges produced by Behring, namely,
g/L for males and 0.1-0.4 g/L
for females. Clearly, this is nonsense
and the reference range for Dakopatts
should be at least 25% less than these
figures. Perhaps the only safe recourse
is to produce a reference range at the
local level, which has always been
0.2-0.5
good practicein the past.
References
1. Hamlin
CR, Pankowsky DA. Turbidimetric determination of transthyretin
(prealbumin) with a centrifugal analyzer. Clin
Chem 1987;33:144-6.
2. Konstantinides
FN, Mitchell
DR,
Blixby E, et al. Imniunoturbidimetry
of
prealbumin (transthyretin) in a microcentrifugal analyzer [Tech Brief]. Clin Chem
1989;35:178-9.
588 CLINICAL CHEMISTRY, Vol. 37, No. 4, 1991
3. Ledue TB, Rifai N, Irish GR, Silverman
LM. Immunoturbidimetry
of transthyretin
(prealbumin) in human serum [Tech Brief].
Clin Chem 1987;33:1260.
J. Coore
J. Ambler
Dept. of Clin. Chem.
Univ. Hospital, Queen’s Med. Ctr.
Nottingham
NG7 2UH, UJC.
Two spokesmen
spond:
from
Behring
re-
To the Editor:
Drs. Coore and Ambler have identified a difference in standardization between two sources of prealbumin
reagents and some ofits consequences.
We agree that the difference exists
and that such a difference
can cause
difficulties
for users in assessing
published reference ranges and in evaluating results of external proficiency surveys; this situation will continue until
an international
reference preparation
for prealbumin becomes available.
Standardization
of Behringwerke
prealbumin
methods is based on
transfer of values in a self-consistent
manner
from highly
purified protein
preparations
to stable, serum-based
secondary materials suitable for routine use. Such standardization
is controlled-from
lot to lot and across
technologies-to
provide a family of
assay methods meeting the various
demands imposed upon clinical laboratories yet able to produce equivalent
results for patients’ samples (within
the “state of the art”). Publications
describe appropriate
reference ranges
for the various proteins. The reference
range for prealbumin
(1), established
for Behringwerke’s
products, is 0.250.45 g/L forboth males and females.
Where international reference preparations are available, Behringwerke
establishes conversion factors relating
internal standards to international