CLIN.CHEM. 32/5, 792-796 (1986)
Two MethodsComparedfor MeasuringLD-liTotaJLD Activityin Serum
Larry H. Bernstein and Patricia Scinto
We presentevidence for the utilityof an improvedassay for
the activity of lactate dehydrogenase (EC 1.1.1.27) isoenzymes 1 and 2 in serum, involvinginhibitionof the H-subunit
of LD by pyruvate at pH 7.1. Resultscorrelate well with the
LD-1/total LD ratio as evaluated by immunologicalassay
and, as an indexto infarct, the method is superiorto either
the change in CK-MB activityor to the LD-1 activity or to a
combinationof these tests, as is the percentage of LD-1 to
total LD activity. Moreover, the percentage inhibition ofLD
activity by pyruvate may have an advantageover other
methods of isoenzyme fractionation because of its smaller
populationCV for patients with acute myocardialinfarction
than is true of other methods. We also demonstrate how,
usinga lineardiscriminantanalysis,we comparedthis method with alternative methods.We determinedthat evaluation
of CK-MB isoenzymecontributesno informationin additionto
that obtainedfrom the LD-1 isoenzyme.
AdditIonal Keyphrases: LD-lltotal LD ratio vs percent inhibition of
lactate dehydrogenase by pyruvate - myocardial infarct
atlne kinase isoenzyme MB
from this laboratory
(1-3) have indicated
that
dehydrogenase
(EC 1.1.1.27) isoenzyme-1
(LD-1)
activity in serum can be determined
by using an assay that
exploits the superiority
of the pyruvate-to-lactate
reaction
(4) that is optimized at two concentrations
of pyruvate.’
LD-1 activity has
Moreover, the importance of determining
been reinforced by the observation that it increases
during
the first 12 h after myocardial
infarction
and even before
creatine
kinase
(CK; EC 2.7.3.2) isoenzyme MB activity
reaches its maximum (5-8), although LD-1 currently is not
relied upon as much as CK-MB is for confirming the
diagnosis
of acute myocardial
infarction.
That the importance of measuring
LD-1 isoenzyme
has
been underestimated
and that its role is not clear (7) is
illustrated
by studies such as the one supporting
measurement of LD-1 activity by the a-hydroxybutyrate
dehydrogenase (HBD) method, along with CK-MB, on two consecutive days (9), thus minimizing
the importance of the assay or
its timing because the combination of tests was considered
adequate
for clinical use. One may then ask whether the
adequacy of CK-MB assay perceived by clinicians accounts
for a different attitude toward LD-1 isoenzyme being held by
clinicians
and by clinical chemists.
Jablonsky
et al. (7)
called attention to a lack of clinical interest in LD-1 isoenzyme because of misperceptions
among cardiologists
about
the methodological
and clini#{235}al
status of this test, but it is
not clear from reviews (10,11) whether the dissatisfaction
is
related to problems of the logistics of performing
the evaluaStudies
lactate
Department of Pathology, Bridgeport Hospital, Bridgeport, CT
06610.
‘Nonstandard abbreviations: U), lactate dehydrogenase; CK,
creatine kinase; AMI, acute myocardial infarction; INH, inhibition;
HBD, a-hydroxybutyrate
dehydrogenase.
Received July 31, 1985; accepted January
792
31, 1986.
CLINICALCHEMISTRY,Vol. 32, No. 5, 1986
tion and reporting results; difficulty in interpreting
a high
cutoff value, resulting in a specificity that is not clinically
useful (isoenzyme-1/isoenzyme-2
ratio); or to analytical
measurements
that are clinically unacceptable
or not acceptable as compared with CK-MB. The accuracy of measuring the percentage of LD-1/total LD or the LD-1/LD-2 ratio
for the early detection and diagnosis of acute myocardial
infarction (AMI) is well documented (6, 7, 12). However, a
misperception about its accuracy does not itself appear to be
the major reason for disinterest in LD-1.
Despite numerous studies defining the efficacy of CK-MB
and LD-1 methods, there still is no consensus concerning the
advantage of any approach over another, except for the fact
that electrophoresis
is widely used for CK-MB or LD-1
analyses (13). In addition, when the isoenzyme-1/isoenzyme2 ratio is measured
by electrophoresis
it is interpreted by
use of an inappropriate standard (12). Results by column
chromatography
(14) and electrophoresis
have been comparable, but there have been continuing reports of assays by
direct spectrophotometric
kinetic assays (15,16). Procedures
for separating
LD-1 have been based on differences in
physical or kinetic properties of the H- and M-type LD, such
as heat denaturation,
stability of urea (17), inhibition by
oxainate or oxalate (18), differences in Michaelis constant
(19-22), and differences in the immunological
properties of
the LD isoenzymes (20, 23), The principal advantage
of
kinetic methods for determining
U) isoenzyme content is
their rapidity, their inexpensiveness,
and their adaptability
to automated instruments.
Here we report on further studies in which we measured
the H-subunit content of LD activity in serum (2), comparing this method with the immunological
assay for LD-1 (23).
The method involves two LD assays under different conditions, and does not require separation of the LD-1 fraction
before assay. We also construct functions, using the results
of CK-MB and LD-1 concentrations
measured in serum, to
patients with and
determine their efficiency in classifying
without infarcts, and we rank these tests according to the
contribution of each as single or joint predictors in classifying patients.
Materials and Methods
Enzyme Assays
Isoenzymes were quantified with kits for immunological
inhibition of CK-MB (SmithKline
Beckman Corp., Brea,
CA) and for immunological
assay of LD-1 (Isomune; Roche
Diagnostics, Nutley, NJ). These were compared with the
assay of LD-1 activity by the method of optimized pyruvate
inhibition (2) with kits manufactured
for Bridgeport Hospital by Diagnostic Chemicals, Ltd, Monroe, CT. Human H4,
H3M, and M4 isoenzymes of U) were purchased from Sigma
Chemical
Co., St. Louis, MO.
We carried out serial CK-MB analyses at 6-h intervals
during the first 24 h after admission of the patient and LD-1
analyses at the peak of CK-MB activity or 6 h thereafter as
previously described (8). For immunochemical
LD-1 assay,
performed at 30#{176}C,
we used a Cobas-Bio centrifugal analyz-
er (Roche Instruments,
Nutley, NJ).
LD-1 inhibition was at pH 7.1 and 30 or 37 #{176}C
by the
method of Bernstein and Everse (2), under the following
conditions. Total U) activity is measured in a 0.34 mmol/L
solution
of the oxidized substrate.
This is followed by a
second assay in a reagent
containing, per liter, 02 mmol of
NADH and 5 mmol (30 #{176}C)
or 5.5 mmol (37 #{176}C)
of pyruvate,
respectively. The NADH concentration
was increased (from
0.14 mmol/L) for the clinical studies to eliminate substrate
depletion at high LD activities, and the standard curve for
the reaction is linear to >500 U/L. The modification
of
pyruvate concentration
at 37#{176}C
is an empirically
determined adjustment
for the increased activity at the higher
4
I-
a
temperature.
Subjects
The 85 patients in this study were admitted to Bridgeport
Hospital for diagnosis or exclusion of AMI. The diagnoses
were established by clinical presentation,
electrocardiography, serial enzyme measurements,
and additional catheterization and angiographic
studies as required.
After grouping the patients into two categories on the
basis of a detailed study of the medical records, we carried
out data analyses, using the “Statistical
Package for the
Social Sciences.” The mean age of the study population
was
well over 60 years, but congestive
heart failure was not a
common complication,
although it is the most significant
factor determining
which patients will require an extended
stay in the hospital. In the absence of arrhythinias
these
patients’ cases are considered uncomplicated
unless pulmonary infiltrates appear in their chest roentgenograms,
appropriate clinical findings are present, and (or) they have
decreased ejection fractions, any of which would lead to their
condition being classified as in Killip Class LU or IV. LD-5 is
massively increased in patients who develop cardiogenic
shock as a complication of AM!, but there were no patients
in Killip Class IV in this study, which would have been
complicated by any effect of severe congestive heart failure
on the LD-1 isoenzyme in their serum. We carried out basic
statistics to measure central tendency, variation, and skewness, and plotted histograms
by group for each variable.
This was followed by one-way analyses of variance and
analyses,
to determine
the best model for
discriminant
predicting
membership
in either the AMI or non-AM!
groups.
pH
I..
I-
a
a
p.
a
a
a
S.
PYIUVATI
(SS,IIL)
0.876) by the equation:
%INH = 25.0 + 0.61378 (% LD-1)
Results
Figure la, a plot Of Km for pyruvate vs pH for both the H4
and M4 isoenzyme of U), shows the decreasing affinity of U)
for pyruvate with increasing pH with an optimum pH at 7.
Figure lb is a plot of the percentage of maximum activities
of H4 and M4 in human LD as a function
of pyruvate
concentration at pH 7.0. On the basis of these data, we chose
a pair of pyruvate concentrations for assay for U) isoenzyme
composition, utilizing the pH-dependent inhibition of U)
activity by pyruvate
to distinguish
the activities of the
isoenzymes at defined pH and temperature.
Measuring U)
activity after the inhibition of U) activity of human 1-14,
H3M, and M4 in different combined ratios gives a parabolic
relationship
between the percent H-subunit
activity and
percent inhibition (r = 0.8957).
We carried out further studies of the 32 AMI and 53 nonAM! patients to compare the percentage of LD-lltotal U)
activity and percent inhibition of U) activity (percent INH).
The percent INN and percent LD-l are correlated
(r =
cQCINTATION
Fig. 1a Plot of ! in millimolesof pyruvate per litervs pHforH4(dosed
squares) and M4 (dosed dries) W isoenzymesof man; b. Plot of
percentma,dmumactivityof H4 (‘en squares) and M4(en dries)
vs pyruvateconcentration in millimoles per liter
LDH4and M4 isoenzymes
were providedby J.Everse afteraffinitych,omatographyand crystallinepuflficallon
of AM! and non-AM! patients by assay at
The comparison
30#{176}C
for inhibition of U) activity in the presence of 5.0
mniol of pyruvate per liter showed the population coefficient
of variation for percent inhibition to be considerably lower
than for percent U)-l isoenzyme, as follows: for AM!, % LD1 = 2626%, %INH = 6.93;for non-AM!, % LD-1 = 33.85%,
%INH
16.37%. Similarly, when the assay was modified
for assay at 37 #{176}C
(55 mmol of pyruvate
per liter) we
obtained an overall population CV of 50.22% for percent LD1 as compared with 25.98% for percent INN.
In Table 1 we compare the patients’ results obtained by
inhibiting
U) activity at 37#{176}C
by use of 5.5 mmol/L
pyruvate
with the measurements
of CK-MB activity and
percentage of LD-1/total
U) activity
in the AM! and the
non-AM! groups. The standard deviation for the populations
was higher for CK-MB and U)-1 activity than for either
percent LD-l or for percent INN. This proved to be important in our other studies.
=
CLINICAL CHEMISTRY, Vol. 32, No.5, 1986
793
Table 1. PercentLD-1as Measuredby Two Methods
Diagnosis
CK-MB,
U/L
Age, yr
Trauma
%LD.1
12
40
CHFN. FIBC
62
3
22
Cirrhosis,
46
43
36
50
8
24
18
48
22
39
65
2
56
30
335
79
11
22k’
56
73
78
21
47
21
50
38
39
65
170
we ob-
Additionally, the change in CK-MB was found to contribute
no additional predictive value in distinguishing
the AM!
and non-AM! groups, mainly because of shared variance
with the other predictors.
The results clearly show the discriminating
power of
15
30
18
71
However,
47
13
14
thrombocytopenia
Angina
Angina
AM!
by pyruvate.
tamed the best separations
in any combination that included U).1/total
LD activity, even using inhibition of subunit
content
of U) activity for proportion
of LD-l activity.
24
Angina
and its inhibition
hl
Ion
20
52
36
Gunshot
Trauma
activity
47
52
the
percent U)-! and reinforce the principle of measuring
total U) activity whenever isoenzyme-l
U) activity is
obtained. Although the results show that the discriminating
power of percent INH with U) activity measured with
pyruvate as substrate is not as great as that of percent LD-1,
the separation is more than adequate.
48
Discussion
CHFC
56
59
43
179
45
75
73
75
54
37
57
38
56
57
60
56
60
#{149}ij:)
increase to 33% representsearly evoMng AMI. bAnterolateral AMI
with absent CK-MB and LD.1 changes. Total LD activity (0.34 mmot/L:P-”L)
increased. cngestlve
heart failure.
In addition to basic statistics, we carried out one-way
analyses of variance and established
that the differences
between the two groups, AM! and non-AM!, were statistically significant (p <0.01) for all of the dependent variables:
U)-l, percent U)-l, and percent NH. The F-value for
percent U)-1 was the highest at 160.6. The correlation
matrix for these predictors showed the highest correlation
and U) activity (pyruvate);
LD-l
between: U)-l activity
and percent U)-1; and percent U)-l and percent !NH. We
also conducted
discriminant
analyses, with results significant at the 0.01 level.
In Table 2 we compare results of discriminant
analyses
that best classify patients, using the candidate predictors.
We entered the tests one at a time to determine combinations of the fewest predictors that would result in low
misclassification
rates and to determine the contribution of
each predictor to the classification functions. The very high
eigenvalue, high canonical correlation, high chi square, and
low Wilk’s lambda for the percent LD-l shows good separation with this single predictor variable, somewhat better
than the results obtained with a combination
of total U)
Table 2. Discrlmlnant Analyses of AMI and Non-AMI
Elgsn-
CanonIcal
value
corr.
Wilk’s
lambda
CliP
%INH
1.99
0.82
0.33
78.9
%LD-1
1.90
0.81
0.34
77.2
1.25
0.74
0.44
58.6
0.93
0.69
0.52
474
Varisbl.s
No. 1
%LD-1
No.2
No.3
LD-
pyruvate
%INH
No.4
%INH
LD-1
794
CLINICALCHEMISTRY, Vol. 32, No. 5, 1986
We conclude that measuring LD-l either as percentage of
isoenzyme-l/total
U) (6) or as percent inhibition
of Hsubunit activity is superior to measuring the activity of U)1 or CK-MB, either individually or in combination,
an
important
conclusion in view of the current reliance on
serial CK and CK-MB isoenzyme measurements
in many
institutions. Indeed, the findings make it extremely difficult
to explain or justify the declining use of U) isoenzyme-1
evaluation for the diagnosis of AM! (7). Perhaps there are
factors related to untimely reporting rather than accuracy of
laboratory results that have provoked this disinterest in LD1. Moreover, there is a need for early recognition of AM! in
to limit the progression
of
order to initiate treatments
infarct and avoid late complications. This has created greater interest in excluding non-AM! patients who have equivocal CK-MB changes.
There is a lack of agreement
about the best method for
measuring LD-l, but the studies of Leung and Henderson
(12) provide some explanation
for the popularity of electrophoresis over all other methods. The value of electrophoretically separating CK-MB and LD-l may be compared with
that of measuring
CK-MB immunochemically
combined
with HBD on the first and second day (9), but the better
efficiency of such an approach has not adequately been
validated with respect to errors of sensitivity estimations
(25). Nevertheless,
the study (9) supports measuring
the
activity of LD-l after 12 hand kinetically, mainly because it
is used with CK-MB isoenzyme in addition to electrocardiographic and clinical criteria. Moreover, the study (9) places
greater importance
in the value of LD-1 used for a late
marker for AM! than in the method of the U)-l assay.
We present evidence for the adequacy of an U)-l assay
that is nearly as accurate as the immunological
LD-1 assay
but is as inexpensive as the HBD assay. The advantages of
this approach have not been previously identified. Other
assays have recently been reported for measuring the activities of U)-! and LD-2 isoenzyme by automated
kinetic
procedures (15, 16). While we do not report as high a
correlation with inhibition by pyruvate and U)-1/total U)
determined
by immunological
assay as that reported for
alkaline inactivation,
the manipulation
required in the
latter method is as great as that required for immunological
assay. We here substantiate
by its clinical performance the
use of substrate inhibition for measuring LD-1 activity. This
kinetic method for determining
the isoenzyme contents of
activity is based on the fact that H-type U) is significantly inhibited by high concentrations
of pyruvate
(1)
associated
with the formation at pH 7 of an abortive
complex between enzyme, NAD, and pyruvate. This assay
takes advantage of the known superiority of the pyruvateto-lactate reaction (3), and it is simpler than those recently
reported
(15, 16).
The comparison of the pyruvate-inhibition
assay with the
iinniunological
assay for U)-1 fraction shows a slight superiority of the latter despite the lower CVs of the inhibition
method within the study populations. The differences between these methods are largely found in examining the
means and variances between the AM! and non-AM! populations. These differences may not be substantiated
in
studies with a larger population. The measurement
of LD-1
by inhibition
of U) subunit activity is dependent on a set of
variables that includes the differences in Km for pyruvate of
the H and M subunits. The means of LD isoenzyme inhibition in non-AM! and patients with small AM! tend to cluster
in a range of between 45 and 48% inhibition. Comparable
results in the U)-1 immunological
assay would be consistently less than 29% of the total U) activity. This would be
similar to the population with an isoenzyme-1/isoenzyme-2
ratio in a range from 0.69-0.85 by electrophoresis,
currently
under investigation
by Henderson and coworkers (7, 12).
For this group it becomes very difficult to assess the
accuracy of these methods because of the limited size of the
subpopulation
under investigation.
However, reduction of
the population variances becomes a desirable feature in
determining
their sensitivities and for comparing the performance of these methods in the clinical subpopulation of
borderline AM! patients. The method of inhibition of subunit content, though it is biased in estimating
the H-type
U) activity because it is affected by the presence of U)-2
isoenzyme, does appear to accurately measure differential
increases in LD-1 isoenzyme activity in serum.
What significance
is there to the distinct and subtle
differences between the immunological
and the enzymatic
assay for LD-1 isoenzyme composition? The results suggest
that the significance is more apparent than real, as judged
from the extremely
good efficiency obtained with either
method. The apparent problem of concern with either method has been the issue of dealing with U)-! activities in the
presence of congestive heart failure. This only applies to the
patient with Killip Class ffl-IV with reduced ejection fractions and severe circulatory
impairment
associated with
increased U)-5 activity in cardiogenic shock or with increased U)-2 and-3 isoenzyme in some patients with pulmonary edema. When patients are found to have a total U)
activity that is out of proportion to the U)-1 activity by any
measurement,
they are usually found to have an increased
LD-5 activity associated with generalized hypoperfusion.
In these patients there is no need for a superior LD-1
assay because of the likelihood that there are electrocardioconsistent
with AM! with
graphic and clinical findings
massive
increase in CK-MB isoenzyme. The HBD would
also be high because of massive increases in U) activity in
the serum, originating not only from the heart but also from
the liver. The most important
clinical problem for the
patient is management
of cardiogenic shock.
The present studies coniirm the validity and discriininating power of LD-1 subunit content measured
as percent
inhibition of U) activity by pyruvate. More importantly,
they demonstrate how we can evaluate these tests and
of each to a vector of observations
establish the contribution
in forming a joint predictor for classifying patients. We
found that, by comparison with LD-1 activity as percent of
U) activity or as percent inhibition,
CK-MB and U)-1
activities provide no additional discrimination
beyond that
of percent inhibition because of the large population variance of enzyme activities in the AM! and non-AM! populations. Finally, we think that the more rapid and accurate
analyses of LD-1 here demonstrated
may contribute to the
improved treatment of AM! because of the current emphasis
on more rapid diagnosis
and treatment
to reduce late
complications,
especially in those patients who have small
infarcts with equivocal changes in the concentration of CKMB activity in their serum.
total
We appreciate the competent statistical assistance provided by
Dr. Susan Carroll, of Words and Numbers, Torrington, CT. Human
IL, and M4 purified by affinity chromatography was kindly provided
for determining assay conditions by Professor Johannes Everse,
Department of Biochemistry, Texas Tech University Health Sciences Center, Lubbock, TX.
References
1. Bernstein LH, Everse J, Shioura N, Russell PJ. Detection of
cardiac damage using a steady state assay for lactate dehydrogenase isoenzymes in serum. J Molec Cell Cardiol 1974;6:297-315.
2. Bernstein LH, Everse J. Determination of the isoenzyme levels
of lactate dehydrogenase. In: Wood WA, ed. Carbohydrate metaboham. New York, NY: Academic Press, 1975;47-53. (Colowick SP,
Kaplan NO, eds. Methods in Enzymology, Vol XLI, Part B).
3. Bernstein LII. Automated kinetic determination of lactate dehydrogenase isoenzymes in serum. Clin Chem 197723:1928-30.
4. Howell BF, McCune S, Schaffer R. Lactate-to-pyruvate or pyruvate-to-lactate assay for lactate dehydrogenase: A reexamination.
Clin Chem 1979;25:269-72.
5. Roe CR. Diagnosis of myocardial infarction by serum isoenzyme
analysis. Ann Clin Lab Sci 1977;7:201-9.
6. Bruns DE, Emerson JC, Intemann S, et al. Lactate dehydrogenaae isoenzyme-1: changes during the first day after acute myocardial infarction. Cliii Chem 1981;27:1821-3.
7. Jablonsky G, Leung FY, Henderson AR. Changes in the ratio of
lactate dehydrogenase isoenzymes 1 and 2 during the first day after
acute myocardial infarction. Clin Chem 1985;31:1621-4.
8. Bernstein LH, Reynoso G. Serum lactate dehydrogenase isoenzyme-1: effect of time of sampling and total serum LI) activity on
diagnostic efficacy [Letter]. Clin Chem 1983;29:589-90.
9. Werner M, Brooks SH, Mohrbacher RJ, Wasserman AG. Diagnostic performance of enzymes in the discrimination of myocardial
infarction. Clin Chem 1982;28:1297-1302.
10. Wagner (18. Optimal use of serum enzyme levels in the
diagnosis of acute myocardial infarction. The perspective in 1980.
Arch Intern Med 1980;140:317-9.
11. Roberta R. The two out of three criteria for the diagnosis of
infarction. Is it pass#{233}?
Chest 1984;86:511-3.
12. Leung FY, Henderson AR. Thin-layer agaroee electrophoresis
of lactate dehydrogenase
isoenzymes in serum: a note on the
method of reporting and on the lactate dehydrogenase iaoenzyme1/isoenzyme-2 ratio in acute myocardial
infarction. Clin Chem
1979;25:209-11.
13. Boone J, Sampson &J, Lewis 5, et al. An interlaboratory study
of creatine kinase and creatune kinase isoenzymes. Clin Chem
198026:513-9.
14. Vasudevan G, Mercer DW, Varat MA. Lactate dehydrogenase
in the diagnosis of acute myocardial
isoenzyme determination
infarction. Circulation 1978;57:1055-7.
15. Eckfeldt Jil, Kershaw MJ, Lewis LA. Automated kinetic assay
for lactate dehydrogenase isoenzymes by centrifugal analysis after
alkaline inactivation. Clin Chem 1984;30:1821-4.
16. Tanishima
K, Hayashi T, Matsushima
M, Mochikowa Y.
Activity of lactate dehydrogenase isoenzymea and LD-1 and LD-2 in
serum as determined by using an inhibitor of the M-subunit. Clin
Chem 1985;31:1175-7.
CLINICAL CHEMISTRY, Vol. 32, No.
5, 1986 195
17. Emery AEH. The determination
of lactate dehydrogenase
isoenzymes in normal human muscle and other tissues. Biochem J
1967;105:599-604.
18. Emerson PM, Wilkinson JH. Urea and oxalate inhibition of the
serum lactate dehydrogenases.
J Cliii Pathol 1965;18:803-7.
19. Veeaell ES, Beam AG. Isozymes of lactic dehydrogenase
in
human tissues. J Clin Invest 1961;40:586-91.
20. Plagemann PGW, Gregory KF, Wroblewaki F. The electrophoretically distinct forms of mnnimnlian
lactic dehydrogenase
U.
Properties and interrelationships of rabbit and human lactic dehydrogenase isozymes. J Biol Chem 1960,235:2288-93.
Rosalki SB, Wilkinson
dehydrogenase in diagnosis.
21.
JH.
Serum
alpha
hydroxybutyrate
J Am Med Assoc 1964;189:61-6.
196 CLINICAL
CHEMISTRY,Vol.
32, No.5, 1986
22. Lane RS, Dekker EE. 2-Keto-4-hydroxybutyrate.
Synthesis
chemical properties, and as a substrate for lactate dehydrogenase of
rabbit muscle. Biochem 1969;8:2958-66.
23. Usategui-Gomez M, Wicks RW, Warshaw M. Immunochemical
determination
of the heart isoenzyme of lactate dehydrogenase (LI)
1) in human serum. Cliii Chem 1979;25:729-34.
24. Bernstein LII, Reynoso G. Creatine kinase B-subunit activity
in serum in cases of suspected myocardial infarction: a prediction
model based on the slope of MB increase and percentage CK-MB
activity [Letter]. Clin Chem 1983;29:591-2.
25.
Linnet K. Precision of sensitivity estimations in diagnostic test
evaluations. Power functions for comparisons of sensitivities of two
tests. Cliii Chem 1985;31:574-80.