1719
Clinical Chemistry 46, No. 10, 2000
in-house ELISAs for antineutrophil cytoplasmic
antibodies directed against proteinase 3 and
myeloperoxidase. Pathology 1999;31:38 – 43.
Anthony A.M. Ermens1*
Angelique J.M. Bayens1
Adriënne Crooymans1
Anita A.M. Broekman-van Hout2
Hans L.P. van Duijnhoven2
1
Klinisch Laboratorium
Diaconessenhuis
Postbus 90052
5600 PD Eindhoven, The Netherlands
2
Algemeen Klinisch Laboratorium
Elkerliek Ziekenhuis
5800 AB Helmond, The Netherlands
*Author for correspondence. Fax 31-402335595; e-mail [email protected].
Dilution Protocols for Detection of
Hook Effects/Prozone Phenomenon
To the Editor:
The prozone or (high-dose) hook effect, documented to cause false-negative assay results ⬎50 years ago (1 ),
still remains a problem in one-step
immunometric assays (2–9 ), immunoturbidimetric assays (10 ), and immunonephelometric assays (11 ) for
immunoglobulins. To detect the prozone effect, samples are often tested
undiluted and after dilution (9 ). If
the result on dilution is higher than
for the undiluted sample, then the
undiluted sample most likely exhibited the prozone effect. Unfortunately, this approach increases labor
and reagent costs for assays that may
only rarely encounter extremely high
analyte concentrations. An alternative approach involves pooling patient samples and measuring the
pool and a 10-fold dilution of the
pool (12 ). If one or more of the samples in the pool is falsely low because
of the prozone effect, then the results
from the undiluted and diluted pools
(after correcting for the 10-fold dilution) will differ significantly (12 ).
Other approaches to eliminate the
prozone effect include using twostep immunoassays that have a wash
step between the addition of sample
and labeled antibody (7 ) and the use
of neural network classifier systems
that analyze reaction kinetics (13 ).
Serum immunoglobulins can be
markedly increased in patients presenting with large myeloma tumor
burdens and may lead to falsely low
results in nephelometric assays (11 ).
We combine 50-␮L aliquots from
each of 10 samples to dilute each
sample 10-fold and eliminate any
prozone effect. The concentrations of
IgG, IgA, and IgM in the pool are
measured using a nephelometer
(BNII; Dade Behring, Inc.) and compared with the mean values when all
samples in the pool are analyzed
(calculated value). When the two values for an immunoglobulin differ by
a specified quantity, all samples in
the pool are reanalyzed after a 10fold dilution.
Criteria for detecting the prozone
effect are based on data obtained
from routine samples during a 10day period. Measured immunoglobulin concentrations for 27 pools (10
samples per pool) were compared
with the mean values of samples in
the pools. The range of values for the
measured serum pools and the differences between the measured pool
value and the value derived from the
mean of individually measured samples in the pool (calculated value) for
each immunoglobulin were as follows: IgG, range 10.20-32.50 g/L,
mean difference 4.6%, SD 4.1%; IgA,
range 0.31-17.90 g/L, mean difference 12.6%, SD 8.6%; and IgM, range
0.27-5.96 g/L, mean difference
13.2%, SD 8.2%. The small SD indicated that none of the samples exhibited the prozone effect. A percentage
difference less than the mean plus 2
SD was considered acceptable and
was determined to be 15% for IgG,
30% for IgA, and 30% for IgM. Large
differences were considered suggestive of a prozone effect.
The ability of this approach to
identify samples exhibiting the prozone effect during routine analysis
was evaluated during a 6-month period. Approximately 750 samples/
month were received, and 460 pools
were analyzed. Ten samples from
five different myeloma patients were
identified as being falsely low because of the prozone effect (Table 1).
Four samples were from patients
with IgA myeloma, and one was
from a patient with IgG myeloma.
The discrepancy between the measured and calculated pool was 6288% (initial difference; Table 1).
When the sample generating the erroneous value was identified and the
“correct” result (obtained after dilution) was used in the calculation, the
difference between the measured
and calculated pool was within the
established limits of 30% for IgA and
15% for IgG (corrected difference;
Table 1). The falsely low values differed from the actual results by as
much as 11-fold for IgA and 40-fold
Table 1. Detection of the prozone effect in nephelometric assays for
immunoglobulin (Ig) by monitoring the percentage of difference between
measured and calculated pool values.
Difference between
measured and calculated
pool values,c %
Patient
Myeloma
class
Original Ig
result,a g/L
Diluted Ig
result,b g/L
Initial
difference
Corrected
difference
1
2
3
4
5
IgA
IgA
IgA
IgA
IgG
8.26
3.91
8.46
7.21
3.08
59.00
43.10
47.60
77.90
126.00
64
77
75
88
62
7
16
5
21
2
a
Result was obtained for either IgA or IgG depending on the myeloma class.
Samples were diluted 10-fold before re-analysis.
c
An aliquot (50 ␮L) from 10 samples was combined and assayed. The measured value for the pool was
compared with the sum of the 10 individual measurements divided by 10. The percentage of difference
between the two values is shown.
b
1720
for IgG (Table 1). The prozone effect
is not restricted to IgA and IgG because we identified samples exhibiting this phenomenon when measuring IgM (data not shown).
A 2% incidence (1 of 46 pools) for
the prozone effect when measuring
immunoglobulins may be higher
than at institutions not specializing
in the treatment of multiple myeloma. However, the incidence of
multiple myeloma over the age of 25
is 30 per 100 000 (14 ), and most laboratories will eventually encounter a
sample exhibiting the prozone effect
when measuring immunoglobulins
by nephelometry. Reporting of an
erroneous result can have serious
medical implications, and sample
pooling is a simple method for detecting falsely low concentrations attributable to the prozone effect. Although this screening approach
increases reagent costs by 10% and
involves additional labor to prepare
and analyze pools, it is considerably
more cost-effective than analyzing all
samples undiluted and after dilution,
which doubles reagent costs. Furthermore, this simple prozone detection
method can be adapted to other nephelometric assays with the potential for
erroneous results from antigen excess.
References
1. Landsteiner K. The specificity of serological
reactions. Cambridge, MA: Harvard University
Press, 1946;240-52.
2. Brensing AK, Dahlmann N, Entzian W, Bidlingmaier F, Klingmuler D. Underestimation of LH
and FSH hormone concentrations in a patient
with a gonadotropin secreting tumor: the high
dose “hook effect” as a methodological and
clinical problem. Horm Metab Res 1989;21:
697-8.
3. Haller BL, Fuller KA, Brown WS, Koenig JW,
Evelend BJ, Scott MG. Two automated prolactin
immunoassays evaluated with demonstration
of a high-dose “hook effect” in one. Clin Chem
1992;38:437-8.
4. Petakov MS, Damjanovic SS, Nikolic-Durovic
MM, Dragojlovic ZL, Obradovic S, Gilgorovic
MS, et al. Pituitary adenomas secreting large
amounts of prolactin may give false low
values in immunoradiometric assays. The
hook effect. J Endocrinol Invest 1998;21:
184-8.
5. Flam F, Hambraeus-Jonzon K, Hansson LO,
Kjaeldgaard A. Hydatidiform mole with nonmetastatic pulmonary complications and a
false low level of hCG. Eur J Obstet Gynecol
Reprod Biol 1998;77:235-7.
6. Zweig MH, Csako G. High-dose hook effect in a
two site IRMA for measuring thyrotropin. Ann
Clin Biochem 1990;27:494-5.
7. Vaidya HC, Wolf BA, Garrett N, Catalona WJ,
Letters
8.
9.
10.
11.
12.
13.
14.
Clayman RV, Nahm MH. Extremely high values
of prostate-specific antigen in patients with
adenocarcinoma of the prostate; demonstration of the “hook effect”. Clin Chem 1988;34:
2175-7.
Pesce MA. “High-dose hook effect” with the
Centocar CA 125 assay. Clin Chem 1993;39:
1347.
Saryan JA, Garrett PE, Kurtz SR. Failure to
detect extremely high levels of serum IgE with
an immunoradiometric assay. Ann Allergy
1989;63:322-4.
Jury DR, Mikkelsen DJ, Dunn PJ. Prozone effect
and the turbidimetric measurement of albumin
in urine. Clin Chem 1990;36:1518-9.
Van Lente F. Light scattering immunoassays.
In: Rose NR, de Macario EC, Folds JD, Lane
HC, Nakamura RM, eds. Manual of clinical
laboratory immunology, 5th ed. Washington,
DC: ASM Press, 1997:13-9.
Cole TG, Johnson D, Eveland BJ, Nahm MH.
Cost-effective method for detection of “hook
effect” in tumor marker immuometric assays.
Clin Chem 1993;39:695-6.
Papik K, Molnar B, Fedorcsak P, Schaefer R,
Lang F, Sreter L, et al. Automated prozone
effect detection in ferritin homogeneous immunoassays using neural network classifiers. Clin
Chem Lab Med 1999;37:471-6.
Cooper MD, Lawton AR. Disorders of the immune system. In: Braunwald E, Isselbacher
KJ, Petersdorf RG, Wilson JD, Martin JB,
Fauci AS, eds. Harrison’s principles of internal medicine. New York: McGraw-Hill, 1987:
1396-403.
Anthony W. Butch
University of Arkansas
for Medical Sciences
Department of Pathology
4301 West Markham
Little Rock, AR 72205
Address correspondence to this author
at: UCLA Medical Center, Department of
Pathology and Laboratory Medicine,
10833 Le Conte Ave., Mailroom A2-179
CHS, Los Angeles, CA 90095-1713. Fax
310-794-4864; e-mail abutch@mednet.
ucla.edu.
To the Editor:
Hook effect is an infrequent event
that is notoriously difficult to detect
in the clinical laboratory (1 ). One of
the best methods of detection is to
run samples both undiluted and diluted (2 ). Any sample that does not
dilute properly may have antigen
excess. This method prevents the reporting of falsely low results but
incurs substantial time and expense.
We report on the effectiveness of a
simple, inexpensive method reported
by Cole et al. (3 ) that uses a pooled
sample to detect “hook effect”.
Cole et al. (3 ) recommend batching
patient samples in groups of 10,
forming a pooled sample with each
sample diluted 10-fold by the other
samples in the batch. In addition, the
pooled sample is diluted 10-fold,
producing a 100-fold final dilution.
“Hook” samples produce a higher
result for the 100-fold dilution pool
than the 10-fold dilution pool. Each
of the 10 samples must then be reanalyzed at a higher dilution to detect
the out-of-range result. Cole et al.
also describe a modification in which
up to 30 patient samples are pooled.
For the past 6 years, we have used
the modified protocol described by
Cole et al. (3 ) in combination with
predilution of samples from patients
known to have extremely high results. We use 50 ␮L of sample from
each patient sample in an analytical
run to create a pooled sample. Each
run usually contains 20 –30 patient
samples, so the final pool dilutes
each patient sample 20- to 30-fold.
The answer for the pooled result
should not exceed the highest patient
result in the analytical run. If the
pooled sample is higher than the rest
of the patients, all samples are repeated after dilution.
With this protocol, we have found
two samples, one for prostate-specific antigen (PSA) and one for CA125, with falsely low values attributable to antigen excess. The most
recent was a CA-125 value that gave
a result of 375 units/mL when analyzed undiluted. The pooled sample
from that analytical run had a value
of ⬎500 units/mL (reportable range,
15–500 units/mL). The final patient
result was 23 000 units/mL. The patient had no previous laboratory results at our institution, so the error
would not have been detected by
delta checking and may not have
been apparent to the ordering physician. Although the manufacturer
claims that this assay will not hook
back into the normal range until concentrations exceed 100 000 units/mL
(CA-125 II product insert; Centocor
Diagnostics Division), erroneous re-