Prevalence of Heterophilic Antibody Interference inEight Automated Tumor Marker Immunoassays, CarolM. Preissner,1 Larry A. Dodge,1 Dennis J. O’Kane,1 RavinderJ. Singh,1 and Stefan K.G. Grebe1,2* (1 Endocrine Laboratory,Department of Laboratory Medicine & Pathology, and2 Department of Medicine, Mayo Clinic, Rochester, MN;* address correspondence to this author at: EndocrineLaboratory, Hilton 730C, Mayo Clinic, 200 1st Street, SW,Rochester, MN 55905; fax 507-284-9758, e-mail grebs@mayo.edu)

Heterophilic antibodies are human antibodies that canbind animal antibodies. They can cause problems inimmunoassays, particularly immunometric assays, wherethey can form a bridge between the capture and detectionantibodies, leading to a false-positive result in the absenceof analyte or, if analyte is also present, to a false increasein measured concentrations. Very rarely, heterophilic an-tibodies can also lead to false-negative or falsely lowresults. By adding blocking reagents, assay manufacturershave reduced the incidence of heterophile interferencesfrom the 2–5% observed in unblocked assays but havebeen unable to completely eliminate the problem (1, 2).During the last 10 years, tumor marker assays for humanchorionic gonadotropin (hCG), prostate-specific antigen(PSA), cancer antigen 125 (CA 125), carcinoembryonicantigen, and calcitonin have all been reported to sufferfrom this problem, frequently with undesirable clinicaloutcomes (3–7).

In our own laboratory we noted that our thyroglobulinassay displayed a higher than acceptable rate of hetero-phile interference (8 ). Although we were able to rectifythis problem through pretreatment of samples in hetero-phile blocking tubes (HBT) (8 ), we became concerned thatsome of our other tumor marker assays may suffer fromsimilar problems. We therefore decided to systematicallyevaluate all of our tumor marker assays for evidence ofinterference from heterophilic antibodies.

Our laboratory performs the following tumor markerassays: calcitonin on the Nichols Advantage, gastrin onthe Immulite2000, CA 125 and cancer antigen 15-3 (CA15-3) on the Vitros ECi, and �-fetoprotein (AFP), hCG,total PSA, and free PSA on the Beckman Access. All ofthese assays use at least one mouse monoclonal antibody(Table 1).

We designated the maximum tolerable heterophile in-terference rate for any of these assays as 0.5% and startedcollecting samples prospectively. Power calculations sug-gested that 500 specimens would give us �80% power todetect a doubling of the heterophile interference rate.Because of much lower test volumes for calcitonin andgastrin, we aimed for 200 samples for these tests, some-what reducing the statistical power for these two assays.All specimens except those with insufficient volume forheterophile interference testing were included in thecollection. We used comparison of paired results of un-treated samples and samples pretreated with HBT orheterophile-blocking reagent (HBR1 nonmurine, hCG as-say only; both from Scantibodies) to determine whether

heterophile interferences had occurred. HBT/HBR block-ing reagents consist of immunoglobulins that have beenselected to bind human heterophilic antibodies, and theyblock the majority of heterophile interferences (5, 8–10).

Before and during our evaluation of heterophile inter-ference rates, we ensured that HBT treatment did notaffect assay performance adversely by comparing themeans, medians, and variability of at least 50 paireduntreated and HBT-treated samples for each assay. Thesevariables were also continuously monitored during theentire study. Assay performance was comparable to thatof untreated samples for all assays except hCG. Eleven of59 samples showed a marked increase in apparent hCGconcentrations after HBT pretreatment. We determinedthrough discussion with Scantibodies that assays config-ured with an antibody complex (goat anti-mouse anti-body–mouse monoclonal antibody in our case) could givespurious results when treated with HBT. HBR1 nonmu-rine reagent is reputed to overcome this limitation. Wetherefore validated the use of HBR1 in our hCG assay byperforming intra- and interassay precision, recovery(from samples with added analyte), and linearity studiesafter HBR1 treatment. The performance characteristicswith HBR1 were comparable to the native hCG assay,with linearity ranging from 95% to 117%, recovery rang-ing from 91% to 99%, and intra- and interassay CVs �10%across the measurement range. The mean CV for paireduntreated and HBR1-treated samples was �4% (95%confidence interval, 0–19%).

For each assay, the mean, median, range, and SD wereestablished for the observed absolute and percentagedifferences between paired untreated and pretreated sam-ples. Results were plotted in modified Bland–Altmanplots with the mean of paired original and post-HBTtreatment measurements on the abscissa and the percent-age difference between paired original and post-HBTtreatment measurements on the ordinate. An examplemodified Bland–Altman plot can be seen in the figure inthe Data Supplement that accompanies the online versionof this Technical Brief at http://www.clinchem.org/con-tent/vol51/issue1/. Our formal criteria for decidingwhether an outlying data point may be attributable toheterophilic antibody interference were as follows:

(a) A percentage difference between the initial resultand HBT/HBR treatment result that exceeded a 3 SDpercentage difference was considered a possible het-erophile interference.

(b) Any result deviating from the mean by more than a5 SD percentage difference or any sample whosevalue decreased by more than 80% after HBT/HBRtreatment was considered definite heterophile inter-ference.

All possible and definite heterophile interferences werefurther evaluated for their potential clinical significance,with all cases that were likely to lead to different patientmanagement being regarded as clinically significant.

The results are summarized in Table 1. We observed nosignificant systematic bias attributable to HBT/HBR treat-

Technical Briefs

208 Clinical Chemistry 51, No. 1, 2005

ment. For each of the eight assays, we observed at leastone result outside the �3 SD percentage difference range.The rates of possible heterophile interferences rangedfrom 0.2% (AFP) to 3.7% (calcitonin), with an overallmean of 1.4% and median of 1%. However, in mostinstances the outlying results fell just outside the �3 SDpercentage boundaries, and the observed changes afterHBT/HBR treatment were not consistently downward, asis typical for samples affected by heterophile interference(Fig. 1). The possible heterophile interference rates signif-icantly exceeded the a priori prevalence threshold of 0.5%only for calcitonin (7 of 192; 3.7%; P �0.037, two-tailedFisher exact test) and hCG (13 of 510; 2.5%; P �0.023,two-tailed �2 test, 1 degree of freedom, Yates-corrected).Changes greater than a 5 SD percentage difference or adecrease in value �80% after HBT/HBR treatment, indic-ative of definite heterophile interference, were observedin only two cases (one each for hCG and free PSA). One ofthese, which decreased the patient’s hCG concentrationfrom 5.5 IU/L to 1 IU/L, was considered potentiallyclinically relevant.

The assays examined do not suffer from major prob-lems with heterophile interference. With the exception ofthe calcitonin and the hCG assay, the observed rates ofpossible heterophile interference were consistent with thea priori-determined acceptable range. However, even forthese two assays, the majority of results outside ourpredefined cutoff lay just outside these limits and did notdisplay the typical �80% decrease in value after HBT/HBR treatment that is indicative of definite heterophileinterferences. It is likely that the majority of these slightoutliers reflect a minimal increase in assay variabilityattributable to the HBT/HBR treatment.

The manufacturers of the assays used in our laboratoryhave been largely successful in minimizing the incidence

Tabl

e1.

Sum

mar

yof

hete

roph

ilein

terf

eren

cete

stin

g.

Ana

lyte

s(u

nits

)A

ssay

conf

igur

atio

nS

era

test

ed,

nR

esul

tra

nge

ofte

sted

sera

Mea

nva

lue

untr

eate

d/m

ean

valu

epr

etre

ated

aM

edia

nva

lue

untr

eate

d/m

edia

nva

lue

pret

reat

eda

Ser

aw

ith

>3

SD

%di

ffer

ence

untr

eate

dvs

pret

reat

ed,a

n(%

)

AFP

(�g/

L)Tw

o-si

te;

one

step

;tw

oM

MAB

sb505

0.5

–9800

55.8

/56.8

3.4

/3.1

1(0

.2)

Cal

cito

nin

(ng/

L)Tw

o-si

te;

one

step

;tw

oM

MAB

s192

�3

to28

390

492/4

84

3.4

/4.0

7(3

.7)

CA

12

5(k

iloun

its/L

)Tw

o-si

te;

one

step

;tw

oM

MAB

s508

1.3

–9775

138.4

/138.3

11/1

15

(1)

CA

15

-3(k

iloun

its/L

)Tw

o-si

te;

two

step

s;tw

oM

MAB

s500

8–4

000

85.7

/89.2

28/3

04

(0.8

)G

astr

in(n

g/L)

Two-

site

;on

est

ep,

two

MM

ABs

and

one

PGAB

203

�20

to711

146.2

/144.9

99.8

/99.1

2(1

)

hCG

(IU/L

)Tw

o-si

te;

one-

step

,on

eG

AM-

MM

ABco

mpl

exan

don

ePR

AB510

�0.5

to464

000

6575/6

396

1.4

/1.4

13

(2.5

)c

Free

PSA

(�g/

L)Tw

o-si

te;

one

step

,tw

oM

MAB

s500

0.1

–11.5

0.9

/0.8

0.7

/0.7

5(1

)c

Tota

lPS

A(�

g/L)

Two-

site

;on

est

ep;

two

MM

ABs

500

0.1

–29

5.0

/4.9

4.0

/4.0

6(1

.2)

aPr

etre

ated

inH

BT

tube

s(a

llan

alyt

esex

cept

hCG

)or

with

HB

R1

(hC

Gon

ly).

bM

MAB

,m

onoc

lona

lmou

sean

tibod

y;PG

AB,

poly

clon

algo

atan

tibod

y;G

AM,

goat

anti-

mou

sean

tibod

y;PR

AB,

poly

clon

alra

bbit

antib

ody.

cO

nesa

mpl

ew

ith�

5S

D%

diff

eren

ceof

untr

eate

dvs

pret

reat

ed,

clin

ical

lysi

gnifi

cant

inth

eca

seof

hCG

.

Fig. 1. Samples with �3 SD percentage difference after HBT pretreat-ment.Changes in analyte concentrations after HBT pretreatment (all analytes excepthCG) and HBR1 treatment (hCG only) for all 43 samples that displayed a �3 SDpercentage difference after HBT/HBR treatment. Untreated values are depictedon the left of each panel, and concentrations after treatment are on the right. Thelines of several samples overlap.

Clinical Chemistry 51, No. 1, 2005 209

of heterophile interference in tumor marker assays. This isa particular achievement considering that cancer patientsoften display tumor-induced activation of their immunesystem or may suffer from infections. Both of theseconditions can lead to polyspecific antibody production.Given sufficient incubation time, these would be boundby blocking reagents, but in modern automated assays,reactions are rarely allowed to reach equilibrium andthere may be insufficient time to achieve complete block-ing. In addition, modern assays are often configured withseveral mouse monoclonal antibodies rather than a mousemonoclonal antibody and a polyclonal antibody fromanother species. With the increasing use of mouse mono-clonal antibodies in diagnostic imaging and medical ther-apy for malignancies, and the resulting immunization ofthe recipients, the potential for interference increasessignificantly (1, 7) for assays that use multiple mousemonoclonal antibodies. Consequently, we cannot assumethat assays other than those included in our study per-form equally well for the same analytes. For example, oneparticular hCG assay continues to experience a consider-able number of clinical problems attributable to hetero-phile interference, representing the majority of such prob-lems reported (11 ).

There were, however, even in our study two results thatfollowed the typical heterophile interference pattern. Inone case, a free PSA result, this would not have led to achange in patient management. By contrast, in the secondcase, a falsely increased hCG, clinical management mighthave changed depending on the clinical circumstances.hCG assays, mainly from one particular manufacturer,continue to be the single most important source of erro-neous test results attributable to heterophile antibodyinterference (11, 12). Clinical correlation is therefore par-ticularly important when interpreting hCG results.

In most suspected cases of heterophile interference,HBT treatment is a convenient way to verify and correctthe problem. However, our experience shows that forassays that use antibody complexes, specialized HBRreagents should be used. A significant number of difficultto interpret results may otherwise ensue, as might havebeen the case in a recent publication (10 ). Laboratoriesthat wish to use HBT pretreatment in suspected cases ofheterophile interference therefore need to first explore theeffects of pretreatment on analytical performance.

References1. Kricka LJ. Human anti-animal antibody interferences in immunological

assays. Clin Chem 1999;45:942–56.2. Levinson SS, Miller JJ. Towards a better understanding of heterophile (and

the like) antibody interference with modern immunoassays. Clin Chim Acta2002;325:1–15.

3. Rotmensch S, Cole LA. False diagnosis and needless therapy of presumedmalignant disease in women with false-positive human chorionic gonadotro-pin concentrations. Lancet 2000;355:712–5.

4. Morgan BR, Tarter TH. Serum heterophile antibodies interfere with prostatespecific antigen test and result in over treatment in a patient with prostatecancer. J Urol 2001;166:2311–2.

5. Tommasi M, Brocchi A, Cappellini A, Raspanti S, Mannelli M. False serumcalcitonin high levels using a non-competitive two-site IRMA. J EndocrinolInvest 2001;24:356–60.

6. Bjerner J, Nustad K, Norum LF, Olsen KH, Bormer OP. Immunometric assayinterference: incidence and prevention. Clin Chem 2002;48:613–21.

7. Bertholf RL, Johannsen L, Guy B. False elevation of serum CA-125 levelcaused by human anti-mouse antibodies. Ann Clin Lab Sci 2002;32:414–8.

8. Preissner CM, O’Kane DJ, Singh RJ, Morris JC, Grebe SK. Phantoms in theassay tube: heterophile antibody interferences in serum thyroglobulin as-says. J Clin Endocrinol Metab 2003;88:3069–74.

9. Butler SA, Cole LA. Use of heterophilic antibody blocking agent (HBT) inreducing false-positive hCG results. Clin Chem 2001;47:1332–3.

10. Emerson JF, Ngo G, Emerson SS. Screening for interference in immunoas-says. Clin Chem 2003;49:1163–9.

11. Cole LA, Khanlian SA. Easy fix for clinical laboratories for the false-positivedefect with the Abbott AxSym total �-hCG test. Clin Biochem 2004;37:344–9.

12. Cole LA, Shahabi S, Butler SA, Mitchell H, Newlands ES, Behrman HR, et al.Utility of commonly used commercial human chorionic gonadotropin immu-noassays in the diagnosis and management of trophoblastic diseases. ClinChem 2001;47:308–15.

DOI: 10.1373/clinchem.2004.040501

Detecting the C282Y and H63D Mutations of the HFEGene by Holliday Junction-Based Allele-Specific Geno-typing Methods, Wendy Yang,2 Takuro Yaoi,1,2* ShurongHuang,2 Qinghong Yang,2 Sandra Hatcher,3 Henrietta Seet,3

and Jeffrey P. Gregg3 (1 Panomics, Inc., Redwood City, CA;2 FreshGene, Inc., Concord, CA; 3 Department of Pathol-ogy, University of California Davis Medical Center, Uni-versity of California Davis MIND Institute, Sacramento,CA; * address correspondence to this author at: Panomics,Inc., 2003 East Bayshore Road, Redwood City, CA 94063;fax 650-216-9790, e-mail tyaoi@panomics.com)

Hereditary hemochromatosis, one of the most commongenetic diseases in Caucasians, is characterized by exces-sive iron deposition secondary to hyperabsorption ofdietary iron and can potentially can lead to multiorganfailure if untreated. The C282Y and H63D mutations ofthe HFE gene are the most common mutations associatedwith symptomatic hemochromatosis. Recently, severalgenotyping methods have been used to identify hemo-chromatosis mutations and other single-nucleotide poly-morphisms (SNPs) (1 ). There are advantages and limita-tions for each methodology in terms of cost and efficiency(2 ). We report here the application of a new SNP/pointmutation genotyping platform developed by our group,the Holliday junction-based allele-specific genotyping(HAS) platform (3 ), to identify the C282Y and H63Dmutations associated with hemochromatosis.

For the HAS technology, we developed two detectionmodalities based on differences in the physical and bio-chemical properties between Holliday junctions (HJs) andduplex DNA or single-stranded DNA. The junctions thatform in an allele-specific manner can be detected hetero-geneously through gel electrophoresis (acrylamide oragarose; Fig. 1A) or homogeneously through a fluores-cence polarization (FP) competition assay (3 ). Using theHAS genotyping platform, we developed an assay forgenotyping the C282Y and H63D mutations with both gelelectrophoresis and FP for detection. Five primers (oneforward, two reference, and two reverse primers) weredesigned for each of the two mutations (see Table 1 in

210 Technical Briefs