Immunometric assay interference: incidence and prevention.
The frequency of interference from heterophilic antibodies has been investigated in several studies with somewhat differing results (9-12). The observed frequency depends on the method of detection, but gender, age, smoking habits, and autoimmune or other diseases in the population studied are also suggested factors, as heterophilic antibodies have features common with rheumatoid factors (RFs) (13). Furthermore, interference may be influenced by the handling of samples.
We use combinations of two murine monoclonal antibodies in all our immunometric assays, and to avoid interference from complement (14, 15), we use only capture antibodies of isotype IgG1. Our sample buffer was chosen after a flow cytometric study of interference in real time (16). When we recently changed assay formats from magnetic microspheres and radiolabeled tracers to microtiter wells and europium labeling, we experienced a high frequency of interference in several tumor marker assays. This high frequency was seen both with the capture antibody directly coated on the microtiter plate and with the use of biotinylated capture antibodies combined with streptavidin-coated microtiter plates. Initial studies showed that this interference mostly could be solved by the removal of the Fc fragment from the capture antibody, an indication of the presence of heterophilic antibodies. However, neither the addition of normal mouse serum nor the native murine monoclonal antibody MAK33, which is supposed to block normal heterophilic antibodies, could sufficiently prevent this interference. Aggregating MAK33 dramatically potentiated its blocking effect. Those findings precipitated a more thorough study of assay interference. For this purpose we chose carcinoembryonic antigen (CEA) assays for several reasons. CEA is a well-characterized and stable protein. In addition, the concentrations of CEA in serum are usually very low, making CEA assays prone to interferences. We also have a large number of samples referred for CEA analysis, and the majority of samples have CEA concentrations within the reference interval, making interference easy to identify. The results are of general interest.
Materials and Methods
From December 20, 1999, to October 17, 2000, we screened all 11 261 samples referred to us for analysis of CEA alone. The study protocol was in accordance with the rules of our hospital ethics committee. Approximately one-third of samples were from patients in this hospital, with the remaining two-thirds from outpatients or patients from other hospitals.
CEA. CEA for calibrators was purified from liver metastases from colorectal carcinomas by perchloric acid extraction followed by ion-exchange and gel-filtration chromatography (17). CEA for calibrators was diluted in a pH 7.4 buffer [50 mmol/L Tris (Sigma), 100 mmol/L NaCl, 1 g/L Germall II (ISP Sutton Laboratories), and 60 g/L bovine serum albumin (BSA; Sigma A 4503)] to concentrations of 0, 3.6, 9.0, 36, 180, and 900 [micro]g/L.
Controls. Controls in automated assays were prepared from serum from a pool of blood donors with patient serum added to CEA concentrations of 2.9 and 86 [micro]g/L, respectively. For manual analyses, a control made from a pool of interference-negative serum was used. Controls and calibrators were stored in aliquots at -30[degrees]C until use.
Antibodies. Monoclonal anti-CEA antibodies 12-140-1 and 12-140-10 (15, 18) were produced in female Balb/c mice. Antibodies were purified on a protein A immunoaffinity chromatography column (Amersham Pharmacia Biotech), as described under the separation of [F(ab').sub.2] and Fc fragments. Both antibodies were of IgG1 subtype.
[F(ab').sub.2] fragmentation of antibodies with bromelain. Essentially using the method of Milenic et al. (19), for each mg of antibody in a pH 7.0 buffer [0.05 mol/L Tris, 0.10 mol/L NaCl, 5 mmol/L EDTA], we added 50 [micro]g of bromelain (ID-Diluent 1; Diamed). After incubation at 37 [degrees]C for 2 h, further enzymatic reactions were stopped by the addition of freshly made 0.2 mol/L ethylene maleimide, corresponding to 1/10 of the reaction volume. Products were purified at 4[degrees]C on a protein A chromatography column (Amersham Pharmacia Biotech) equilibrated with a pH 8.2 buffer (0.1 mol/L Na[H.sub.2]P[O.sub.4]-NaOH, 0.1 g/L Na[N.sub.3]) and eluted with a linear pH gradient from the pH 8.2 application buffer to a pH 3.2 buffer [0.025 mol/L citric acid, 0.025 mol/L Na[H.sub.2]P[O.sub.4], 0.025 mol/L NaCl, and 0.1 g/L Na[N.sub.3]]. After chromatography, products were dialyzed against 0.15 mol/L NaCl. The purities and sizes of the fragments were checked by sodium dodecyl sulfate--polyacrylamide gel electrophoresis, which yielded a [F(ab').sub.2] fragment of ~112 kDa and a Fc fragment of 36 kDa (two parts of 18 kDa each). Native antibody 12-140-10 was used as a control and had a size of 151 kDa. No presence of the native antibody could be detected after the bromelain cleavage.
Biotinylation of [F(ab').sub.2] and IgG of capture antibodies. According to the guidelines given by Bayer and Wilchek (20), we added N-hydroxysuccinimidyl-6-biotinamido hexanoate (Vector) in a fivefold molar excess to antibody in 0.15 mol/L NaCl. Sodium borate (pH 8.0) was added to 0.1 mol/L, and the solution was incubated for 120 min at room temperature. Free biotin was removed by dialysis against a pH 7.8 buffer consisting of 0.05 mol/L Tris, 0.15 mol/L NaCl, and 0.5 g/L Na[N.sub.3].
Europium labeling of tracer. Europium chelate (Perkin-Elmer Life Sciences) was added in a 12.5-fold molar excess to antibody in 0.15 mol/L NaCl, and then 0.1 mol/L sodium borate (pH 8.6) was added at a volume ratio of 1:9 (1 volume of borate buffer to 9 volumes of sample). The solution was incubated at room temperature (22[degrees]C) for 48 h. Free europium chelate was removed by gel-filtration chromatography on a PD-10 column (Amersham Pharmacia Biotech) equilibrated with a pH 7.8 buffer consisting of 0.05 mol/L Tris, 0.15 mol/L NaCl, and 0.5 g/L Na[N.sub.3]. Finally, tracer antibody was diluted in the same buffer to a stock solution of 25 mg/L, with 1 g/L BSA treated with diethylenetriamine pentaacetic acid (Perkin-Elmer Life Sciences) added for stability.
Assay buffer. A basic pH 7.4 buffer consisting of 0.05 mol/L Tris, 0.15 mol/L NaCl, 0.02 mmol/L diethylene-triamine pentaacetic acid (Sigma), 5 mg/L tartrazine (Aldrich), 1 g/L Germall II, 10 mg/L Triton X100 (Sigma), 5 g/L BSA (Sigma; cat. no. A 4503), and 0.5 g/L bovine IgG (Sigma; cat. no. G 7516) was filtered through a 0.22 [micro]m filter.
Heat treatment of nonspecific murine antibody for blocking of interference. Murine IgG1[kappa] antibody MAK33 (Roche Molecular Biochemicals) was stored at a concentration of 2 g/L in 0.15 mol/L NaCl-0.01 mol/L [Na.sub.2]HP[O.sub.4], pH 7.4, at -30[degrees]C. Just before being used, 7.5 mL of the antibody solution was thawed, transferred to 10-mL Vacutainer Tubes (Becton Dickinson), and incubated in a 60[degrees]C water bath for 10 min. Untreated, the MAK33 solution has an absorbance at 280 nm ([A.sub.280]) of 2.71. During the study period, we performed the heat treatment 23 times, with a mean final [A.sub.280] of 7.02 (SD, 0.63).
Washing buffer. Washing buffer (pH 7.4) consisted of 0.05 mol/L Tris, 0.15 mol/L NaCl, 1 g/L Germall II, and 0.5 g/L Tween 20 (Sigma).
Microtiter wells. DELFIA streptavidin microtitration strips (Perkin-Elmer Life Sciences) were used.
Samples. After routine assays, serum samples were stored in 0.5-mL aliquots at -30[degrees]C. Before test assays, samples were thawed and mixed manually.
Assays. In all automated assays, buffers, biotinylated capture antibody, tracers, and plates were equilibrated to room temperature for 1 h before use, and 25-[micro]L duplicates of samples, calibrators, and controls were pipetted into the microtitration wells for the routine assays. For test assays, samples were analyzed as singletons. We added 175 [micro]L of buffer containing 2 mg/L biotinylated capture antibody 12-140-10 to each well, and plates were incubated for 120 min with continuous shaking in the AutoDELFIA instrument (Perkin-Elmer Life Sciences). Plates were then washed three times in the AutoDELFIA instrument. Tracer antibody 12-140-1 (0.1 [micro]g in 200 [micro]L of buffer) was added to each well, and plates were incubated with shaking for another 30 min. After six additional washes, 200 [micro]L of DELFIA enhancement solution (Perkin-Elmer Life Sciences) was added to each well. After 10 min of incubation, the time-resolved fluorescence was measured in the AutoDELFIA instrument. Results were calculated with MultiCALC (Perkin-Elmer Life Sciences), using the spline curve fit. The manual assays were performed with the same reagents and under the same conditions. Samples, calibrators, controls, and reagents were pipetted manually. In the manual assays, the AutoDELFIA instrument was replaced by a Wallac PlateShaker, a Wallac PlateWasher, and a Victor 1420 multilabel counter (all from Perkin-Elmer Life Sciences).
For both automated assays, the monthly total CV was <6.3% at 2.9 [micro]g/L and <4.7% at 86 [micro]g/L during the test period. The detection limits (defined as concentrations corresponding to three times the analytical imprecision at zero concentration) were <0.1 [micro]g/L.
For all manual assays together, the total CV, including between-assay variation, was 7.3% for controls (n = 102) at 123.8 [micro]g/L. The detection limits were always <0.5 [micro]g/L. Both assays were linear up to 900 [micro]g/L, the highest calibrator concentration.
The screening consisted of parallel runs of two CEA assays, in which our routine assay with biotinylated [F(ab').sub.2] fragments of monoclonal antibody 12-140-10 as capture antibody and 15 mg/L heat-treated nonspecific murine monoclonal antibody MAK33 added to assay buffer was the reference assay and a test assay using biotinylated whole-IgG monoclonal antibody 12-140-10 without any murine monoclonal antibodies in the assay buffer was the screening assay for interference. An overview of the assays used is given in Table 1.
The aim of the study was to find samples with interference high enough to lead to a different clinical interpretation of the results. The smallest interference considered to be clinically important was an interference with an apparent CEA concentration of ~2 [micro]g/L, which caused a result of 3.2 [micro]g/L to be reported as 5.2 [micro]g/L, slightly above the reference limit. We also had to consider the assay variations, so that the observed interference was not an effect of assay variation alone. To meet those requirements, we classified samples as positive for interference when the screening assay had an apparent CEA result above the reference limit of 5 [micro]g/L and had a 60% higher result than the reference assay. With a CV for both assays <6.3% and no other differences between the assays, the risk of detection of a "false" interference attributable to assay variation alone would be <0.001. Two patients showed markedly higher values in the reference assay than in the test assay. When their samples were diluted (not shown), the values fell more than expected. This was interpreted as interference in both the screening and the reference assay, and samples from those two patients were classified as positive.
After screening, interference-positive samples were frozen and rechecked in manual assays.
To calculate confidence intervals for the frequency of interference, we had to consider that the frequency is a quotient between the small number of patients with interference (this number will have a Poisson distribution) and the large number of samples screened for interference that were from patients not seen previously. This large number could unfortunately not be directly assessed from the Laboratory Information System and had to be estimated from a sample control group of patients (192 of 407 randomly selected samples were from new patients) and the total number of samples screened (11 261 samples). This random selection of the sample control group had a binomial distribution. We then used Excel to generate 1000 random numbers with the expected Poisson distribution of the number of patients with interference and 1000 random numbers of new patients screened for interference, calculated using the control group with its binomial distribution. This generated 1000 random quotients, and the 95% confidence interval was calculated as the 2.5 and 97.5 percentiles.
HEAT TREATMENT OF ANTIBODIES
Murine monoclonal antibody MAK33 was subjected to heat treatment for various times and at various temperatures. Apparent CEA concentrations from three patients with known interference were analyzed by the screening assay with 15 mg/L of the various preparations of heat-treated MAK33 added. The results from these samples together with the screening and reference assays are shown in Fig. 1. Samples treated at 63[degrees]C showed some precipitation after storage, and treatment at 60[degrees]C was chosen for further studies.
FREQUENCY OF INTERFERENCE WITH BOVINE IgG ONLY IN BUFFER
We screened a total of 11 261 samples for interference. Among these samples we randomly selected 407 samples and found that 192 of these samples were from patients not previously investigated for CEA in our laboratory. This gives an estimate of 5312 new patients being screened. The figures in the data supplement (available in the online version of this article at http://www.clinchem.org/content/vol48/issue4) show the distributions of CEA concentration and age among the 192 randomly selected samples, which should be representative of the population studied. In 80% of patients, CEA was below the reference limit of 5 [micro]g/L, suggesting that few patients have a heavy tumor burden (Fig. 1 in the data supplement). The low CEA concentrations indicate that parallel assays may be useful to find interferences. More than 80% of patients were >50 years of age (Fig. 2 in the data supplement).
In the study we found 210 patients, or 4.0% (95% confidence interval, 3.3-4.7%), who were positive for interference and had not undergone any previous CEA investigation in our laboratory. Available sera from those patients were frozen and thawed and rechecked manually with (a) basic buffer (the screening assay); (b) basic buffer plus 15 mg/L native murine monoclonal MAK33 added; (c) basic buffer plus 15 mg/L heat-treated MAK33; (d) the Fc fragment removed plus basic buffer; (e) the Fc fragment removed plus basic buffer plus 15 mg/L native MAK33; and (f) Fc fragment removed plus basic buffer plus 15 mg/L heat-treated MAK33 (the reference assay). We were then able to classify interference into six types, based on the effects of adding native and heat-treated MAK33 to buffer as well as the effect of removing the Fc fragment from the capture antibody. The results are shown in Table 2.
[FIGURE 1 OMITTED]
FREQUENCY OF INTERFERENCE WITHOUT BOVINE IgG IN BUFFER
In this study we concentrated on samples with interference that could not be solved by the addition of 0.5 g/L bovine IgG to the buffer. To estimate the frequency without this additive, we made a modified screening assay in which bovine IgG had also been omitted. In this assay we also replaced BSA with human serum albumin because low concentrations of bovine IgG remaining in the BSA may give a falsely low frequency of interference (5). When sera from 22 consecutive new patients found with the screening assay (with bovine IgG) were first tested in the modified screening assay (without bovine IgG), all showed interference. Sera from 63 consecutive new patients identified by the modified screening assay (without bovine IgG) were tested with the screening assay (with bovine IgG). Of those 63 patients, 49 patients still had interference in the screening assay (with bovine IgG), but for 14 patients the results were normalized. This gives an estimated frequency of interference of 5.1% (4.3-6.5%, by computer simulation) if no bovine IgG was added to the buffer.
MEDIAN INTERFERENCE AND RANGE OF INTERFERENCES
With bovine IgG added to the buffer, 209 consecutive interference-positive patients (1 patient with irregular behavior excluded) who had not been investigated for CEA during the previous months had a median apparent increase in CEA attributable to interference (difference between screening and reference assays) corresponding to 11.3 [micro]g/L (range, 2.0-307.5 [micro]g/L).
Without bovine IgG added to the buffer, 69 consecutive interference-positive patients who had not been investigated for CEA during the previous months had a median apparent increase in CEA concentration attributable to interference (difference between screening and reference assays) corresponding to 8.6 [micro]g/L (range, 3.0-158.3 [micro]g/L). The apparent increase in CEA is plotted in Fig. 2.
REQUIREMENTS OF HEAT-TREATED MAK33 FOR TYPE III SAMPLES
For type III samples, removal of the Fc fragment normalized CEA concentrations, but not the addition of 15 mg/L heat-treated murine monoclonal MAK33. To determine whether higher concentrations of heat-treated murine monoclonal MAK33 would normalize the values, we analyzed sera from all 76 patients (not just the 41 patients who had not been investigated previously for CEA) classified as type III, with higher concentrations of heat-treated MAK33. Results were normalized for 66 patients after 25 mg/L heat-treated MAK33 was added, for 8 additional patients after 50 mg/L was added, and for 1 patient after 100 mg/L was added. Notably, in one patient, not even 100 mg/L heat-treated MAK33 normalized the result.
AGE AND GENDER AS FACTORS FOR INTERFERENCE
Age and gender distributions were compared between the interference-positive group and a randomly selected control group of 186 patients. There were no statistically significant differences. The results are shown in Table 3.
[FIGURE 2 OMITTED]
VARIATIONS IN INTERFERENCE OVER TIME
In the total material we found 25 patients positive for interference and with four or more samples screened for interference during the period of investigation. For 15 patients we found samples without interference before or after the peak of interference, leaving only 10 patients positive for interference in all samples. The results are plotted in Fig. 3 with peak values located to day 0.
SELECTION OF THE POPULATION
The aim of this study was to find ways to reduce interference in our in-house assays. It was natural for us to study interference in the population of samples sent to us for assessment of CEA. Unfortunately, we lacked the high-quality clinical information needed to tell to what extent the results are representative of a population of healthy individuals or of a specific population of tumor patients. Age and CEA values were the only information easily available to us on the control population and are provided in Figs. 1 and 2 of the data supplement (available at http://www.clinchem.org/content/vol48/issue4). Most of patients were elderly and had CEA values within the reference interval, suggesting an absence of tumor or a small tumor burden. From the clinical information provided with the referral, we could add that the prevalence of other diseases is high in this group of patients.
[FIGURE 3 OMITTED]
If a physician receives report showing an unexpectedly high CEA result (which may be attributable to interference), he or she often orders new tests to confirm the results. The prevalence of interference may thus be higher among second and third samples sent to the laboratory for testing. For an unbiased estimation of frequency in this study, we considered only samples from patients who had not been investigated for CEA during the 18 months before sampling (18 months is the memory limit of our Laboratory Information System).
METHODS AND DETECTION OF INTERFERENCES
To detect interference in the study, we chose parallel runs using a pair of monoclonal antibodies instead of an interference assay using the same monoclonal antibody as both capture and tracer antibody (11). Because parallel runs are sensitive to analytical variations, interference is hard to find in the presence of high concentrations of analyte, and it is difficult to estimate the remaining frequency of interference when the reference method is used. However, an interference assay will detect not only human antibodies against relevant isotypic and idiotypic epitopes shared by both assay antibodies, but also irrelevant human antibodies against idiotypic epitopes unique to the single antibody used as both capture and tracer antibody (21). CEA complexes and possible repetitive epitopes on CEA-like molecules (22) will also give an apparent interference, and there is a validation problem for the relevance to the original assay.
HEAT TREATMENT OF MAK33
We currently use heat-treated MAK33 in our routine assays and have not experienced any problems to date. However, we must emphasize that we have not yet performed any systematic studies of the reproducibility and the shelf-life of this product. We also do not know whether those results can be generalized to the heat aggregation of other monoclonal antibodies.
FREQUENCY OF INTERFERENCE AND THE EFFECT OF BLOCKERS
In this study, we found a high frequency of interference in an immunometric assay for CEA when bovine immunoglobulins only were added to the buffer. Removing the Fc fragment from the capture antibody as the only measure nearly eliminated interference. This is highly indicative of interfering agents, such as heterophilic antibodies, that bind to the Fc fragment. In this study we used bromelain, rather than the more common pepsin, to fragment immunoglobulins into [F(ab').sub.2] and Fc fragments because [F(ab').sub.2] fragments produced by bromelain cleavage better retain their immunoreactivity (19, 23) and because we have a long experience with bromelain giving high-quality, clear-cut fragments.
Contrary to our expectations, addition of native murine monoclonal antibody MAK33 reduced interference only slightly. Heat treatment of MAK33 dramatically improved its ability to reduce interference. The superiority of chemically aggregated MAK33 over native MAK33 in reducing interference caused by heterophilic antibodies has been shown by Lenz et al. (24). They also showed that chemical aggregation of MAK33 actually produced a mixture of aggregates of different sizes where the larger aggregates more efficiently reduced interferences than did smaller ones. If the interferences seen in this study are attributable to heterophilic antibodies, two explanations for the effectiveness of heat treatment of MAK33 may be suggested: a size effect and biotinylation.
Size effect. A heterophilic antibody has a low affinity, and in a solution it may form only one weak, and thus temporary, bond with a monomer of MAK33. The aggregation of MAK33 caused by heat treatment may produce a larger complex with several binding sites. A heterophilic antibody may thus form not one, but several weak (but together strong) bonds with such a complex. Although monomeric, capture antibodies in this assay format are crowded together at the well wall and may behave more like a complex than a monomer regarding potential binding sites for heterophilic antibodies.
Biotinylation may alter the configuration of the capture antibody. The heterophilic antibody may bind to epitopes not expressed on native murine monoclonals but exposed by both biotinylation and heat treatment.
The frequency of interference, 5.1%, found in this study without bovine IgG in buffer was considerably lower than the interference found in two other studies, which reported rates of 40% (11) and 52% (16). In both studies the interference distributions were also different, with most of the interferences close to the cutoff limits, whereas in our study, most of the interferences were well above the cutoff limit. Both of those studies used one-step immunoassays, whereas we performed a two-step immunoassay.
GENDER AS A FACTOR FOR INTERFERENCE
We found a slight overrepresentation of men among interference-positive patients, in accordance with previously published results (9), but no statistically significant differences in age and gender distribution between the interference-positive group and the control group were detected. An overrepresentation of men may seem unexpected, considering that autoimmune diseases are more common among women. There are reports indicating that smokers with rheumatoid arthritis more often are RF-positive (25) and that there is a correlation between the number of years smoked and the concentration of RF (26). Smoking habits may thus be a major confounding factor in risk factor calculation for gender and age.
VARIATIONS IN INTERFERENCE OVER TIME
There are case reports of individuals with no signs of interference who in a short period of time developed interference by heterophilic antibodies and later reverted to normal (7, 8). In our study, 15 of 25 individuals with four or more samples had an interference-negative sample either before or after the peak of interference. Our results support those observations of the transient nature of interferences.
We do not know the exact causes of the interferences that we have found, but we suggest human heterophilic antibodies. The blocking effect of immunoglobulins, variations over time, and the reduction of interference seen after removal of Fc fragments are all consistent with effects of heterophilic antibodies or HAMAs. Because the in vivo medical use of animal antibodies has been low in Norway and because we have no indications of any patient in the study being exposed to animal antibodies in a clinical setting, HAMAs are not a major cause of interferences in our study.
There are many ways of avoiding interference from heterophilic antibodies or HAMAs in immunoassays. A combination of assay antibodies from different species and with little cross-reactivity with human anti-animal antibodies is the best theoretical solution (27, 28). However, most assays today are made by combining two murine monoclonal antibodies. The first choice is then an appropriate assay buffer composition. Interference can be blocked if the human antibodies bind nonspecific animal immunoglobulins in the buffer instead of binding the assay antibodies. In assays that use a pair of murine monoclonal antibodies, murine and bovine antibodies are the nonspecific immunoglobulins that reduce interference in the highest percentage of patient samples (12, 29), and they also have the highest avidity for heterophilic antibodies (16). An explanation for the good blocking effect of bovine IgG found in those studies may be the formation of heterophilic antibodies after ingestion of bovine IgG in milk (30). Aggregation of the nonspecific immunoglobulins can enhance their binding to heterophilic antibodies or HAMAs (31). Such aggregation could be achieved by both heat and chemical treatment and is patented (24). A second choice is to reduce interference by removing the Fc fragment (29) or by changing the structure of (32, 33) or chemically modifying (34) the assay antibodies. A third possibility is to identify samples with interference for individual treatment. The assay format can be modified to simultaneously detect the presence of interference (11, 16). Several reagent sets are also commercially available for detecting heterophilic antibodies or HAMAs of sufficient titer and affinity to have an analytically significant effect (1, 35). Systematic dilution, recovery testing, or parallel use of a different reagent set (e.g., a competitive assay) could also rule out interference. The need to test every sample has limited the use of those methods.
When a sample with possible interference from HAMAs or heterophilic antibodies has been identified, the true result could be established in many ways: removal of interference by gel filtration or chromatography (21), the addition of specially designed "HAMA-blocking substances" (36, 37), precipitation of antibodies with polyethylene glycol (38), or simply by use of an assay not sensitive to such interference.
RF is also a known source of interference in immunoassays. Tests for RF are heterogeneous and have changed from simple agglutination assays to more complicated immunometric assays with the possibility of both quantifying RF and defining RF isotype specificity. There is poor agreement among these tests (39). From the clinical information provided, we know that some of the patients positive for interference in our study were RF-positive, whereas some were not; therefore, some correlation between RF positivity and assay interference may be expected. Further studies to explore the relationship between RF and heterophilic antibodies are warranted.
Finally, recombinant technology is becoming more accessible, and the use of single-chain fragments and humanized or chimeric antibodies is surging for therapy and diagnostic in vivo procedures (in addition to in vitro diagnostics). We can therefore expect assay interference from HAMAs to be more common in the future. In a study of HAMA interference in an immunometric assay, polyclonal antibodies were shown to be better blockers than monoclonal antibodies (40). In vivo use of murine antibodies may also lead to the production of anti-anti-idiotypic antibodies, which could possibly interfere in assays by preventing the immunoassay antibodies from binding (41-43).
In conclusion, our large study shows that adding only native bovine or murine immunoglobulins to the buffer does not sufficiently block interferences. Heat-treated nonspecific murine immunoglobulin in the buffer and removal of the Fc fragment from the capture antibody both by themselves reduce interference in immunometric assays. Combining those measures further reduces interference, and we have chosen this combination for our in-house immunometric assays.
(1.) Kricka LJ. Human anti-animal antibody interferences in immunological assays. Clin Chem 1999;45:942-56.
(2.) Despres N, Grant AM. Antibody interference in thyroid assays: a potential for clinical misinformation. Clin Chem 1998;44:440-54.
(3.) Levinson SS. Antibody multispecificity in immunoassay interference. Clin Biochem 1992;25:77-87.
(4.) Kricka LJ. Interferences in immunoassay--still a threat. Clin Chem 2000;46:1037-8.
(5.) Boscato LM, Stuart MC. Heterophilic antibodies: a problem for all immunoassays. Clin Chem 1988;34:27-33.
(6.) Kaplan IV, Levinson SS. When is a heterophile antibody not a heterophile antibody? When is it an antibody against a specific immunogen? Clin Chem 1999;45:616-8.
(7.) Wickus GG, Caplan RH, Mathews EA, Pehling GB. Sudden appearance and subsequent disappearance of interference in immunometric assays of thyrotropin neutralizable with purified mouse IgG. Clin Chem 1991;37:595-6.
(8.) Kazmierczak SC, Catrou PG, Briley KP. Transient nature of interference effects from heterophile antibodies: examples of interference with cardiac marker measurements. Clin Chem Lab Med 2000;38:33-9.
(9.) Hawkins BR, Saueracker GC, Dawkins RL, Davey MG, O'Connor KJ. Population study of heterophile antibodies. Vox Sang 1980; 39:339-42.
(10.) Ward G, McKinnon L, Badrick T, Hickman PE. Heterophilic antibodies remain a problem for the immunoassay laboratory. Am J Clin Pathol 1997;108:417-21.
(11.) Boscato LM, Stuart MC. Incidence and specificity of interference in two-site immunoassays. Clin Chem 1986;32:1491-5.
(12.) Thompson RJ, Jackson AP, Langlois N. Circulating antibodies to mouse monoclonal immunoglobulins in normal subjects--incidence, species specificity, and effects on a two-site assay for creatine kinase-MB isoenzyme. Clin Chem 1986;32:476-81.
(13.) Courtenay-Luck NS, Epenetos AA, Winearls CG, Ritter MA. Preexisting human anti-murine immunoglobulin reactivity due to polyclonal rheumatoid factors. Cancer Res 1987;47:4520-5.
(14.) Baatrup G, Svehag SE, Jensenius JC. The attachment of serum-and plasma-derived C3 to solid-phase immune aggregates and its relation to complement-mediated solubilization of immune complexes. Scand J Immunol 1986;23:397-406.
(15.) Bormer OP. Interference of complement with the binding of carcinoembryonic antigen to solid-phase monoclonal antibodies. J Immunol Methods 1989;121:85-93.
(16.) Frengen J, Kierulf B, Schmid R, Lindmo T, Nustad K. Demonstration and minimization of serum interference in flow cytometric two-site immunoassays. Clin Chem 1994;40:420-5.
(17.) Bormer O. A direct assay for carcinoembryonic antigen in serum and its diagnostic value in metastatic breast cancer. Clin Biochem 1982;15:128-32.
(18.) Bormer OP, Nustad K. Selection of monoclonal antibodies for use in an immunometric assay for carcinoembryonic antigen. J Immunol Methods 1990;127:171-8.
(19.) Milenic DE, Esteban JM, Colcher D. Comparison of methods for the generation of immunoreactive fragments of a monoclonal antibody (B72.3) reactive with human carcinomas. J Immunol Methods 1989;120:71-83.
(20.) Bayer EA, Wilchek M. Protein biotinylation. Methods Enzymol 1990;184:138-60.
(21.) Turpeinen U, Lehtovirta P, Alfthan H, Stenman UH. Interference by human anti-mouse antibodies in CA 125 assay after immunoscintigraphy: anti-idiotypic antibodies not neutralized by mouse IgG but removed by chromatography. Clin Chem 1990;36:1333-8.
(22.) Hammarstrom S, Shively JE, Paxton RJ, Beatty BG, Larsson A, Ghosh R, et al. Antigenic sites in carcinoembryonic antigen. Cancer Res 1989;49:4852-8.
(23.) Mariani M, Camagna M, Tarditi L, Seccamani E. A new enzymatic method to obtain high-yield [F(ab').sub.2] suitable for clinical use from mouse IgG1. Mol Immunol 1991;28:69-77.
(24.) Lenz H, Mossner E, Stock W, Roder A, Haug H, McCarthy RC, inventors. Reagent and method for determination of a polyvalent substance using an immunoaggregate. (Boehringer Mannheim GmbH, Mannheim DE, assignee). US Patent 164054 (4914040), April 3, 1990.
(25.) Jonsson T, Thorsteinsson J, Valdimarsson H. Does smoking stimulate rheumatoid factor production in non-rheumatic individuals? APMIS 1998;106:970-4.
(26.) Wolfe F. The effect of smoking on clinical, laboratory, and radiographic status in rheumatoid arthritis. J Rheumatol 2000;27: 630-7.
(27.) Hennig C, Rink L, Fagin U, Jabs WJ, Kirchner H. The influence of naturally occurring heterophilic anti-immunoglobulin antibodies on direct measurement of serum proteins using sandwich ELISAs. J Immunol Methods 2000;235:71-80.
(28.) Larsson A, Mellstedt H. Chicken antibodies: a tool to avoid interference by human anti-mouse antibodies in ELISA after in vivo treatment with murine monoclonal antibodies. Hybridoma 1992; 11:33-9.
(29.) Csako G, Weintraub BD, Zweig MH. The potency of immunoglobulin G fragments for inhibition of interference caused by anti-immunoglobulin antibodies in a monoclonal immunoradiometric assay for thyrotropin. Clin Chem 1988;34:1481-3.
(30.) Hunter WM, Budd PS. Circulating antibodies to ovine and bovine immunoglobulin in healthy subjects: a hazard for immunoassays. Lancet 1980;2:1136.
(31.) Vaidya HC, Beatty BG. Eliminating interference from heterophilic antibodies in a two-site immunoassay for creatine kinase MB by using [F(ab').sub.2] conjugate and polyclonal mouse IgG. Clin Chem 1992;38:1737-42.
(32.) Kuroki M, Matsumoto Y, Arakawa F, Haruno M, Murakami M, Kuwahara M, et al. Reducing interference from heterophilic antibodies in a two-site immunoassay for carcinoembryonic antigen (CEA) by using a human/mouse chimeric antibody to CEA as the tracer. J Immunol Methods 1995;180:81-91.
(33.) Buist MR, Kenemans P, van-Kamp GJ, Haisma HJ. Minor human antibody response to a mouse and chimeric monoclonal antibody after a single i.v. infusion in ovarian carcinoma patients: a comparison of five assays. Cancer Immunol Immunother 1995; 40:24-30.
(34.) Delgado C, Pedley RB, Herraez A, Boden R, Boden JA, Keep PA, et al. Enhanced tumour specificity of an anti-carcinoembrionic antigen Fab' fragment by polyethylene glycol (PEG) modification. Br J Cancer 1996;73:175-82.
(35.) Hasholzner U, Stieber P, Meier W, Lamerz R. Value of HAMA-determination in clinical practice. An overview. Anticancer Res 1997;17:3055-8.
(36.) Madry N, Auerbach B, Schelp C. Measures to overcome HAMA interferences in immunoassays. Anticancer Res 1997;17:2883-6.
(37.) Reinsberg J. Different efficacy of various blocking reagents to eliminate interferences by human antimouse antibodies with a two-site immunoassay. Clin Biochem 1996;29:145-8.
(38.) Primus FJ, Kelley EA, Hansen HJ, Goldenberg DM. "Sandwich"-type immunoassay of carcinoembryonic antigen in patients receiving murine monoclonal antibodies for diagnosis and therapy. Clin Chem 1988;34:261-4.
(39.) Houssien DA, Jonsson T, Davies E, Scott DL. Rheumatoid factor isotypes, disease activity and the outcome of rheumatoid arthritis: comparative effects of different antigens. Scand J Rheumatol 1998;27:46-53.
(40.) Reinsberg J. Interferences with two-site immunoassays by human anti-mouse antibodies formed by patients treated with monoclonal antibodies: comparison of different blocking reagents. Clin Chem 1998;44:1742-4.
(41.) Reinsberg J. Interference by human antibodies with tumor marker assays. Hybridoma 1995;14:205-8.
(42.) Fagerberg J, Frodin JE, Wigzell H, Mellstedt H. Induction of an immune network cascade in cancer patients treated with monoclonal antibodies (ab1). I. May induction of ab1-reactive T cells and anti-anti-idiotypic antibodies (ab3) lead to tumor regression after mAb therapy? Cancer Immunol Immunother 1993;37:264-70.
(43.) Schmolling J, Wagner U, Reinsberg J, Biersack HJ, Krebs D. Immune reactions and survival of patients with ovarian carcinomas after administration of [131I-F(Ab').sub.2] fragments of the OC 125 monoclonal antibody. Geburtshilfe Frauenheilkd 1995;55:200-3.
JOHAN BJERNER, * KJELL NUSTAD, LARS F. NORUM, KARI HAUGE OLSEN, and OLE P. BORMER
 Nonstandard abbreviations: HAMA, human anti-mouse antibody; RF, rheumatoid factor; CEA, carcinoembryonic antigen; and BSA, bovine serum albumin.
Central Laboratory, Norwegian Radium Hospital, Montebello, N-0310 Oslo, Norway.
* Author for correspondence. Fax 47-22-730725; e-mail firstname.lastname@example.org.
Received September 28, 2001; accepted January 10, 2002.
Table 1. Reagents used in the present study Buffers Basic buffer 0.05 mol/L Tris (pH 7.4), 0.15 mol/ L NaCl, 0.02 mol/L diethylenetriamine pentaacetic acid, 5 mg/L tartrazine, 1 g/L Germall II, 10 mg/L Triton X-100, 5 g/L BSA, 0.5 g/L bovine IgG Modified basic buffer Basic buffer without bovine IgG, and BSA replaced with human serum albumin Basic buffer with native MAK33 Basic buffer with 15 mg/L of native mouse monoclonal antibody MAK33 added Basic buffer with heat-treated Basic buffer with 15 mg/L of MAK33 heat-treated (60[degrees]C for 10 min) mouse monoclonal antibody MAK33 added Solid-phase antibody Streptavidin plates coated with: Whole IgG Biotinylated whole IgG of monoclonal antibody 12-140-10 F[(ab').sub.2] fragments Biotinylated [F(ab').sub.2] fragments of monoclonal antibody 12-140-10 Assay combinations Screening assay Whole IgG + basic buffer (automated) Reference assay F[(ab').sub.2] fragments + basic buffer with heat-treated MAK33 (automated) Modified screening assay Whole IgG + modified basic buffer (automated) Manual assays 1. Whole IgG + basic buffer 2. Whole IgG + basic buffer with native MAK33 3. Whole IgG + basic buffer with heat-treated MAK33 4. F[(ab').sub.2] fragments + basic buffer 5. F[(ab').sub.2] fragments + basic buffer with native MAK33 6. F[(ab').sub.2] fragments + basic buffer with heat-treated MAK33 Table 2. Classification of interference and the measures needed to eliminate it. No. of Frequency, (a) patients % Type I Results normalized after 4 0.08 (0-0.15) removal of Fc fragment or after addition of 15 mg-L native MAK33 or 15 mg-L heat-treated MAK33 Type II Results normalized after 148 3.0 (2.5-3.6) removal of Fc fragment or after addition of 15 mg/L heat/treated MAK33, but not after addition of 15 mg/L native MAK33 Type III Results normalized after 41 0.82 (0.58-1.09) removal of Fc fragment, but not after addition of 15 mg-L heat-treated MAK33 (results for higher concentrations in the text) Type IV Results normalized after 3 0.06 (0-0.13) addition of 15 mg-L heat- treated MAK33, but not after removal of Fc fragment Type V Results normalized only after 1 0.02 (0-0.062) both removal of Fc fragment and addition of 15 mg-L heat-treated MAK33 Type VI Higher results in the 1 0.02 (0-0.062) reference assay than in the screening assay. Dilution experiment indicated interference still after both removal of Fc fragment and addition of 15 mg-L heat-treated MAK33 Serum missing for 12 classification Total 210 4.0 (3.3-4.7) (a) Confidence intervals (95%) were achieved by computer simulation, assigning a Poisson distribution for the occurrence of interference and a binomial distribution for the random selection of patients. Table 3. Comparison of the sample control group and the group of interference-positive patients. Men Women Total Test group, n 108 102 210 Control group, n 87 99 186 Total, n 195 201 Relative risk 1.10 0.91 (0.88-1.39) (a) (0.72-1.14) (a) Logistic regression Age P <0.23 Sex P <0.40 (a) 95% confidence interval.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Enzymes and Protein Markers|
|Author:||Bjerner, Johan; Nustad, Kjell; Norum, Lars F.; Olsen, Kari Hauge; Bormer, Ole P.|
|Date:||Apr 1, 2002|
|Previous Article:||Erythrocyte galactose 1-phosphate quantified by isotope-dilution gas chromatography--mass spectrometry.|
|Next Article:||Biochemical and genetic markers of iron status and the risk of coronary artery disease: an angiography-based study.|