Techniques for Identifying Heterophile Antibody Inter- ference Are Assay Specific: Study of Seven Analytes on Two Automated Immunoassay Analyzers, Margarette

Interference by human anti-animal immunoglobulins, commonly referred to as heterophile antibodies, in immunologic assays is known to be an important consideration for medical testing laboratories (1–3). Although automated immunometric assays are formulated to reduce these effects, it is unlikely that complete elimination occurs (2 ), and artifactual results attributable to heterophile antibodies have been reported for some assays (4, 5). Such results often are identified by addition of blocking agents to the samples before assay. A simple sample pretreatment method uses a commercially available blocking tube, HBT. An alternative technique uses polyethylene glycol (PEG) to precipitate immunoglobulin-sized molecules before assay. For both of these techniques, a difference between values for the treated and untreated specimens is interpreted as evidence for heterophile antibody interference. It is not clear that either of these methods is appropriate for every immunoassay. It is important to know the effect of sample pretreatment on the results of each assay. This has been done recently for HBT and a thyroglobulin assay (4 ). We examined seven automated analyzer hormone assays, using pretreatments with both HBT and PEG of samples from healthy adults with the aim of determining the expected change in results post treatment for both techniques for each assay. We investigated luteinizing hormone (LH), folliclestimulating hormone (FSH), prolactin (PRL), and growth hormone (GH) on the Access2 analyzer and insulin, parathyroid hormone (PTH), and cross-linked C-terminal telopeptide of type I collagen ( -CTx; CrossLaps) assays on the Elecsys analyzer. Analyzer reagents and consumables were obtained from Beckman-Coulter and Roche Diagnostics, respectively. Plasma (EDTA) samples were collected from healthy individuals in a study approved by the Canterbury Ethics Committee, New Zealand. Samples (n 103–113) were assayed directly and after pretreatment of a 0.5-mL aliquot in blocking tubes (HBT; Scantibodies) at room temperature for 1 h. A third aliquot was treated with an equal volume of PEG solution (PEG 6000; 250 g/L in 0.05 mol/L phosphate buffer, pH 7.4, containing 0.5 g/L Triton X-100), and the supernatant obtained after centrifugation was assayed. All tests for each sample were performed on the same day. Results are expressed as percentage recovery of hormone in the treated compared with the untreated sample aliquots. A dilution factor of 2 was used for PEG recovery calculations. We tested additional samples serum and EDTA-plasma specimens from an individual whose plasma had shown extremely high values for PRL, LH, FSH, and GH after HBT treatment. These were also tested for LH, FSH, and PRL by a different methodology (Abbott Architect) and with the Access assay after 1-in-2, 1-in-4, and 1-in-8 dilution of the untreated and HBT-treated samples. The concentrations in treated samples as percentages of the concentrations in the nontreated samples are shown in Fig. 1. For both HBT (Fig. 1A) and PEG treatment (Fig. 1B), the values differed from 100%. The median (CV) values after HBT treatment were 89 (8)% for PTH, 100 (7)% for -CTx, 99 (6)% for insulin, 81 (130)% for LH, 73 (38)% for FSH, 92 (24)% for PRL, and 98 (400)% for GH. The median values after PEG treatment were 179 (22)% for PTH, 188 (22)% for -CTx, 111 (7)% for insulin, 53 (15)% for LH, 95 (15)% for FSH, 100 (14)% for PRL, and 125 (21)% for GH. These data can be used to establish the expected values applicable to each pretreatment technique. Such reference information can be derived even when there is not “recovery” of 100%; for example, a target value for PTH recovery after HBT treatment would be 89%, and a reference interval could be calculated nonparametrically or by other appropriate techniques. A reasonably tight distribution of recovery values is, however, required. If a CV of 10% is regarded as acceptable, our data suggest that HBT pretreatment is a suitable method of testing for heterophile antibody interference in the Elecsys PTH, -CTx, and insulin assays. Similarly, PEG treatment is acceptable for the Elecsys insulin and the Access FSH assays. For both techniques, validation using specimens with true positive or negative interference would be desirable. These conclusions assume that there was no contribution to the distributions from interfering heterophile antibodies or other substances in these presumably normal specimens. The Access PRL assay after PEG treatment may be suitable despite the CV of 14% because PEG treatment is frequently used to detect macroprolactin interference (6 ). It is possible that two samples with low recoveries contained macroprolactin. Our experience with this assay is that samples with PEG recovery values 60% contain high-molecular-weight prolactin immunoreactivity after gel filtration. The considerable variability seen in the Elecsys PTH and -CTx and the Access GH assays after PEG treatment probably reflects interference by the PEG in the antibody– antigen reactions of these assays. Similarly, although we do not know the mechanism, it appears that HBT treatment can produce a spuriously high recovery of hormone in some assays. The manufacturers state that HBT contains a “unique blocking agent” limited to use for antigen assays to confirm or disqualify an original result in conjunction with other data (such as symptoms and other testing). Our data suggest a further limitation that its use is assay specific, possibly dependent on the assay configuration. We did not observe overrecovery in the three Elecsys assays (all configured with mouse monoclonal antibodies for both capture and detection). It was, however, apparent in the Access assays (LH, FSH, GH, and to a lesser extent, PRL), which contain solid-phase goat Technical Briefs