Interference From Anti-Streptavidin Antibody

All testing was performed for patient-care purposes using 4 serum samples obtained from a single patient during a 7-month period. Control sera were obtained from presumably unrelated individuals.

The TSH, total thyroxine, T-uptake, testosterone, prolactin, follicle-stimulating hormone, luteinizing hormone, and cortisol levels were measured using a Roche Diagnostics Elecsys 2010 or E170. Assays on these platforms use streptavidin-biotin capture and ruthenium complex electrochemiluminescent detection (Figure, A through D). Additional TSH, total thyroxine, testosterone, and prolactin measurements were obtained using a Siemens ADVIA Centaur XP and an Ortho Vitros ECiQ. The Centaur uses antibodies covalently linked to paramagnetic particles for separation with an acridinium ester label to generate a chemiluminescent signal. Immunoassays on the Vitros ECiQ use either streptavidin-coated wells to capture biotinylated and horseradish peroxidase-labeled immune complexes or antibody-coated wells to capture horseradish peroxidase-labeled analyte. Alkaline phosphatase was measured on the Beckman Coulter DxC (Beckman Coulter Inc, Brea, California), using a kinetic rate method on p-nitrophenylphosphate substrate in 2-amino-2-methyl-1-propanol buffer (Beckman Coulter Inc, Brea, California).

Serum protein electrophoresis was performed on a SPIFE Split Beta system (Helena Laboratories, Beaumont, Texas).

Heterophile blocking tubes (HBT; Scantibodies Laboratories Inc, Santee, California) were used to investigate possible interference because of antibody directed against the animal reagent antibodies used in immunoassays. To ensure that the magnitude of the interfering antibody did not exceed the binding capacity of the HBT, the experiment was repeated with manually diluted (1:10) specimen. An additional procedure was provided by Roche Diagnostics to specifically investigate human anti-mouse antibodies interference in the TSH assay. This Roche procedure involved the incubation of 25 μL (0.025 mg) mouse immunoglobulin G (IgG; Sigma Aldrich, St. Louis, Missouri), or Roche Diluent MultiAssay as a control, with 250μl patient sample for one hour at room temperature before analysis on the Roche platform.

To characterize the nature of the assay interference, patient or control sera were treated with protein A–linked sepharose, streptavidin-linked agarose, or agarose (control). All sepharose/agarose materials were prepared for use as follows: 0.25 g of lyophilized 4% cross-linked protein-A sepharose (PAS; GE Healthcare Bio-Sciences Corp, Piscataway, New Jersey) was washed 4 times with 50 mL of deionized water followed by phosphate-buffered saline before use with patient serum. The 4% cross-linked, control agarose and the 6% cross-linked, high-capacity streptavidin agarose (Pierce Biotechnology, Thermo Scientific Inc, Rockford, Illinois) were purchased as 50% slurries and were similarly washed repeatedly in deionized water, followed by phosphate-buffered saline, before use, and the 50:50 slurry of agarose to phosphate-buffered saline was centrifuged (500g for 4% cross-linked beads, 2000g for 6% cross-linked beads), and the supernatant was decanted, yielding an agarose pellet to be incubated with serum.

For initial experimentation, PAS was incubated with serum for 1 hour at room temperature. In all subsequent experiments, incubations were performed overnight on a rocker table. Time of incubation was increased because the 1-hour incubation did not appear to remove all interfering antibodies.

Manufacturer specifications included with the PAS suggested that the maximum-binding capacity of PAS was 20 mg of IgG per milliliter of sepharose. Based on a measured immunoglobulin concentration of 8.4 g/L, 0.5 mL of PAS was used to treat 1 mL of patient serum. This ratio of PAS to serum represents the 1× PAS treatment shown. Specimen was also treated with 2× (Table 1), 5×, and 10× (data not shown) PAS ratios. Streptavidin agarose binds at least 4 × 10⁻⁷ mol biotin per milliliter of settled resin. Assuming 4 biotin binding sites per streptavidin tetramer, that is equivalent to 1 × 10⁻⁷ mol/mL of resin, even in the conservative case where the antibody only recognizes the tetrameric form. Because the IgG concentration of the patient's serum was approximately 6 × 10⁻⁸ mol/mL, we used the same ratio of streptavidin agarose to patient serum as was used for PAS.

A 61-year-old man was referred to an endocrinologist at our institution after a diagnosis of hyperthyroidism subsequent to multiple measurements of TSH and free thyroxine on a Roche platform. Review of tests previously performed showed that, before treatment, he had repeatedly elevated free thyroxine and low, but not suppressed, TSH concentrations. Following treatment with methimazole, the free thyroxine decreased, and the TSH increased, but the TSH did not become elevated even when the free thyroxine was subnormal, and he had symptoms of hypothyroidism. The discrepancies between free thyroxine and TSH suggested the possibility that one or more of the laboratory results did not accurately represent his biologic status.

An initial T-uptake result of less than 0.2 thyroxine-binding index and a nonlinear TSH dilution suggested an immunoassay interference (Table 2). The possibility of heterophile antibody interference was investigated with an HBT. Incubation with HBT removed some, but not all interference, as seen in the posttreatment TSH results (Table 2).

To determine whether the interference was antibody-mediated, serum was treated with 1× PAS and incubated for 1 hour at room temperature (Table 2). Pretreatment and posttreatment TSH, TSH (1:10 dilution), total thyroxine, prolactin, luteinizing hormone, and follicle-stimulating hormone results showed significant differences posttreatment, suggesting an antibody-mediated interference with all of these assays. Of note, interference was seen in assays using monoclonal mouse as well as polyclonal sheep antibodies, providing further evidence against a single human anti-animal antibody. Treated specimen still showed a nonlinear dilution of TSH, suggesting that PAS treatment for 1 hour was not sufficient to remove all interfering antibody. However, no evidence of a monoclonal paraprotein was observed by serum protein electrophoresis, and γ-globulin concentration was determined to be 8.4 g/L (reference range, 5–16 g/L).

Protein-A sepharose treatment was repeated using increasing PAS concentrations and incubation time (Table 1). Because treatment of serum with the washed PAS pellet causes dilution with residual phosphate-buffered saline, alkaline phosphatase was measured using a different assay methodology and platform (enzymatic method on Beckman DxC) to estimate and correct for dilution of the sample. Linear dilution of TSH was observed after overnight PAS treatment, confirming the removal of all interfering antibody. In contrast, serum from control samples (obtained from an unrelated individual) showed equivalent results before and after sepharose treatment, after correction for dilution by alkaline phosphatase (data not shown).

Interestingly, a comparison of the PAS-treated and untreated specimens revealed an interference trend of falsely decreased results for all analytes measured by sandwich assay and falsely elevated results for analytes measured by competitive assay. This pattern was noted for protein and small molecule hormones of thyroid, nonthyroid function, pituitary, and nonpituitary origin (Tables 1 through 3).

To further characterize the nature of this interference, TSH was measured using a different immunoassay platform (Siemens Centaur). By this methodology, no significant, nonlinear dilution trend was observed for untreated serum (Table 1). Results from PAS-treated specimen on the Roche platform, when corrected for dilution, correlated well with the results from the untreated specimen on the Siemens Centaur, further supporting the conclusion that the Centaur was not affected by this interference (Table 1). These Centaur results were reported to the clinician, who discontinued treatment for hyperthyroidism, after which the patient's clinical status improved.

Untreated serum was also analyzed using an Ortho Vitros ECi. The TSH results showed a trend of nonlinear dilution and suppression relative to the Centaur results. This suggested the likelihood that the ECi TSH assay was subject to the same interferences as that of the Roche platform. In contrast, however, ECi T4 results correlated well with Centaur results and with results of the PAS-treated specimen analyzed on the Roche system, indicating that the ECi T4 was not susceptible to that interference (Table 1).

Roche Elecsys and E170 assays use a streptavidin-biotin capture in conjunction with ruthenium complex electrochemiluminescent detection. Similarly, the Ortho Vitros ECi TSH assay uses streptavidin-coated wells to capture biotinylated and horseradish peroxidase–labeled immune complexes. In contrast, neither the Centaur assays nor the ECi total T4 use streptavidin. This suggested the possibility that the patient's interference was due to an endogenous anti-streptavidin antibody. To test that hypothesis, patient serum was precleared using streptavidin-agarose (Table 4). Thyroid-stimulating hormone, TSH (1:10 dilution), T4, and cortisol were assayed on the Roche platform. The results of streptavidin-sepharose adsorption were comparable to those obtained by preclearing with PAS (Table 4), providing evidence that the specificity of the interfering antibody is against streptavidin. In contrast, results after treating with control agarose beads showed both suppressed TSH with a nonlinear dilution as well as elevated T4 after alkaline phosphatase correction, similar to an untreated specimen on this platform. This ruled out nonspecific binding to solid-phase agarose beads as the cause for the removal of the interference.

To confirm that streptavidin preclearing corrects the affected assays on multiple streptavidin-based platforms, treatment was repeated on a separate sample from the same patient (Table 3). Pretreatment and posttreatment TSH, total thyroxine, testosterone, and prolactin results were obtained on the Roche platform and the Ortho Vitros ECiQ for comparison against the Siemens Centaur results. As was previously seen for untreated specimens, both the Ortho and Roche TSH results were suppressed. Additionally, ECiQ testosterone results were increased and prolactin results were decreased compared with the Centaur results, consistent with the use of streptavidin in a competitive and sandwich assay, respectively. In all cases, the predicted suppression trends between treated and untreated specimens on the ECiQ and Centaur platforms were obtained, although the absolute concentrations of prolactin were noted to be higher on the ECiQ. To address this discrepancy, correlations of the prolactin assay between the Siemens Centaur and the Ortho Vitros ECiQ were obtained from technical representatives from both instrument manufacturers. Using the appropriate regression line, a Centaur result of 5 ng/mL was calculated to be equivalent to a mean result of 8.5 ng/mL on the Vitros ECiQ, thus explaining the difference in precleared/corrected prolactin results between the platforms.

For a final investigation of the possibility of heterophile antibodies, the HBT treatment and Roche human anti-mouse antibodies procedure were done on a separate patient sample (Table 5). Again, HBT treatment seemed to reduce the suppression of TSH results. However, incubation with mouse IgG had no effect, other than the expected, nonlinear dilution observed in all results from the patient's untreated specimens. This suggests that the interference is not an anti-mouse heterophile, and that the Scantibodies HBT may be nonspecifically blocking human immunoglobulin.

Sandwich and competitive immunoassays are commonly used in the laboratory for clinical diagnosis, although interferences have been documented.¹  On the Roche electrochemiluminescent platform, interference against the ruthenium signal complex has been previously reported.^2,3 

In this case, we have identified an anti-streptavidin antibody interference. Pretreatment with protein A corrected the nonlinear dilution in TSH results, characterizing the interference as immunoglobulin. The nature of interference with signal generation supports the identification of anti-streptavidin. For all assays evaluated, the patients' interference caused a falsely decreased signal, leading to low results from sandwich assays and elevated results from competitive assays. Decreased signal is likely cause by the interfering antibody limiting the availability for biotinylated components to bind the solid phase.

The patient's specimen demonstrated interference across manufacturer platforms, affecting all Roche assays tested and all streptavidin-mediated Ortho Diagnostics assays tested. Platforms and assays that do not employ streptavidin did not demonstrate nonlinear dilution or elevated competitive and suppressed sandwich results. Given the striking effect of this antibody on laboratory tests for an extended time, the clinical laboratory provided a formal letter cautioning the patient on the potential effects of this antibody on other laboratory testing and suggesting that any provider of future medical services be informed because of the risk of inappropriate treatment.

One notable element of this case is that the interference caused reciprocal changes in TSH and T4. It is important for the clinical laboratory to recognize that a single interference can yield opposite effects in different assays, depending on whether a sandwich or competitive format is used. A traditional algorithm for investigating immunoassay interference from species heterophiles or human anti-mouse antibodies would have failed to resolve these results because the interference was not fully removed by heterophile blocking, and affected assays spanned multiple species of reagent. This case also demonstrates the value of having diverse immunoassay platforms available to aid in the investigation of problematic or suspicious results, as interference against common laboratory assay components can have significant clinical consequences.

To our knowledge, this is the first confirmed incidence of endogenous streptavidin antibody reported in the literature. Khieng and Stevens⁸  reported a falsely elevated competitive assay result on the Roche platform, although their work did not demonstrate that the interference was immunoglobulin-mediated, and they could not rule out the possibility of endogenous biotin. Further, their report did not demonstrate that the interference affected more than one platform.⁸  The Roche assay package insert does caution that rare cases of human anti-mouse antibodies, anti-ruthenium or anti-streptavidin antibodies are possible limitations of the assays. Subsequent to our investigation, an aliquot of the patient's serum was sent to Roche Diagnostics in Germany, where streptavidin antibody interference in TSH, T4, T-uptake, prolactin, luteinizing hormone, follicle-stimulating hormone, cortisol, and prostate-specific antigen assays was confirmed by pretreatment with streptavidin beads. In all assays, pretreatment with streptavidin beads increased the signal generation, whereas treatment with uncoated beads did not alter the signal generation (data not shown).

Although it appears a streptavidin antibody is a rare occurrence, this interference may easily go undetected as it did for more than 2 years in this patient. A seroprevalence study may be indicated to evaluate the magnitude of this problem.