Elecsys Insulin Assay: Free Insulin Determination and the Absence of Cross-Reactivity with Insulin Lispro

Insulin assays are used in the identification of rare hypoglycemic syndromes, and may be useful in classifying the different types of diabetes (1)(2)(3). Hemolysis (4)(5)(6) and circulating anti-insulin antibodies (7)(8) are sources of interference in insulin determinations. Anti-insulin antibodies may be present in the serum of insulin-treated diabetic patients, even when they are treated with biosynthetic human insulin, and may also be found in the serum of type 1 diabetic patients before insulin administration. These antibodies interfere in competitive and noncompetitive insulin immunoassays. In the increasingly used two-site noncompetitive assays, anti-insulin antibodies may yield increased results. Thus, anti-insulin antibodies must be eliminated, usually through precipitation with polyethylene glycol, before the concentration of free (active) insulin is determined (9). In addition to these considerations, both biosynthetic human insulin and insulin lispro, a fully potent analog with a faster absorption rate, are used therapeutically so that not only human endogenous and exogenous insulin, but also insulin lispro may be present in samples in which the insulin assay is to be used. In the present study we performed a preliminary evaluation of an insulin electrochemiluminescence immunoassay recently developed by Roche Diagnostics (Mannheim, Germany) for the Elecsys® analyzer. We studied the analytical performance of this new insulin assay and assessed the results of insulin measurements in serum samples with and without anti-insulin antibodies. The Elecsys assay uses an 18-min incubation time and a 20-μL sample volume, and has a stated dynamic range of 0.2–1000 mIU/L (1.39–6945 pmol/L). Assays were performed on the Elecsys 2010 analyzer in singleton, according to the manufacturer’s instructions, with two different lot numbers. The Elecsys results were compared with those of our routine method (Bi-Insulin IRMA; Bio-Rad) in EDTA-plasma samples from 110 diabetic patients treated with oral hypoglycemic drugs or human insulin, and samples from 7 patients treated with insulin lispro. The Elecsys and Bio-Rad assays are two-site noncompetitive assays, using monoclonal antibodies (mAbs), with stated negligible cross-reactivity with intact proinsulin. The Bio-Rad assay is a one-step immunometric assay with one antibody coated on the assay tube and another labeled with 125I. In the Elecsys assay, in a first 9-min incubation, plasma insulin, a biotinylated anti-insulin antibody, and a second monoclonal anti-insulin antibody labeled with an electrochemiluminescent ruthenium complex react to form a sandwich complex. After the addition of streptavidin-coated microparticles, during a 9-min incubation phase, the complex binds to the solid phase. Both assays are calibrated against the WHO 66/304 reference preparation. For free insulin measurements by both methods, we used the supernatant obtained after polyethylene glycol precipitation with the Bio-Rad precipitating solution according to the Bio-Rad free-insulin protocol. The presence of free anti-insulin antibodies (not complexed with circulating insulin) was determined using a radiobinding assay in liquid phase (Bio-Rad). We fixed the positivity threshold at 2.5%, the +3 SD value of the negative control serum included with the assay. The procedures were in accordance with the Helsinki Declaration of 1975 and the subsequent 1996 amendments. Interassay reproducibility was assessed over a 3-month period through repeated analysis (n = 11) of three control sera from Roche Diagnostics. CVs were 2.2–2.7% at mean values of 14.8, 44.6, and 104 mIU/L. One serum sample (142 mIU/L) was diluted 2- to 16-fold in Universal Diluent (Roche Diagnostics). Measured values were 93–101% of expected. The proinsulin cross-reactivity indicated by the manufacturer is 0.05%. The insulin concentration of the proinsulin standard in the Dako intact proinsulin reagent set (76.4 pmol/L) measured with the Elecsys method was <0.2 mIU/L. For the Elecsys and Bio-Rad assays, respectively, the manufacturers indicate cross-reactivities of 20% and <0.1% with des-31,32-proinsulin, the major proinsulin conversion intermediate fragment present in serum, and 0.1% and 100% with the barely detectable des-64,65-proinsulin fragment (Bio-Rad package insert and Roche Diagnostics, personal communication). Among the 110 samples from the patients treated with oral hypoglycemic drugs or human insulin, 30 did not contain anti-insulin antibodies. For these 30 samples, the correlation coefficients for the Elecsys direct and free insulin results with the Bio-Rad results were 0.98 (Sy|x = 2.94) and 0.97 (Sy|x =2.95), respectively (Fig. 1A ), and the direct:free insulin ratio (mean ± SE) was close to 1 (1.05 ± 0.01 for the Elecsys assay; 0.95 ± 0.03 for the Bio-Rad assay). The Elecsys assay yielded slightly higher results (∼15–20%) than the Bio-Rad assay. This difference might in part be attributable to the cross-reactivity of the Elecsys assay with des-31,32-proinsulin. Elecsys vs Bio-Rad insulin measurements. (A), direct (▴) and free (▵) insulin concentrations in 30 EDTA-plasma samples without anti-insulin antibodies. The solid line represents the least-squares regression line for Elecsys vs Bio-Rad direct insulin measurements: Elecsys = [1.23 (1.13–1.33) × Bio-Rad] − 0.67 (−2.3 to 1.0). The dashed line corresponds to the least-squares regression line for Elecsys vs Bio-Rad free insulin measurements: Elecsys = [1.16 (1.05–1.27) × Bio-Rad] − 1.10 (−2.9 to 0.7). (B), direct (▴) and free (▵) insulin concentrations in 80 EDTA-plasma samples with anti-insulin antibodies. The solid line represents the least-squares regression line for Elecsys vs Bio-Rad direct insulin measurements: Elecsys = [2.45 (2.04–2.86) × Bio-Rad] − 9.7 (−32.1 to 12.6). The dashed line corresponds to the least-squares regression line for Elecsys vs Bio-Rad free insulin measurements: Elecsys = [1.13 (1.05–1.21) × Bio-Rad] + 0.10 (−1.0 to 1.2). Three direct insulin data points with the following coordinates were not reported on the graph: Bio-Rad 111, 154, and 306 mIU/L; Elecsys 502, 930, and 428 mIU/L. Standardization of immunoassays for insulin remains a problem despite the availability of large quantities of human insulin through recombinant DNA technology (2). A task force on the standardization of insulin assays noted a significant variability in insulin results and suggested a process for the assessment and certification of insulin assays. To date, in the absence of a reference method, this process has not been followed (10). Because direct and free insulin measurements were equivalent in samples without anti-insulin antibodies, free insulin measurements are unnecessary in such samples. In the 80 samples with anti-insulin antibodies, the correlation (r) for direct insulin measurements by the two methods was 0.80 (Sy|x = 19.0), but the correlation of free insulin concentrations measured by the two assays was much better (r = 0.96; Sy|x = 1.60; Fig. 1B ). The direct:free insulin ratios (mean ± SE) were >1 (Elecsys, 5.0 ± 1.3; Bio-Rad, 2.6 ± 0.3), consistent with the anti-insulin antibody interference in the two direct assays. This interference was more marked in the Elecsys assay than in the Bio-Rad assay, and its intersample variability decreased the correlation between the results of the two direct determinations. To correctly estimate active insulin concentrations, free insulin measurements are necessary in these samples. In the seven samples from patients treated with insulin lispro, the Elecsys and Bio-Rad results for direct or free insulin measurements were discordant. Elecsys results were very low or below the detection limit (Table 1 ). This observation prompted us to investigate the cross-reactivity of insulin lispro in the Elecsys assay. For this purpose, we added up to 1000 mIU/L insulin lispro (Humalog®; Lilly France) to the Roche Universal Diluent (Elecsys insulin concentration <0.2 mIU/L). The cross-reactivity of insulin lispro was <0.02%. The different cross-reactivity of insulin lispro in the Bio-Rad assay (75%) (3) explains the discrepancies we observed in samples from patients treated with this insulin analog. Direct and free insulin concentrations measured with the Elecsys and Bio-Rad immunometric assays in seven EDTA-plasma samples1 from patients treated with insulin lispro. The seven samples contained anti-insulin antibodies as determined by the Bio-Rad radiobinding assay result (positivity threshold, 2.5%). Direct and free insulin concentrations measured with the Elecsys and Bio-Rad immunometric assays in seven EDTA-plasma samples1 from patients treated with insulin lispro. The seven samples contained anti-insulin antibodies as determined by the Bio-Rad radiobinding assay result (positivity threshold, 2.5%). The difference in cross-reactivities reflects the antibodies used in the assays. The Bio-Rad insulin assay uses two mAbs, 19 and P10, which recognize two distinct regions on the insulin molecule (11). mAb 19 binds to a highly conserved region, including the A10–A17 residues. It is worth noting that the cross-reactivities of the two assays with pork insulin are also different: 100% for the Bio-Rad assay and 19.2% for the Elecsys assay (from package inserts). Pork insulin differs from the human hormone by only a single residue (B30) and insulin lispro by two residues from the same C-terminal region of the B chain (permutation of B28 and B29 residues). The Elecsys method, which poorly recognized pork and insulin lispro, uses two mAbs: one, MAK-Bi (R1), recognizes the A7–A10 portion of the A chain; and the other, Fab-Ru (R2), recognizes the C-terminal part of the B chain (Roche Diagnostics, personal communication). The specificity of the latter antibody is consistent with poor recognition of pork insulin and insulin lispro by the Elecsys assay. The automated Elecsys insulin immunoassay is a rapid method for insulin determination in serum samples without anti-insulin antibodies. The low imprecision of this method could be useful in insulin pulsatility studies. In samples containing anti-insulin antibodies, free insulin measurements can be performed with the Elecsys method after antibodies have been eliminated by precipitation with polyethylene glycol. Clinicians should be aware of the absence of cross-reactivity of insulin lispro in this new assay. This insulin analog can be measured with a specific assay (12). We are grateful to Roche Diagnostics for providing us with the insulin reagent sets free of charge. This work was supported in part by the Hôpitaux Universitaires de Strasbourg. We thank N. Heider for carefully reviewing the English in the manuscript, and N. Martz and E. Fischbach for excellent technical assistance.