Assessment of Complement Interference in Anti-Müllerian Hormone (AMH) Immunoassays

Anti-Müllerian hormone (AMH) is a dimeric glycoprotein produced by ovarian granulosa cells and used as a marker of ovarian reserve. Since 2011, the Beckman AMH Gen II assay has been widely used in research and clinical settings. This assay was reported to be affected by complement interference due to the binding of C1q to the capture antibody, resulting in C3 recruitment and activation of complement cascade. Steric hindrance from this complex prevents AMH binding, resulting in falsely lowered values in freshly drawn or freshly frozen samples. Introduction of a sample predilution step by the manufacturer prior to incubation with the capture antibody resolved the interference by preventing complement binding. Recently, automated high-throughput AMH immunoassays for Beckman Access and Roche Elecsys have been introduced. According to the 2017 CAP survey, 83% of labs that offer AMH testing use one of these automated platforms. Employing the same antibody pairs as the Gen II assay, these platforms may be susceptible to complement interference; however, no direct evaluation of complement interference has been published. In this study, we assessed complement interference in the Beckman and Roche immunoassays. Spike-recovery experiments were performed in serum samples from females (>55 years old) with known C1q activity (n = 24). AMH recovery was compared to C1q activity in fresh and heat-inactivated samples. In samples with C1q activity within or above the reference interval (34–63 U/mL), complement interference in the Gen II assay was evident by a mean recovery of 41.0% (range, 28.1%-57.9%) followed by an increase in recovery to 77.0% (71.7%-84.2%) upon heat inactivation (P < .01). Complement inactivation was confirmed by measuring C1q activity. C1q activity was 60.3 U/mL (54.9–67.1 U/mL) before and 3.1 U/mL (0–12 U/mL) after heat inactivation (P < .01). Heat-inactivated recoveries corresponded with the recoveries seen in the Gen II assay with the predilution step, 71.6% (38.9%-81.5%) (P = .08). In samples with C1q activity below the reference interval, similar recoveries were observed with or without the predilution step with mean recoveries of 71.7% (67.5%-72.9%) and 77.7% (75.8%-77.7%), respectively (P = .07). In the Beckman and Roche assays, AMH recoveries were 111.0% (90.0%-125.0%) and 97.3% (74.6%-109.5%), respectively. Heat inactivation did not affect AMH recoveries significantly with post–heat inactivation recoveries of 93.3% (82.3%-99.9%, P < .01) for Beckman and 93.1% (83.6%-107.3%, P = .2) for Roche. Our study shows that C1q activity is responsible for AMH underrecovery on the Gen II assay when samples are not prediluted. Although there appears to be a C1q threshold for complement interference, the degree of underrecovery was independent of the level of complement activity once the activity was within the reference interval. Furthermore, we provide direct evidence that the automated Beckman and Roche AMH assays are not affected by complement interference despite using the same antibody pair as the Gen II assay.