Anouk Donners

93 LC-MS/MS-method validation quantifying emicizumab The influence of anti-FVIII antibodies and FVIII on the emicizumab concentration differences obtained by both methods was assessed. The absolute differences were similar (p = 0.30) for samples in presence (n = 19) and absence (n = 58) of antiFVIII antibodies. The absolute differences were also similar (p = 0.17) for samples in presence (n = 19) and absence (n = 58) of FVIII. The presence of both covariates resulted in minor increases of mcOSA results (positive absolute differences), but these were neither statistically significant nor clinically relevant. This is in line with reported spike experiments; despite a 1:80 dilution, mcOSA remains sensitive to the presence of replacement FVIII or endogenous FVIII and cannot be made completely specific to emicizumab by using higher dilutions [33]. Especially in a clinical setting, patients receive large amounts of FVIII products during bleeding episodes or peri-operative periods during emicizumab therapy [34]. Fortunately, the LC-MS/MS method is unaffected by FVIII interference owing to its principle, which is one of the strengths of this method. The ideal comparator for the LC-MS/MS method would be a total ELISA; that is, an ELISA with a preceding dissociation step to release the drug from any other potential binding target. This classical cross-validation approach cannot be applied in our study, because the sole existing ELISA, which was used in the HAVEN studies [20-23], detects only the free, dual-binding competent drug and cannot detect emicizumab in complex with either FIXa or FX. Nevertheless, a cross-validation can still be of value to determine whether the data obtained are reliable and can be compared between laboratories. As the ELISA from the HAVEN studies was not commercially available, the LC-MS/MS method was compared with the standard mcOSA. The principle of this type of OSA-based assay relies upon measuring emicizumab activity as a factor VIII mimetic and is based on clotting (enzymatic) reactions in FVIII-deficient plasma [24]. In contrast, the principle of the LC-MS/MS method relies upon measuring the exact amount of a signature peptide of emicizumab per sample using SIL-IS for quantification. These different principles explain the slight negative trend in absolute differences above 50 µg/mL (Figure 2A). Despite the fundamental differences, a very strong correlation between the methods was found. The 95% CI of the intercept contained ‘zero’ in the weighted Deming regression fit, and the 95% CI of the slope contained ‘one’ (Figure 1). The relative differences were well within the EMA’s acceptance criteria for cross-validation (Figure 2B). Therefore, the application of a correction factor for interchangeable method use is not required. The six samples with relative difference of >20% had emicizumab concentrations below 34 µg/mL. These six samples were obtained during the loading phase, because concentrations in maintenance phase range between 38 and 67 µg/mL [11, 35]. While relative differences are of analytical importance, they have low clinical value. The absolute differences were small, especially from a clinicians’ perspective, and relevant outliers or trends were lacking. The EMA criterion was evidently met, making this finding not clinically relevant. 5