Inter-run variation of luminex single-antigen assay for the detection of antibodies to HLA

Abstract

Aim Luminex single-antigen assay (SA) utilizes multiplexed beads to detect specific antibodies to HLA. The approach to quality control during the daily operation in an HLA laboratory has not been well defined. This study aims to examine the inter-run variation of MFIs using a stable positive control (PC) specimen, and to explore possible control rules (CRs) for judging the acceptability of each run. Methods To control for multiple beads with MFIs flanking an internal MFI cutoff of 2000, sera of highly sensitized patients were mixed, aliquoted and frozen as PC material. During a 7-week period, one aliquot was thawed each week and tested at 1:100 dilution in all runs. MFIs adjusted for background were compiled and analyzed. Bead-specific means and standard deviations (SDs) from the first 20 runs were used to develop empirical CRs, which were applied to subsequent runs to calculate rejection rate. The means and coefficients of variation (CVs) were also compared among different operators. Results SDs increased and CVs decreased in linear relations with mean MFIs on log-log plot for both class I (Fig. 1A) and II beads (p 0.0001). For the first 20 runs, the CVs (median and interquartile range) for all class I and II beads were 24% (21–28%) and 21% (14–34%) respectively; for beads with MFIs of 1000–3000, the CVs were 22.5% (21–23%) and 14% (13–19%) respectively. If accepting runs with 2 beads exceeding the mean ± 2SD limit, 1 each of class I (Fig. 1B) and II run must be rejected out of 19 and 17 runs respectively. If accepting runs with 4 beads exceeding the mean ± 1SD limit, 53% and 41% of class I and II runs must be rejected. Among 6 operators, the mean MFIs and CVs for class I and II beads were different from each other (p 0.0001); 3 operators with CVs significantly higher than the overall CVs were identified. Conclusion Our approach allows evaluation of inter-run variations over a range of MFIs and the performance of different operators. A CR of mean ± 2SD may help reject occasional, unqualified runs while maintaining analytical quality within a stable range. Download : Download full-size image