Postdoctoral Fellow University of North Carolina, Chapel Hill Chapel Hill, North Carolina
Aim: Our study aims to bioinformatically identify discordant lot-to-lot variation and assess a functional modality to reduce non-specific reactivity (NSR) in the One Lambda single antigen bead (SAB) assay.
Methods: Patient sera was incubated with 25mM EDTA and 10% fetal bovine serum (FBS), shaking at 1000 RPM, for 15 minutes, at room temperature. The SAB assay was performed via the rapid protocol described by Liwiski et al. LabScreen Mixed was performed per vendor instructions. To evaluate lot-to-lot variability more systematically, we employed an unbiased statistical approach known as principal component analysis (PCA), a method that reduces dimensionality while preserving variance.
Results: Although One Lambda’s Lot 17 passed our initial QC validation, we observed multiple beads that appeared falsely reactive, most notably DR1, DR10, and DR12. Using PCA, the beads with the largest Euclidean distance between Lot 16 and Lots 15, 17, and 17+FBS were identified (Fig 1a). Next, we investigated strategies to reduce NSR. FBS is commonly used to improve specificity in protein-based assays. Accordingly, pre-treatment with FBS eliminated the NSR observed with Lot 17. This effect was not attributable to dilution alone as PBS had no impact (Fig 1b). Importantly, the addition of FBS did not impair detection of true DR1, DR10, or DR12-specific antibodies. Sera from patients with confirmed antibodies to these antigens was tested with and without FBS, and no reduction in signal was observed (Fig 1c). Next, we applied PCA to Lot 17+FBS. Our data demonstrate a significant reduction in the Euclidean distance with pre-treatment of FBS, indicative of improved concordance (Fig 1a, d). Finally, we asked if FBS was sufficient to limit NSR in the mixed screen assay in an effort to decrease the number of positive samples reflexed to SAB. Indeed, FBS significantly reduced the number of reflex samples, thus improving assay specificity (Fig 1e).
Conclusion: As more labs rely solely on virtual crossmatching for transplant risk assessments, ensuring lot-to-lot consistency will be paramount. PCA offers the advantage of quantitatively and unbiasedly detecting outliers between lots, suggesting its utility as an added layer of quality control. Collectively, we show PCA can identify lot-to-lot variability and that FBS reduces NSR and may offer cost savings for the laboratory by reducing the number of repeat and reflex samples.
