Serological reactivity of HLA alleles is dependent on race/ethnicity

Aim: Many HLA alleles, especially those common in non-European groups, do not have a clear serological assignment, or are not represented on routine solid phase antibody tests. We determined the concordance in reactivity between serological equivalents, across patients from different racial and ethnic backgrounds.

Methods: 169 sera in were tested by ExPlex (OneLambda). Validation Cohort: 15 reference sera from multiparous donors; 26 patients with known allele-specific reactivity. Cross-sectional Cohort: adult renal transplant candidates, cPRA 48-100%; 81 from UCLA (22 Asian/Pacific Islander [A], 13 Black/African American [B], 49 Hispanic/Latino [H], and 23 European [E]), 42 from Emory (100% B).
Serological equivalency was merged from HLA Dictionary and UNOS Equivalency Tables. Correlation in MFI between two alleles assigned the same antigen was assessed using simple linear regression, and correlation coefficient (R2) was calculated.

Results: Across 131 allele pairs assigned the same antigen, 59 alleles showed excellent MFI correlation (R2≥0.95). 23 alleles showed good but not perfect (R2 0.90-0.95) correlation, with outliers in specific sera. 34 alleles had poorer correlation (R2 0.70-0.90), with more individual sera substantially diverging. And, 15 alleles demonstrated no concordance (R2 <0.70) (Fig 1). Although these outlier alleles with highly discordant reactivity had a greater number of eplet differences as determined by HLAMatchmaker, overall there was no association between the number of disparate eplets between two alleles and the R2 of serological reactivity.

We tested if different populations of patients exhibit unique patterns of serological reactivity to alleles only frequent in their own population. For B71 (B*15:10 vs. 15:18), correlation was found in B and H patients (R2>0.90), but less in E and A patients (R2=0.7883 and 0.5938, respectively). Similarly, the correlation between A*02:01 and A*02:05 sera from B patients was only R2=0.600, vs. a much better correlation in non-B (R2=0.963). The same trend was observed for other alleles (Fig 2).

Conclusion: Alleles with the lowest concordance in real-world serology are nonetheless considered equivalent to the parent antigen according to UNOS tables. Further, workshops defining HLA serology need to include sera from individuals representative of all global populations. Finally, the current routine SAB panels are not fully able to profile reactivity in non-E patients.