Effect of amino acid mismatch in the UNOS dataset.

This study began with the 2010 UNOS data-set of 181,653 deceased donor kidney transplant cases and 92,577 living donor cases. Cases with ambiguous or missing HLA typing were excluded, and the remaining cases were split into subgroups by the number of previous transplants and ethnic groups of donor-patient pairs. 41,128 Caucasian donor-patient pairs that were primary living-donor transplant cases were used as the pilot population to identify potential epitope groups that have a negative effect on graft outcome. Sixty four of the most common HLA-A and -B antigens were selected. Amino acid sequences of the most frequently corresponding allele in the Caucasian population were used to build the starting theoretical epitope table. Amino acids of the 115 polymorphic positions, analyzed in one, two or three positions, resulted in 15,801,920 combinations. After eliminating combinations shared by no allele or by all 64 alleles, the table was trimmed to 1,635,044. Grouping combinations according to their antigen list (antigens that share the combinations/epitopes), 40,830 epitope groups were left. Based on the distances between amino acid positions of each epitope, and the requirement that each epitope must be shared by at least one allele, but not all 64, the number of theoretical epitopes was reduced to 39,670 and 3,703 epitope groups of unique antigen lists. The pilot population was composed of 41,128 primary living-donor transplants with Caucasian donor-patient pairs. For each of the 40,830 epitope groups in the non-distance-restricted table, 15-year death-censored graft survival was computed for epitope-group mismatches--i.e., cases with a BMQ0001 mismatch, with a BMQ0002 mismatch, etc.. Results were compared, using the log rank test, with average graft survival. Of the 3,703 epitope groups, 2,487 appeared in over 1000 cases, but only 88 of them had significant p-values, which ranged from 0.006 to 0.049, with 76 of the 88 significantly below average, 12 above average (Fig. 1). We then ran survival analyses taking the 76 epitope groups that were below average two at a time--i.e., cases with mismatches of the first and second epitope group, the first and third, first and fourth, etc. Of more than 2,500 pairs, 148 resulted in significantly (p < 0.01) lower than average survival in those primary living-donor cases. The effect of the 76 epitope-group mismatches that showed below-average results was then analyzed for other transplant populations--Caucasian donor-patient pair cases with deceased donors , Caucasian donor-patient pair cases with re-transplant living donors, Caucasian donor-patient pair cases with re-transplant deceased donors, and African-American donor-patient pair cases with primary living donors. None of these four populations exhibited any significant effect due to the 76 epitope- group mismatches. Likewise, the effect of the 148 epitope-group combination mismatches detailed in paragraph 5, above, was analyzed for the other four other transplant populations, detailed in paragraph 6. That significant effect was absent in all four. The analyses were repeated on the 40,830 epitope groups without the 27 angstrom distance constraint. Of the 40,830 epitope groups, 439 exhibited a significantly lower 15-year graft survival, with p-value ranging from 0.0053 to 0.0498. Again, none of these had any negatively significant effects on graft survival for the other four transplant populations. In the pilot population, the negative effect of the epitope group mismatches was clearly seen, but the significant differences did not carry across to the other four populations. That absence may be partially explained by the allele level differences in the HLA-A and -B typing of the donors and patients. Past studies indicate that the appearance of DSA has a negative effect on the graft outcome, so the mismatch of epitopes recognized by these DSA could also have similar negative effects. With the data available at present, and with the currently available assays for antibody detection, we are not able to analyze the impact of specific epitope mismatches. We will need the development of new methods of antibody detection that specifically indicate the exact epitope to which an antibody binds before we can continue this effort to determine the negative impact on graft survival due to epitope mismatches.