P159 Evaluation of differing methods for calculating cpra for kidney allocation

Aim Currently, there are two different approaches to calculating cPRA. The first, used in USA since 2009, is the expectation maximization (EM) algorithm. The second, utilized in Europe/Canada, is to simply observe the percentage of donors with a particular antigen (“count method”). Recently, the UNOS/OPTN Histocompatibility Committee has been re-evaluating the optimal method for determining cPRA. This has been driven by the need to consider incorporating DP (and other) locus antibodies. The benefit of the EM method, based on haplotype frequencies, is to estimate population frequencies that would include unobserved, rare phenotypes. However, because DP is not in linkage disequilibrium, it may no longer be feasible to use. Here, we evaluate the use of the count method in comparison with the EM approach for determining HLA antigen frequencies in the U.S. donor population. Methods Using the UNOS/OPTN data, we evaluated HLA typings from 39,568 deceased donors from 1/2015–1/2018. We then compared the observed antigen frequencies (count method) with those derived from the UNet cPRA calculator (EM method). Results See table. Download high-res image (81KB) Download full-size image Conclusions There was high concordance between the two methods of calculating cPRA in antigen frequencies within the A, B, and DR loci. Instances where there were large discrepancies were primarily due to parent antigens that were subsequently split; the UNet cPRA was based upon a pool of donors from 2007–9. This suggests, at the very least, that updates to the cPRA calculator are needed. Since DPB is not in linkage disequilibrium with the other loci, thereby negating our ability to use the EM algorithm, we suggest a new updated cPRA calculator be based upon the count method of calculating antigen frequencies.