P OPULATION FREQUENCIES DETERMINED BY NEXT - GENERATION SEQUENCING PROVIDE 1 STRATEGIES FOR PROSPECTIVE HLA EPITOPE MATCHING FOR TRANSPLANTATION

Canada Transplant Members of the research team are listed in the 12 appendix. Research is supported Canada, and the Canadian Institutes Health and funded by awards LSARP 273AMR 14 and GP1-155871. of matching perfect identity important HLA class


INTRODUCTION
Transplantation is the treatment of choice for irreversible renal failure, offering superior survival, 40 quality of life and economic costs compared to alternative options 1,2 . But despite superb initial 41 success (1 year kidney graft survival often exceeds 95%), many grafts fail within the first decade     84 A total of 2,000 subjects from the BC renal transplant program had full NGS sequence data at all 85 11 allelic HLA loci for the study. Of these, 154 subjects expressed alleles that were not yet present 86 in the HLAMatchmaker database (47 alleles, average carrier rate ≤0.18%, Supplemental Figure 1) 87 and were excluded from this analysis. Carrier rates of other HLA alleles in these subjects were 88 otherwise comparable to the overall population (Supplemental Figure 1) The 564 class I and 290 class II common and well-documented alleles in HLAMatchmaker were 102 used to define a string of eplets for each allele. The network diagrams in Figure 2 show the 103 extensive sharing of eplets by alleles within and between HLA class I A, B and C gene loci, while 104 class II eplets were shared only by alleles within the same gene locus. DQA1 and DPA1 contained 105 two mutually exclusive allele groups of eplet expression ( Fig. 2e and 2f conversion of HLA alleles to eplets is depicted in Figure 3. 115 Numerous intra-locus eplets were identified, encoded by multiple alleles within the same gene. 116 The class I eplet 131S, for example, was encoded by 90 class I alleles whereas the 163RG eplet 117 was encoded by only 2 alleles (A*01:01 and A*01:02), and the class II eplet 25R by 67 alleles 118 while the 25Q eplet was encoded by only 1 allele (DRB1*07:01). Multiple inter-locus eplets were 119 also present, encoded by more than 1 gene within each gene region (Supplemental Table 2). 120 Thirteen of these (22%) were encoded by more than one class I gene, with one eplet (163EW) 121 occurring by all three class I genes and the remainder by two genes and fourteen (15%) were 122 7 encoded by more than one class II gene, restricted to the DRB1/3/4/5 alleles. No eplets were shared 123 between class I and class II alleles. 124 Relative frequencies of HLA alleles and eplets 125 Most of the 361 alleles observed occurred with low population frequencies (Figure 4a and 4b, 126 Supplemental Table 3). Less than 2% (n = 7) were carried by more than 30% of subjects (class I:    Table for eplet frequencies in table   155 format.

157
A total of 1800 discrete genotypes, comprising the 16 -18 alleles encoded at each of the 11 HLA 158 gene loci on both chromosomes (note that DRB3/4/5 may be absent or hemizygous in an individual 159 genotype) were identified in the 1846 study subjects and combinations of these are shown in Table   160 1. The 206 class I alleles identified were combined in 1572 discrete genotypes and the 155 class 161 II alleles in 1509 discrete genotypes. Diversity at a single gene locus ranged from a maximum of 162 107 alleles and 602 genotypes at HLA-B to 7 alleles and 14 genotypes at DPA1 (Table 1). (1.5%) occurring in both groups ( Figure 6a). The number of shared genotypes increased as fewer 166 gene loci were considered; for example, 6% of class I genotypes and 7.4% of class II genotypes 167 were shared between patients and donors. The specific HLA gene locus was of primary 168 importance: 30% of genotypes were shared at DRB1/3/4/5+DQB1, 37% at DRB1/3/4/5, 51% at 169 DPB1, 64% at DPA1, and 78% at DQA1. Genotype sharing was most common at DQB1, with 90  for the full epitype; for class I from 10.2 (0 -27) to 3 (0 -11) and for class II from 16.8 (0 -45) 233 to 1 (0 -13); and for DRB1/3/4/5 from 6 (0 -20) to 0 (0 -4), for DQB1 from 6 (0 -21) to 0 (0 -  Extension of these models to other organs showed that regional or national sharing may be required 251 to enable epitope compatibility for heart, lung and liver transplants whose national waiting lists at these class II loci is substantially higher than for the full epitype, ranging from 30% at HLA-319 DRB1/3/4/5 + DQB1 to 79% at DQB1 alone, so providing a logistical basis for deliberate matching 320 at these loci.

321
In focusing on AMR, we have primarily examined antibody-verified eplets and employed 322 mismatch counts/scores as a measure of incompatibility. However, since our understanding of list numbers for kidney, and that regional or national sharing will be required for non-renal organs.

364
But graft success or economic costs are not the sole arbiters of policy and the utility donated organs 365 must be balanced by equality of access to them 24 . Matching at the eplet level may more closely 366 approximate this latter goal than the simple use of allele compatibility, though accommodation 367 must still be considered for recipients with uncommon eplets of high biological importance.

368
Limited approaches to eplet-matching have been incorporated by other programs. Eurotransplant 369 has successfully used class I eplet-matching to expand the donor pool for highly-sensitized patients  This study has certain limitations which we are working to address. We restricted this analysis to HLAMatchmaker. We are working to update HLAMatchmaker to include these alleles and will 379 re-evaluate once this process is complete. Subject selection and sequencing were performed in a 380 single provincial program, raising potential concern for both precision and representativeness. B.C.