Abstract
Carbamazepine triggers dermatologic hypersensitivity reactions, associated with specific human leukocyte antigens (HLAs), especially HLA-B*15:02 and HLA-A*31:01. Previous efforts to identify single nucleotide variants (SNVs) with high predictive value as HLA proxies, revealed that rs1061235 and rs17179220 fulfill these requirements for HLA-A*31:01 in some but not all populations. Herein we explored the predictive performance of rs1061235 and rs17179220 as HLA-A*31:01 tags in populations of Native American ancestry, which are largely underrepresented in pharmacogenomic studies. The study cohorts comprised the overall Admixed American superpopulation of the 1000 Genomes Project (1 KG_AMR), a subcohort of individuals with predominant Native American ancestry (1 KG_NAT), the Native American population of the Human Genome Diversity Project (HGDP), plus Kaingang (KRC) and Guarani (GRC and GKW) adults from indigenous reservation areas in Brazil. The diversity of cohorts is reflected in the range of frequencies of HLA-A*31:01 (0.02-0.65), rs1061235 (0.03-0.13) and rs17179220 (0.12-0.66), as well as in the predictive performance of these SNVs as HLA-A*31:01 proxies. NPV (negative predictive value), the metric of primary interest for pharmacogenetic-informed carbamazepine prescription was maximal (NPV = 1.0) for both SNVs in 1 KG_AMR and 1 KG_NAT, for rs17179220, but not rs1061235 (NPV = 0.91) in HGDP, and for rs17179220 in GWK, but not GRC (NPV = 0.88) or KRC (NPV = 0.80). Collectively, the data support the notion that rs1061235 and rs17179220 are not optimal proxies for HLA-A*31:01 across populations of Native American ancestry.