Abstract
Background: Multiple class I and II HLA associations have been described in association with nevirapine (NVP) hypersensitivity reaction (HSR) phenotypes. We tested the hypothesis that peptide binding (PB) properties may be shared between these alleles. Methodology: HLA-A,-B-,-C, -DR typing was performed on stored DNA from a retrospective case controlled analysis of NVP HSR (ClinicalTrials.gov NCT00310843) using the 454 FLX platform. Univariate and multivariate analyses stratified for race were performed according to HLA class I/II alleles, HLA supertypes and HLA alleles according to PB (Sidney J, Lund O), Kir ligand groupings and HLA B/C haplotypes for cutaneous and hepatitis phenotypes of NVP HSR. In silico modelling to simulate HLA binding to NVP was performed with the highest ranked candidates. Results: HLA -A,-B,-C and -DR typing resolved to four digits (794 samples (controls =524, cutaneous NVP HSR cases =170, hepatitis NVP HSR cases = 100)). Multivariate analysis of cutaneous NVP HSR in Southeast Asians (SEA) associated DR4 supertype(OR=2.9, p=0.015) and alleles with HLA-35/18 PB properties(OR=6.4, p=0.002). HLA-DRB1*01:02 was associated with hepatitis NVP HSR in Caucasians (OR=2.7,p=0.01) whereas carriage of alleles of the PB B46 were protective (OR=0.3, p=0.04). HLA-C*04:01 was associated with cutaneous NVP HSR, including SJS/TEN across all races (p<0.0001, Mantel-Haenszel test; Caucasians: OR=2.8 [1.3-5.9], p=0.009; African Americans: OR=4.0 [1.4-13.0], p=0.02; SEA: OR=9.0 [3.2-24.9], p<0.0001). However, haplotype analysis of HLA-B/C showed pairing of HLA-C*04:01 with HLA-B alleles with B35 and B18 like PB (HLA-B*35:01,B*35:05,B*35:08,B*53:01,B*18:01,B*18:02, B*44:02,B*4403), and this effect was strongest in SEA where carriage of HLA-C*04:01 when paired with the HLA-B alleles(OR=11.8,p=0.0003) was more strongly associated with cutaneous NVP HSR than HLA-C*04:01 carried alone(OR=4.8,p=0.047). This suggests that risk of cutaneous NVP HSR attributed to carriage of HLA-C*04:01 may be enhanced by HLA-B alleles which are in strong linkage disequilibrium. An in silico and peptide binding model both suggest that NVP non-covalently binds in the F pocket of HLA-B*35:05 and near the B pocket of HLA-C*04:01. In multivariate analyses, Kir ligand groupings Bw4/Bw6 and C1/C2 did not significantly contribute to the modelling of associations with cutaneous hypersensitivity or hepatitis. Conclusions: Cutaneous and hepatitis phenotypes of NVP HSR associate with different HLA-B and DR-alleles respectively that share PB characteristics.The pairing of these HLA-B alleles with HLA-C*04:01 appears important for the development of cutaneous NVP HSR, providing a testable model for the immunopathogenesis of NVP HSR.