Background:  NNRTI hypersusceptibility has been associated with improved outcomes to NNRTI-based therapy.  Although preliminary analyses have shown an association with NRTI resistance mutations, specifically the "classic" thymidine analog mutations (TAM) at codons 67, 41, 215, and 219, viral genetic determinants of NNRTI hypersusceptibility have not been thoroughly defined. 

Methods:  Paired baseline genotypes (VGI or ABI) and phenotypes (ViroLogic) were obtained from 444 subjects entering one of five ACTG studies: 290, 359, 364, 370, and 398.  All subjects were NRTI-experienced but NNRTI-naïve at study entry.  Fisher’s exact tests, recursive partitioning (CART), and stepwise binary regression were used to identify specific RT mutations associated with NNRTI HS (defined as a >2.5 fold-increase in susceptibility vs a reference virus).  A mutation was considered anything different from consensus B, including mixtures.  The initial models were constructed with EFV and then run for NVP and DLV.

Results:  The overall prevalence of EFV hypersusceptibility was 34%.  In univariate analyses, 23 different RT codons were associated with EFV hypersusceptibility (p <0.05), with the top 5 in order of p-value significance being 215 > 41 > 210 > 118 > 208 (all p<0.000001).  Although mutation at 215 was significantly associated with EFV hypersusceptibility, this was the case only for the more common 215Y (tyrosine) allele (p<0.000001, 53% EFV hypersusceptibility among 176 isolates with 215Y) and not for the less frequent 215F (phenylalanine) allele (p = 0.13, 44% EFV hypersusceptibility among 54 isolates with 215F).  From stepwise model selection, the 215, 208, and 118 mutations remained independently predictive of EFV hypersusceptibility.  A final binary regression model to predict EFV hypersusceptibility included one covariate for the 215 mutation (relative risk [RR] = 2.6, p <0.0001) and a second covariate representing either the 208 or 118 mutation (RR=1.8, p <0.0001).  Similarly, in a CART analysis, mutation at codon 215 was the first split selected, followed by mutations at 208 and 118.  Mutations at these 3 codons were also highly associated with DLV and NVP hypersusceptibility in binary regression models.

Conclusions:  Mutations at 215 and either 208 or 118 were independently associated with NNRTI hypersusceptibility.  The more frequent 41/210/215Y TAM pattern was more associated with hypersusceptibility than the 67/70/219/215F.  Incorporating this into genotype interpretation algorithms is likely to improve prediction of NNRTI hypersusceptibility.

Keywords: NNRTI; Hypersusceptibility; Genotype