Background: Theoretically, HIV-1 subtype C viruses have a greater propensity to develop a K65R mutation due to silent nucleotide polymorphisms at nearby codons (64-66) and/or TAM codons which affect the probability of GÇA transition. Although a high prevalence of K65R has been documented in Southern Africa cohorts, it is unclear whether this is explained by subtype per se or more general characteristics of these cohorts e.g. prolonged time on a failing regimen. We have exploited the diversity of viral subtypes within the UK to undertake a direct comparative analysis of this issue.
Methods: We analysed patients infected with HIV-1 subtype B or C virus who had a resistance test (1996-2013) following virological failure (VF), regardless of type or line of regimen. Sequences with no major IAS-defined mutations were excluded; for patients with ≥2 tests, we selected the first test if K65R was ever detected or the last test if never detected. Prevalence of K65R was related to subtype and exposure to the NRTIs that primarily select for this mutation (TDF, ABC, ddI, d4T). Exposure was considered both as “current” (regimen at time of VF) and any previous exposure. A multivariate logistic regression model quantified the effect of subtype on the prevalence of K65R, adjusting for previous and current exposure to all four specified drugs.
Results: 4,242 patients (3,439 subtype B, 803 subtype C) met the inclusion criteria. Subtype B patients were mostly MSM (77%), and those with subtype C mostly heterosexual (82%, F:M ratio=1.8). Overall, K65R was detected in 7.8% subtype B patients (median 5.0 years [IQR 1.6-7.8] after initiating ART) compared with 14.5% subtype C patients (2.5 years [0.8-5.1]). The subtype difference in K65R prevalence was observed irrespective of NRTI exposure, and K65R was frequently selected by ABC, ddI, and d4T in patients with no previous exposure to TDF (Figure). Multivariate logistic regression confirmed that K65R was significantly more common in subtype C viruses (odds ratio 1.95, 95% CI: 1.51-2.51, P<0.001).
Conclusions: These clinical data complement experimental evidence that K65R is more likely to be detected at VF for subtype C viruses compared with subtype B viruses.