Background:  The influence and predictors of K65R mutation on virological response to salvage therapy has not been definitively elucidated.

Methods:  We retrospectively identified all K65R mutations in virus samples from 6 Italian clinical centers (4 centers from Genotypic Networking Organization Multicentric Observation). Clinical, viro-immunological, and genotypic data were collected. Probability of virological response (time to first HIV RNA <50 copies/mL) and its determinants were calculated using KM and Cox regression.

Results:  Overall, 145 patients were included. The mean number of nucleoside reverse transcriptase inhibitors (NRTI) used was 4 (range 2 to 7). At the time of genotyping, the most frequently administered NRTI were:  tenofovir (TDF; 56.5%), lamivudine (3TC; 56.5), didanosine (ddI; 55.7%), and stavudine (d4T; 22.9%). Median HIV RNA and CD4 count were 4.17 log₁₀ copies/mL (IQR 3.62 to 4.73) and 312 cells/mm³ (IQR 148 to 471). M184V was detected in 44.8%, and S68G/N/R/K in 27.6%; at least 1 mutation from Q151M-complex, thymidine analogue mutation (TAM)1, and TAM2 pathways were found in 31.7%, 15.2%, and 22.8%, respectively. After genotyping, the most frequent NRTI was 3TC (59.7%), zidovudine (AZT; 29.0%), d4T (27.4%), and ddI (23.4%). Over a total of 197 person-years of follow-up, the 12-month probability of virological response was 60% (SE 5%); this probability was strongly reduced by the Q151M pathway (52% vs 65%; p  =0.02 at log-rank test). By multivariate Cox model the strongest predictors of virological response was adding a thymidine analogue in the salvage regimen (AHR 2.55; 95%CI 1.25 to 5.19; p = 0.01):  in a separate model the relative risk of virological response by thymidine analogue was higher for d4T (2.66; 1.05 to 6.72; p = 0.04) than for AZT (1.64; 0.63 to 4.24; p = 0.31). Moreover, M184V (1.97; 1.002 to 3.86; p = 0.049) was associated with a better outcome, even there was no evidence of interaction with 3TC use (p = 0.56 at interaction test). Furthermore, the presence of  lopinavir/ritonavir (LPV/r) as a new drug was marginally associated with better outcome (1.73; 0.95 to 3.18; p = 0.07). Excluding subjects with Q151M, use of thymidine analogue as a new drug had a slight effect (2.07; 0.96 to 4.46; p = 0.06) on virological response, that was strongly reduced by viremia at baseline (0.59; 0.37 to 0.94; p = 0.03). Even in this model, the effect of M184V remained favourable (2.26; 1.003 to 5.08; p = 0.049).

Conclusions:  Development of K65R mutation does not preclude the effectiveness of rescue therapy. Inclusion of a thymidine analogue in the salvage regimen (d4T rather than AZT) as well as low-level viremia at change predict better response. Concomitant presence of M184V may have a favorable effect on virological outcome.