Objective: Stavudine-based antiretroviral combinations are less effective in zidovudine-experienced patients than in naive subjects and recently, mutations have been described to be associated to the use of both stavudine and zidovudine. In the ALTIS 2 trial, it was shown that a combination of stavudine and lamivudine is less effective in zidovudine-experienced patients than in naive patients. We conducted a retrospective genotypic and phenotypic resistance study (expressed as stavudine phenotypic index, calculated by dividing the inhibitory concentrations 50% [IC50] by the mean value of the sensitive viruses) to evaluate the factors associated with decrease in plasma HIV-1 RNA.
Design: Associations with continuous variables were studied using non-parametric Spearman correlation coefficients. Associations with categorical variables were studied using non-parametric Mann-Whitney tests. Multivariate stepwise regression analyses were used to determine independent prognostic factors of the virological response.
Results: At baseline, most of the subjects harboured zidovudine-associated mutations in plasma and peripheral blood mononuclear cells. Zidovudine and stavudine IC50 and IC50 were strongly associated with response. It appears that a cut-off of stavudine phenotypic index of 1.8-fold of IC50, much lower than the usually used value, could be clinically significant for response to stavudine. In the multivariate analysis, the stepwise model with the higher multiple correlation coefficient (R2 = 0.742) included the presence of a 215 Y/F mutation, the number of previously used nucleoside analogues and a resistant stavudine phenotype.
Conclusion: These results argue for a phenotypic and genotypic cross resistance between stavudine and zidovudine. Modest increases of IC50 and IC50 for stavudine had an important impact on the virological response during the trial and plead for a new definition of the threshold value for stavudine phenotypic index.