Prevalence and impact of HIV-1 protease codon 33 mutations and polymorphisms in treatment-naive and treatment-experienced patients

Background: HIV-1 protease gene mutations at codon 33 have been associated with resistance to some but not all protease inhibitors (PIs). Little is known about the difference in prevalence of codon 33 mutations and polymorphisms between treatment-naive and treatment-experienced patients, and the effect of codon 33F on PI phenotypic resistance patterns. Methods: Baseline genotypes (TRUGENE) from 772 patients participating in two different randomized clinical trials [504 antiretroviral treatment-naive patients and 268 antiretroviral treatment-experienced patients] were evaluated for the presence of protease codon 33 mutations and polymorphisms. Baseline phenotypes (Antivirogram), including fold-change in resistance for 16 antiretroviral drugs, were available for the 268 treatment-experienced patients. Multivariate linear regression models were used to determine factors associated with phenotypic fold-change for PIs. Results: The prevalence of codon 33 mutations and polymorphisms was 5.2% in the naive cohort (0.2% 33F, 2.5% 33V, 2.5% 33I) and 34.7% in the experienced cohort (30.2% 33F, 1.5% 33V, 3.0% 33I). In the antiretroviral-experienced cohort (mean = 4.2 prior PIs, 10.6 prior antiretroviral drugs overall), a model adjusting for the presence of specific major protease and multi-PI resistance conferring mutations, the number of other minor PI mutations, prior PI drug exposure (current, prior only, never), and HIV transmission risk factor was used to estimate the phenotypic fold-change in resistance for those with and without mutation 33F. Those with 33F had a significantly higher fold-change for amprenavir (33 vs 19, P<0.0001). ritonavir (162 vs 82, P<0.0001), lopinavir (49 vs 38, P=0.04), and saquinavir (47 vs 41, P=0.02). The presence of the 33F was not a significant predictor of fold change in susceptibility for indinavir or nelfinavir. Conclusions: At protease codon 33, the prevalences of polymorphisms 33V and 33I were similar for PI-naive and PI-experienced patients (<3.0%), but the prevalence of 33F was significantly different (0.2% versus 30.2%). In the treatment-experienced cohort, the differences in phenotypic fold-change for amprenavir, lopinavir, saquinavir, and ritonavir between those with and without 33F persist after adjustment for the presence of other major PI mutations and PI drug exposure history. Given the availability of newer PIs that may select for 33F, monitoring for the presence of this mutation should be ongoing for both treatment-naive and treatment-experienced patients.