Background:  Few longitudinal resistance studies are performed outside clinical trials. We wished to quantify the rate and patterns of antiretroviral resistance as a function of initial therapy in a large population-based drug-naïve cohort starting triple combination therapy.
Methods:  In British Columbia, Canada, antiretrovirals are distributed centrally at no cost to eligible HIV-infected individuals. Our analyses were based on a retrospective cohort of >1200 therapy naïve HIV-positive adults starting triple therapy between August 1996 and September 2000 (the Homer cohort). Genotypic HIV protease and RT resistance analyses (n >3000 assays) were attempted on all samples with a viral load >1000 copies/mL as long as 30 months after starting therapy. Four resistance categories (3TC, any other nRTI, any NNRTI, or any PI) were assigned based upon the IAS-USA table. Samples with HIV-1 RNA <1000 copies were assumed not resistant. Data were analysed on an intent-to-treat basis.

Results:  Resistance data were available from 1091 individuals. The most commonly used initial regimens were D4T/3TC/IND (n = 321); AZT/3TC/IND (n = 254); D4T/3TC/NEV (n = 136); D4T/3TC/NFV (n = 65); AZT/3TC/SQV (n = 47), and D4T/ddI/NEV (n = 45). 3TC resistance occurred with a ~3-fold hazard ratio compared with the other 3 categories in PI-containing regimens, but in NNRTI-based regimens NNRTI resistance was more common than 3TC resistance. Patients were assigned therapy non-randomly, so direct comparisons cannot be made between regimens. Nonetheless, resistance development after starting AZT/3TC/SQV (primarily unboosted Invirase) was startlingly high. For example, by 30 months, 68%, 34%, and 19% of individuals starting initial AZT/3TC/SQV had developed resistance to 1, 2, or 3 resistance categories, respectively (see left figure). This compares to 20%, 7%, and 2% for those starting AZT/3TC/IND (right figure). We have also described the rate of selection of each individual mutation with each therapy. For example, 10% of patients starting D4T/3TC/NEV had developed an M184V, K103N, and Y181C by 12 and 11 and 11 months, respectively.

Conclusions:  These data represent one of the first systematic efforts to describe the emergence of drug resistance and time-course of the selection of specific mutations in a clinical setting. They should also serve as a benchmark for comparison of more recently introduced therapies.

INSERT GRAPHS

Keywords: resistance; cohort