Objective: To determine whether analysis of sequence variation in reverse transcriptase at baseline can explain differences in response to combination antiretroviral therapy.
Methods: Amino acid sequences of reverse transcriptase obtained from baseline isolates from 55 patients included in a trial of zidovudine and didanosine versus zidovudine/didanosine/nevirapine (ACTG241) were analysed. Simple and multiple linear regression were used to determine the relationship between numbers and identity of mutations at baseline and virological response after 8 and 48 weeks.
Results: Numbers of baseline zidovudine resistance mutations were predictive of short-term response (week 8). Amino acid identity at position 215 explained > 20% of the variation in response at week 8, but less at week 48. Multiple regression identified the combinations: 215 + 44 and 41 + 202, each of which explained about 30% of the variation in week 8 response. A model incorporating amino acids 214 + 215 + 60 + 202 + baseline viral load explained > 40% of the variation in response at week 48. Unexpectedly, the mutant combination 601 + 215Y/F responded threefold better than 60V + 215Y/F over 48 weeks.
Conclusions: Use of clinical data to analyse virological response to combination therapy has revealed effects of baseline amino acid mutations at sites not previously identified as being important in antiretroviral resistance. Predictors of long-term responses were different from those involved in the short term and may require more complex analysis.