Background:  Quantitative analysis of dynamic shifts in the HIV-1 quasi-species that occur under changing selective pressure can be used to estimate the in vivo fitness cost of drug resistance mutations. 

Methods:  A sensitive allele-specific real-time polymerase chain reaction (PCR) (ASPCR) assay was used to study decay and re-emergence of enfuvirtide (ENF)-resistant viruses during interruption and reinstitution of ENF. ART-experienced subjects with incomplete viral suppression on ENF-based regimens interrupted ENF but remained on a stable background regimen (partial treatment interruption). The proportion of the plasma viral quasi-species carrying a V38A mutation in gp41 was quantified by ASPCR in serial samples from 3 patients collected at 1-4 week intervals. The ASPCR assay had a sensitivity of 0.8% for detection of V38A. The fitness coefficients were calculated by linear regression of the log (mutant/wild type) ratio (conventional method) and by a growth and error corrected method (GECM) that corrects for time-dependence of the viral replication rate. 

Results:  The V38A mutant made up ≥85% of the quasi-species at baseline and decayed to <1 to 5% over 16 to 36 weeks. Fitness estimates using the GECM gave the best fit with the following fitness differences for mutant vs wild type:  patient 1, –24% (±2%); patient 2, –34% (±3%); patient 3, –50% (±4%). Reintroduction of ENF during a 4-week pulse-intensification phase resulted in rapid rebound of V38A, which increased to 25 to 75% of the quasi-species within 4 weeks. The estimated replicative fitness advantage of V38A vs wild type in presence of ENF ranged from 157 to 226%; estimates were less reliable due to the small number of samples available from the pulse phase. Samples obtained from patient 1 during the second ENF partial treatment interruption showed faster decay of V38A, with a fitness difference of –59% (±6%) vs wild type. 

Conclusions:  The relative fitness of V38A mutants vs wild type varied according to conditions:  wild type had a fitness advantage over V38A in the absence of ENF, but V38A showed a much greater fitness advantage over wild type under positive selection pressure during ENF pulse-intensification. Faster decay of V38A during the second partial treatment interruption could be due to additional ENF-resistance mutations or smaller size of the ENF-resistant population after ENF pulse therapy. Careful measurements during and after pulse-intensification may provide more precise in vivo fitness estimates as both wild type and mutant are present in the quasi-species.