No abstract

Detection of drug resistance is critical for determining antiretroviral treatment options. Ultradeep pyrosequencing (UDPS; 454 Life Sciences) is capable of detecting virus variant subpopulations with much greater sensitivity than population sequencing, which typically has a detection limit around 20%. UDPS of the HIV-1 reverse transcriptase (RT) (amino acids 56-120) was performed to detect the key mutations K65R and L74V associated with tenofovir and abacavir use. Plasma specimens from subjects with persistent rebound viremia following suppression on tenofovir (n = 8) or abacavir (n = 9)-based therapy were studied. Samples from a subject treated with zidovudine/lamivudine/efavirenz with a similar loss of virologic response served as a control. HIV-1 plasma RNA was ≥3.68 log(10) copies/ml at all time points sequenced. The median number of UDPS sequences analyzed/time point was 33,246. Among the eight tenofovir-treated subjects, three showed high-frequency (>20%) RT K65R at the time of failure, whereas one showed low-frequency (<20%) L74V; no low-frequency K65R was detected in these subjects. Among the nine abacavir-treated subjects, three showed low-frequency K65R; no L74V was detected in these patients. No K65R or L74V was detected in the samples from the control subject. At failure, other RT mutations were detected, including low-frequency NNRTI-resistant species detected at ≥1 time point in nine subjects; the key NNRTI mutation K103N, however, was always observed at >20% frequency. Although UDPS is useful in the detection of low-frequency subpopulations with transmitted resistance in antiviral-naive patients, it may have less utility in treatment-experienced patients with persistent viremia on therapy.

Thymidine-sparing triple-nucleoside regimens have exhibited poor virologic response despite apparent phenotypic susceptibility to 2 of 3 regimen components at early time points. Phenotypic resistance masking by wild-type virus may explain this discrepancy.Consistent with this notion were (1) the presence of low level nucleoside reverse-transcriptase inhibitor-resistant human immunodeficiency virus in subjects receiving failing first-line regimens consisting of tenofovir (TDF), abacavir (ABC), and lamivudine (3TC); (2) lower fold resistance associated with mixtures versus mutants in a clinical-isolate database; and (3) dose dependent changes in susceptibility to ABC, 3TC, TDF, and didanosine on titration of K65R and/or M184V with wild-type virus. These findings underscore the limitations of stand-alone phenotypic susceptibility measures and emphasize the importance of complementary and/or more sensitive techniques.