Abstract
A case-cohort study was used to determine the effect of baseline nonnucleoside reverse-transcriptase inhibitor (NNRTI) resistance, as assessed by viral genotyping, on the response to efavirenz-containing regimens in AIDS Clinical Trials Group A5095. The sample included a random cohort of efavirenz-treated subjects plus unselected subjects who experienced virologic failure. Of 220 subjects in the random cohort, 57 (26%) had virologic failure. The prevalence of baseline NNRTI resistance was 5%. The risk of virologic failure for subjects with baseline NNRTI resistance was higher than that for subjects without such resistance (hazard ratio 2.27 [95% confidence interval], 1.15–4.49; P = .018). These results support resistance testing before starting antiretroviral therapy
Treatment guidelines recommend efavirenz (a nonnucleoside reverse-transcriptase inhibitor [NNRTI]) or a ritonavir-boosted protease inhibitor (PI) plus a fixed-dose combination of nucleoside (or nucleotide) reverse-transcriptase inhibitors (NRTIs) for initial anti-HIV therapy [1]. The efficacy of NNRTI-based regimens is threatened, however, by the spread of antiretroviral drug resistance. Data from the US Centers for Disease Control and Prevention show that the prevalence of drug-resistant HIV-1 is increasing among newly infected or newly diagnosed persons who have not received antiretroviral therapy [2,3]. Most striking is the substantial increase in NNRTI resistance, which rose from 0.4% to 8.4% from 1998 to 2004. In response to these trends, treatment guidelines now recommend routine drug-resistance testing before initiating antiretroviral therapy [1,4]. Although models suggest that such a strategy is cost-effective [5], data on the impact of primary resistance on the response to NNRTI-based regimens in chronically infected patients are scant
The AIDS Clinical Trials Group (ACTG) A5095 study was a randomized, controlled trial comparing the efficacy of efavirenz plus a fixed-dose combination of zidovudine/lamivudine (3-drug arm) or zidovudine/lamivudine/abacavir (4-drug arm) in previously untreated HIV-1–infected subjects [6]. With a median 144 weeks of follow-up, time to virologic failure was not significantly different between the 3- and 4-drug arms. Black non-Hispanic race/ethnicity and recent nonadherence were associated with a significantly increased risk of virologic failure [6]. Genotypic testing of stored baseline plasma samples from efavirenz-treated subjects who experienced protocol-defined virologic failure showed a prevalence of NNRTI resistance of 10%
Because this subset represented a biased cohort, we developed a case-cohort study to determine the prevalence of NNRTI resistance and its impact on treatment outcome. A case-cohort study design uses a subsampling technique in survival data for estimating the relative risk of disease in a cohort study without collecting data from the entire cohort. This design is an efficient and economical way to study risk factors for infrequent disease in a large cohort. It involves the collection of covariate data for all disease cases observed in the entire cohort and for the members of a random subcohort. In addition to ascertaining the relative risk of disease in relation to the risk factors of interest, the random subcohort also offers the ability to assess the prevalence of risk factors that may be too costly to evaluate in the entire cohort
MethodsACTG A5095 enrolled 765 HIV-1–infected adults with plasma HIV-1 RNA levels ⩾400 copies/mL into the efavirenz-containing arms between March 2001 and November 2002 at 33 AIDS Clinical Trials Units (ClinicalTrials.gov identifier NCT00013520). A total of 193 subjects (25%) experienced virologic failure (2 consecutive measurements of HIV-1 RNA level ⩾200 copies/mL, with the first measurement at least 16 weeks after study entry) [6]. Genotyping of HIV-1 protease and reverse trancriptase was performed on stored plasma samples obtained at study entry by use of the HIV-1 TRUGENE assay (version 10; Bayer HealthCare Diagnostics). For the case-cohort study, the primary outcome measure was the occurrence of virologic failure, and the primary variable of interest was the presence or absence of mutations associated with resistance to efavirenz or nevirapine (NNRTI resistance). In the case of the PIs, samples were considered to be resistant if categorized by the TRUGENE algorithm as being resistant to at least 1 PI. The case-cohort sample consisted of a random sample (subcohort) stratified by and drawn from the efavirenz-containing arms of A5095 plus the case subjects (those who experienced virologic failure) who were not selected to be in the subcohort. Assuming a 25% failure rate and 5% prevalence of NNRTI resistance, it was estimated that a subcohort size of 220 subjects would give 80% power to detect a hazard ratio (HR) for virologic failure of 3.56 between subjects with and without NNRTI-resistant virus
On the basis of the random sample, the prevalence of baseline NNRTI resistance was estimated; the demographics of subjects with or without preexisting NNRTI resistance were compared by Fisher’s exact and Wilcoxon tests for categorical and continuous variables, respectively. Weighted Cox proportional hazards models [7] were used to estimate the risk of virologic failure in the presence and absence of NNRTI resistance at baseline. Analyses were adjusted for race/ethnicity and recent self-reported adherence. The presence of an interaction between recent self-reported adherence and baseline NNRTI resistance was examined. Such an interaction would imply a different impact of recent adherence dependent on the presence and absence of NNRTI resistance mutations at baseline. Results are presented from intent-to-treat or as-treated analyses as described. All P values and confidence intervals (CIs) presented are nominal, unadjusted for multiple comparisons
ResultsOf the 220 randomly sampled subjects, 57 (26%) were case patients (subjects with virologic failure) and 163 (74%) were control subjects (those without virological failure). The additional 136 subjects with virologic failure were added to the subcohort to make up a case-cohort sample of 356 (193 with virologic failure and 163 without virologic failure). Viral genotypes could not be obtained for 1 (nonfailure) subject from the randomly selected subcohort and for 2 subjects from the additional subjects with virologic failure. Of the 353 subjects in the case-cohort sample, 11 subjects in the control group had <16 weeks of follow-up and therefore could not have reached protocol-defined virologic failure; these subjects were excluded from the analyses of virologic failure
Of the 219 subjects in the randomly sampled subcohort with resistance data, 12 (5%) showed resistance to NNRTIs, 6 (3%) to NRTIs, and 6 (3%) to PIs. Table 1 summarizes the baseline demographics of the random subcohort by presence or absence of preexisting NNRTI resistance. Subjects with NNRTI-resistant virus were somewhat more likely to be previous or current injection drug users (P=.091) and to have higher median baseline CD4 cell counts (P=.048). Demographics are also presented for the additional subjects with virologic failure
Baseline characteristics of the study sample
Baseline characteristics of the study sample
Of the 342 subjects with viral genotype data and ⩾16 weeks of follow-up, 16 (8%) who had experienced virologic failure and 3 (2%) who had not carried NNRTI-resistant virus at baseline. By contrast, virus samples from 6 (3%) with virologic failure and 3 (2%) without carried mutations associated with PI resistance. In the 3-drug arm, baseline NNRTI resistance was detected in 13 (13%) of 97 subjects with virologic failure versus in none (0%) of 72 of those without failure. By comparison, in the 4-drug arm, baseline NNRTI resistance was detected in 3 (3%) of 94 subjects with virologic failure and in 3 (4%) of 79 without
The K103N mutation was present in samples from 12 of the 16 subjects with virologic failure (75%) and NNRTI resistance at baseline; in 5 cases it was present together with a second NNRTI mutation (Y181C in 2, Y188L in 1, and G190A in 2). The K103T, Y181C, Y188L, and G190A mutations were each present individually in samples from the remaining 4 subjects with virologic failure. All of these mutations have been associated with resistance to efavirenz (http://hivdb.stanford.edu). K103N, Y188L, and M230L were each present individually in samples from 3 control subjects. None of the control subjects had >1 NNRTI resistance mutation
Time to first virologic failure in the subcohort was substantially shorter for subjects with preexisting NNRTI-resistant virus than for those without (figure 1). Weighted Cox proportional hazard models including baseline NNRTI resistance showed a significantly increased risk of virologic failure for subjects with NNRTI-resistant virus at baseline compared with those without (intent-to-treat: HR, 2.27 [95% CI, 1.15–4.49]; P=.018) (as-treated: HR, 2.61 [95% CI, 1.30–5.20]; P=.007). The above results persisted after adjusting for race/ethnicity and recent self-reported adherence (as-treated: adjusted HR, 3.13 [95% CI, 1.36–7.22]; P=.007). The estimated effects of race/ethnicity and recent adherence were consistent with those previously reported [6]. There was no significant evidence of an interaction between recent self-reported adherence and baseline NNRTI resistance (P=.46)
Kaplan-Meier plot of time to protocol-defined virologic failure among subjects in the randomly selected subcohort with and without virus resistant to nonnucleoside reverse-transcriptase inhibitors at baseline. Virologic failure was defined as a confirmed plasma HIV-1 RNA level ⩾200 copies/mL at week 16 or later
Kaplan-Meier plot of time to protocol-defined virologic failure among subjects in the randomly selected subcohort with and without virus resistant to nonnucleoside reverse-transcriptase inhibitors at baseline. Virologic failure was defined as a confirmed plasma HIV-1 RNA level ⩾200 copies/mL at week 16 or later
DiscussionWe used a case-cohort design to compare the prevalence of NNRTI resistance at baseline in samples from subjects who started receiving an efavirenz-containing regimen in a randomized clinical trial that compared the efficacy of efavirenz plus 2 or 3 NRTIs. The overall prevalence of NNRTI resistance at baseline was 5%. Resistance to NNRTI was present at baseline in samples from 8% of subjects with virologic failure and from 2% of subjects without failures; the presence of NNRTI-resistance mutations more than doubled the risk of virologic failure. In this study, all subjects with genotypic evidence of NNRTI resistance at baseline who received a standard regimen of 2 NRTIs plus efavirenz eventually experienced virologic failure, although a few subjects maintained virologic suppression for 2 to 3 years before failure. Thus, these results provide a strong rationale for performing resistance testing before selecting a regimen for initial antiretroviral therapy. Moreover, these results are highly relevant to health care providers and policymakers because NNRTI-containing regimens such as the one used in our study are the most commonly used regimens for the initial treatment of HIV-1 infection globally
Although most current guidelines for the treatment of HIV-1 call for resistance testing before the initiation of antiretroviral therapy, few published reports document the impact of transmitted drug resistance on response to first-line regimens outside the setting of acute or recent HIV-1 infection [8–10]. One study found a longer time to virologic suppression and a shorter time to virologic failure among recently infected patients whose regimen included 1 or more drugs to which their virus was resistant [8]. Another study found a nearly 3-fold increased risk of virologic failure in chronically infected patients with evidence of transmitted zidovudine resistance [9]. By contrast, a third study found no effect of transmitted resistance on response to an initial 3-drug regimen [10]. None of these studies specifically investigated the issue of transmitted NNRTI resistance
Similarly, few randomized clinical trials of initial antiretroviral therapy in treatment-naive patients have considered the effect of preexisting drug resistance on treatment outcome. In most cases, because resistance testing is performed only on samples from subjects who experience treatment failure, it has not been possible to determine the risk of treatment failure associated with baseline drug resistance. For example, one study detected NNRTI resistance in baseline samples from 35% of subjects with virologic failure receiving an initial regimen of emtricitabine or stavudine plus didanosine and efavirenz [11]. No data were available, however, on the prevalence of NNRTI resistance in subjects classified as virologic responders
In the absence of antiretroviral therapy, most transmitted drug-resistance mutations revert to wild type over time [8, 12]. Certain NNRTI resistance mutations (e.g., K103N) remain detectable by routine genotypic resistance tests for at least 2 to 3 years after transmission, but data on the longer-term persistence of these mutations are limited. An earlier study of chronically infected antiretroviral-naive patients with a median CD4 cell count of 0.296×109 cells/L found an overall prevalence of resistance of 9%, suggesting persistence of transmitted drug-resistance mutations [13]. Although we do not have data on the duration of HIV-1 infection in the subjects we studied, the relatively low median baseline CD4 cell count (0.203×109 cells/L) suggests that they had been infected for many years before entry into ACTG A5095. Our results therefore support current guidelines that recommend performing baseline drug-resistance testing even in patients who have been infected for >2 years
In the present study, detection of NNRTI resistance relied on a standard genotyping assay (TRUGENE) that is commonly used in clinical practice. This approach may have failed to detect the presence of NNRTI resistance when such variants made up a minority of the virus population. Although other methods such as allele-specific polymerase chain reaction and single-genome sequence offer greater sensitivity in detecting drug-resistance mutations, at present they are not available for routine clinical use [14]
Three subjects maintained virologic suppression despite the presence of NNRTI-resistant virus at study entry. Of note, all 3 received the 4-drug regimen of abacavir, zidovudine, and lamivudine plus efavirenz. Although no difference was observed between the 3- and 4-drug arms in the main study [6], it is possible that in these 3 subjects virologic failure was prevented by the triple-NRTI component of the regimen. However, triple-NRTI regimens have been shown to be inferior to NNRTI-containing regimens as initial antiretroviral therapy [15]. The optimal choice of initial regimen for patients with NNRTI-resistant virus will need to be defined through randomized clinical trials that target this growing population
Acknowledgments
We thank the 33 participating AIDS Clinical Trials Units and the additional AIDS Clinical Trials Group (ACTG) A5095 study team members, including: Carol Bick (Indiana University School of Medicine) and Alison Boyle (University of California, Los Angeles), study laboratory technologists; Valery Hughes (Weill Medical College of Cornell University), study field representative; Anne Kmack and Sandra Oyola (Frontier Science and Technology Foundation), study data managers; Ana Martinez (Division of AIDS, National Institute of Allergy and Infectious Disease), study pharmacist; Monica Murphy and Nancy Webb (Frontier Science and Technology Foundation), laboratory data coordinators; Vinny Parillo (AIDS Clinical Trials Group Community Constituency Group), community representative; Sally Snyder and Barbara Bastow (Social & Scientific Systems, Inc.), clinical trials specialists; Kirk Ryan and Shulin Wang (Bristol-Myers Squibb); and Keith Pappa and Qiming Liao (GlaxoSmithKline). Danielle Smith (Massachusetts General Hospital), Russell Young (University of Colorado Health Sciences Center), Lorraine Sutton (Vanderbilt University), J. Darren Hazelwood (University of Alabama, Birmingham) and Leslie Petch (University of North Carolina at Chapel Hill) performed HIV-1 genotyping assays for ACTG A5095
