In HIV-1-infected treatment-naive participants enrolled in A5262, an RTI-free, two-drug regimen comprising DRV/r plus RAL met the protocol definition of acceptable virologic efficacy (at week 24), but only 71 and 61% of participants had viral load less than 50 copies/ml (86 and 73% <200 copies/ml) at week 48 in ITT and modified ITT analyses, respectively. Baseline viral load more than 100 000 copies/ml was associated with an increased risk of virologic failure. Baseline CD4 cell count per 100-cell increase was associated with reduced risk of virologic failure. In multivariable models fitted with DRV or RAL Ctrough BDL, baseline viral load more than 100 000 copies/ml remained strongly associated with increased risk of virologic failure. Having RAL Ctrough BDL at the visit immediately before or at one or more previous visits was also associated with an increased hazard of failure.
Potential explanations for our findings were explored. Self-reported adherence (4-day recall) was not significantly different between those with and without virologic failure, or between those with baseline viral load less than or equal to or more than 100 000 copies/ml. However, having one or more DRV and RAL plasma concentrations below detection limits was significantly more common in those with virologic failure, possibly related to unreported suboptimal adherence. Other investigators have demonstrated discordance between self-reported adherence and objectively measured adherence [12
]. Adverse effects of therapy are unlikely to have been the major determinant of adherence or virologic efficacy as RAL and DRV were well tolerated. An alternative hypothesis is that asymmetrical dosing of DRV/r (once daily) and RAL (twice daily) predisposed to suboptimal adherence and virologic failure, but such association has not emerged as a concern with RAL twice daily plus TDF/FTC, a similarly asymmetrically dosed regimen [13
]. It was suggested recently that RAL–DRV interactions may lower plasma concentrations of DRV [14
], but DRV Ctrough
observed in this study (Supplementary Table 1
) are within the range reported in an intensive pharmacokinetic study of DRV 800/100 mg daily [15
]. Finally, because over half of the patients who experienced virologic failure had low-level viremia (51–200 copies/ml) at the time of failure, we considered virologic failure artifacts due to assay variability [16
]. This possibility was excluded because viral load determinations during low-level viremia were similar with the Abbott m2000 and the Roche Amplicor Ultrasensitive Assays in the first 10 participants with low-level virologic failure.
An association between efficacy and baseline viral load has been demonstrated with other antiretroviral regimens. In some but not all studies of two NRTIs plus a third preferred agent, smaller proportions of patients with baseline viral load more than 100 000 copies/ml achieved HIV RNA less than 50 copies/ml at 48 weeks [5
], but these differences tend to be small, are in part related to tolerability and associated with low CD4 cell count [16
], and may not be synonymous with subsequent virologic failure [22
]. In our RTI-sparing study in which we specifically examined virologic failure, as opposed to a combined endpoint, the differences in virologic outcomes between the high and low viral load strata were striking and the results were consistent or even more evident in multivariable analyses that included baseline CD4 cell count or assessments of drug concentrations. A pilot study evaluating twice-daily atazanavir (ATV) plus RAL was prematurely terminated at week 24 due in part to adverse events and frequent RAL resistance in those with virologic failure [23
]. A larger randomized study, however, found no significant difference in viral load less than 40 copies/ml at week 48 in patients treated with RAL plus lopinavir/ritonavir (LPV/r) compared with TDF/FTC plus LPV/r [24
]. The mean baseline viral load in the latter study was 4.25 log10
copies/ml, which is lower than the 4.9 log10
copies/ml in the twice-daily RAL plus ATV study and 4.83 log10
copies/ml in our study. Our study is also the first to report virologic outcomes by baseline viral load less than or equal to or more than 100 000 copies/ml separately from nonvirologic treatment discontinuations, further limiting cross-study comparisons. The mechanisms underlying the poorer virologic outcomes in some patients with high baseline viral load, as observed in this study, are uncertain. One possibility is that high baseline viral load may be associated with more extensive reservoir of infected cells and prolonged viral decay time to levels below 50 copies/ml. However, only one of the 28 virologic failures in the current study had viral dynamics that may be explained solely by this specific hypothesis. Another possibility is that high baseline viral load may predispose to greater diversity of HIV-1 quasispecies and an increased opportunity to select drug-resistant mutants. Q148R minority variants were detected at very low levels (median 0.46%) in 86% of treatment-naive patients in one study [25
]. The effect of pretherapy RAL-resistant minority variants on virologic outcome in treatment-experienced patients has not been clearly demonstrated [25
] and to our knowledge has not yet been reported in treatment-naive patients.
Baseline viral load more than 100 000 copies/ml appears to increase the risk of RAL resistance in patients receiving DRV/r plus RAL. All the patients with evidence of RAL resistance mutations at virologic failure (20% of those genotyped) had baseline viral load more than 100 000 copies/ml. None of these patients had documented treatment interruption, and no significant difference in RAL Ctrough
was observed between those with or without RAL resistance. Viral load at the time of integrase genotyping in the five participants ranged from 62 to 685 copies/ml. Notably, a participant who achieved HIV RNA less than 50 copies/ml at week 12 and had no subsequent documentation of viral load level more than 100 copies/ml experienced virologic failure at week 48 with detection of Q148K/Q and N155H/N. Thus, like NNRTI and NRTI resistance mutations [27
], RAL resistance mutations may be present during low-level viremia [28
], an important observation as recent guidelines state that viral load more than 200 copies/ml can be considered the threshold for virologic failure in clinical practice [1
]. Protease inhibitor resistance was not detected in any participant experiencing virologic failure, consistent with evidence that boosted protease inhibitor resistance seldom develops early in virologic failure [29
Interpretation of this study should take into account its single-arm design as a randomized trial could have reduced the potential impact of patient characteristics and other variables. Also, patients were not screened for pretreatment RAL resistance, but primary mutations that confer resistance to RAL are uncommon in RAL-naive patients [30
]. Despite these limitations, the results of A5262 raise important issues that should be examined carefully in future clinical trials evaluating DRV/r plus RAL and perhaps in all RTI-sparing two-drug regimen trials. We urge caution in patients with baseline viral load more than 100 000 copies/ml and emphasize a need to further elucidate the implications of low-level viremia in patients receiving the regimen.