There have been few opportunities for direct comparison of the virologic efficacy and safety of regimens from each of the 3 major antiretroviral drug classes. At the XV International AIDS Conference, 2-year data from the CLASS Study, which is a randomized, open-label trial designed to compare initial therapy with 1 of 3 regimens representative of each major class of antiretrovirals, was presented and included the results of lipid and body shape investigations (posters TuPeB4544 and WePeB5944).

The study was designed to compare the strategy of starting HIV treatment with a protease inhibitor (PI)-, non-nucleoside reverse transcriptase inhibitor (NNRTI)- or nucleoside reverse transcriptase inhibitor (NRTI)-based regimen. Subjects naive to HIV therapy with a CD4+ cell count of at least 50 cells/mm³ and a viral load of 5,000 copies/mL or more were randomized 1:1:1 to one of the following initial regimens:

There is no misprint. The triple-NRTI regimen was not the fixed-dose combination of zidovudine/lamivudine/abacavir (AZT/3TC/ABC, Trizivir) manufactured (along with amprenavir) by GlaxoSmithKline (GSK), the sponsor of this trial. One suspects that zidovudine/lamivudine/abacavir was not included as the triple NRTI as a hedge by GSK. If the triple NRTI underperforms, an argument could be made that it is not exactly zidovudine/lamivudine/abacavir which is impugned, particularly if toxicity drives poor performance. However, the success of this regimen could be easily extended to zidovudine/lamivudine/abacavir. Admittedly, this may be a cynical view and an alternative, less jaded, rationale may exist.

Subjects in Arm B were allowed to change to fosamprenavir (908, Lexiva) and ritonavir (1,400 mg and 200 mg, respectively) once a day, when this became available.

Subjects experiencing virologic failure or treatment-limiting toxicity were switched to study-arm-specific second-line therapy and continued to be followed for the primary endpoint of 96-week efficacy and safety. Second-line therapies for each arm were:

All subjects starting second-line therapy (except for those who had experienced an abacavir hypersensitivity reaction) underwent a second randomization to have or not have abacavir added to the regimen. Failure of second-line therapy led to best medical therapy crafted by the patient's provider.

Efficacy and Safety

A total of 291 subjects were enrolled. Following 96 weeks, 60-70% of the subjects in the study had viral load levels below 50 copies/mL, using intent-to-treat analysis that included all randomized subjects completing the baseline visit and patients with missing data at subsequent time points considered to be failures (so-called "missing = failure"); there were no differences between the study arms in the proportion of subjects with a viral load result below 50 copies/mL. However, when the analysis was performed exclusively with the data available and missing data ignored, the efavirenz arm was found to have a greater proportion of subjects with undetectable (<50 copies/mL) levels of viremia at 96 weeks compared to the other arms.

Only 2 subjects who started with efavirenz (Arm A) switched to second-line therapy, compared to 12 subjects in the amprenavir arm (Arm B) and 14 subjects in the triple-nucleoside arm (Arm C). In the case of efavirenz, the switch in both cases followed toxicity development; remarkably, no subject in this arm switched due to virologic failure.

For the other arms, virologic failure prompted the switch in approximately half of the subjects requiring a switch (7 in PI arm and 8 in NRTI arm). The rate of switching among study arms is an important aspect of this trial, as the study was designed to test therapeutic strategies rather than purely specific regimens. Thus, subjects who switched from their original regimen assignment to second-line therapy are included in the primary analysis at 96 weeks. In fact, at the study end, close to 95% of the study participants were receiving efavirenz either as part of their original randomization (Arm A) or following a switch to second-line therapy after virologic failure (Arms B and C). Therefore, the 96-week results should not be misunderstood to indicate that there was no difference between the virologic efficacy of the 3 regimens subjects received at the start of the study.

How the 3 initial regimens themselves performed can be gleaned from the rate of switching from initial to second-line therapy and a secondary analysis of the durability of viral suppression. This demonstrated a significantly longer effect with abacavir plus lamivudine plus efavirenz compared to abacavir plus lamivudine plus stavudine. Such a significant difference was not seen between the efavirenz arm and those randomized to start with abacavir plus lamivudine plus amprenavir and ritonavir, although the study was underpowered to address this comparison.

Additional support for the better performance of the efavirenz arm comes with examination of the proportion of subjects with viral loads below 50 copies/mL at time points prior to 96 weeks. Early in the study, when subjects tended to be on their original treatment assignments, subjects in the efavirenz arm had significantly greater reductions in their viral load than those in the other 2 arms.

Approximately 40% of the subjects in each arm had viral load levels >100,000 copies/mL at baseline. Among these subjects, the efavirenz and amprenavir arms had a higher proportion of subjects with undetectable viremia at week 96 (65% efavirenz, 64% amprenavir and 49% triple NRTI). This is an interesting and notable finding as it suggests that among patients with high viral loads, starting with this triple-nucleoside regimen handicapped subsequent response compared to that achieved when therapy was led off with the NNRTI or PI regimens.

Drug-related adverse events leading to treatment discontinuation occurred in no more than 3 subjects in each arm. Abacavir hypersensitivity reactions were diagnosed in 7.3% of the subjects across the arms taking this agent.

Metabolic Assessment

Fasting total cholesterol and triglyceride levels were measured at baseline and every 24 weeks. In addition, at baseline and weeks 4, 12, 48, 72 and 96, body shape measurement was performed using anthropometrics (mid-arm, mid-thigh, mid-waist, abdomen at the umbilicus, neck and chest circumferences measured with a tape measure). Analysis of the metabolic parameters was done on the as-treated population; that is, only the subjects still on their randomized initial treatment regimen were included. The numerators here again tell a story of the staying power of the efavirenz regimen (and perhaps the amprenavir combination) compared to the NRTI arm. For the efavirenz arm, 61 of the 97 subjects were still on the original treatment assignment at week 96 compared to 53 of 96 in the amprenavir arm and 49 of 98 in the triple-NRTI arm.

Of course, the particular triple-NRTI combination chosen for this study was doomed to fare poorly in any morphology match-up due to the inclusion of stavudine. This was potentially compounded by the open-label design of this study, as there may have been unwittingly closer scrutiny for body shape changes among subjects who were receiving stavudine.

The triple-NRTI regimen led to a net reduction of hip, arm and thigh circumferences compared to overall modest gains in these and all the other areas measured among the subjects in the other arms. Further, at week 96, more subjects in the triple-NRTI arm had facial fat wasting (6 versus 1 in the efavirenz arm and none in the amprenavir arm). Buffalo hump was reported in 1 subject receiving amprenavir, and gynecomastia was seen in 1 subject in each of the efavirenz and NRTI arms.

Also predictable was the finding that the NNRTI and PI arms had greater increases in total cholesterol than the triple-NRTI arm (NNRTI 49 mg/dL, PI 54 mg/dL and NRTI 28 mg/dL -- P<.001). Unfortunately, and inexplicably, low-density lipoprotein (LDL) and high-density lipoprotein (HDL) cholesterol were not measured, so it is impossible to know if the rise in total cholesterol in the efavirenz arm may have been due to a relatively greater increase in HDL cholesterol, as has been seen in other studies. As for triglycerides, they also tended to rise in all arms, with a relatively smaller increase in the efavirenz arm.

Lipid-lowering therapies were initiated during the study by 12 subjects in the PI arm, 5 in the efavirenz arm and 4 in the NRTI arm.

Summary

Overall, this study, unique in its comparison of 3 major strategies for therapy initiation, points out the strengths and weaknesses of each. After almost 2 years, there were no significant differences between the study arms in rates of achieving viral suppression. However, more subjects in the PI and triple-NRTI arms required a switch to a second-line regimen prior to week 96. Thus, while the different "strategies" led to a similar outcome at study's end, more subjects who had not started with an efavirenz-based regimen had to switch regimens in order to get undetectable at this point.

There is no arguing with the success of the efavirenz regimen during this study. Subjects assigned to this arm tended to have a longer time to first virologic failure and greater declines in viral load, as well as significantly more durable virologic suppression. Further, not 1 subject in the efavirenz arm had to switch to second-line therapy due to virologic failure.

In this relatively small study, it was difficult to discern significant differences in efficacy between the NNRTI and PI arms in terms of time to first virologic failure or proportion with undetectable virus at week 96, although arguably the trends seem to place the PI-first strategy behind efavirenz. The triple NRTI in this study did underperform, and the poor response among those with higher viral loads assigned to triple-NRTI therapy reinforces the need for cautious use of this and similar combinations in such patients, if not all treatment-naive patients.

There was no difference in rates of serious adverse events or treatment-limiting toxicity between study arms. The body shape changes seen with the stavudine-containing arm were unsurprising. The lipid data suggested the PI and NNRTI arms had a similar impact on total cholesterol, but whether this was the case with the dread LDL cholesterol was not examined. The smart money would bet that HDL levels would have accounted for more of the total cholesterol rise that was seen in the NNRTI arm compared to the PI arm.

Given recent data¹ demonstrating zidovudine/lamivudine/abacavir's inferiority as an initial regimen when compared to zidovudine plus lamivudine plus efavirenz or zidovudine/lamivudine/abacavir plus efavirenz, the question still remains whether it is preferable to start treatment with a boosted PI or start with efavirenz plus 2 NRTIs. A large AIDS Clinical Trials Group (ACTG) study comparing efavirenz versus lopinavir/ritonavir (LPV/r, Kaletra), when each is combined with 2 NRTIs, is currently under way and will address the relative potencies of these remaining first-line strategies.

Footnote

1. Gulick RM, Ribaudo HJ, Shikuma CM, et al. Triple-nucleoside regimens versus efavirenz-containing regimens for the initial treatment of HIV-1 infection. N Engl J Med. April 29, 2004;350(18):1850-1861.