New data from clinical trials of antiretrovirals

Charles Hicks MD, for

As has been true for most of the major HIV conferences over the past 10 years, presentations from the XIV International AIDS Conference in Barcelona included results from several important antiretroviral therapy clinical trials.

While there were no blockbuster breakthrough studies, newly presented data helped clarify some important issues related to antiretroviral therapy. Most of the studies involved either treatment-naïve patients or fairly heavily pre-treated patients.

Clinical trials in treatment-naïve patients

The CLASS study

John Bartlett from Duke University presented results from the GlaxoSmithKline-sponsored ESS40001 trial, also known as the CLASS study. The primary objective of this trial was to compare drugs from different classes of antiretroviral therapy when paired with a nucleoside backbone of abacavir (ABC) and lamivudine (3TC). The third drug was stavudine (d4T), efavirenz (EFV), or ritonavir-boosted amprenavir (APV/r). Data presented were from a planned 48-week analysis. [Importantly, the ultimate objective of this study is to assess two-year outcomes including the efficacy of preplanned sequencing of regimens for those experiencing virologic failure of initial therapy. This portion of the study is ongoing.]

A total of 297 patients enrolled in the study, and the treatment arms were well matched demographically (notably, about 70% of the patients enrolled were black or Hispanic). Median baseline viral load at entry was 4.9 log10 copies/ml and median entry CD4 count was around 300 cells/mm3. Intent-to-treat (ITT) and on treatment (OT) analyses are shown below. For the <50 copy assay, there was a significant difference favoring the EFV arm over either of the other treatment arms in the ITT analysis.

Treatment Arm <400 copies (ITT) <400 copies (OT)
d4T (NRTI) 80% 90%
APV/r (PI) 75% 91%
EFV (NNRTI) 83% 98%
Treatment Arm <50 copies (ITT) <50 copies (OT)
d4T (NRTI) 62% 71%
APV/r (PI) 59% 72%
EFV (NNRTI) 76% 89%

Virologic suppression (using a 50 copy assay) among patients with baseline viral loads above 100,000 copies/ml were also significantly better in the NNRTI arm (EFV = 77%, d4T = 55%, APV/r = 53%). CD4 lymphocyte increases were in the range of 173 to 196 cells/mm3 and were not significantly different among the three arms of the study.

While these 48-week data support the use of efavirenz-based regimens for treatment-naïve patients, it is worth noting that the two-year outcome is the primary objective of this trial. Important treatment strategy issues including successful sequencing and “salvageability” have yet to be delineated.

ACTG 384 study

This study in many ways illustrates the unique strengths of the AIDS Clinical Trials Group (ACTG) as well as some of its weaknesses. Strengths include the large numbers of patients enrolled, the considerable scientific rigour in the design and conduct of the trials, and the meticulous impartiality of the results. Weaknesses primarily relate to the considerable bureaucracy that must be accommodated to get a big study underway.

This latter issue often means that there are substantial delays from study concept to patient accrual to study analysis. Moreover, the review process for ACTG proposals often adds considerable complexity to study designs. Thus, by the time some ACTG trials are presented, study designs often seem surprisingly outdated, and interpretation of results can be challenging.

ACTG 384 was a study with a fairly complex design that was intended to address three main questions: 1) Is zidovudine (ZDV) + 3TC better than d4T + didanosine (ddI) as the nucleoside backbone of ART in treatment-naïve patients? 2) Is an NNRTI (EFV)-based regimen better than a PI (nelfinavir [NFV])-based regimen? 3) Is a four-drug regimen (using both EFV + NFV) that uses drugs from all three classes superior to a three-drug, two-class regimen?

The study enrolled 980 ART-naïve patients in the U.S. and Italy to try to answer these questions. The study was set up as a factorial design, which means that there were six treatment arms that contributed to the results with the various questions being answered by combining results across the regimens.

The design specified second-line regimens for patients failing initial therapy if they were randomised to a three-drug regimen initially. The primary outcome measurement was the time to failure of the initial regimen for patients on the four-drug, three-class arms, but was time to failure of a second regimen for those initially treated with a three-drug, two-class regimen. This outcome was chosen to assess the durability of treatment in terms of the time until patients experience failure to all three classes of ART.

At entry, median baseline viral load was around 87,000 copies of RNA/ml and median baseline CD4 count was 278 cells/mm3. Median follow-up was 28 months. The statistical analysis of the factorial design was quite complex.

The results suggested that better responses were seen with ZDV + 3TC as the nucleoside backbone and with EFV as the third drug of the three-drug regimens. Much of the difference in outcome appeared to be a consequence of excess toxicity in the d4T + ddI arm.

Importantly, the overall interpretation is conditional, however, in that the superiority of the ZDV + 3TC combination was only seen when EFV was the third drug, and the superiority of EFV was only seen when it was given with ZDV + 3TC, and not when given with d4T + ddI. Moreover, the four-drug arms were superior to the three-drug arms, but only when all the three-drug arms were combined for comparison.

The four-drug arms were not superior to the best of the three-drug arms (ZDV + 3TC + EFV). There were no significant differences in CD4 lymphocyte changes among any of the arms.

Thus, this large and complicated study produced an increasingly familiar outcome – ZDV + 3TC + EFV is an excellent initial regimen for ART-naïve patients. This was an initial analysis of a wealth of important data and we are sure to hear more results from this trial in the future.

Gilead 903 study

A review of virologic outcomes in HIV treatment trials over the last five years shows that the proportion of patients achieving viral suppression below the level of detection has slowly but surely increased with the passage of time. This probably reflects improvements in drug potency, in ease of administration, and in drug pharmacokinetics, as well as an increased emphasis on patient adherence.

Results presented at this meeting from the Gilead 903 study continued the trend of improving HIV suppression and raised the bar even higher in terms of what the target ought to be for virologic success among treatment-naïve patients initiating a first HAART regimen.

The study was conducted at 81 sites in the U.S., Europe, and South America and was designed to compare tenofovir (TDF) to stavudine (d4T) when combined with 3TC and efavirenz. Six hundred patients were enrolled in this randomized, double-blind, placebo-controlled study to receive either d4T + 3TC + EFV or TDF + 3TC + EFV.

The mean viral load was 4.9 log10 and the mean CD4 count was 279 cells/mm3. By an ITT analysis in which missing values were counted as failures, both arms achieved 87% suppression to <400 copies RNA/ml at week 48 of follow-up. Using a 50 copy RNA assay, 82% of the TDF patients and 81% of the d4T patients achieved undetectable viral load at week 48. CD4 increases were 169 and 167 cells respectively. Both arms had about 9% of patients discontinuing study therapy, in most cases because of “lost to follow-up” or withdrawal of consent.

Discontinuations due to study drug-related adverse events were uncommon. Efficacy in patients with viral loads above 100,000 copies and with CD4 counts below 200 cells/mm3 was similar to those with less advanced infection.

These data are quite impressive and attest to the tolerability and convenience of both of these regimens. This study adds further evidence of the value of efavirenz and lamivudine as part of initial antiretroviral regimens, and establishes TDF as equivalent to d4T as part of the nuceloside/tide backbone of HAART regimens.

Future trials of antiretroviral regimens for treatment-naïve patients will need to achieve similarly impressive results to be placed alongside these therapeutic choices.

EFAVIP-2 study

In addition to the data reviewed above regarding the use of efavirenz in treatment-naïve patients, there was one additional study that was of considerable interest to clinicians treating patients with advanced HIV infection. This trial, called “EFAvirenz in Very Immunosuppressed Patients” [or, EVAVIP-2] assessed a population of treatment-naïve patients with baseline CD4 counts <100 cells/mm3 who initiated therapy with efavirenz (n = 92) or a non-boosted PI (n = 218).

The data was drawn from three large patient cohorts in Madrid, Spain, and the study was intended to help determine the appropriate treatment strategy for patients with advanced HIV infection as determined by low CD4 lymphocyte counts at HAART initiation. Median CD4 counts were 34-39 cells/mm3, and median viral load was >250,000 copies RNA/ml.

Of potential importance in interpreting the results, the patients treated with efavirenz were significantly more likely to also be on ZDV + 3TC while those on PI therapy were more likely to be receiving ddI + d4T. Probably at least in part due to this difference in nucleoside backbone, there were significantly more discontinuations due to adverse events in the PI arm (23.8% versus 8% in the EFV arm).

There were also more discontinuations due to virologic failure in the PI group (12.8% in the PI arm versus 4.3% in the EFV arm). When analysed by time to treatment failure, the two treatment strategies diverged by about six months with improved outcomes being observed in the EFV arm at all subsequent time points. Similarly, CD4 increases were significantly greater in patients receiving EFV.

While the differences observed were impressive, the study was a retrospective analysis, and, as such, is subject to several limitations. Most importantly, the study was not randomised, which makes selection bias a very real potential issue. Additionally, the PI regimens were not ritonavir-boosted and thus not reflective of most PI-based HAART currently used. Despite these reservations, this study provided valuable data supporting the efficacy of EFV in patients with advanced degrees of immunosuppression.

Clinical trials in treatment-experienced patients

Data from trials of treatment-experienced patients was not generally as compelling as that from treatment-naïve patients. There were, however, some very interesting data presented on the use of T-20, the first fusion inhibitor, as well as on a PI-only regimen that avoided the use of RT inhibitors.

TORO-1 and TORO-2

T-20 (now also known by the tongue-twisting name “enfuvirtide”) is the first entry inhibitor to reach late stages of clinical development. The drug works by blocking viral fusion with cell envelope and is administered by subcutaneous injection. It is close to achieving regulatory approval in parts of the developed world.

Two important trials (termed the TORO studies for “T-20 versus Optimised Regimen Only”) were undertaken to obtain this approval and the results from these studies were presented in Barcelona. Both studies tested the efficacy of T-20 in heavily pre-treated patients experiencing virologic failure. Entry criteria required treatment experience with all three classes of drugs and a viral load >5000 copies RNA/ml.

All patients received an optimised salvage regimen of antiretroviral agents based on clinical history and resistance testing and were randomised to an optimised ART regimen alone or to an optimised regimen plus open label T-20 given at a dose of 90mg SQ q12h.

TORO-1 was conducted in the U.S. and enrolled 491 patients. Patients were very treatment-experienced and had advanced HIV infection. They had received a median of 12 prior antiretroviral agents over seven years of therapy, and 87% had experienced an AIDS-defining condition. Median viral load was 5.2 log10 and median CD4 count was 80 cells/mm3.

A total of 326 patients were randomised to the T-20 arm and 165 to the optimised therapy only arm. There was a significantly greater reduction in viral load in the T-20 arm: -1.70 log10 copies versus –0.76 log10 copies in the optimised therapy only arm. At 24 weeks, 37% of the T-20 patients had viral load <400 copies RNA/ml compared to 16% in the optimised therapy only arm.

Although injection site reactions were near universal, treatment discontinuations did not differ between the two arms: 11.3% in the T-20 arm and 10.9% in the optimised therapy only arm.

TORO-2 was almost identical in design. Five hundred and four patients with more than three months of experience with all three classes of antiretroviral agents were randomised to optimal therapy alone or optimal therapy plus T-20. At week 24, the patients receiving T-20 had a mean decrease in viral load of –1.43 log10 copies compared to –0.65 log10 copies in the optimised therapy alone arm. Study discontinuation rates in this trial were greater in the T-20 arm: 17% versus 5% by week 24. Only 3% of the patients withdrew due to injection site reactions.

These two studies were consistent with one another and demonstrated impressive antiretroviral activity through 24 weeks. This is welcome news for the ever-growing population of HIV-infected patients who have continued viral replication despite having received therapy with all currently available classes of agents. It remains to be seen what the long-term efficacy and tolerability of this injectable agent will be, but for now, the substantial potency of this drug is encouraging.


A variety of approaches have been advocated to manage patients who have experienced multiple drug failure including treatment interruption on one extreme and “mega-HAART” on the other.

A small study by Schlomo Staszewski and colleagues suggested another option, that of a dual boosted PI, RTI-sparing regimen. In this study, 33 patients with multiple drug failure were given lopinavir/r + saquinavir. The rationale for this approach came from resistance testing in these patients which demonstrated potential susceptibility to both protease inhibitors with little evidence for susceptibility to any of the available reverse transcriptase inhibitors.

At the time of the switch, the median viral load was 5.2 log10 copies and the median CD4 count was 157 cells/mm3. After 24 weeks of follow-up, the median decline in viral load was –3.5 log10 and the median increase in CD4 count was +159 cells/mm3. An amazing 82% achieved viral suppression to below 400 copies and 58% were below 50 copies.

The dual PI therapy was relatively well tolerated with 73% still on the drug regimen after a median of 29 weeks of follow-up. While this is certainly a very selected population, the impact of this regimen was quite dramatic in these patients.

Dr. Hicks is Associate Professor of Medicine in the Division of Infectious Diseases, Associate Director of the Duke University AIDS Research and Treatment Center, Durham, NC, and a Contributing Editor to HIV and



Links to other websites are current at date of posting but not maintained.