ACTG 384: Major HAART strategy trial a closing day highlight in Barcelona

The choice of a trio of clowns to lead the closing ceremony of the Barcelona International AIDS Conference was, one imagines, an ironic gesture on the part of the organisers of this biannual extravaganza which, now mammoth in scope, has more sideshows than the average British seaside resort.

The closing morning’s Late Breaker session was, however, unusually replete with clinical presentations of immediate relevance – amongst them data on tenofovir in naïve patients, on T-20 in salvage therapy, nevirapine plus AZT for perinatal transmission prophylaxis, and growth hormone for lipodystrophy.

ACTG 384 is a large US/Italian trial investigating strategies for initial antiretroviral treatment of HIV infection. This large, randomised trial was designed to answer three key questions about how to start HAART:

1. Is it better to begin with ddI/d4T or AZT/3TC as the nucleoside analogue backbone?

2. Is it better to start with a protease inhibitor (in this case nelfinavir, NFV) or an NNRTI (efavirenz, EFV)?

3. Is it better to use two sequential three drug combinations, or a single four drug combination?

To answer these questions, the researchers recruited 980 people who had taken no prior antiretroviral therapy, and randomised them into one of six study arms:

Participants in arms A to D were therefore to begin a three drug regimen (Step 1) and switch to an alternative second regimen (Step 2) on the failure of their first regimen. The primary trial endpoint for these four arms was the failure of the second regimen. Time to first regimen failure or first virological failure were secondary endpoints.

Participants in arms E and F began a four drug regimen. Their primary endpoint was reached when this regimen failed. So, at the time of reaching the primary endpoint, all trial participants would have been exposed to drugs from all three major classes: nucleoside analogues, NNRTIS and protease inhibitors. However, those in Arms A to D would have taken six different drugs compared to four in Arms E and F.

Regimen failure was defined as regimen intolerance/toxicity; premature study treatment discontinuation (for any reason); or using the following virological criteria (based on two consecutive viral load tests):

• less than one log decrease and above 200 copies at week 8

• greater than one log increase from nadir, and above 2,000 copies

• above 200 copies after two viral loads below 200 copies

• above 200 copies after week 24 of a regimen.

Nucleoside analogues were given open-label, but both efavirenz and nelfinavir were given blinded by placebo.

All participants had viral load above 500 copies on entry, the median value being 4.9 log. Median CD4 count was 278 cells. Participants were followed for a median of 28 months. During this time, there were 263 premature treatment discontinuations at Step 1 or 2 which resulted in a primary endpoint.

Regarding the comparison between the two nucleoside analogue backbones, there was a trend towards delaying the primary endpoint in favour of AZT/3TC versus ddI/d4T when taken in combination with efavirenz, but not with neflinavir. In relation to the secondary endpoints, AZT/3TC significantly delayed both first regimen failure and first virological failure when combined with efavirenz but not with nelfinavir.

Turning to the protease inhibitor versus NNRTI comparison, it follows that efavirenz out-performed neflinavir with regard to delayed first regimen failure and first virological failure when combined with AZT/3TC, but not with ddI/d4T.

ACTG 384 has not provided evidence that beginning with four drugs is preferable to beginning with three (though participants were stratified at entry for viral load, and future analyses will presumably provide more data on responses in those with high viral load, the group this strategy is sometimes favoured for). Overall, there was no difference in time to the primary endpoint between those starting with a four drug combination and those who began with three drugs and switched to a second three drug regimen. Whilst the four drug regimens delayed time to first regimen failure and first virological failure compared to the nelfinavir-based three drug regimens, and compared to ddI/d4T/efavirenz, they performed no better than AZT/3TC/efavirenz here.

There were no significant differences in CD4 response across study arms. The median increase was 168 cells at week 48, 251 cells at week 96, and 295 at week 144.

Given that overall toxicity and (unsurprisingly) peripheral neuropathy were greater in the ddI/d4T-containing arms, the study group concluded in favour of AZT/3TC/efavirenz as first-line therapy for treatment of HIV infection.

Future analyses will assess the degree to which resistance, pharmacology and adherence may explain these data.