How does abacavir/3TC compare to tenofovir/FTC? ACTG5202 and GSK studies disagree

Contradictory findings regarding the safety and efficacy of the dual-nucleoside combination of abacavir and 3TC were presented in a late-breaker session at the XVII International AIDS Conference on Thursday. A secondary analysis of 48-week data from the ACTG5202 study reasserted that the dual-nucleoside backbone of abacavir plus 3TC was more likely to lead to toxicity and earlier virologic failure than tenofovir plus FTC. However, a presentation from GlaxoSmithKline (GSK) indicated that, when data from other well-designed clinical trials was analysed using the same study endpoints as ACTG5202, abacavir/3TC proved just as tolerable and effective as tenofovir/FTC.

Two dual-nucleoside "backbones" – abacavir(ABC)/3TC and tenofovir (TDF)/FTC – are both recommended as part of first-line antiretroviral therapy, and are both available in fixed-dose combinations. (Abacavir/3TC is marketed as Kivexa in Europe and the EU and Epzicom in the US; TDF/FTC as Truvada.)

In the ongoing US AIDS Clinical Trials Group (ACTG) Study A5202, each of these two nucleoside pairs has been paired with either efavirenz (Sustiva) or ritonavir-boosted atazanavir (Reyataz). This randomised study was designed to compare the safety and effectiveness of the four resultant first-line antiretroviral regimens:

• ABC/3TC plus efavirenz,
• ABC/3TC plus ritonavir-boosted atazanavir,
• TDF/FTC plus efavirenz, and
• TDF/FTC plus ritonavir-boosted atazanavir.

Participants (1858 treatment-naïve patients) were randomised to one of the four treatment arms. For analysis, participants were further divided into two groups according to their pre-treatment viral load – either above or below 100,000 copies/ml.

This trial enrolled patients between 2005 and 2007 and is intended to run for 96 weeks. In January 2008 (after a median follow-up time of 60 weeks), a planned interim safety monitoring review found that participants with higher viral loads (> 100,000 copies/ml) were failing virologically more quickly on ABC/3TC than on TDF/FTC. (See the earlier aidsmap report here).

Originally, investigators and participants had not known which nucleosides the participants were taking. Due to this unexpected finding, the study nucleosides were unblinded (i.e., revealed to the investigators and participants) for participants with higher viral loads. These participants continued in the trial while having the option of staying on ABC/3TC, or switching to TDF/FTC.

For ACTG5202 participants with viral loads below 100,000 copies/ml, the study remained blinded and is continuing as originally planned; those results will be presented at a later date.

In this late-breaker session, Paul Sax of Boston's Harvard Medical School reported new data from the 797 participants with viral loads greater than 100,000 copies/ml. Characteristics of these 797 were: 85% male, 47% white, 26% black and 25% Hispanic; mean baseline viral load 5.1 log copies/ml and median CD4 cell count 181 cells/mm³. Discontinuations were 10% overall, with no difference between treatment arms.

Overall, for this high-viral load group, people failed virologically more than twice as quickly on ABC/3TC than on TDF/FTC (hazard ratio [HR], 2.33; 95% confidence interval [CI], 1.46 to 3.72; p=0.0003). The investigators looked at this finding in more detail, separately analysing virologic failures that happened before and after 24 weeks, failures to ever achieve undetectable viral load, and viral rebounds after initial virologic success. The rates of failure on ABC/3TC were higher than those for TDF/FTC in all of these scenarios.

However, this secondary analysis also looked at a cross-section of patients at the 48-week point, finding that after allowing for treatment changes, outcomes became similar. This analysis counted all cases of virologic failure, either before or after the treatment change, and compared all participants according to the treatment they began with, whether or not they had changed treatment. At 48 weeks, similar proportions of each group showed undetectable viral loads: 75% of patients who began on ABC/3TC (95% CI, 69% to 80%), and 80% of those who started on TDF/FTC (95% CI, 74% to 85%; p=0.20).

Serious (grade 3 or 4) adverse events (AEs) occurred significantly sooner in people receiving ABC/3TC (HR, 1.87; 95% CI; 1.43 to 2.43; p<0.0001). Grade 3 or 4 AEs were reported in 33% on ABC/3TC and 19% on TDF/FTC. These were predominantly increases in blood lipid levels (10% vs 3%) and general body pains (14% vs 10%); there were no myocardial infarctions in either arm.

Dr Sax concluded that, "in subjects entering study ACTG5202 with screening HIV RNA ≥ 100,000 copies/ml, ABC/3TC was associated with a significantly shorter time to virologic failure and grade 3/4 adverse events than TDF/FTC."

The remainder of ACTG5202, in which the third drug (efavirenz or boosted atazanavir) is blinded for all participants, and NRTIs remain blinded for those in the lower HIV viral load stratum, is ongoing and scheduled to continue until November 2009. Other analyses are also ongoing, including analyses of resistance and adherence.

In the following session, Keith Pappa challenged the ACTG5202 findings with data from a study by GSK (the manufacturer of abacavir and Kivexa). This GSK study examined 48-week data from five other clinical trials in which ABC/3TC-containing regimens were used: the KLEAN, SHARE, CNA30021, CNA30024 and ESS30009 studies.

The actual treatment regimens used in each trial were as follows:

• ABC/3TC + atazanavir/r (SHARE)
• ABC/3TC + lopinavir/r and ABC/3TC + fosamprenavir/r (KLEAN)
• ABC/3TC + efavirenz (CNA30024 and ESS30009)
• ABC/3TC (qd) + EFV (CNA30021)

As in ACTG5202, participants were split into two groups according to viral load – less than, and greater than or equal to, 100,000 copies/ml. The study also used the same primary efficacy endpoint as ACTG5202 – time to virologic failure, defined as either a confirmed viral load of at least 1,000 copies/ml between 16 and 24 weeks ("early failure"), or a viral load at least 200 copies/ml at or after 24 weeks("late failure").

The analysis found that, in the existing six clinical trials, between 87% and 95% of participants had not experienced virologic failure by 48 weeks. Importantly, there was no clarification as to whether these were on-treatment or intent-to-treat values; the extremely high success rates would suggest on-treatment values. The rates of people who had not experienced virologic failure by 48 weeks in the studies examined, by high (≥100,000 copies/ml) and low (<100,000) viral loads, were as follows:

• SHARE: 93% (high), 93% (low)
• KLEAN, lopinavir/r: 92% (high), 95% (high)
• KLEAN, fosamprenavir/r: 92% (high), 94% (high)
• CNA30024: 93% (high), 95% (high)
• ESS30009: 95% (high), 95% (high)
• CNA30021: 89% (high), 94% (high)

Pappas also presented data from the HEAT trial, data from which were previously presented at the Fifteenth Conference on Retroviruses and Opportunistic Infections in Boston earlier this year. HEAT showed statistical non-inferiority of ABC/3TC to TDF/FTC in terms of virologic success at 48 weeks, in patients at all baseline viral loads.

For patients with viral loads < 100,000 copies/ml, 90% were virologically suppressed at 48 weeks vs 87% on TDF/FTC, and for those with viral loads ≥ 100,000 copies/ml, the figures were 87% vs 90% respectively.

The HEAT study also assessed the time of occurrence of the first serious adverse event – any symptom or lab toxicity of either grade 3 or 4, or at least one grade higher than at baseline. Occurrences of any such AE were similar and infrequent in both treatment arms, at 3% or less. Discontinuations due to AEs were 5% for ABC/3TC and 8% for TDF/FTC for people in the higher viral load group. (People with abnormal creatinine kinase or bilirubin levels were excluded from the study.)

This study team concluded that, "using the efficacy and safety endpoint defined in ACTG5202, this analysis of six clinical trials showed that the efficacy of ABC/3TC-containing regimens was robust in ART-naive pts irrespective of baseline viral loads."

Why the difference?

The question remains as to how to reconcile the ACTG5202 findings with the HEAT and other figures presented by GSK. Pappas indicated that the ACTG findings were "unexpected" (a point the ACTG investigators do not dispute) and "different from clinical experience". However, as Sax pointed out, the sample size of the HEAT study – 688 patients – is smaller than that of ACTG 5202.

One delegate at the presentation pointed out that, in each of the six studies cited by GSK, there appeared to be a trend (although small) toward poorer virologic performance with ABC/3TC. As a pooled meta-analysis was not performed, there is a question as to what pooled data from all six studies might show.