First-line antiretroviral therapy has improved over time, meta-analysis shows

The effectiveness of initial antiretroviral treatment has improved markedly over the years, but many people still do not achieve full HIV suppression and a majority end up going off their initial regimen for various reasons, according to a large meta-analysis presented Wednesday at the 7th International AIDS Society Conference on HIV Pathogenesis, Treatment and Prevention in Kuala Lumpur.

Countless clinical trials of first-line antiretroviral drugs have been conducted over the course of the HIV epidemic, but it can be hard to see the bigger picture of how well initial antiretroviral therapy (ART) has performed overall, and difficult to compare individual studies with variable patient inclusion criteria and outcome measures. A broader overview could help inform the development of treatment guidelines.

Frederick Lee of St. Vincent's Centre for Applied Medical Research in Sydney and colleagues therefore undertook a large meta-analysis of antiretroviral clinical trials through to the end of December 2012, gathering study data from MEDLINE, clinical trial registries, conference abstracts, drug label information and data provided by pharmaceutical companies.

The researchers included prospective trials that enrolled treatment-naive adults with HIV, followed participants for at least 48 weeks and used an intent-to-treat (ITT) efficacy analysis. The analysis excluded retrospective studies and those that evaluated regimens not currently recommended due to toxicity or poor effectiveness.

The primary measure was overall efficacy, or the proportion of patients with undetectable viral load <50 copies/ml in all studies over the maximum follow-up period. Secondary analyses included changes in efficacy over time, efficacy for individuals with high vs low viral load, efficacy according to the latest US treatment guidelines and predictors of efficacy and treatment failure.

The researchers identified 114 relevant studies, which included 216 study arms and a total of 40,124 participants. Most (196 studies) were randomised and 68 included a placebo control group. There were 50 phase 2 trials, 96 phase 3 studies and 70 phase 4 or post-market studies. Looking at study funding, 123 were sponsored by the pharmaceutical industry, 55 were academic and 38 were classified as 'both'.

Only 10 of the included studies were conducted prior to the widespread availability of effective combination ART in the mid-1990s. A total of 53 studies were conducted during 1997 to 1999, 43 during 2000 to 2002, 61 during 2003 to 2005, 31 during 2006 to 2008 and 17 during the most recent period, 2009 to 2010. Most studies (131) had a duration of 48 weeks, 60 lasted for 96 weeks and 25 continued for 144 weeks; the overall mean duration was 82 weeks.

Overall, most participants (76%) were men, 65% were white or Caucasian and 27% were black or of African descent. By HIV risk group, about half were men who have sex with men, 38% were heterosexual and 10% had a history of injection drug use. Collectively, the average CD4 count was quite high at 248 cells/mm³, but 12% had previously received a diagnosis of AIDS; 10% were co-infected with hepatitis C.

Lee explained that his team did not have access to data about individual study participants, but he did note that patient characteristics had shifted over time (including more women and people with less advanced disease in later years, for example).

Looking at the regimens evaluated in the trials, 49% included a non-nucleoside reverse transcriptase inhibitor (NNRTI), 24% included ritonavir-boosted protease inhibitors and 14% included unboosted protease inhibitors. In addition, nine studies (5%) included the newer integrase inhibitor class, while 12 (4%) looked at regimens containing only nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs).

About one-third of participants were using regimens classified as 'preferred' or 'alternative' in the 2012 US Department of Health and Human Services (DHHS) antiretroviral therapy guidelines. Two-thirds of study regimens had some sort of food restrictions, and they included an average of six pills per day taken in two daily doses.

Focusing on the NRTIs, the most frequently used 'backbone' was AZT (zidovudine or Retrovir) plus 3TC (lamivudine or Epivir), the drugs in the Combivir fixed-dose combination pill. This was followed by tenofovir (Viread) plus FTC (emtricitabine or Emtriva), the NRTIs in the Truvada, Atripla, Complera/Eviplera and Stribild co-formulations. In addition, 27 studies evaluated d4T (stavudine or Zerit) plus 3TC, 26 looked at abacavir (Ziagen) plus 3TC (the drugs in Kivexa or Epzicom), 20 tested d4T plus ddI (didanosine or Videx) and nine tested tenofovir plus 3TC.

Efficacy over time

Taken as a whole, the overall efficacy of first-line ART evaluated in the studies was 60%. Effectiveness decreased with longer time on a regimen. Among participants evaluated for 48 weeks, efficacy was 66%, falling to 60% in 96-week evaluations and 52% in 144-week evaluations.

Overall, 25% of participants stopped ART during their studies. Rates rose with longer time on therapy, from 20% in 48-week studies to 28% in 96-week studies to 34% in 144-week studies. Eleven per cent of discontinuations were due to 'patient decision', 8% were due to adverse events or side-effects, and nearly 4% were due to virological failure.

While the overall 60% rate of viral suppression was surprisingly low, treatment effectiveness did improve over time. Before 2000 the combined efficacy was just 47%, rising to 52% between 2000 and 2002, 64% between 2003 and 2005, 74% between 2006 and 2008 and, finally, 82% after 2008.

A variety of factors were associated with better ART effectiveness.

Regimens containing integrase inhibitors – which were all conducted in later years – worked best, with an efficacy of 84%. Boosted protease inhibitor and NNRTI regimens performed similarly, with efficacy of 67% and 61%, respectively. Bringing up the rear were all-NRTI regimens (51%) and unboosted protease inhibitors (42%). Looking at widely used modern NRTI backbones, tenofovir/FTC (73%), tenofovir/3TC (69%) and abacavir/3TC (63%) performed best, followed by AZT/3TC (48%).

Regimen components classified as 'preferred' in US treatment guidelines produced an overall efficacy of 75%, compared with 65% for those classified as 'alternative'. Preferred regimens were also less likely to result in treatment discontinuation (20% vs 25%, respectively).

Based on these findings, the researchers concluded, "overall mean efficacy of initial ART is low," at just 60%, and "most patients will interrupt initial ART." Furthermore, there was a "significant 8.4% difference in efficacy" between study participants with higher (>100,000 copies/ml) vs lower pre-treatment viral load.

Factors that predicted greater treatment success included use of tenofovir/FTC vs abacavir/3TC and use of integrase inhibitors vs NNRTIs or boosted protease inhibitors – a finding that held for people with high viral load.

"Despite focus on co-formulation, fewer daily pills and doses [are] not independent predictors of overall efficacy," they added.

This analysis "calls into question antiviral potency at high viral loads", said Lee.

Although US and European treatment guidelines have not included viral load ART initiation thresholds for several years, the investigators suggested that they should perhaps recommend that patients start therapy before viral load reaches 100,000 copies/ml.

Speaking from the audience, Daniel Kuritzkes of Brigham and Women's Hospital, a member of the DHHS guidelines panel, opined that these findings validated current treatment advice. "It turns out a bunch of experts sitting around looking at data actually get it right," he said.