In a meta-analysis of data from nearly 8000 patients reported in Clinical Infectious Diseases, researchers found significantly fewer drug resistance mutations, including NRTI-associated mutations, in people who had experienced treatment failure while taking first-line HIV treatment based on boosted PIs than in people taking NNRTI-based therapy.
Genotypic NNRTI resistance was much more common than PI resistance. The NRTI-associated M184V and (in non-AZT-containing regimens) K65R mutation were also more common in cases of NNRTI failure than of PI failure. These data were drawn from a meta-analysis of participants in 20 clinical trials for whom first-line treatment failure was seen after 48 weeks or less.
Previous studies have suggested that genotypic resistance is more frequent after failure of non-nucleoside reverse transcriptase inhibitor-(NNRTI) based antiretroviral therapy than antiretroviral therapy based on boosted protease inhibitors (PIs). (See, for instance, the reports here and here.
However, studies have not always been based on the most common present-day combinations, and have not produced many conclusions about the relative rates of nucleoside reverse transcriptase inhibitor (NRTI) resistance.
In this study, a UK/Thai research team conducted a meta-analysis of clinical trials involving antiretroviral regimens consisting of two NRTIs plus either an NNRTI or a ritonavir-boosted PI. The objective was to compare genotypic resistance profiles after cases of virologic failure of first-line antiretroviral therapy within the first 48 weeks of treatment.
The researchers included 20 trials (such as 2NN, Gemini, KLEAN, MONARK, ACTG5142 and others) with a minimum of 48 weeks follow-up data and genotypic data on virologic failures. This comprised 7970 adult patients (over thirteen years of age) in a total of 30 treatment arms. All but one were randomised trials with a control arm; fifteen were open-label and five were double-blind. The studies were conducted and presented or published between 2001 and 2007.
At baseline, participants in NNRTI vs boosted protease inhibitor arms were a mean 35 vs 37 years old, with mean CD4 cell counts of 250 vs 202 cells/mm3 and viral load of approximately 75,000 copies/ml in both groups of patients; most of the patients (77%) were men. Very few studies determined baseline genotypes.
Virologic failure was defined as any of the following: viral load above 400 copies/ml with no further data given, viral load above 400 copies/ml after achieving suppression or failure to fall below 400 copies/ml by 24 weeks, or viral load above 200 copies/ml after achieving suppression or failure to fall below 200 copies/ml by 32 weeks.
Major NNRTI and PI resistance mutations were identified as defined by the International AIDS Society–USA guidelines; NRTI resistance was defined as either the K65R or 3TC-resistance-associated M184V mutation or at least one thymidine analogue mutation (TAM).
Results – virologic failure and resistance rates
Virologic failure at 48 weeks occurred in approximately 5% of NNRTI recipients and 5% of patients treated with a PI. Of the patients whose NNRTI-based regimens failed, 80% were successfully genotyped, and NNRTI resistance mutations were seen in 53% of these (95% confidence interval [CI], 46%–60%). Of the 83% of the successfully genotyped boosted PI-based regimen failures, PI resistance was seen in approximately 1% (95% CI, 0.0%–6.2%; p
The M184V mutation occurred in 35% (95% CI, 29.3%–41.6%) of failures on NNRTI-based treatment vs 21% (95% CI, 14.4%–28.8%; p
In the subset of patients on non-AZT-containing regimens, the multi-nucleoside-associated K65R mutation was seen in 5% (95% CI, 2.4%–9.9%) of NNRTI failures but hardly any patients taking a boosted PI (0%, 95% CI, 0.0%–3.6%; p = 0.01).
Boosted PI-based treatments were more likely to use newer NRTIs such as tenofovir and abacavir: the thymidine analogues (d4T and AZT) were used in 61% of NNRTI regimens but only 15% of boosted PI regimens. The use of thymidine analogues vs tenofovir or abacavir did not affect rates of NNRTI resistance (36% vs 35% respectively) or of 3TC resistance (54% vs 53%). This analysis could not be done for boosted PIs due to the low number receiving thymidine analogues.
Interpretations
The investigators stress that, while no detectable resistance mutations were found in roughly 30% of the virologic failures in this study, genotyping may significantly underreport true levels of resistance: “perhaps most importantly, our data from clinical trials substantially underestimate resistance rates seen in clinical practice.”
These findings are consistent with the 96-week results of ACTG 5142 and other published meta-analyses, and the researchers state that results from individual contributing studies were generally consistent – a fact that would tend to argue against any bias or confounding.
While the rates of PI vs NNRTI mutations are not surprising, the differences in NRTI resistance are “intriguing”. The investigators speculate that, “because bPIs retain activity despite low-level emergence or preexistence of NRTI-resistant species, then such resistant species [would] remain suppressed. Thus, a larger proportion of failures during bPI treatment [would] be associated with poor adherence” (a factor that was not assessed).
They conclude that these findings are “of particular significance for the developing world, where rates of resistance to NRTIs and NNRTIs at 48 weeks are much higher than has been seen in both cohorts and clinical trials in well-resourced countries.”
Reference:
Gupta R et al. Emergence of drug resistance in HIV type 1-infected patients after receipt of first-line highly active antiretroviral therapy: a systematic review of clinical trials. CID 47:712–722, 2008.