Initial NNRTI-based HAART regimens are sustained longer than PI-based HAART, Canadian study shows

Initiating highly active antiretroviral therapy (HAART) based on a non-nucleoside (NNRTI) may provide a better chance to achieve durable treatment and to avoid the need for multiple salvage therapies than protease inhibitor (PI)-based HAART, according to a prospective clinical cohort study from Montreal, Canada.The study is published in the September 24^th issue of the journal AIDS, now available online.

All patients attending the Immunodeficiency Service at the Montreal Chest Institute who initiated their first HAART regimen between January 1998 and June 2003, were included in the study; however, those who began with triple nucleoside-based HAART or a regimen containing both a PI and NNRTI were excluded. A total of 440 patients initiated PI- (66%) or NNRTI-based (34%) HAART during the study period and were followed for a total of 1256 person-years.

The PIs used were nelfinavir (Viracept; 44%), indinavir (Crixivan; 31%), lopinavir/ritonavir (Kaletra; 12%), soft-gel saquinavir (Fortovase; 5%), hard-gel saquinavir (Invirase) plus ritonavir (Norvir; 5%), full dose ritonavir alone (2%) and amprenavir (Agenerase; 1%). The NNRTIs used were efavirenz (Sustiva; 72%), nevirapine (Viramune; 27%) and delavirdine (Rescriptor; 1%). The most common nucleoside backbones were AZT (zidovudine, Retrovir)/3TC (lamivudine, Epivir; 51%) and d4T (stavudine, Zerit)/3TC (35%), and PI-based HAART included d-drugs (ddI [didanosine, Videx / VidexEC], ddC [zalcitabine, Hivid], d4T) more often than NNRTI-based HAART (50% vs. 40%, p = 0.04). With the exception of 2001, more patients began a PI-based HAART regimen during each year of the study period than an NNRTI-based regimen. The median length of follow-up was 3.07 years for PI-based HAART and 2.27 years for NNRTI-based HAART.

Almost one in three patients (58%) modified their initial HAART regimen during the study period: 184 (63%) of those on PI-based HAART and 70 (47%) of those on NNRTI-based HAART. Although the majority switched to a different regimen, a significant minority stopped HAART altogether: 28 (10%) of those initiating PI-based HAART and 17 (12%) of those initiating NNRTI-based HAART.

Overall, the median time to treatment modification was 2.1 years for those initiating with NNRTI-based HAART compared with 1.6 years for those starting PI-based HAART (p = 0.03). Patients starting PI-based HAART were a third more likely to change their initial HAART regimen than those starting NNRTI-based HAART (crude and adjusted hazard ratio [HR], 0.74 for NNRTI vs. PI). This advantage persisted even after controlling for variables that included the year of HAART initiation, previous nucleoside exposure, previous AIDS diagnosis, and baseline CD4 cell count and viral load levels. However, this advantage did not translate into a clinical finding: there were no significant differences seen in rates of new AIDS-defining illnesses, hospitalisation or death.

There was a difference seen in the reasons given for treatment discontinuation, however. Those on NNRTI-based HAART were more likely to stop for reasons of intolerance (especially allergic, psychological and liver toxicities), whereas those on PI-based HAART were more likely to stop for reasons of virological failure. Given that unboosted nelfinavir and indinavir made up 75% of initial PI-based therapies in this study, and head-to-head studies comparing nelfinavir with nevirapine and indinavir with efavirenz found the NNRTIs to be superior to these unboosted PIs, these results are not surprising.

Other factors associated with the likelihood of changing treatment regimens included: previous nucleoside experience (adjusted HR, 1.41), injection drug use (adjusted HR, 1.32), a previous AIDS diagnosis (adjusted HR, 1.32), viral load (adjusted HR, 1.16 per log₁₀ copies/ml) and CD4 cell count (adjusted HR, 1.13 per 100 cells/mm³).

Given the shorter duration of follow-up for individuals initiating NNRTI-based HAART, an analysis restricted to the first two years of follow-up, and adjusting for factors including previous nucleoside use, found that the hazard ratio was no longer statistically significant at 0.86 for those initiating NNRTI-based HAART versus PI-based HAART. Additionally, since only 17% starting PI-based therapy initiated with boosted PIs - which, alongside NNRTIs, are now recommended by both United Kingdom and United States guidelines as an initial treatment option - these findings may not represent state-of-the-art HAART at it is today.

However, in this cohort, the initial use of NNRTI-based HAART was associated with less regimen change than PI-based HAART. The study’s authors note that more follow-up is necessary to determine if this translates into improved clinical outcomes, and these findings will need to be confirmed with larger cohort studies and randomised, controlled clinical trials.