We report here on an analysis describing the outcomes of 70 HIV patients on LPV/r-based second-line HAART regimens for more than 24 months, followed at the ESTHER programme in Phnom Penh, Cambodia. After 24 months of follow up on second-line regimens, 65 (92.8%, CI: 84.3- 96.8) remained on treatment and 60 (92.3%, CI: 83.2- 96.6) had undetectable viral loads, giving an 85.7% (CI: 75.6- 92.0) treatment success rate in intent-to-treat analysis. A strong immune reconstitution was observed at 24 months (+258 cells/mm3 [IQR: 136-425]) of follow up on second-line HAART regimens. While no specific information on adherence was available, these very good virological data revealed that these patients were indeed adhering well despite the complexity (number of pills) and the difficulties of storage conditions (at 2-8°C for LPV/r) of such regimens used at that time, demonstrating the feasibility of PI-based second-line regimens in such resource-limited settings.
Most of these patients were switched to second-line regimens following cross-sectional virological evaluations performed at the ESTHER/Calmette programme. Thus, it is important to note that this modality of switch is not representative of what is currently done in Cambodia, where clinico-immunological criteria are predominantly used. On the other hand, we could reasonably assume that such a cross-sectional survey might have detected treatment failures earlier than if routine immunological detection was applied.
Clearly, some patients at switch were already resistant to some second-line NRTI drugs recommended and available in Cambodia. Indeed, we found that all 41 patients with resistance genotype available before switching to LPV/r based second-line ARV regimens were resistant to 3TC/FTC, and that 10% (four of 41) were resistant to at least one other second-line drug, including ddI, ABC or TDF. Thus, we can assume that the vast majority of the 70 patients on PI-based regimens and followed up for 24 months were resistant to 3TC.
Because of national programme recommendations, a large majority of our patients (57%) initiated 3TC-based triple combinations as second-line HAART regimens. It is thus noteworthy to observe such excellent immuno-virological outcomes after two years of follow up despite the fact that such second-line regimens might have worked only as dual therapy or monotherapy from their initiation. This might be not surprising because all patients were apparently naïve for protease inhibitors.
On the other hand, the M184V mutation, in addition to conferring high levels of resistance to 3TC, has also been shown to reduce viral replication capacity [34
], to increase the faithfulness of the RT [35
], and to maintain more sensitivity to 3TC than previously believed [36
]. This led to the speculation that 3TC might still be beneficial in patients harbouring the M184V mutation. Recent data from a pilot study comparing complete HAART interruption with the maintenance of 3TC despite the presence of M184V mutation in salvage therapy revealed better clinical and immunological outcomes at 48 weeks when maintaining 3TC, with no further accumulated mutations [38
]. Whether such a benefit of 3TC also applies to patients on second-line regimens is still unknown and needs to be investigated.
In addition, it remains to be shown that the effectiveness of such "non-optimal" second-line regimens after two years could be maintained on longer periods of follow up. In any case, the fact that such patterns of resistance with limited options for second-line regimens was already observed following the initial cross-sectional survey advocates for the use of monitoring strategies allowing early virological failure detection, as well as the use of stronger first-line regimens with higher genetic barriers.
Considering the five patients presenting detectable VL after 24 months on LPV/r-based second-line regimens, we found that two of them (patients 4 and 5) harboured PI resistance-associated mutations. Further clinical investigations revealed that both of these two patients were exposed to several PIs and other RTIs before enrolment into the ESTHER cohort. Another patient (patient 3, Table ), displayed no drug resistance mutation at all, which could be explained by treatment interruption. The PR gene could not be amplified for the two remaining patients (patients 1 and 2), probably because of their low-level viral loads (2.8 and 3.6 Log10). As for first-line ARV regimens, these data suggest that not being PI naïve could also be a risk factor for treatment failure on second-line regimens.
Recently, a multicohort study analyzing patients on second-line ARV therapy for more than six months in 27 Médecins Sans Frontières (MSF) ART programmes in Africa and Asia reported that 18.8% of 632 patients displayed WHO clinical, immunological or virological criteria of failure after a median of 11.9 months of follow up, with cumulative probabilities showing up to 28% failures at two years [20
]. Our current study found only 7.7% (CI: 3.4- 16.8) virological failure among actively followed patients after 24 months, which is in apparent disagreement with the results reported by the MSF study. In fact, it is difficult to compare these studies since each used very different criteria to define second-line treatment failures.
Indeed, not all patients with clinical or immunological criteria of failure had virological confirmation in the MSF study, while all of our patients were defined virologically only. As it is known that patients with clinical or immunological WHO criteria of failure could display undetectable viral loads, it would be of interest to know the rates of viological failure among those patients with virological confirmation in the MSF study. In addition, the differences observed might be cohort specific since we analyzed a limited number of patients from one cohort only, while the MSF study analyzed 27 distinct cohorts and a much larger number of patients. Further studies addressing the effectiveness of boosted protease inhibitor-based regimens in resource-limited settings are still needed. Detailed description of cohort characteristics, including patient follow-up procedures and second-line adherence support, will be critical for identifying programmatic factors that might influence second-line outcomes in such settings.