Antiretroviral Therapy (ART) has undoubtedly changed lives, both in resource-rich and in resource-limited settings (RLS). While a large majority of patients on antiretroviral (ARV) drugs achieve sustained virological suppression, some experience a treatment failure eventually leading to treatment change. Treatment switches to second–line ARV regimens in RLS occur at a lower rate than expected due to several constraints: limited access to virological monitoring, limited availability of suitable ARV drugs, pill burden, toxicity and cost, thus limiting the use of second-line ART regimens even in failing patients 1 2 3 4 . Overall, it is estimated that second–line ART accounts for <5% of total ARV use in RLS at the present time 5 .

Currently, most treatment failures are diagnosed by a CD4 cell count decrease, or by a clinical event as per the 2006 World Health Organization (WHO) guidelines 6 . Cohort analyses suggested that non-virological criteria do not accurately predict virological failure 7 8 .

Two randomised trials, (Home-Based AIDS Care [HBAC] and Development of Antiretroviral Therapy in Africa [DART] 9 10 , have addressed the question of the efficacy of monitoring strategies in RLS. A meta-analysis of these two trials concluded that clinical monitoring alone (compared to combined immunological and clinical monitoring or to combined virological, immunological, and clinical monitoring) resulted in unnecessary treatment switches, disease progression, and increased mortality 11 . These results were not confirmed by a recent study conducted in Cameroon in which no differences between laboratory-based or clinical strategies were demonstrated at 24 months 12 . Another clinical trial in Thailand failed to demonstrate a significant difference in clinical failure among patients whose switching decision was based on CD4 cell counts versus HIV-RNA at 3 years 13 . Nevertheless, newer WHO recommendations do recommend using HIV-RNA plasma viral load (HIV-RNA) to confirm a suspected treatment failure 14 .

The optimal threshold of HIV-RNA and the optimal frequency of HIV-RNA determination that best inform antiretroviral switches in RLS are still being debated. In Europe, Australia and the USA, most clinicians continue to rely on the approach in which a virological failure is defined by an HIV-RNA level above the detection threshold; and complete viral suppression is the goal of ART, with HIV-RNA level being monitored every three to six months. Thus, the American guidelines and the guidelines from various European countries all propose that HIV-RNA detection is the essential tool by which to monitor the need to switch ART 15 16 .

This strategy of maximal viral suppression may be less appropriate in a context in which the second-line treatment may be considered as the last chance for a suppressive regimen. The PLATO study suggested that ART regimen that maintains a viral load of <10,000 copies/ml was not associated with an appreciable CD4 cell count decline in triple-class experienced patients after a median follow-up period of two years 17 . Similarly, an Italian cohort study on 3023 patients over a median clinical follow-up time of 46 months showed that the risk of clinical progression was lower in patients with moderate viremia (<10,000 copies/ml), compared with patients with higher HIV-RNA levels 18 .

We aimed to investigate the effect of on-going HIV-RNA replication at low levels in routinely monitored patients on a non-nucleoside reverse transcriptase inhibitor (NNRTI) based first-line ARV regimen in RLS. For this purpose, we analysed changes in CD4 cell counts in patients with an incomplete virological response or with a virological failure to two ARV drug classes.

Our data demonstrates that among patients initiating therapy in RLS with (mainly) an NNRTI-based regimen, CD4 cell count recovery is impaired in the presence of on-going HIV-RNA replication over an 18-month period. We were not able to find a minimal HIV-RNA threshold below which viral replication had no impact on immunological restoration. Our analysis showed an impact of HIV-RNA replication at 12 months on the direction of the CD4 cell count slope; thresholds of 10,000 or 5,000 copies were equally predictive of the direction of the CD4 cell count slope in the adjusted analysis.

Current ART guidelines recommend changing therapy when virological failure occurs 14 15 16 . The WHO latest recommendations reinforce the role of virological monitoring in informing the switching decision 14 , a substantial change compared to the earlier versions. We used data from a multicentre cohort collaboration capturing routine data in various RLS on three continents, in which we only included the subset of patients with regular HIV-RNA and CD4 cell count monitoring. In these patients, we confirmed that the percentage of patients with short term virological success on an NNRTI-based first-line therapy was excellent 20 21 .

The optimal response to low or moderate-level viremia in patients on ART in RLS has been controversial. Analyses performed in countries with regular and frequent HIV-RNA monitoring and using PI-based ARV regimens tend to suggest that the use of an HIV-RNA threshold of 5,000-10,000 copies/mL to define failure in a regimen-adherent patient with no other reasons for an elevated HIV-RNA (e.g., drug-drug interactions, poor absorption, and concurrent illness) is not associated with clinical and immunological deterioration in cohort studies 17 18 22 . There are substantial differences between patients in high-income countries and patients in RLS that prevent us from generalising findings from Western cohorts. Firstly, patients starting ART in RLS have much lower CD4 cell counts at treatment initiation when compared with high-income countries 23 . Secondly, most first-line regimens include a combination of two nucleoside analogues (NRTI) and one NNRTI, unlike high-income countries in which half of new patients are put on a boosted protease inhibitor-based ART. Finally, one might infer that immunity is affected by numerous antigenic stimulations in countries endemic for malaria or tuberculosis 24 .

It is thus critical to accumulate data from countries where the majority of patients are now receiving ART, where the treatment switch is going to occur late and where first-line ARV combinations are different from the standard practice in high-income countries from where most clinical guidelines have been issued.

Because the WHO now recommends using targeted or sometimes routine use of HIV-RNA measurements in public-health oriented programs, much effort needs to go into making HIV-RNA determination available in RLS. However, the cut-off describing when to switch patients failing an NNRTI first-line regimen with few remaining therapeutic options has not been firmly established and depends on the objective: if the aim is to impact on HIV transmission or to prevent viral resistance 25 26 27 28 , then viral load measurement has to be widely implemented over all ART programs and the lowest threshold must be used to inform the switching decision. However, it is critical to determine whether there is a viral threshold below which the goal of maintaining a reasonable level of CD4 cells and a good clinical outcome is achievable without switching treatment.

In this large multicohort study, we observed that in patients experiencing virological failure, the impact on immune restoration was observed after the first year of ARV treatment; CD4 cell count slope one year after ART initiation was altered by any (even low-level) HIV-RNA replication, regardless of the threshold considered (5,000 or 10000 copies/mL). By contrast, all patients who had a suppressed viral load 12 months after ART commencement experienced a rise in CD4 cell counts (Table 2).

While our study lacked the statistical power necessary to categorise what threshold would best stratify the decision for changing ART regimens, our analysis suggests that any viral replication had a deleterious impact on immune recovery in patients remaining on a failing first-line regimen.

Interestingly, patients who had residual HIV-RNA replication in the first six months on treatment, but were then virologically suppressed at M12, experienced an immune restoration with a positive CD4 cell count slope from M6 up to M18. This could suggest that routine, compared to random, viral load determination may allow for therapeutic intervention leading to successful immune restoration. Hoffmann et al. nicely showed that patients on a first line regimen in South Africa who developed viremia were successfully re-suppressed after implementing intensive adherence support 29 .

The individual risk of switching treatment late after recognition of failure in first line treatment has been controversial in resource-limited settings. Seyler et al. showed in Côte d’Ivoire that patients with a detectable viral load had an increased risk of immunological failure, but there was no difference in morbidity at 18 months between the patients who switched during early regimen failure and those who remained on a failing regimen 26 . In contrast, other studies have shown that delays in switching to second-line regimens resulted in increased mortality when the failing regimen was a non-protease inhibitor-based regimen 30 . More recently, Keiser et al. demonstrated high mortality prior to and following a switch in treatment in South Africa: the cumulative mortality at 1 year was 4.2% (95% CI 2.2-7.8%) in patients who switched to a second-line regimen, 11.7% (7.3%-18.5%) in patients who remained on a failing first-line regimen, and 2.2% (1.6-3.0%) in patients on a non-failing first-line regimen (P < 0.0001). These results suggest that there are significant consequences to delayed switching 31 .

Our analysis also confirmed some interesting determinants of the CD4 cell count slope; subjects with more extreme ages (below 30 years old or above 40 years old) had a better immune recovery when compared with the middle-age group. Additionally, the calendar year of starting ART and the stage at ART initiation were good predictors of early immunological outcome. Considering an HIV-RNA threshold of 10,000 copies/mL, which provided in our multivariate model the best prognostic value for immunological response, elevated HIV-RNA levels were always associated with impaired immune reconstitution.

We recognise several limitations in our analysis. Patients have been included in our study according to the existence of frequent HIV-RNA monitoring, but only a minority of patients fulfilled our criteria of biological monitoring. Selection bias is therefore possible, although we verified that none of the patients in our study had entered specific therapeutic trial protocols but were rather prescribed viral load testing during specific periods when this laboratory exam was available in the clinic. In addition, we were unable to detect differences between treatment regimens, as patients in RLS are generally prescribed with standard NNRTI-based first-line regimens. However, treatment homogeneity is helpful in controlling confounding factors related to different drug regimens. Finally, we could not consider clinical endpoints, as we had a very limited number of deaths in the selected patient population, possibly due to the short follow-up. For this reason, the clinical significance of the differences in CD4 slope reported in our analysis cannot be evaluated – although many cohorts from resource limited settings clearly assessed the relationship between AIDS-related events, non AIDS related events and death and CD4 cells 32 .

Our data suggest that a failing regimen with documented HIV-RNA replication should be switched onto a second-line regimen, in line with recently issued WHO guidelines requesting a decrease in the HIV-RNA threshold to inform the switching decision (from 10,000 copies in the 2006 guidelines down to 5,000 in the 2009–2010 revision of the guidelines). We recognise that immune recovery is not the only factor to consider regarding HIV disease management in the short-term after ART initiation. Whenever possible, HIV-RNA levels should be fully suppressed, not only to prevent accumulation of resistance mutations but also to prevent HIV transmission 28 . It is therefore critical to promote access to low cost, context-adapted viral load tools.

We have no competing interests to declare in relation to this paper. The content of this publication is solely the responsibility of the authors and does not necessarily represent the official views of any of the institutions mentioned above.

AB, ES, RW, ED, EM, JM, KMEF, MS, NK, DB, PM, SVDB, and CZ contributed the necessary data from their health facilities. EB collected the data. AC, FB, FD contributed to the study design. EB and RT carried out the statistical analysis and EB, RT, AC, FB and FD interpreted the results. FD, AC and RT contributed to the study coordination. AC first drafted the manuscript. All authors had a critical revision of the manuscript for important intellectual content. All authors have read and approved the final manuscript.

The ART-LINC collaboration of the International epidemiological Databases to Evaluate AIDS (IeDEA) is funded by the US National Institutes of Health (Office of AIDS Research and National Institute of Allergy and Infectious Diseases) and the French Agence Nationale de Recherches sur le Sida et les hepatites virales (ANRS). Centre de Prise en Charge de Recherches et de Formation (CEPREF)/Agence Nationale de Recherches sur le Sida et les hepatitis virales (ANRS) 1203 COTRAME Cohort (Abidjan, Côte d’Ivoire); Senegalese Antiretroviral Access Initiative (ISAARV), ANRS 1290 (Dakar, Senegal); Academic Model for the Prevention and Treatment of HIV/AIDS (AMPATH), Moi University College of Health Sciences/University of Indiana (Eldoret, Kenya); Adherence Monitoring Uganda (AMU) cohort, Makerere-University of California in San Francisco (UCSF; Kampala, Uganda); Kamuzu Central Hospital/Lighthouse Trust (Lilongwe, Malawi); Connaught Clinic (Harare, Zimbabwe); Gugulethu ART Programme, (Cape Town, South Africa); Khayelitsha ART Programme, (Cape Town, South Africa); Operational Research on ART (OPERA), Perinatal HIV Research Unit (Soweto, South Africa); Morocco Antiretroviral Treatment 14 Cohort, Centre Hospitalier Universitaire (Casablanca, Morocco); MTCT-Plus Initiative, International Center for AIDS Care and Treatment Programs, Mailman School of Public Health, Columbia University, New York, USA; Prospective Evaluation in the Use and Monitoring of Antiretrovirals in Argentina (PUMA), Buenos Aires, Argentina; South Brazil HIV Cohort (SOBRHIV), Hospital de Clinicas (Porto Alegre, Brazil); Rio de Janeiro HIV Cohort, Hospital Universitario Clementino Fraga Filho (Rio de Janeiro, Brazil); Y R Gaitonde Centre for AIDS Research and Education (YRG) Care Cohort (Chennai, India); HIV-NAT, Thai Red Cross AIDS Research Centre (Bangkok, Thailand).

The ART-LINC of IeDEA Collaboration was funded by the United States National Institute of Health (NIH - Office of AIDS Research) together with the French Agence Nationale de Recherches sur le Sida et les hépatites virales (ANRS – grants 12101 and 12138). The IeDEA network is now funded through collaborative agreements by the National Cancer Institute (NCI), the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) and the National Institute of Allergy And Infectious Diseases (NIAID).

Eric Balestre, Martin Brinkhof, François Dabis (principal investigator), Matthias Egger (principal investigator), Claire Graber, Beatrice Fatzer, Olivia Keiser, Charlotte Lewden, Mar Pujades, Mauro Schechter (principal investigator).