Mozambique, a sub-Saharan country with HIV prevalence of 10.6% provides antiretroviral therapy based on a public health approach [
3,
11,
12]. Treatment options for HIV infected people in low-middle income countries (LMIC) are based on WHO guidelines [
3]. First-line regimen includes two NRTI (tenofovir or zidovudine and lamivudine or emtricitabine) plus one NNRTI (nevirapine or efavirenz) whereas a boosted protease inhibitors-based ART is used as second line regimen with a substitution of the NRTIs: tenofovir is given after zidovudine failure, while zidovudine is recommended in the second-line after tenofovir has failed in the first-line regimen [
3].This treatment sequencing strategy is based on the rationale that after virological failure with 2 NRTI + a NNRTI, the activity of the protease inhibitors is preserved and the cross-resistance of the alternate NRTIs is limited [
3,
6,
13]. However, after tenofovir failure, HIV-1 usually selects for RAMs such as K65R that do not affect zidovudine activity; on the contrary zidovudine selects for TAMs that accumulate and progressively confer increasing cross-resistance to tenofovir. This is one of the reasons why WHO now recommends to prefer tenofovir as first-line regimen, but many countries still use zidovudine due to cost and procurement issues. Viral load monitoring to detect treatment failures, is now recommended by WHO and, although not yet available in many areas, is becoming increasingly accessible [
3]. Consideration of the timing of treatment switch after 1st-line ART failure is particularly important in resource-limited settings where salvage regimens are scarce and costly [
9]. For these reasons in LMIC, the correct switching time should also be informed by the probability of accumulating resistance to the subsequent treatment lines at a given time of virological failure [
7,
9,
13]. In this study, we report how RAM accumulate after virological failure of a thymidine analogue-based first-line regimen in a resource-limited setting. In particular, we show that cross-resistance to tenofovir was still limited when failure was detected at 1 year after ART initiation. After 2 years and, in projection, after 3 years, cross-resistance to tenofovir accumulated significantly. This lead to a significant accumulation of resistance to drugs that, based on WHO guidelines, would have been used for the second-line regimen: a predicted low-level or higher resistance to at least 1 drug of the second-line regimen rose from 75% of cases at t1 to 89% at t2, while 10% at t1 and 23% at t2 showed a predicted resistance to 2 second-line drugs. Previous studies have anayzed the accumulation of drug resistance mutations in patients failing first-line regimens in Sub-Saharan Africa. In a retrospective study in South Africa, in 43 patients performing sequential resistance tests with a median interval of 5 months, RAMs accumulated at a mean of 0.07/month of drug exposure [
14]. In a prospective cohort of Zambian children on first-line ART, 6 had sequential genotypes while failing on stavudine/lamivudine/nevirapine and showed an accumulation rate of 0.59 TAMs/year [
15]. In a retrospective analysis of a randomized study performed in African countries on 36 genotype pairs from weeks 48 and 96 of first-line ART, the mean TAMs accumulation rate was 1.50/year in nevirapine-treated participants and 1.82/year in abacavir-treated participants [
16]. In a retrospective analysis of adults and children failing NNRTI-based first-line ART, NNRTI resistance mutations accumulated at 0.62/year and NRTI resistance mutations at 0.84/year [
17]. In our study we observed a mean TAM accumulation rate of 0.32/year; which rose to 0.49/year in patients with a pharmacy refill adherence >90%. The yearly accumulation rate of NNRTI resistance mutations was 0.15, rising at 0.17 in adherent patients. An hypothesis for the reason for the lower rate of resistance accumulation in this cohort as compared to previous reports may be the availability of viral load monitoring. Indeed, patients included here were selected among those not switching to second-line for one year despite documented virologic failure. The population included showed relatively low viral loads at failure, and this might have been a reason for keeping them on first-line, prompting adherence interventions before switching to second-line. This probably selected a population at lower risk of resistance accumulation, as reported by other studies relating drug resistance accumulation to viral load [
4,
18]. In agreement with this, patients with an HIV-1 RNA >10,000 copies/mL in this study showed higher yearly accumulation rates for TAMs and NNRTI resistance mutations (0.50 and 0.40, respectively), values that are closer to those provided by previous reports. Our findings may represent a practical indication for the management of patients with virological failure in this settings. In particular, after initial detection of virological failure patients may still benefit from adherence counselling strategies without major risk of accumulating significant cross-resistance to second-line drugs. However, the risk of resistance accumulation is higher in patients with an HIV-1 RNA above 10,000 copies/ml, and if virological failure persists subsequently, despite adherence implementation, a switch to second-line ART should be recommended, as RAM will accumulate in the majority of patients. Our findings should be interpreted with caution given the limited sample size and the retrospective design of the study and require validation in a prospective study. Moreover, our sequencing technique did not cover the mutation N348I in the connection domain, which is frequently selected by nevirapine and thymidine analogues in subtype C, and slightly reduces susceptibility to NNRTIs and zidovudine, so our analyses may have slightly underestimated resistance to these agents [
19]. However, the accurate patients selection and follow up and the contemporary longitudinal assessment of viral load, resistance and adherence at specified times on ART represent strong points of this cohort study. It is important to consider that studies in Sub-Saharan Africa have shown that viral load could be successfully re-suppressed with boosted PI-based second-line regimens despite the presence of NRTI resistance after first-line failure. One observational cohort study showed successful virologic suppression with second-line drugs, despite 53% were predicted to receive partially active regimens due to drug resistance [
20]; moreoever, in a randomized clinical trial [
21] response to a second-line regimen based on boosted PI +2–3 NRTIs was better as compared to a boosted PI + raltegravir or boosted PI monotherapy despite NRTIs had no predicted activity due to resistance mutations; finally, in another randomized trial, the major determinant of response to boosted PI +2–3 NRTIs as second-line regimen was adherence but not baseline resistance [
22]. Therefore, our findings on NRTI RAM accumulation on first-line failing patients may have limited implications for clinical practice given the residual activity of a regimen based on boosted PI + NRTIs. It remains to be established whether the introduction of dolutegravir in the recommended first-line or second-line regimens will change this scenario.