Even though several countries in southern Africa have introduced TDF in first-line ART, most patients who will be failing their first-line regimen in the coming years will not have been exposed to TDF. As a consequence, these patients might benefit from this drug in their second-line regimen. The comparative effectiveness of second-line regimens including or excluding TDF in southern Africa is therefore of great interest. We compared clinical outcomes between patients receiving TDF-containing second-line ART and patients treated with other second-line regimens in six ART programmes in Zambia and RSA. Overall, mortality and the rate of treatment failure were low in this population, underlining the benefit of PI-based second-line ART in patients failing first-line treatment in the region20
. In Zambia, LTFU was similar in patients on second-line ART containing and not containing TDF, but mortality and immunological failure were lower in patients on TDF. In the five South African cohorts with access to routine viral load monitoring the rate of virological failure was also lower in the TDF group.
In contrast to Zambia the use of TDF was not associated with reduced mortality in South Africa. This finding could be the result of differences in the capacity of the health system in South Africa compared to Zambia or reflect differences in ascertainment of deaths and tracing of patients LTFU. Confounding by indication could be another explanation: the relatively few patients who were prescribed TDF in South Africa before 2010 might have been a selected group of sicker patients. Finally, the difference between the two countries could reflect the play of chance: the confidence intervals overlapped widely and a formal test of heterogeneity gave a p value of 0.13. We can thus not exclude a similar reduction in mortality in RSA.
In both groups mortality after one year of second-line ART was somewhat lower than the 5.4% mortality observed after a median of 15.1 months in the Médecins sans Frontiéres (MSF) multi-cohort study of patients on second-line ART 3
. In contrast, in a study of patients virologically failing first-line ART in Malawi, 10% of patients on second-line ART died during the first six months 6
. The higher mortality in the latter study might be explained by the presence of virological failure in all patients and the very low median CD4 count at the start of second-line ART. Furthermore, patients who are treated in settings without access to routine viral load monitoring are at risk of remaining on failing first-line regimens for long periods before switching to second-line ART2,21
, and of accumulating drug resistance mutations which might limit the efficacy of some second line regimens11
Studies from different regions in sub-Saharan Africa showed a high prevalence of TDF-related resistance mutations in patients failing first-line ART 10,11,22
. This raised concerns on the efficacy of TDF in second-line regimens for populations infected with subtype C HIV-1 variants. In high-income countries, the K65R mutation is present only in 2–5% of HIV-1 subtype B infected patients failing first-line ART 23
. In contrast, over 20% of patients failing first-line ART in an urban public-sector ART clinic in Malawi had developed this resistance mutation, without prior exposure to TDF 11
. Interestingly, in the Malawian study, and the PharmAccess African Studies to Evaluate Resistance (PASER), clinical outcomes one year after initiation of second-line ART were not affected by resistance to TDF 6,24
. Another report from PASER nevertheless argued that in light of the high prevalence of the K65R mutation in patients failing a D4T-containing regimen, AZT might be a better option for second-line ART than TDF12
. Prolonged treatment with a failing D4T-containing first-line regimen might explain the high levels of TDF resistance mutations in the region 23
. We found little evidence for an association of the risk of second-line treatment failure with the presence of D4T in the first-line regimen, however, the power of our study to detect smaller effects was limited.
There are several possible explanations for the superior effect of TDF in second-line ART in Zambia and South Africa. More than 80% of the patients not on TDF were treated with either DDI/AZT or DDI/ABC as the NRTI backbone. Due to its better tolerability and once-daily dosing, treatment adherence might be higher in patients receiving TDF compared to other NRTI combinations, especially those including DDI: the higher toxicity of DDI-based regimens might have led to poorer adherence. Wallis et al. and Van Zyl et al. reported a low prevalence of PI mutations in patients failing second-line ART in the Republic of South Africa 25,26
, indicating that failure was due to insufficient drug levels following non-adherence, rather than resistance. Finally, the high potency of LPV/r monotherapy in patients without prior PI exposure could have masked larger differences in treatment outcomes between the two groups 27,28
. Patients on TDF-containing second-line regimens might have had favourable outcomes despite potential NRTI mutations, including thymidine-analogue mutations (TAM’s) and K65R. Of note, the difference in treatment failure between the two second-line regimens emerged already after one year in RSA, whereas it was only apparent later during follow-up in Zambia. This could be explained by the earlier diagnosis of treatment failure with virological monitoring in RSA compared to CD4 monitoring in Zambia.
To our knowledge this is the first study to compare second-line regimens in southern Africa. In particular, there are no randomized trials of second-line ART tailored to regions where non-B HIV subtypes dominate. The main limitation of multi-cohort data comparing treatments lies in the lack of randomization and the heterogeneity between the different treatment sites. Confounding by indication and differences between settings in background mortality, monitoring and treatment strategies, and health systems may have biased our results. Of note, the proportion of patients on a TDF-containing second-line regimen varied widely across countries and calendar time, reflecting national treatment guidelines. However, the association of TDF with reduced rates of treatment failure was consistent within countries and cohorts, which adds strength to our findings. Furthermore, over 99% of the patients received the same PI, thus effectively removing one potential source of confounding. We had no data on treatment adherence, which is known to influence ART outcomes29–31
. Young age and male gender were risk factors for second-line failure, probably as a consequence of the lower adherence to ART in younger patients and men 32,33
. Finally, as no genotypic resistance data is routinely collected in southern Africa, we could not assess the relationship between treatment failure, resistance patterns and clinical outcomes.
We did not evaluate toxicity and side-effects related to the different regimen. Most patients on non-TDF second-line ART had DDI in their backbone. The toxicities of DDI, including lipodystrophy, gastrointestinal intolerance, peripheral neuropathy and pancreatitis, will have influenced our results.34,35
TDF is associated with nephrotoxicity, including an increased risk of loss of kidney function, acute renal failure and tubulopathy36,37
. A recent study from South Africa showed that pre-existing renal disease was frequently exacerbated by the use of tenofovir38
. Furthermore, patients on PI-based regimens may be at increased risk of renal failure39
. Screening for renal dysfunction before the initiation of TDF-containing regimen and close monitoring during treatment is part of treatment guidelines and should be performed routinely.
In conclusion, we found that patients on TDF-containing second-line ART were less likely to develop treatment failure in all cohorts and less likely to die in Zambia than patients on other regimens. Despite the increased prevalence of TDF-related resistance mutations in patients failing first-line ART in southern Africa, TDF seems to be an effective component of second-line ART for many patients who have not been exposed to this drug previously. This finding is of considerable importance, as an increasing number of TDF-unexposed patients failing their first-line treatment will be switched to TDF-containing second-line regimen in the coming years. Second-line ART is becoming more available in sub-Saharan Africa, but most ART programmes in the region do not have access to individual genotypic resistance data. Thus, randomized trials comparing the efficacy and toxicities of different second-line regimens are urgently needed to inform clinical practice and guidelines.