Secondary Logo

Journal Logo

Clinical Science

Brief Report: Assessing the Association Between Changing NRTIs When Initiating Second-Line ART and Treatment Outcomes

Rohr, Julia K. PhD*; Ive, Prudence MD; Horsburgh, Charles Robert MD‡,§; Berhanu, Rebecca MD; Hoffmann, Christopher J. MD¶,#; Wood, Robin DSc**; Boulle, Andrew MBChB††; Giddy, Janet MBChB‡‡; Prozesky, Hans MD§§; Vinikoor, Michael MD‖‖,¶¶,##; Mwanza, Mwanza wa MBChB¶¶; Wandeler, Gilles MD***,†††; Davies, Mary-Ann MBChB, PhD††; Fox, Matthew P. DSc*,§,‡‡‡

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: April 1, 2018 - Volume 77 - Issue 4 - p 413-416
doi: 10.1097/QAI.0000000000001611



Although drug resistance is rare at antiretroviral therapy (ART) initiation, most patients who fail first-line ART have some resistance to nucleoside reverse transcriptase inhibitors (NRTIs) that are also used in second-line regimens.1–6 To optimize the efficacy of NRTIs in second line, national treatment guidelines in South Africa and Zambia recommend patients change at least 1 NRTI when switching to second line.7–11 In practice, changing NRTIs does not always occur, usually due to contraindications to specific NRTIs or lack of availability.

The importance of the choice of NRTIs used in second-line is questioned because of the high potency of standard protease inhibitors (PIs) that are a part of second-line regimens (lopinavir/ritonavir). Second-line ART can successfully suppress HIV in the presence of NRTI drug resistance mutations12–17; yet, the number of NRTIs switched from first-to second-line ART is associated with improved outcomes on second line.18 We investigated the impact of switching NRTIs on virologic and immunologic outcomes as patients in South Africa and Zambia reach multiple years on second line.


We used data from the International Epidemiologic Databases to Evaluate AIDS Southern Africa (IeDEA-SA) cohort, a National Institutes of Health–funded initiative pooling data to address HIV treatment research questions. IeDEA-SA data included medical records from 2004 to 2013 from 175,933 patients in Zambia and 79,908 patients in South Africa, with information on basic patient demographics, height, weight, date of visits, diagnoses, ART drugs, and laboratory values, including CD4 and, in South African sites, HIV viral load.

The study population included adult patients aged 18 years and older, initiated on standard first-line ART [2 NRTIs plus a non-NRTI], and had evidence of first-line failure: either a viral load >1000 copies/mL among South African patients after at least 6 months on treatment, or 2 consecutive CD4 counts <100 or a 30% drop from highest CD4 count among Zambian patients. The first NRTI used in first line and second line was categorized into zidovudine (AZT), stavudine (d4T), tenofovir (TDF), abacavir (ABC), or other. We evaluated switch to each type of NRTI for patients: (1) initiated on AZT, (2) initiated on d4T, and (3) initiated on TDF.

Treatment failure on second line was our primary outcome and was modeled with crude and propensity score-adjusted Cox proportional-hazards regression where propensity score was included in the models as a covariate. Our primary analysis used data from South Africa, where viral load monitoring was available, and treatment failure was defined as 2 consecutive viral loads >1000 copies/mL. We also evaluated immunologic failure on second-line ART in Zambia, where treatment failure was persistent CD4 levels below 100 cells/mm3, defined as 2 consecutive CD4 counts <100 cells/mm3.19,20 Potential confounders included year of starting second line, age, sex, duration on first line, and first-line non-NRTI, as well as CD4 count, viral load, hemoglobin, and creatinine clearance at first-line initiation and at switch to second line. Multiple imputation was used for missing covariates unless they were missing for >50% of patients.21,22 Propensity scores for switching to each type of NRTI vs. remaining on the same NRTI were calculated for each model with logistic regression, which allowed us to control for as many confounders as possible while maximizing statistical efficiency.23

For South African models, propensity scores were adjusted for year of switch to second line, sex, age, baseline alanine aminotransferase, hemoglobin, CD4, creatinine clearance, weight, and viral load and CD4 count at time of switch to second line. Models for changing from TDF were not adjusted for year because it was only available in more recent years. For Zambian models, propensity scores were adjusted for year of switch to second line, sex, age, baseline alanine aminotransferase, CD4, hemoglobin, weight, creatinine clearance, and second-line values of CD4, creatinine clearance, and hemoglobin. Models for changing from d4T were not adjusted for year because the number initiated on d4T who remained on d4T in second line was too small and propensity score models did not converge when year was adjusted. In sensitivity analyses, we adjusted for loss to follow-up through inverse probability of treatment weighting.24


Study Sample

In South Africa, 4614 patients had evidence of first-line virologic failure and switched to a PI-based regimen and in Zambia, 2061 patients experienced first-line immunologic failure and switched to a PI-based regimen. Patients without visits after date of switch to second line (2.5%) and patients who did not have any NRTI identified in the regimen at time of second-line initiation (3.2%) were excluded. The total sample included 6290 patients (4275 in South Africa and 2015 in Zambia).

Among the 6290 patients in analysis, the majority were female (61%) and the median age was 34 years [interquartile range (IQR): 29–40]. A description of the patients is shown by country and by change in NRTI in second line in Supplemental Digital Content Table 1, In both countries, CD4 counts at ART initiation were low (median <100 cells/mm3), and CD4 count at switch to second line was slightly lower for patients who had a change in second-line NRTI in South Africa, but not in Zambia (Supplemental Digital Content Table 1, Overall, 90% of patients changed NRTIs at second line (Table 1). In South Africa, the proportion who changed NRTI in second line was mostly constant over time (92.5% in 2004–2006, 96.3% in 2007–2008, 94.5% in 2009–2010, and 95.5% in 2011–2013). In Zambia, the proportion changing NRTIs was lower than in South Africa, and changing NRTI was more common in earlier periods (85.1% in 2004–2006, 91.8% in 2007–2008, 80.3% in 2009–2010, and 72.4% in 2011–2013).

NRTI in Second-Line Regimen Stratified by First-Line NRTI and Country

Follow-up time on second-line was 18 months (IQR: 14–28) for South African patients and 23 months (IQR: 15–33) for Zambian patients. Second-line virologic failure occurred among 15% of patients in South Africa, and second-line immunologic failure occurred among 7% of patients in Zambia. Death was recorded for 4% in South Africa and 2% in Zambia. On average, in South Africa, patients who did not switch NRTIs received slightly more viral load monitoring measurements on second line (1.6 per year) compared with patients who did switch NRTIs (1.4 per year). Yet, in recent years, viral load monitoring was performed less frequently, with a mean of >1 measures per year before 2011, and a mean of 1.0 measures per year in 2011 and later.

Second-Line Outcomes

Propensity score-adjusted Cox proportional-hazards models for virologic failure in South Africa are shown in Table 2 (summary of propensity scores displayed in Supplemental Digital Content Table 2, Among patients initiated on AZT, we observed an association between switching to TDF and reduced second-line failure compared with staying on AZT {adjusted hazard ratio [aHR] 0.25 [95% confidence interval (CI): 0.11 to 0.57]}. Switching from AZT to ABC was not associated with reduced failure. Among patients initiated on first-line d4T, there was weak evidence for reduced hazards of failure on second line associated with switching to TDF vs. remaining on d4T [aHR = 0.70 (95% CI: 0.42 to 1.16)]. Switching from d4T to AZT did not have an association with reduced hazards of second-line failure. For patients initiating TDF in first line, follow-up time was more limited because the drug was introduced into South Africa's national program later than d4T and AZT. Changing to AZT in second line vs. remaining on TDF was associated with decreased second-line failure (aHR: 0.35; 95% CI: 0.13 to 0.96). Although loss to follow-up was common (34% of patients), weighting models using inverse probability weights to account for loss to follow-up did not impact the hazard ratio point estimates.

Hazard Ratios and 95% Confidence Intervals of Second-Line Virologic Failure Among South African Patients, Using Cox Proportional-Hazards Models Adjusted for Propensity Scores

Using patient data from Zambia to evaluate second-line immunologic failure showed similar trends, but all hazard ratio estimates had wide confidence intervals (Supplemental Digital Content Table 3,


Among patients in South Africa who failed first-line ART and switched to second-line ART, change in NRTI was associated with reduced virologic failure on second line for changes to TDF and for change from TDF to AZT. Changes to other NRTIs had no association with second-line failure. In Zambia, where virologic monitoring was not available and treatment failure on second line was more poorly defined and likely underestimated,25–28 we did not see strong evidence of a benefit from changing NRTIs in second line.

It is possible that drug resistance to NRTIs had an effect on second-line failure; yet, previous research has shown that PI-based second-line ART can successfully suppress HIV in the presence of NRTI drug resistance mutations and that poor adherence is a more likely cause of second-line failure.12–17,29 Given the importance of adherence, it is possible that patients on TDF in second line often had better outcomes because the drug was better tolerated. Interestingly, remaining on TDF in second line after initiating TDF in first line was common in Zambia, perhaps because of physician preference for this drug. This trend may explain why there was less switching of NRTIs in Zambia in recent years as TDF became available. We did not see any strong evidence for reduction in second-line immunologic failure associated with switching away from TDF in Zambia. Because most patients now initiate TDF-based regimens as first-line ART, further exploration into switching to AZT, which was associated with better second-line outcomes, compared with remaining on TDF in South Africa, is warranted.

AZT is no longer the preferred NRTI for first line in Zambia or South Africa; however, it may still be used for patients with contraindications to TDF (eg, renal failure). First-line AZT in South Africa would typically only have been prescribed over d4T (under 2004 guidelines) when patients had preexisting peripheral neuropathy or were at increased risk of hyperlactatemia. Since TDF became available in South Africa in 2010, patients normally are only initiated on AZT if they have renal failure. In Zambia, AZT was a more common option in first line, and was part of national guidelines before 2010, but has now also been replaced by TDF. Our results showed evidence that switching from AZT to TDF was associated with reduced second-line failure but the impact of switching from AZT to ABC was not clear. For patients initiated on AZT-based regimens because of renal failure, switching to TDF in second line may not be an option if contraindications remain, and more detailed research into treatment choices for this population is important.

One of the main obstacles in this study is the potential for confounding by indication, which is common in observational studies of drug prescriptions. Although propensity scores were used to make groups as comparable as possible with respect to their clinical profile, because of limited data available in medical records, there is likely residual confounding. Patients who stay on the same NRTI because of complications that prevent them from taking certain second-line drugs may have worse outcomes due to these contraindications. Alternatively, patients who switch NRTIs may include more patients who truly failed first-line ART rather than switching regimens for other reasons, and who may have worse outcomes on second-line ART. In addition, lack of virologic monitoring in Zambia made it difficult to accurately identify second-line treatment failure and draw more conclusive results from these data. Another potential problem is differential surveillance of patients who remain on the same NRTIs compared with those who switch. We did not see large differences in monitoring between these groups, but monitoring frequency did change over time, along with use of NRTIs in second-line regimens, with modifications to national treatment guidelines; so, we controlled for calendar time where possible. Finally, although we had a large initial sample size, stratification by NRTI used limited the numbers in the models, and with a relatively short follow-up time on second line, led to some imprecise results.

Our results support that the NRTI in second line plays a role in second-line outcomes and provide limited evidence in support of current guidelines to change NRTI in second line, although the impact of NRTI on second-line activity may act through drug resistance, drug side effects, or better tolerance of drugs associated with improved adherence. This study supports the need for more research regarding NRTI choices for patients with renal failure who fail AZT first-line regimens, ideally with drug resistance data, and more follow-up of patients initiated on TDF who must switch to second line. Observational patient cohorts in South Africa and Zambia are challenging settings for answering these complex questions comparing prescription of different drugs, and more information from clinical trials is necessary.


1. Wallis CL, Mellors JW, Venter WDF, et al. Varied patterns of HIV-1 drug resistance on failing first-line antiretroviral therapy in South Africa. J Acquir Immune Defic Syndr. 2010;53:480–484.
2. Mansana J, Lessells R, Skingsley A, et al. High-levels of acquired drug resistance in adult patients failing first-line antiretroviral therapy in a rural HIV treatment programme in KwaZulu-Natal, South Africa. PLoS One. 2013;8:e72152.
3. Marconi VC, Sunpath H, Lu Z, et al. Prevalence of HIV-1 drug resistance after failure of a first highly active antiretroviral therapy regimen in KwaZulu Natal, South Africa. Clin Infect Dis. 2008;46:1589–1597.
4. Van Zyl GU, van der Merwe L, Claassen M, et al. Antiretroviral resistance patterns and factors associated with resistance in adult patients failing NNRTI-based regimens in the Western Cape, South Africa. J Med Virol. 2011;83:1764–1769.
5. Wallis C, Aga E, Ribaudo H, et al. Drug susceptibility and resistance mutations after first-line failure in resource limited settings. Clin Infect Dis. 2014;59:706–715.
6. Hamers RL, Sigaloff KCE, Wensing AM, et al. Patterns of HIV-1 drug resistance after first-line antiretroviral therapy (ART) failure in 6 sub-Saharn African countries: implications for second line ART strategies. Clin Infect Dis. 2012;54:1660–1669.
7. Republic of South Africa Department of Health. National Antiretroviral Treatment Guidelines. Pretoria, South Africa: National Department of Health; 2004.
8. Republic of South Africa Department of Health. Clinical Guidelines for the Management of HIV & AIDS in Adults and Adolescents. Pretoria, South Africa: National Department of Health; 2010.
9. Zambian National AIDS Council. National Guidelines for Management and Care of Patients With HIV/AIDS. Lusaka, Zambia: Printech Press; 2004.
10. Government of the Republic of Zambia, Ministry of Health. Antiretroviral therapy for Chronic HIV infection in adults and adolescents: New ART protocols. Ministry of Health Zambia, 2007.
11. Government of the Republic of Zambia, Ministry of Health. Adult and adolescent antiretroviral therapy protocols. Ministry of Health Zambia, 2010.
12. Dlamini J, Hu Z, Ledwaba J, et al. Genotypic resistance at viral rebound among patients who received lopinavir/ritonavir- or efavirenz-based first antiretroviral therapy in South Africa. J Acquir Immune Defic Syndr. 2011;58:304–308.
13. Ajose O, Mookerjee S, Mills EJ, et al. Treatment outcomes of patients on second-line antiretroviral therapy in resource-limited settings: a systematic review and meta analysis. AIDS. 2012;26:929–938.
14. Sigaloff KCE, Hamers RL, Wallis CL, et al. Second-line antiretroviral treatment successfully re-suppresses drug-resistant HIV-1 after first-line failure: prospective cohort in sub-Saharan Africa. J Infect Dis. 2012;205:1739–1744.
15. Scherrer A, Boni J, Yerly S, et al. Long-lasting protection of activity of nucleoside reverse transcriptase inhibitors and protease inhibitors (PIs) by boosted PI containing regimens. PLoS One. 2012;7:e50307.
16. Levison JH, Orrell C, Gallien S, et al. Virologic failure of protease inhibitor-based second-line antiretroviral therapy without resistance in a large HIV treatment program in South Africa. PLoS One. 2012;7:e32144.
17. Johnston V, Cohen K, Wiesner L, et al. Viral suppression following switch to second-line antiretroviral therapy: associations with nucleoside reverse transcriptase inhibitor resistance and subtherapeutic drug concentrations prior to switch. J Infect Dis. 2014;209:711–720.
18. Pujades-Rodríguez M, Balkan S, Arnould L, et al. Treatment failure and mortality factors in patients receiving second-line HIV therapy in resource-limited countries. JAMA. 2010;304:303–312.
19. World Health Organization. Antiretroviral therapy for HIV infection in adults and adolescents: recommendations for a public health approach, 2006. Available at: Accessed June 1, 2017.
20. Meya D, Spacek LA, Tibenderana H, et al. Development and evaluation of a clinical algorithm to monitor patients on antiretrovirals in resource-limited settings using adherence, clinical and CD4 cell count criteria. J Int AIDS Soc. 2009;12:3.
21. Von Hippel P. Regression with missing Ys: an improved strategy for analyzing multiply imputed data. Sociological Methodol. 2007;37:83–117.
22. Allison PD. Multiple imputation for missing data: a cautionary tale. Sociological Methods Res. 2000; 28:301–309.
23. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46:399–424.
24. Cole SR, Hernán MA. Constructing inverse probability weights for marginal structural models. Am J Epidemiol. 2008;168:656–664.
25. Moore DM, Awor A, Downing R, et al. CD4+ T-cell count monitoring does not accurately identify HIV infected adults with virologic failure receiving antiretroviral therapy. J Acquir Immune Defic Syndr. 2008;49:477–484.
26. Kanapathipillai R, McGuire M, Mogha R, et al. Benefit of viral load testing for confirmation of immunological failure in HIV patients treated in rural Malawi. Trop Med Int Health. 2011;16:1495–1500.
27. Sigaloff KCE, Hamers RL, Wallis CL. Unnecessary antiretroviral treatment switches and accumulation of HIV resistance mutations; two arguments for viral load monitoring in Africa. J Acquir Immune Defic Syndr. 2011;58:23–31.
28. Keiser O, Tweya H, Boulle A, et al. Switching to second-line antiretroviral therapy in resource-limited settings: comparison of programmes with and without viral load monitoring. AIDS. 2009;23:1867–1874.
29. Wallis CL, Mellors JW, Venter WDF, et al. Protease inhibitor resistance is uncommon in HIV-1 subtype subtype C infected patients on failing second-line lopinavir/r—containing antiretroviral therapy in South Africa. AIDS Res Treat. 2011;2011:769627.

antiretroviral therapy; South Africa; Zambia; second-line; nucleoside reverse transcriptase inhibitor; NRTI

Supplemental Digital Content

Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.