High mortality in people taking antiretroviral therapy with delayed switching following virologic failure

Delayed switch of antiretroviral therapy after virologic failure associated with elevated mortality among HIV-infected adults in Africa.

Petersen ML, Tran L, Geng EH, Reynolds SJ, Kambugu A, Wood R, Bangsberg DR, Yiannoutsos CT, Deeks SG, Martin JN. AIDS. 2014 Jun 28. [Epub ahead of print].

Objective: Routine monitoring of plasma HIV RNA among HIV-infected patients on antiretroviral therapy (ART) is unavailable in many resource-limited settings. Alternative monitoring approaches correlate poorly with virologic failure and can substantially delay switch to second-line therapy. We evaluated the impact of delayed switch on mortality among patients with virologic failure in Africa.

Design: A cohort.

Methods: We examined patients with confirmed virologic failure on first-line nonnucleoside reverse transcriptase inhibitor (NNRTI)-based regimens from four cohorts with serial HIV RNA monitoring in Uganda and South Africa. Marginal structural models aimed to estimate the effect of delayed switch on mortality in a hypothetical trial in which switch time was randomly assigned. Inverse probability weights adjusted for measured confounders including time-updated CD4 cell count and HIV RNA.

Results: Among 823 patients with confirmed virologic failure, the cumulative incidence of switch 180 days after failure was 30% [95% confidence interval (95% CI 27-33]. The majority of patients (74%) had not failed immunologically (as defined by WHO criteria) by the time of virologic failure. Adjusted mortality was higher for individuals who remained on first-line therapy than for those who had switched [odds ratio (OR) 2.1, 95% CI 1.1-4.2]. Among those without immunologic failure, the relative harm of failure to switch was similar (OR 2.4; 95% CI 0.99-5.8) to that of the entire cohort, although of borderline statistical significance.

Conclusion: Among HIV-infected patients with confirmed virologic failure on first-line ART, remaining on first-line therapy led to an increase in mortality relative to switching. Our results suggest that detection and response to confirmed virologic failure could decrease mortality.

Abstract access 

Editor’s notes: The World Health Organization recommends scaling up access to routine viral load monitoring. This will enable healthcare workers to detect non-adherence and virologic failure earlier and to intervene to re-establish virologic control. This could be either on first-line antiretroviral therapy (ART) or by switching to second-line ART. If successful, this should limit the duration of viraemia, thereby limiting accumulation of resistance mutations and conserving future treatment options. The effect of viral load monitoring on mortality is less certain. To date, no randomised controlled trial has demonstrated a survival benefit of viral load monitoring over and above CD4 count monitoring. This may be due to study design - short follow-up time and intensive adherence support meant that few people experienced virologic failure. Previous observational studies described lower mortality in a South African cohort monitored using viral load monitoring as compared to a Malawian and Zambian cohort using CD4 count monitoring. However, subsequent mathematical modelling indicated that viral load monitoring only accounted for a small proportion of this difference. The rest was due to differences in resources, infrastructure and other unmeasured confounders. 

This study explored the impact on mortality of delayed switching following confirmed virologic failure. Among 7 975 people initiated on ART between 2002-2011, some 823 experienced confirmed virologic failure. As described by others, even in the context of routine viral load monitoring and access to second-line ART, marked delays in switching occur. The cumulative incidence of switching to second-line ART within six months of confirmed failure in this study was only 30%. Some of these ‘delays’ may have been due to the study definition of virologic failure. A lower viral load threshold was used in this study than was likely to have been used in the clinical guidelines.  Delays could also have been due to healthcare system factors such as delayed turn-around-time for results or difficulties in recalling people. However, healthcare worker factors, such as delayed switching to address adherence barriers or to avoid drug interactions, particularly with treatment for tuberculosis, are also likely to also have played a part. Regardless of the reasons for delays, after adjusting for measured confounders, mortality was higher for people who experienced delayed switching. Longer delay was associated with higher probability of death.

This study indicates that viral load monitoring alone may not be sufficient to reduce mortality. A greater understanding of the reasons for delays, together with innovative ways to ensure virologic failure is detected and managed in an effective, timely manner, is needed. 

HIV Treatment
South Africa, Uganda
  • share