Virologic rebound occurred in only one in six patients during the first 18 months of antiretroviral treatment at five major public hospitals in South Africa, according to an analysis of close to 20,000 patients published in the advance online edition of the Journal of Acquired Immune Deficiency Syndromes. However, the study found big variations between clinics in the speed at which patients with virologic failure were switched to second-line treatment, despite clear national guidelines.
The median delay between confirmation of failure and switch to second-line treatment was five months, yet time to switching varied by up to two-fold between cohorts. Lower CD4 cell counts at failure and rapidly declining CD4 cell counts were predictive of switching. The authors found that close to three-quarters of those with confirmation of treatment failure and at least six months of follow-up switched.
During the study period, clinics were following the 2004 South African national guidelines, which recommended a switch to second-line treatment if patients with a detectable viral load measurement above 400 copies/ml had a viral load above 5000 copies/ml three months later, despite intensive adherence support. This guidance is consistent with WHO guidance on the use of viral load for determining when to switch to second-line treatment in resource-limited settings.
In 2010, South African guidance was updated to recommend a switch if viral load was above 1000 copies/ml three months after the first detectable measurement.
The authors chose to describe, for the first time, how WHO guidance works at the national level and what the impact of different thresholds for determining virological failure is at the programme level.
In this observational cohort study, the authors included adult patients who began public-sector antiretroviral therapy (ART) between January 2000 and July 2008 at five sites in South Africa and who had completed at least six months of follow-up. Guidelines at the time, now revised, included all patients with CD4 cell counts under 250 cells/mm3 or with WHO clinical stage 4 as being eligible for ART. Once on ART, guidelines supported six-monthly viral load testing and CD4 cell count monitoring.
In total, 19,645 people were followed for 29,935 person years, with a median of 1.3 years (1.1-1.4) in the study and 1.8 years on ART and with a median CD4 cell count at the start of ART of 96 (IQR: 40-159). Eighty-eight per cent achieved viral suppression on first-line therapy.
9.9% experienced treatment failure at what was determined to be a common failure definition (viral load at or above 1000 copies/ml) at a median time from starting ART of 16 months (IQR: 12-23 months); of which 62% (833/1348) switched to second-line ART.
Using a confirmatory threshold of 400 copies/ml, 16.9% (95% CI: 15.4-18.6%) of the sub-group treated for five years had experienced treatment failure by five years whereas using a threshold of 10,000 copies/ml only 7.8% (95% CI: 6.6-9.3%) did according to survival analysis.
Advanced disease at the start of ART increased the risks of failure by 60% and two or more treatment interruptions were associated with a seven-fold increased risk of failure (aHR: 7.25, 95% CI: 4.95-10.6).
These findings mirror those found in individual cohorts.
Overall 10.1% (9-11.4%) of the combined cohort switched to second-line treatment between six months and five years on ART. The expected delays between first and second detectable viral loads and subsequent switching were a median time of 2.7 months (IQR:1.6-4.7) and 4.6 months (IQR: 2.1-8.7), respectively.
This public health approach, note the authors, has been shown to successfully identify those patients with high levels of drug resistance needing to switch to second-line therapy but with low levels of cross-resistance between first- and second-line regimens.
But, as these findings show, how this approach is applied will have a considerable impact on the numbers who experience treatment failure and need second-line regimens.
The authors note that subsequent changes to South Africa’s guidelines may affect how these findings are interpreted. The initial nucleoside reverse transcriptase inhibitor (NRTI) backbone has been changed from stavudine (d4T) and lamivudine (3TC) to tenofovir and lamivudine. Routine viral load testing is now done annually rather than six monthly after the first year on ART. Confirmation of virologic failure was lowered from 5000 to 1000 copies/ml and the confirmatory test has to be done within three months of the first.
While less frequent monitoring may delay identifying those on failing treatment, use of tenofovir in first-line treatment causes less concern for the accumulation of thymidine analogue mutations that can compromise second-line regimens.
This study found a high proportion of those needing to be switched did switch treatment. Nonetheless, the nearly five-month delay between confirmation and switching and the link with a decline in CD4 counts suggest clinical and administrative factors contributing to the variations among the cohorts, note the authors.
Nevirapine, used as a non-nucleoside reverse transcriptase inhibitor (NNRTI), was associated with virological failure as commonly found in observational studies and with clinical trial data from African sites, the authors add.
Limitations include the absence of appropriate data on prevention of mother-to-child transmission to look at single-dose nevirapine and treatment failure. There were no data on adherence or prior resistance, two important potential predictors of failure.
The authors conclude “future treatment guidelines should make explicit the rationale for thresholds chosen to define and confirm virologic failure in light of our finding these profoundly affect the proportion of patients meeting failure definitions, and the resultant costs of second-line treatment…Future research should look at the impact failure definitions and delays in switching have on treatment outcomes.”
Fox MP et al. Rates and predictors of failure of first-line antiretroviral therapy and switch to second-line ART in South Africa. Advance online edition J Acquir Immune Defic Syndr, doi: 10.1097/QAI.0b013e3182557785, 2012.