Background:  Patients who have virologically failed all 3 original antiretroviral drug classes have limited treatment options. It is therefore important to evaluate the rate at which patients experience extensive triple-class failure.

Methods:  Data were merged through COHERE, a collaboration of 33 European cohorts. Included patients started ART from 1998 onward with 2 NRTI and either a NNRTI or a PI/r, and were followed from starting ART until their last viral load. Failure of a drug was defined by a viral load >500 copies/mL despite ≥4 months of continuous use. extensive triple-class failure was defined as failure of ≥2 NRTI, ≥1 NNRTI, and ≥1 PI/r. Kaplan-Meier and Cox regression methods were used to investigate the risk of extensive triple-class failure after starting ART. In further analyses we focussed on patients starting their third drug class after following one of the treatment paradigms in Figure 1.

Results:  Of 45,977 patients contributing a total of 157,013 person-years, 980 (2.1%) experienced extensive triple-class failure. Most patients (63.7%) started ART with an NNRTI-based regimen. The estimated cumulative proportions (95%CI) with extensive triple-class failure by 5 and 9 years after starting ART were 3.4% (3.1% to 3.6%) and 8.6% (7.5% to 9.8%). Baseline factors found to be associated with an increased risk of extensive triple-class failure included lower CD4 count (350≤ CD4 <500 compared to 200≤ CD4 <350 cells/mm³: adjusted HR 0.65, 95%CI 0.50 to 0.85) and higher viral load. There was no evidence that the risk of extensive triple-class failure varied by initial regimen (Figure 2) (PI/r compared to NNRTI: adjusted HR 0.89, 95%CI 0.77 to 1.04). In the 2047 patients who followed treatment paradigm A, the mean person-time from starting ART to failing an NNRTI was 1.5 years, and the mean person-time from failing an NNRTI to starting a PI/r was 1.4 years. In the 557 patients who followed paradigm B, the mean person-time from starting ART to failing a PI/r was 1.0 year, and the mean person-time from failing a PI/r to starting an NNRTI was 1.2 years. The estimated cumulative proportions (95%CI) with extensive triple-class failure 1 and 5 years after starting the third drug class were 20% (18 to 22%) and 46% (43 to 50%) in patients who followed paradigm A, and 27% (23 to 31%) and 43% (37 to 48%) in patients who followed paradigm B.

Conclusions:  The risk of extensive virologic failure of the original 3 classes is low and does not appear to depend on the order in which the PI/r and NNRTI classes are used. A substantial proportion of the time which elapsed before starting the third drug class occurred after virologic failure of the first 2 classes.