Background:  Virologic failure (VF) is a key endpoint in evaluating ART in clinical trials. A VF threshold (VFT) of 50 copies/mL has been used in recent trials. Noting that patients with viral load >50 copies/mL often return to <50 copies/mL without change in ART, we evaluated the behavior of VFD with alternative VFT.

Methods:  The composite US Food and Drug Administration (FDA) time to loss of virologic response VFD with VFT of 50 (TLOVR50) and 200 copies/mL (TLOVR200) and the following purely virologic endpoints were considered. A:   confirmed viral load >1000 copies/mL at or after week 16 and before week 24; or confirmed viral load >200 copies/mL at or after week 24; B:  confirmed viral load >1000 copies/mL at or after week 16 and before week 24; or confirmed viral load  >50 copies/mL at or after week 24; C:  as A, or confirmed viral load >50 copies/mL at or after week 48; D:  confirmed viral load >50 copies/mL after confirmed viral load<50 c/ml; patients without confirmed viral load ≤50 copies/mL before death or last viral load have VF at week 0. Each VFD was applied to data from 2 recent ACTG ART-naïve trials (A5095 and A5142). Since both trials allowed patients to continue randomized ART (rART) after VF, viral load after VF could be studied in the context of ART changes driven by clinical practice. Patients without VF at the end of follow-up were censored. False positive VF was defined as confirmed viral load <50 copies/mL after VF and while on rART.

Results:  We included 1237 patients with at least 1 post-baseline viral load. Across VFD, 26 to42% had VF with 49 to 78% of these occurring on rART (see the table). Of note, 47 to 50% of the VF by TLOVR were due to discontinuation of rART. Across VFD, a VFT of 50 copies/mL gave a higher false positive rate (B-D, TLOVR50; 23 to 30%) than 200 copies/mL (A, TLOVR200; 13 to 14%). Since VFD-A and C differ only in the VFT after week 48, a comparison of their results after week 48 directly examined the use of a lower VFT; 65 more patients on rART had VF with VFD-C than with VFD-A. Of these, 75% had a viral load of 51 to 200 copies/mL at VF and 52% were false positives. Of the 65 non-VF with VFD-A, 89% were censored on rART; 55% were censored on rART with viral load <50 copies/mL; 14% (<1% of all patients) were censored without ever achieving confirmed viral load <50 copies/mL.

¹ VF due to rART discontinuation was considered off rART.

Conclusions:  For both purely virologic and TLOVR VF definitions a threshold of 50 copies/mL falsely declares VF for an unacceptably high number of patients who ultimately re-suppress <50 copies/mL without a change in ART. A threshold of 200 copies/mL may be preferable.