Background:  The risk for selecting drug-resistant variants during therapy interruptions remains a major concern. We assessed the emergence of genotypic drug-resistance mutations in 87 patients from the Tibet Study.

Methods:  Patients on HAART with a viral load of < 50 copies/mL for ³ 1 year, CD4 ³ 500 cells/mm³ for ³ 6months before study entry and a nadir CD4 count of > 50 cells/mm³ were eligible. Patients were randomized to interrupt therapy (n = 100) or to continue (n = 101) with the same treatment. The criteria for restarting therapy in patients assigned to the interruption arm were:  a viral load increase > 100,000 copies/mL or a drop of CD4 < 350 cells/mm³ or the appearance of any opportunistic infection. Patients were planned to stop therapy again when:  CD4 ³ 500 cells/mm³ and viral load < 50 copies/mL. The HIV-1 drug-resistance genotypes were determined on plasma samples 1 month after each therapy interruption and on peripheral blood mononuclear cells (PBMC) at baseline (study entry).

Results:  Drug-resistance mutations were detected in 31 of 87 (36%) patients allocated to the interruption arm. Most of them (84%) received sequential suboptimal treatment before HAART; 19 received non-nucleoside reverse transcriptase inhibitor- (NNRTI)-containing regimens and the remaining 12 received protease inhibitor- (PI)-containing regimens at the study entry. In 23 patients, thymidine analog mutations (TAM) were detected during the first, second, or third therapy interruption. These mutations, linked to the previous ART regimen, and persisted in those patients who underwent subsequent therapy interruptions although the frequency varied among subjects. Similar NRTI-related resistance mutation patterns were detected in PBMC at baseline. The most frequent single TAM were:  41L (26%), 67N (35%), 70R (39%), 210W (19%), and 215Y/F (35%). The mutation 184V was detected during the first therapy interruption in 8 patients and was associated with patients’ regimen. This mutation did not persist in the 2 patients who underwent a second therapy interruption because both switched to non-lamivudine HAART regimen. NNRTI-related mutations were selected de novo (not detected in DNA at baseline) during the first or second therapy interruption or shortly after ART resumption in 8 of 19 (42%) patients receiving NNRTI HAART regimens. In the protease gene, 46L mutation was detected in only 1 case.

Conclusions:  Drug-resistance mutations were detected in 36% of the patients during therapy interruptions. Among detected mutations, pre-existence rather than evolution as a consequence of therapy interruption was noted for TAM. However, NNRTI-resistant mutations were selected de novo. Treatment interruption in patients on NNRTI-based HAART should not be recommended. These patients should be switched to strategies containing drugs with shorter plasma half-lives.

Keywords: HIV drug-resistance; Treatment interruption; NNRTI