ABSTRACT: Introduction: HIV resistance testing is increasingly available in clinical practice, but the benefits of resistance testing in a heavily pre-treated population remain relatively unproven.
Method: A prospective resistance testing program was established providing real-time (<1 month) genotypic results (by Vircogen sequencing) for treatment experienced HIV-1 patients with viral load values >1000 copies/ml who were switching onto different therapies. Treatment response was defined as a >1 log fall from baseline values or a viral load <50 copies/ml at week 12; factors related to treatment response were analyzed by logistic regression analysis.
Results: 52 patients have „12 week follow-up values from date of therapy change. All patients were nucleoside analogue (NA) experienced. 42 (81%) were protease inhibitor (PI)-experienced and 25 (48%) NNRTI-experienced with 15 (29%) exposed to all classes. Median CD4 and viral load (VL) at treatment change were 185 cells/mm3 (range 1-590 cells/mm3) and 4.31 log10 (range 2.84-6.40 log10). Patients switched to regimens including either an NNRTI (n=22), PI (n=18), both NNRTI and PI (n=10) or NA only (n=2). Proportion of patients receiving 0, 1, 2 or 3/4 drugs to which they were not previously exposed were 8%, 27%, 27% and 38% respectively. After a median follow-up of 14 weeks, the viral load dropped by a median 1.6 log10 (range 1.42 drop to -3.60 increase) and CD4 count rose by a median 29 cells/mm3. Thirty-two patients (62%) had responded to treatment at 12 weeks. Factors associated with treatment response in all patients were a higher CD4 count (P=0.04) and the number of key NNRTI mutations (defined as Y181C and K103N, P=0.02) present at switch. The number of key NNRTI mutations predicted treatment response independently of previous exposure to NNRTI. When the analysis was repeated only in those who switched to an NNRTI containing regimen (n=32), the number of key NNRTI mutations remained associated with treatment response. However, when the analysis was restricted to those starting a PI-containing regimen (n=28), the number of key PI mutations, rather than the number of NNRTI mutations, was associated with response.
Conclusion: Presence of mutations on genotyping at the time of treatment switch may help define risk of not responding to salvage regimens. A higher CD4 count at the time of switch is also associated with higher treatment response rates.
