Non-nucleoside reverse transcriptase inhibitor failure impairs HIV-RNA responses to efavirenz-containing salvage antiretroviral therapy

In a retrospective cohort study of salvage antiretroviral combination therapy including efavirenz, 60% of 51 patients were able to suppress HIV RNA by at least 1 log10 or to less than 50 copies/ml. A lack of the previous use of non-nucleoside reverse transcriptase inhibitors was the only factor predictive of response on multivariate analysis. No patient with a viral isolate with an increased IC50 to efavirenz by virtual phenotype had a virological response. Non-nucleoside reverse transcriptase inhibitors (NNRTI) are potent inhibitors of HIV reverse transcriptase [1–3]. High-level resistance, usually as a result of a single step mutation, develops rapidly with NNRTI monotherapy [4–7]. Because of the high potential for NNRTI cross-resistance, expert panels recommend that if resistance develops to one NNRTI, it is unlikely that a second NNRTI will be a useful component of salvage therapy [8,9]. However, there are few published clinical data to support this [10], and at least two studies have suggested that if the Y181C mutation develops during failure with nevirapine, efavirenz might still be effective in a subsequent combination [11,12], at least in the short term. In addition, NNRTI have been included as a part of multidrug rescue therapies despite previous exposure [13]. This cohort included patients with a previously documented CD4 cell count of less than 200/mm3 and virological failure to at least one previous protease inhibitor (PI)-containing regimen. All patients received efavirenz through a compassionate access program between March 1998 and April 1999, with follow-up to April 2000. To be included, subjects required at least one HIV-RNA measurement after the switch to the efavirenz-containing combination. Baseline data were obtained by chart review. Plasma HIV-RNA levels were measured using a branched chain DNA assay (Chiron 3.0, Chiron Corporation, Emeryville, CA, USA) with the lower limit of detection 50 copies/ml. Follow-up data were collected until efavirenz discontinuation, antiretroviral regimen modification, or until April 2000. Genotype (ABI 3700, Applied Biosystems, Foster City, CA, USA) and ‘virtual’ phenotype analysis of the reverse transcriptase was performed for all NNRTI-experienced patients. The ‘primary’ genotype mutations reported included: Y181C/I/Y/Q, K103N, V106A, V108I, Y188C/L, G190A/S/E, and P236L. An estimation of the degree of efavirenz resistance from the sequence data was performed using the Virtual-Phenotype (Virco, Belgium) [14]. Statistical comparisons were performed on an intent-to-treat basis by univariate and multivariate regression models using the SPSS statistical package for Windows (SPSS Inc., release 9.0). Complete response was defined as suppression of HIV RNA to less than 50 copies/ml and partial response was defined as at least a 1 log decrease in plasma HIV RNA but 50 copies/ml or more at any time after the initiation of the salvage regimen. Categorical variables were compared by chi-square or Fisher's exact test and continuous variables by Student's t-test. Multivariate logistic regression was used to identify baseline predictors of HIV-RNA suppression. Fifty-one patients (94% men) with a mean age of 40 years were included. The median baseline CD4 cell count and HIV-RNA levels were 184/mm3 (range < 10–811) and 4.69 log10 copies/ml (range 2.37–5.66), respectively. All patients were nucleoside reverse transcriptase inhibitor therapy experienced, with a mean of 3.7 compounds. A total of 27, 38 and 35% had received one, two, three or more PI, respectively, for a mean of 23 months. Sixteen patients (31% had received at least one previous NNRTI for a mean duration of 7.5 months. In addition to efavirenz, 29 patients (57%) initiated a new PI concurrently. The salvage regimen consisted of one, two, three or more new agents in 10, 33, and 57% of the cohort, respectively. Seventeen of the cohort (33%) achieved complete HIV-RNA suppression, and 14 patients (27%) had at least a 1 log10 reduction in HIV RNA, for a total complete or partial response rate of 61% Corresponding complete and partial response rates (intent-to-treat) at 24 weeks were 28 and 10% respectively. Of the 17 complete responders, two discontinued therapy because of adverse events, one stopped therapy by choice, one had virological failure at week 20, and the remaining 13 (76%) continue to have less than 50 copies/ml at a mean of 53 weeks’ (range 24–80) follow-up. Partial or complete responses were seen in 15 out of 29 patients (52%) who initiated a new PI as part of the salvage regimen and 16 out of 22 patients (73%) who did not (P = 0.22). For NNRTI-naive patients, 16 (47%) were complete responders and 12 partial responders (35%). Responders were compared with non-responders on the basis of baseline CD4 cell count or HIV-RNA level, the mean number of previous PI, the use of a new PI in the salvage regimen, the number of new antiretroviral agents in the salvage regimen and previous use of an NNRTI. Using logistic regression analysis, baseline HIV RNA and the lack of previous NNRTI exposure were predictive of a complete virological response, but only the lack of previous NNRTI exposure was predictive of a complete or partial response, odds ratio 16.7 (2.8–101.1), P = 0.0022. In total, 16 patients were NNRTI experienced, and nine were on NNRTI at the time efavirenz was initiated. The baseline genotype was available for 14 patients. In the non-responders (n = 11), mutations Y181C, K103N/Q or both were seen in five, three, and one patient, respectively, and two patients also had the G190S mutation. Of the responders (n = 3), two patients had no baseline NNRTI mutations. The other responder had a baseline Y181C mutation. She had previously received delavirdine for a 2 week period, but it was discontinued because of a rash. On the salvage therapy, a partial response of a 2.9 log reduction in HIV RNA occurred at 2 months, but rebounded to a baseline of more than 500 000 copies/ml at 4 months (Table 1).Table 1: Predictors of virological response to an efavirenz-containing combination. A ‘virtual’ phenotype for efavirenz was determined for these 14 patients. Of the four patients without NNRTI primary mutations, efavirenz IC50 was estimated to be be less than two times the IC50 of wild-type (WT) virus. Of these, one patient was a complete responder, one a partial responder and two were non-responders. For the patients with the Y181C mutation but no K103N (n = 6) the estimated efavirenz IC50 varied from three to 25 times greater than WT, depending on the presence of additional mutations; of these five patients were non-responders and one was a partial responder (IC50 three times the WT). The single patient with the double mutation (Y181C, K103N) had an estimated IC50 more than 50 times the WT and was a non-responder. The study limitations included the relatively small, non-randomized nature and varied treatment histories of our patients, which frequently limited the drug options in the salvage combination. In addition, it is possible that some individuals in the ‘non-responder’ category had a very brief period of virological suppression, which was achieved and lost between visits. This cohort study demonstrates that despite considerable antiretroviral exposure, a significant proportion of patients can respond to a salvage therapy including efavirenz, and supports the hypothesis that PI failures can be salvaged with an NNRTI-containing combination. Previous NNRTI failure seriously compromised the antiviral effect of efavirenz-containing therapy, as partial or complete responses were seen in 82% of NNRTI-naive patients versus 18% of NNRTI-experienced patients. A decreased response rate was probably the consequence of cross-resistance. All non-responders had a decreased IC50 to efavirenz; however, the true break-point for resistance requires further study. Sharon L. Walmsleyab Deborah V. Kellyag Alice L. Tsengac Atul Humarb P. Richard Harrigandef