To the Editor:
Development of HIV-1 resistance to the antiretroviral drug class of protease inhibitors (PIs) is caused by the selection of PI resistance-associated mutations in the protease coding region. High-level resistance to PI requires accumulation of resistance mutations in protease.1 To predict viral susceptibility to a drug, genotypic resistance scores are generated, based on the assessment of the impact of genotypic patterns at baseline on the subsequent virological response. Different drug-resistance interpretation algorithms, all based on genotypic resistance scores, are mostly used in clinical practice to reliably select and adapt a new antiretroviral-based regimen after a virological failure, particularly in antiretroviral experienced patients.2,3
Several studies have reported genotypic resistance scores for darunavir (DRV), the latest PI to be licensed.4-6 The study of Descamps et al4 determined a set of 8 protease mutations (K14R, K20I, E34Q, I47V, I54M, K55R, T74P, and I84V) that predict the virological response to darunavir/ritonavir (DRV/r) in PI-experienced HIV-1-infected patients (P = 2.54 10−8). The virological response rates were 78% (n = 46), 59% (n = 56), 36% (n = 36), 18% (n = 11), and 0% (n = 4) in patients with 0, 1, 2, 3, and ≥4 virus DRV/r mutations, respectively.4
In this substudy, we assessed the impact of coprescription of enfuvirtide (ENF) and DRV/r in a salvage regimen on a previously validated DRV/r genotypic resistance score based on virological response.
PATIENTS AND METHODS
This work is a substudy from a previous study designed to assess protease mutations associated with virological response to DRV/r in PI-experienced patients.4 We analyzed 153 highly antiretroviral experienced patients receiving a DRV/r salvage regimen, whose characteristics have been previously described.4 ENF was included for the first time in the regimen in 43 patients (28%). Virological response was defined as HIV-1 RNA <200 copies per milliliter at month 3. This latter threshold may represent a potential limitation of the study as the current guidelines recommend that the goal of the antiretroviral therapy is to suppress HIV-1 RNA to below 50 copies per milliliter; however, our value represents the limit of detection of the HIV-1 RNA quantification assay used at time of this study.
At baseline, the median plasma HIV-1 RNA level was 4.7 log10 copies per milliliter [interquartile range (IQR): 4.3-5.2] and the median CD4 cell count was 142 cells per cubic millimeter (IQR: 28-264). Before this study, the patients had been exposed to a median of 12 antiretroviral drugs (range: 10-14), including a median of 6 nucleoside reverse transcriptase inhibitors (IQR: 5-7) and 4 PIs (IQR: 3-5). At baseline, the median numbers (IQR) of major and minor PI-resistance mutations, based on the International AIDS Society-USA (IAS-USA) list, were 4 (3-8) and 9 (7-10), respectively. The median number of potentially active drugs used along with DRV/r and ENF was 1 (range: 0-5), according to the ANRS algorithm.4
A logistic regression was performed to discern if there was a relationship between the number of mutations from the mutation score (MS) and the probability of having a virological response (π). The logistic model was obtained after using a logarithm link on π, that is,
where (1) X is the number of mutations from the MS reported in the virus from each patient, considered as a continuous variable; (2) μ is the intercept that can be interpreted as the value of π when X = 0 (no mutation); and (3) β is the slope, which can be interpreted as an adjustment of the probability of virological response depending on the number of mutations reported. The interest in the logistic model, compared with traditional linear trend tests (Cohran-Armitage test performed to determine the genotypic resistance score), was that other covariates can be included in the model to assess their influence on virological response conjointly to the MS. The covariates were screened as variables using a univariate approach for those that had an impact on the virological response at a statistical level of 5%. In this study, the following covariates were screened: (1) baseline HIV-1 RNA level, (2) baseline CD4 cell count, (3) first intake of ENF, and (4) number of protease mutations as assessed by the MS. After reverse elimination, only the number of protease mutations, the first intake of ENF, and baseline HIV-1 RNA levels were kept as significant predictors of a virological response. Thus, the final model retained was as follows:
where VL0 = baseline HIV-1 RNA level (log10 copies/mL); 1ENF = indicator variable (1 = first intake of ENF at baseline, 0 = no intake of ENF); and β2 and β3 were the slope for baseline HIV-1 RNA level and the increment for first ENF intake at baseline, respectively.
This substudy is based on the DRV/r genotypic resistance score described by Descamps et al,4 which included 8 protease mutations that have a negative impact on virological response and were as follows: K14R, K20I, E34Q, I47V, I54M, K55R, T74P, and I84V. As expected, the probability of presenting with a virological response decreases as the number of protease mutations increases, as shown in Figure 1. Furthermore, the first intake of ENF at baseline, independent of the number of DRV resistance-associated mutations, increased the probability of obtaining a virological response (P = 0.0104). Thus, in patients exhibiting plasma viruses with 2 DRV resistance-associated mutations according to the genotypic resistance score, the probability of displaying a virological response with a first intake of ENF at baseline was approximately 70%, whereas it was only 40% without ENF in the salvage regimen (Fig. 1). Similarly, the probability of a virological response in patients displaying viruses with 3 DRV resistance-associated mutations was approximately 50% and approximately 20% in the groups with and without ENF in the salvage regimen, respectively (Fig. 1). This phenomenon was also observed with highly mutated viruses. Indeed, adding ENF increased the probability of a virological response by 16% and 8% in patients who had viruses with 4 and 5 mutations as assessed by the DRV/r genotypic resistance score, respectively.
In this substudy on DRV/r genotypic resistance score, we have shown that including ENF in the DRV/r-based regimen improves the predictive value of the DRV/r genotypic resistance score when assessing the rate of virological response. According to our results, the coprescription of ENF with DRV/r in the present series compensates for the effect of 1 additional mutation in the DRV/r genotypic resistance score.
It is well described that the introduction of a compound from a new antiretroviral drug class into a salvage regimen increases the rate of virological response, as observed in different clinical trials conducted in highly antiretroviral experienced patients (BENCHMRK, DUET).7,8 In the BENCHMRK studies, the subgroup of patients who were coprescribed ENF (first use) and DRV/r displayed an 89% virological response rate (39 of 44 patients) compared with 69% of patients only receiving DRV/r (52 of 75 patients).7
In our study, as expected, we reported that ENF coprescription to ENF-naive patients was independently associated with virological response by month 3 (P = 0.001).4 Moreover, the present substudy, which included highly PI-experienced patients, allowed us to assess the impact of first intake of ENF in a salvage DRV-based regimen on DRV/r genotypic resistance score.
Further studies are needed to assess virological response using resistance-interpretation algorithms that take into account the impact of the combination of the different antiretroviral drugs comprised in the regimen.
Charlotte Charpentier, PharmD, PhD*†
Sidonie Lambert-Niclot, PharmD, PhD‡§
Lucile Larrouy, PharmD*†
Gilles Peytavin, PharmD, PhD†‖
Marc-Antoine Valantin, MD, PhD¶
Roland Landman, MD, PhD#
Christine Katlama MD, PhD¶
Patrick Yeni, MD, PhD#
Mathieu Felices, MD**
Françoise Brun-Vézinet, MD, PhD*†
Vincent Calvez MD, PhD‡§
Anne-Geneviève Marcelin, PharmD, PhD‡§
Diane Descamps, MD, PhD*†
*AP-HP, Hôpital Bichat-Claude Bernard, Laboratoire de Virologie, Paris, France †Université Paris 7, Paris, France ‡AP-HP, Laboratoire de Virologie, Groupe hospitalier Pitié-Salpêtrière, Paris, France §UMR_S 943,UPMC Université Paris 6, Paris, France ‖AP-HP, Laboratoire de Pharmacologie, Hôpital Bichat-Claude Bernard, Paris, France ¶AP-HP, Groupe hospitalier Pitié-Salpêtrière, Service de Maladies Infectieuses et Tropicales, Paris, France #AP-HP, Hôpital Bichat-Claude Bernard, Service de Maladies Infectieuses et Tropicales, Paris, France **SGS, Aster, S.A.S, Paris, France