Nonnucleoside reverse transcriptase inhibitors (NNRTIs) are a cornerstone of first-line HAART. However, first-generation NNRTIs are limited by their low genetic barrier to the development of resistance and substantial cross-resistance . For this reason, NNRTIs are not used sequentially in treatment-experienced patients in routine practice. Potential strategies to overcome this, such as the combination of first-generation NNRTIs with boosted protease inhibitors, have been investigated; however, toxicity issues have prevented full exploration of this combination of antiretroviral agents .
Etravirine (TMC125) is a recently introduced NNRTI with potent in-vitro activity against wild-type and NNRTI-resistant HIV-1 , and a higher genetic barrier to the development of viral resistance than the first-generation NNRTIs , which may be explained by its molecular flexibility . As such, etravirine offers the potential to achieve antiviral activity in patients with NNRTI resistance. Following the preclinical and early clinical studies, the efficacy and safety of etravirine given for 48 weeks in treatment-experienced HIV-1-infected patients has been demonstrated in the double-blind and placebo-controlled phase III clinical studies with TMC125 to Demonstrate Undetectable viral load in patients Experienced with ARV Therapy (DUET)-1 and DUET-2 [6,7]. The design of the DUET studies was characterized by a background therapy with the ritonavir-boosted protease inhibitor darunavir (which has shown substantial virologic efficacy in treatment-experienced patients [8–10]), a planned pooling of data from the two studies, and the stringency of the primary endpoint (attainment of confirmed viral load <50 HIV-1 RNA copies/ml). The 24-week results of the DUET studies [11,12] were the basis of the indication for the use of etravirine in combination with other antiretroviral agents for the treatment of HIV-1 infection in treatment-experienced adult patients [13,14].
Improved virologic responses to combination therapy may be obtained if antiretroviral drugs are selected on the basis of their activity as determined by resistance testing, and such testing is recommended to guide the choice of new drug regimens after treatment failure [15–17].
The impact of baseline genotype on virologic response to etravirine was initially investigated in 2007, utilizing a comprehensive list of 44 resistance-associated mutations (RAMs) compiled on the basis of their association with NNRTI resistance [18–20]. Of the 44 mutations, a total of 13 (etravirine RAMs: V90I, A98G, L100I, K101E/P, V106I, V179D/F, Y181C/I/V and G190A/S) were identified by their association with at least a 25% reduction in response to etravirine as compared with that of a subgroup of patients with no detectable NNRTI RAMs at baseline . Multiple etravirine RAMs (≥3) at baseline were generally required to confer a reduced virologic response.
Phenotypic susceptibility is generally determined using clinical cutoff (CCO) values. By applying these CCOs to the dynamic range of measurable fold change in EC50 values phenotypic susceptibility to a given antiretroviral, it is possible to predict how response is affected. Two phenotypic CCOs are typically derived; the lower CCO usually defines the fold change in EC50 above which a reduction in antiviral activity starts to be observed, whereas the upper CCO can be defined as the fold change in EC50 value above which there is little or no antiviral activity. Fold change values between the lower and upper CCO are usually indicative of an intermediate response.
Here, we present the establishment of CCOs for etravirine as well as the genotypic analysis, including the refinement of the etravirine RAMs list and the development of a weighted scoring system. These additional analyses were performed to improve the genotypic prediction of virologic response, the relationship between phenotypic and genotypic susceptibility and thus, ultimately, to provide better guidance in the interpretation of etravirine resistance.
Genotyping and phenotyping
Genotypic analyses were performed by automated population sequencing (Virco BVBA, Mechelen, Belgium). Individual data were reported as amino acid changes along the HIV-1 protease and reverse transcriptase genes as compared with the HIV-1/HXB2CG (Acc Nr K03455) wild-type reference .
For phenotypic analyses, recombinant clinical isolates were constructed according to the Antivirogram method (Virco BVBA) [22,23]. Briefly, protease and reverse transcriptase coding sequences were amplified from patient-derived viral RNA with HIV-1-specific primers. After homologous recombination of amplicons into a protease–reverse transcriptase-deleted proviral clone, the resulting recombinant viruses were harvested and used for in-vitro susceptibility testing.
The resistance profile of etravirine was analyzed using a panel of HIV-1/HXB2 mutants harboring a selection of well defined single or multiple amino acid substitutions in the reverse transcriptase . Amino acid substitutions were introduced into the HIV-1/HXB2 backbone by site-directed mutagenesis (SDM) using the QuikChange SDM kit (Stratagene, La Jolla, California, USA) or a commercial supplier (Eurofins Medigenomix GmbH, Martinsried, Germany). The resulting strains were genotyped and phenotyped.
DUET study design
In both the DUET-1 and DUET-2 studies, patients were randomized on a 1: 1 basis to etravirine 200 mg or placebo twice daily. Full details of the design of these studies have been published previously [11,12]. All patients also received background therapy with darunavir 600 mg along with ritonavir 100 mg twice daily, investigator-selected NRTIs and optional enfuvirtide.
The studies were carried out in 185 centers in 19 countries in men and women aged at least 18 years with documented HIV-1 infection being treated with a stable but failing regimen with plasma viral load of at least 5000 HIV-1 RNA copies/ml. Patients also had at least one documented NNRTI RAM from screening genotype or historical resistance data and at least three primary protease inhibitor mutations at screening . Patients were stratified according to enfuvirtide use [re-use, no use or use for the first time (de novo)], previous use of darunavir (yes or no) and screening plasma viral load (< or ≥30 000 HIV-1 RNA copies/ml).
The primary endpoint was the proportion of patients who achieved confirmed viral load of less than 50 HIV-1 RNA copies/ml at week 24; secondary endpoints included the proportion of patients with confirmed viral load of less than 400 HIV-1 RNA copies/ml or at least 1 log10 viral load reduction, changes in viral load and CD4 cell count from baseline, proportion of patients in whom new AIDS-defining illness or death was reported, change in HIV-1 genotype and drug susceptibility and safety and tolerability.
Plasma viral loads were determined with the Roche COBAS Amplicor HIV-1 monitor ultra sensitive test (version 1.5; Roche Diagnostics, Basel, Switzerland). Confirmation of virologic response or loss of response required two consecutive values below or above the relevant threshold, respectively.
Study population analyzed
This analysis reports pooled results from the DUET-1 and DUET-2 studies for patients who initiated treatment with etravirine in combination with a background regimen.
At the time of the resistance analysis presented here, all patients had received at least 24 weeks of treatment or had discontinued earlier. To reduce bias in the resistance analyses, patients who discontinued the study for reasons other than virologic failure before week 24 were excluded (the non virologic-failure excluded population) unless stated otherwise [e.g. for the LOWESS or LOESS (LOcally WEighted Scatterplot Smoothing) smoothed spline model]. De-novo enfuvirtide users were also excluded in order to avoid the confounding effect of the activity contributed to the regimen by a new antiretroviral class. This non virologic-failure excluded and not de-novo enfuvirtide population included 406 patients, of whom 403 had both genotypic and phenotypic data available.
In order to determine the effect of baseline genotype or phenotype on virologic response, week 24 data were predominantly used. This timepoint was selected because, with increasing time from baseline, there is likely to be more interference of factors that are not related to resistance. Nonetheless, week 48 data are also presented to illustrate the consistency of the results and the durability of the virologic response.
Effect of baseline phenotype on virologic response: determination of etravirine clinical cutoffs
The LOESS smoothed spline and analysis of covariance (ANCOVA) models were used to determine phenotypic CCOs for etravirine, using virologic response data from the pooled DUET studies. Briefly, data were graphically presented as a scatter plot of the etravirine fold change in EC50 versus the change in log10 viral load from baseline at week 24 together with the LOESS spline analyzing the relationship between both parameters. The LOESS spline was added in two enfuvirtide strata: de-novo enfuvirtide use and not de-novo enfuvirtide use (i.e. enfuvirtide re-used or not used). The change in log10 viral load in the etravirine group was analyzed using an ANCOVA model with the factors enfuvirtide use (not used, re-using), number of active nucleoside/nucleotide reverse transcriptase inhibitors in the background regimen and the covariates baseline viral load, baseline CD4 cell count and log10 transformed darunavir and etravirine fold change in EC50. A lower CCO was determined by graphical exploration of the data, using sliding baseline fold change in EC50 intervals in relation to virologic response and change in log10 viral load from baseline at week 24.
Effect of baseline genotype on virologic response
The list of NNRTI mutations upon which the analyses leading to the identification of the 17 etravirine RAMs was based comprised 57 mutations. These included RAMs listed by International AIDS Society-USA or other sources as being associated with NNRTI resistance , RAMs associated with an increased fold change in EC50 to etravirine or mutations at NNRTI resistance-associated positions observed at baseline in the pooled DUET data. The full list of 57 RAMs is V90I, A98G/S, L100I, K101E/H/N/P/R/Q, K103H/N/R/S/T, V106A/I/M, V108I, E138A/G/K/Q, V179A/D/E/F/G/I/T, Y181C/F/I/V, Y188C/F/H/L, V189I, G190A/C/E/Q/R/S, H221Y, P225H, F227C/L, M230I/L, P236L, K238N/T, Y318F and N348I/T.
Multivariate analyses were carried out to refine the initial etravirine RAM list and to develop a weighted genotypic score for etravirine, with the goal of improving the predictive value of resistance testing with regard to virologic response and improving the relationship between genotypic and phenotypic interpretation of etravirine resistance. Etravirine RAMs were selected if they were associated with decreased virologic response to etravirine at week 24 in the DUET studies and if they were present at baseline in at least five patients. In addition, an increased etravirine fold change in EC50 in SDMs or NNRTI-resistant recombinant HIV-1 clinical isolates was taken into account. Decreased virologic response was defined as a response at least 25% lower than a reference value . This reference value was determined by using the study population analyzed (non virologic-failure excluded) and by determining the proportion of responders among those patients harboring none of the initially identified 44 NNRTI RAMs at baseline [69.2% (36/52)]. The 25% reduction from this response led to a defined response threshold of 51.9% or lower.
Assignment of relative weight factors to the etravirine resistance-associated mutations
The following two statistical methods were applied to determine the relative weight factors: random forest and linear modeling.
The random forest methodology was used to predict the effect of specific mutations on resistance (fold change in EC50) and to determine which mutations were the best predictors of response. One hundred bootstrap samples were generated from the dataset by sampling with replacement, keeping the original sample size of the dataset constant. For each of these 100 bootstrap samples, a regression tree was constructed using all available mutations. Grouping of samples according to the level of fold change in EC50 was based on the lowest ‘within group’ variance for each group. Subsets of mutations were chosen randomly. Using the 100 different regression trees, an importance measure was calculated for every mutation. These calculations were repeated 100 times to generate a reliable estimate of the impact of each mutation on etravirine fold change in EC50.
The impact of mutations on etravirine fold change in EC50 was also estimated by multiple linear regression modeling using the following equation :
Equation (Uncited)Image Tools
The log-transformed fold change in EC50 (as the dependent variable) was predicted using the independent variables muti, representing the presence of mutations in a generic sequence. Values 0 and 1 were assigned for absence or presence of mutations in a sequence, respectively. If a mutation was present in a mixture, the assigned value was weighted accordingly, 1/2 or 1/3 for mixtures of two or three amino acids, respectively. The linear regression model was calculated in two steps. A first model was calculated using single mutations only. Only those mutations occurring in at least 20 samples of the dataset were considered. Stepwise regression was performed in which mutations were added alternately to, and removed from, the growing model based on a statistical threshold. The P-value threshold to enter the model was set at 0.5, and at 0.0001 to remain in the model. Therefore, in each forward step, the most predictive parameter was chosen, and in the subsequent backward step, more stringent P values allowed elimination of the least predictive variables.
Second-order interactions were added to the regression model using the equation:
Equation (Uncited)Image Tools
The mutations that were significant in the first model were used in the second model, together with mutation pairs, to assess the importance of interaction terms that accounted for synergistic and antagonistic effects between mutations. Interactions were only taken into account if at least 20 samples had the combination and each constituent mutation of the pair was present at least 20 times without the other. The P-value threshold to enter the model was set at 0.5, and at 0.001 to remain in the model.
DUET study results
In the combined DUET studies, 1203 patients were randomized, of whom 599 were treated with etravirine and 604 with placebo . Overall baseline characteristics were similar between the treatment groups, with median viral loads of approximately 4.8 log10 HIV-1 RNA copies/ml and median CD4 cell counts of approximately 100 cells/μl. The de-novo use of enfuvirtide was the same (26%) in both groups.
Pooled data from the week 24 analysis showed that significantly more patients achieved viral load less than 50 HIV-1 RNA copies/ml with etravirine than with placebo (58.9 versus 41.1%, P < 0.0001) . Patients who achieved viral load less than 50 HIV-1 RNA copies/ml at week 24 sustained viral suppression to week 96 in 83.2 and 77.6% of cases in the etravirine and placebo groups, respectively . Etravirine was associated with a significant increase in CD4 cell counts.
Effect of baseline etravirine fold change in EC50 on virologic response to etravirine and determination of clinical cutoff values
Baseline etravirine fold change in EC50 was found to be a strong predictor of virologic response at week 24 . Using the LOESS model, no clear relation was found between baseline etravirine fold change in EC50 and virologic response in patients using de-novo enfuvirtide due to the high overall response rate (RR) seen in this patient subpopulation (66.7% had <50 HIV-1 RNA copies/ml at week 24). In patients in the not de-novo enfuvirtide group, a gradual decline in virologic response achieved at week 24 was observed, as baseline etravirine fold change in EC50 increased. However, the LOESS smoothed spline model did not provide evidence of clear breakpoints associated with changes in virologic response trends. In the population of patients not using de-novo enfuvirtide, baseline fold change in EC50 and virologic responses to etravirine were characterized by a continuum rather than a bimodal bivariate distribution (i.e. samples have either a low or high fold change in EC50). As a result, further analyses using covariance models were performed on patients not using de-novo enfuvirtide.
Via inverse prediction, the estimates from this model determined the etravirine fold change in EC50 that would result in a certain threshold change in log10 viral load from baseline. In line with a previous study , this threshold was set at 1 log10 reduction in viral load for etravirine in addition to the reduction in viral load observed for the comparable placebo group at week 24. The resulting viral load threshold was a −2.4 log10 change from baseline. The lower limit of the 95% confidence interval around the estimated change in viral load was found to correspond to an etravirine fold change in EC50 value of 13.6. Therefore, a fold change in EC50 value of 13.0 was used as one of the breakpoints to predict virologic response based on baseline etravirine phenotypic susceptibility (Fig. 1a). A formal upper CCO above which etravirine provided little or no additional efficacy benefit could not reliably be determined because of the small number of patients with fold change in EC50 above 13.0 (Table 1). Thus, the etravirine fold change in EC50 is equal to 13.0 should be regarded as a preliminary upper CCO.
As the majority of patients in the pooled DUET trial population (85%) had an etravirine fold change in EC50 below 13.0, analyses were conducted to determine whether responses could be further distinguished using lower etravirine fold change in EC50 thresholds. A lower etravirine fold change in EC50 above which virologic responses started to decline was investigated, using graphical exploration of sliding fold change in EC50 windows (Fig. 1b).
The subgroups of patients with a baseline etravirine fold change in EC50 value 3.0 or less demonstrated similar virologic response levels (<50 HIV-1 RNA copies/ml) at week 24, and higher response levels than those with fold change in EC50 above 3.0. On the basis of these results, a lower CCO was determined at 3.0 (Table 1). The highest RR (70.6%) was observed in the group of patients in DUET, with a baseline etravirine fold change in EC50 of 3.0 or less (67% of patients). An intermediate RR of 50.0% was observed in patients with baseline etravirine fold change in EC50 between 3.0 and 13.0 (18% of patients), and the lowest RR (36.7%) was recorded in patients with etravirine fold change in EC50 above 13.0 (15% of patients). The results were similar when week 48 virologic responses in DUET were analyzed by baseline fold change in EC50 (Table 1).
Effect of baseline genotype on virologic response to etravirine
An analysis of the pooled week 24 DUET dataset was performed in 2008 to refine the list of etravirine RAMs and to develop a weighted genotypic score for etravirine to improve the predictive value of resistance testing with regard to virologic response, and the relationship between genotypic and phenotypic results . The mutation list assessed in this new analysis was expanded from 44 to 57 mutations by inclusion of all mutations at NNRTI resistance-associated amino acid positions observed at baseline in the pooled DUET dataset [18,30]. In this instance, mutations were selected if they were present in at least five patients and were associated with decreased virologic response (<51.9%), increased etravirine fold change in EC50 or both. Four additional etravirine RAMs were identified through this process (K101H, E138A, V179T and M230L), resulting in a total of 17 etravirine RAMs. Even though it was present in only four patients at baseline, M230L was included on the basis of its strong effect on etravirine fold change in EC50 combined with a RR, in patients carrying this mutation, below 51.9%. Notably, V106I, V179F/T, Y181V and G190S were all present in fewer than 6% of patients in the pooled DUET studies at baseline (Table 2).
Relationship between etravirine-weighted genotypic score and virologic response
A weighted genotypic scoring system was developed, whereby etravirine RAMs were assigned relative weight factors according to their impact on response, fold change in EC50 or both. The purpose of this was to assess the variable effect that individual etravirine RAMs have on virologic response, fold change in EC50 or both. The relative weight factors were determined with random forest and linear modeling techniques, using matched genotypic and phenotypic data from the nonvirologic failure excluded population using not de-novo enfuvirtide in the etravirine treatment group in the pooled DUET studies (n = 403) and data from a panel of NNRTI-resistant recombinant HIV-1 clinical isolates (n = 4248). These weight factors were summed for each patient to generate a weighted genotypic score reflecting the effect of all etravirine RAMs on virologic response. Among the 17 etravirine RAMs, Y181I and Y181V were assigned the highest weight factor (3) followed by K101P, L100I, Y181C and M230L (2.5). Overall, as shown in Table 2, the etravirine RAMs with the highest weight factors generally had low prevalence at baseline in patients in the DUET studies (all below 10%, with the exception of Y181C, which was present in 32% of patients).
Previously, data were reported on the prevalence and effect on virologic response of the most frequently observed combinations of etravirine RAMs . The data showed that virologic responses in etravirine-treated patients harboring the combination of mutations Y181C along with G190A (n = 29) and K101E along with G190A (n = 5), the two most prevalent combinations, were 58.6 and 80.0%, respectively.
As shown in Fig. 1(c), application of the genotypic scoring system showed that virologic response was a function of the baseline etravirine-weighted genotypic score. Accordingly, three response categories were defined: weighted genotypic scores of 0–2, 2.5–3.5 and at least 4, corresponding to RRs of 74.4% (highest response), 52.0% (intermediate response) and 37.7% (reduced response), respectively (Table 1 and Fig. 1c). In patients with a weighted genotypic score below 4, virologic RRs were substantially greater in patients receiving etravirine than those in the corresponding subgroups receiving placebo. The robustness of the weighted genotypic score is illustrated by the concordance of the results for week 48 virologic responses with the week 24 data (Table 1).
Use of the updated list of 17 etravirine RAMs along with the relative weighting improved the relationship between genotypic and phenotypic susceptibility interpretations as compared with the list of 13 etravirine RAMs (Tables 1 and 3). Table 1 shows similarities in virologic RRs in patient subgroups according to CCO and weighted genotypic score. Table 3 shows results illustrating that the 2008 etravirine-weighted genotypic score provides better agreement with the phenotypic susceptibility categorization than the count of the 2007 etravirine RAMs. In particular, a higher proportion of samples within the genotypic category of ‘highest response (92.4%)’ according to the weighted score were also classified as phenotypically susceptible than with the 2007 etravirine RAMs classification system (73.2%). The association between etravirine susceptibility, as defined by the old and new genotypic interpretation system, and virologic response is shown in Fig. 1(d). The new weighted genotypic score showed strong concordance with two other independent genotypic interpretation systems (Fig. 1d), as illustrated by similar area under the curve (AUC) values [0.683, 0.644 and 0.636 for the Tibotec (Tibotec BVBA, Mechelen, Belgium), Monogram (Monogram Biosciences, Inc., South San Francisco, California, USA) and Stanford algorithms (Stanford University, Stanford, California, USA), respectively]. Statistical comparison of these AUC values is hampered by the fact that for both Tibotec interpretation systems, the AUC values were determined using the same dataset as that used to develop the Tibotec-weighted score. However, although comparisons with other (non-DUET-based) interpretation systems are difficult to make, the weighted genotypic score described here (based on 17 RAMs) showed a statistically significant improvement in predicting virologic response versus the initial system, using 13 RAMs without weighting (P = 0.0087).
Effect of K103N on virologic response to etravirine
The NNRTI RAM K103N has been identified as one of the most prevalent NNRTI RAMs; therefore, it was important to analyze the effect of K103N on virologic response to etravirine.
K103N was also the most prevalent NNRTI RAM in patients participating in the DUET studies (n = 388, 32%). Of note, however, the list of etravirine RAMs does not include K103N, as the presence of K103N had no effect on virologic response to etravirine. Proportions of responders in the overall non virologic-failure excluded, not de-novo enfuvirtide population receiving etravirine and in the subgroup with the K103N mutation were 62.0 and 69.0%, respectively. RRs of 95.0% (19/20) and 87.5% (42/48) were reported in patients with the K103N mutation without any of the other NNRTI RAMs and in those with the K103N mutation without any of the 17 etravirine RAMs, respectively (Fig. 1e). These RRs are higher than the RR observed in the subgroup of patients with no detectable NNRTI RAMs at baseline.
It is important to be able to predict whether a treatment-experienced, HIV-infected patient will respond to a particular antiretroviral drug. Resistance testing has been shown to be a valuable tool in the management of antiretroviral therapy, with studies [32,33] showing a beneficial influence on the selection of therapy in treatment-experienced patients, and such testing is now recommended internationally [15–17]. Whereas, the presence of a single mutation is sufficient to affect the virologic response to efavirenz or nevirapine, the DUET data presented here have demonstrated that the resistance profile of etravirine is more complex.
The impact of baseline resistance parameters on virologic response to etravirine was determined, using genotypic and phenotypic data from the pooled DUET trials together with data from a panel of NNRTI-resistant HIV-1 recombinant clinical isolates. The analysis from the DUET studies has enabled us to define CCOs for etravirine and to determine the in-vivo genotypic profile of etravirine by identifying 17 RAMs and establishing a weighted genotypic score to show the relationship between genotypic and phenotypic interpretation systems, and to relate both approaches to virologic response.
CCOs provide the most relevant method for interpreting phenotypic resistance data, whereas characterization and clinical interpretation of specific mutations are important to determine the extent of resistance. The lower and preliminary upper CCO identified for etravirine (3.0 and 13.0, respectively) using Antivirogram (Virco BVBA) phenotypic susceptibility testing provide phenotypic guidance for the optimal use of etravirine in treatment-experienced, HIV-1-infected patients. As a formal upper CCO could not be established, the fold change in EC50 value of 13.0 can be used for practical purposes, as it is not clear whether continued activity would be observed above this value in a setting outside of DUET. CCO analyses have also been performed on samples from DUET patients using Monogram's PhenoSense HIV test (Monogram Biosciences, Inc.) using different statistical approaches, resulting in a lower and upper CCO of 2.9 and 10.0, respectively, for etravirine . This is the first time to our knowledge that CCOs have been established for an NNRTI. One of the reasons why this has been possible is that the dynamic range of measurable phenotypic resistance to etravirine is characterized by a continuum of fold change in EC50 values, rather than by two discrete groups (i.e. low fold change in EC50 = susceptible and high fold change in EC50 = resistant), as is characteristic for efavirenz and nevirapine.
Initially, 13 etravirine RAMs were identified, with four additional RAMs added upon further analysis. To assess the different impact of each of these etravirine RAMs on the virologic response to etravirine, relative weight factors were assigned to each mutation. Addition of all the individual weight factors for each etravirine RAM present in a patient's virus generates the etravirine-weighted genotypic score. The development of this score may improve the predictive value of genotypic resistance testing with regard to virologic response to etravirine, with an AUC value (receiver operating characteristic curve) of 0.683 as compared with 0.654 for the 13 etravirine RAMs.
The K103N and Y181C mutations frequently emerge in patients failing on first-generation NNRTIs, with K103N tending to emerge more in patients failing efavirenz and Y181C in patients failing nevirapine . Although K103N was the most prevalent NNRTI RAM noted at baseline in the DUET studies, overall it had no effect on the virologic response to etravirine. Furthermore, when it occurred in isolation, K103N was associated with a greater virologic response to etravirine than when it occurred with other NNRTI or etravirine RAMs. A preliminary analysis of the potential hypersusceptibility effect of K103N on etravirine phenotypic susceptibility yielded negative results, but the number of samples analyzed was limited . An independent study in NNRTI-experienced patients treated with an etravirine-containing regimen demonstrated that the presence of K103N was associated with a better virologic response to etravirine . The increased RR may be relevant in settings of early treatment failure with efavirenz or nevirapine, in which K103N may be the only NNRTI mutation expressed, and deserves further study. The mutation Y181C was allocated a weight factor of 2.5. Occurring on its own, virologic response remained substantially above that for placebo (62.0 versus 37.5% with <50 HIV-1 RNA copies/ml); virologic response then progressively declined when Y181C was combined with other etravirine RAMs (data not shown).
Other scoring systems based on the relationship between genotype and phenotype have been developed [e.g. Monogram, which included 30 etravirine RAMs , and Stanford, which included 41 etravirine RAMs ], each with their respective weight factors. Comparison of the weighted genotypic score system developed by Tibotec with the scores developed by Monogram and Stanford indicates strong concordance between these genotypic interpretation systems and with the etravirine phenotype [18,19].
The Tibotec, Monogram and Stanford scoring systems all effectively predicted virologic response to etravirine, with 74.2, 73.2 and 71.8% of patients with a sensitive score, respectively, achieving viral load below 50 HIV-1 RNA copies/ml at week 24 in DUET. For comparison, the proportion of responders among those with a sensitive phenotype (etravirine fold change in EC50 ≤3) was 70.6%. Each system can be effectively used to predict virologic response to an etravirine-containing regimen in treatment-experienced patients. Recently, an independent evaluation of the list of 13 etravirine RAMs (which is used by the French ANRS) and the two etravirine-weighted genotypic scores developed by Tibotec and Monogram was reported, using a database containing sequences from treatment-naive and NNRTI-experienced patients . The authors concluded that the weighted scores perform better especially for samples with partial etravirine resistance.
An important limitation of this study includes the fact that the etravirine-weighted genotypic score and the CCOs have been derived from clinical studies in which darunavir or ritonavir was always a component of the regimen. Hence, caution should be exercised in extrapolating these findings to clinical settings in which etravirine-based regimens are to be constructed that include protease inhibitors other than darunavir or ritonavir, or new classes of anti-HIV drugs. Another limitation of this analysis is that mutations occurring at very low prevalence that could nonetheless potentially impact etravirine resistance are not accounted for in the weighted genotypic score. Ongoing and future clinical trials, as well as independent analyses in various clinical settings, will further contribute to the refinement of the etravirine resistance profile. Further studies are also required to assess the performance of the etravirine-weighted genotypic score in different HIV-1 subtypes (94% of patients in DUET harbored HIV-1 subtype B), although it was shown that the HIV subtype did not affect virologic outcome . Finally, further studies would be required to analyze whether specific combinations of certain mutations may have synergistic (or antagonistic) effects on etravirine resistance.
The present study has shown the relationship between phenotypic resistance and virologic response to etravirine based on the pooled results of two major phase III clinical studies. In addition, the updated etravirine RAMs list and developed weighted scoring system described for etravirine have both improved the predictive value of genotypic resistance testing with regard to virologic response and the discordance between interpretations of genotypic and phenotypic susceptibility to etravirine. Overall, the etravirine CCOs and the etravirine-weighted genotypic scoring system described here optimize interpretation of the phenotypic and genotypic susceptibility to etravirine and will be valuable tools for clinicians seeking guidance for the use of etravirine in treatment-experienced patients.
The authors received medical writing support from Gardiner-Caldwell Communications (GCC) Ltd., Macclesfield, UK, which was funded by Tibotec. All authors are employees of Tibotec.
J.V., L.T., H.A., A.H., S.N., M.P., M.-P.deB., G.DeS., B.W. and G.P. contributed to the study concept. J.V., L.T., H.A., A.H., S.N., M.P., M.-P.deB. and G.P. contributed to the analysis and interpretation of the data. A.H., S.N. and M.P. performed the statistical analyses. J.V., L.T., A.H. and G.P. contributed to the writing of the manuscript with the support of Gardiner-Caldwell Communications (this support was funded by Tibotec). All authors contributed to the critical reviewing of the manuscript.
Data previously presented in part at the 9th International Congress on Drug Therapy in HIV Infection in 2008 and the 16th International HIV Drug Resistance Workshop in 2008.
1. Delaugerre C, Rohban R, Simon A, Mouroux M, Tricot C, Agher R, et al
. Resistance profile and cross-resistance of HIV-1 among patients failing a nonnucleoside reverse transcriptase inhibitor-containing regimen. J Med Virol 2001; 65:445–448.
2. Riddler SA, Haubrich R, DiRienzo AG, Peeples L, Powderly WG, Klingman KL, et al
. Class-sparing regimens for initial treatment of HIV-1 infection. N Engl J Med 2008; 358:2095–2105.
3. Andries K, Azijn H, Thielemans T, Ludovici D, Kukla M, Heeres J, et al
. TMC125, a novel next-generation nonnucleoside reverse transcriptase inhibitor active against nonnucleoside reverse transcriptase inhibitor-resistant human immunodeficiency virus type 1. Antimicrob Agents Chemother 2004; 48:4680–4686.
4. Vingerhoets J, Azijn H, Fransen E, De Baere I, Smeulders L, Jochmans D, et al
. TMC125 displays a high genetic barrier to the development of resistance: evidence from in vitro selection experiments. J Virol 2005; 79:12773–12782.
5. Das K, Clark AD Jr, Lewi PJ, Heeres J, De Jonge MR, Koymans LM, et al
. Roles of conformational and positional adaptability in structure-based design of TMC125-R165335 (etravirine) and related nonnucleoside reverse transcriptase inhibitors that are highly potent and effective against wild-type and drug-resistant HIV-1 variants. J Med Chem 2004; 47:2550–2560.
6. Katlama C, Haubrich R, Lalezari J, Lazzarin A, Madruga JV, Molina JM, et al
. Efficacy and safety of etravirine in treatment-experienced HIV-1 patients: pooled 48 week analysis of two randomized controlled trials. AIDS 2009; 23:2289–2300.
7. Haubrich R, Schapiro JM, Vingerhoets J, Nijs S, Peeters M, De Smedt G, et al
. Combined darunavir and etravirine resistance analysis of the pooled DUET trials: when can we spare the other new classes
? [Abstract #TUPE0048]. XVIIth International AIDS Conference
; 3–8 August 2008; Mexico City, Mexico; 2008.
8. Clotet B, Bellos N, Molina JM, Cooper D, Goffard JC, Lazzarin A, et al
. Efficacy and safety of darunavir-ritonavir at week 48 in treatment-experienced patients with HIV-1 infection in POWER 1 and 2: a pooled subgroup analysis of data from two randomised trials. Lancet 2007; 369:1169–1178.
9. Molina JM, Cohen C, Katlama C, Grinsztejn B, Timerman A, Pedro Rde J, et al
. Safety and efficacy of darunavir (TMC114) with low-dose ritonavir in treatment-experienced patients: 24-week results of POWER 3. J Acquir Immune Defic Syndr 2007; 46:24–31.
10. Madruga JV, Berger D, McMurchie M, Suter F, Banhegyi D, Ruxrungtham K, et al
. Efficacy and safety of darunavir–ritonavir compared with that of lopinavir–ritonavir at 48 weeks in treatment-experienced HIV-infected patients in TITAN: a randomised controlled phase III trial. Lancet 2007; 370:49–58.
11. Madruga JV, Cahn P, Grinsztejn B, Haubrich R, Lalezari J, Mills A, et al
. Efficacy and safety of TMC125 (etravirine) in treatment-experienced HIV-1-infected patients in DUET-1: 24-week results from a randomised, double-blind, placebo-controlled trial. Lancet 2007; 370:29–38.
12. Lazzarin A, Campbell T, Clotet B, Johnson M, Katlama C, Moll A, et al
. Efficacy and safety of TMC125 (etravirine) in treatment-experienced HIV-1-infected patients in DUET-2: 24-week results from a randomised, double-blind, placebo-controlled trial. Lancet 2007; 370:39–48.
13. Tibotec, Inc. Intelence (etravirine) tablets. Full prescribing information. Raritan, New Jersey, USA: Tibotec, Inc.; 2008.
15. Panel on Antiretroviral Guidelines for Adults and Adolescents. Guidelines for the use of antiretroviral agents in HIV-1-infected adults and adolescents. DHHS Panel on Clinical Practices for Treatment of HIV Infection; 2008. www.aidsinfo.nih.gov/Guidelines
. [Accessed 17 December 2008]
16. Hammer SM, Eron JJ Jr, Reiss P, Schooley RT, Thompson MA, Walmsley S, et al
. Antiretroviral treatment of adult HIV infection: 2008 recommendations of the International AIDS Society-USA panel. JAMA 2008; 300:555–570.
18. Tambuyzer L, Azijn H, Rimsky LT, Vingerhoets J, Lecocq P, Kraus G, et al
. Compilation and prevalence of mutations associated with resistance to nonnucleoside reverse transcriptase inhibitors. Antivir Ther 2009; 14:103–109.
19. Tambuyzer L, Vingerhoets J, Azijn H, Hoogstoel A, Nijs S, Picchio G. Comparison of two ETR weighted genotypic scores with phenotypic susceptibility and virologic response data
[abstract #114]. 7th European HIV Drug Resistance Workshop
; 25–27 March 2009; Stockholm, Sweden; 2009.
20. Johnson VA, Brun-Vézinet F, Clotet B, Günthard HF, Kuritzkes DR, Pillay D, et al
. Update of the drug resistance mutations in HIV-1. Top HIV Med 2007; 15:119–125.
21. Vingerhoets J, Buelens A, Peeters M, Picchio G, Tambuyzer L, Van Marck H, et al
. Impact of baseline NNRTI mutations on the virological response to TMC125 in the Phase III clinical trials DUET-1 and DUET-2 [abstract #32]. Antivir Ther 2007; 12(Suppl 1):S34.
22. Hertogs K, de Béthune MP, Miller V, Ivens T, Schel P, Van Cauwenberge A, et al
. A rapid method for simultaneous detection of phenotypic resistance to inhibitors of protease and reverse transcriptase in recombinant human immunodeficiency virus type 1 isolates from patients treated with antiretroviral drugs. Antimicrob Agents Chemother 1998; 42:269–276.
23. Pauwels R, Hertogs K, Kemp S, Bloor S, Van Acker K, Hansen J, et al
. Comprehensive HIV drug resistance monitoring using rapid, high-throughput phenotypic and genotypic assays with correlative data analysis
[abstract #51]. 2nd International Workshop HIV Drug Resistance
; 24–27 June 1998; Lake Maggiore, Italy; 1998.
24. Johnson VA, Brun-Vézinet F, Clotet B, Kuritzkes DR, Pillay D, Schapiro JM, et al
. Update of the drug resistance mutations in HIV-1. Top HIV Med 2005; 13:125–131.
25. Vermeiren H, Van Craenenbroeck E, Alen P, Bacheler L, Picchio G, Lecocq P. Prediction of HIV-1 drug susceptibility phenotype from the viral genotype using linear regression modeling. J Virol Methods 2007; 145:47–55.
26. Katlama C, Gatell JM, Molina JM, Peeters M, Vingerhoets J, Woodfall B. Pooled 24-week results of DUET-1 and DUET-2: efficacy of TMC125 (etravirine; ETR) in treatment-experienced HIV-1-infected patients
[abstract #73/11]. 11th European AIDS Conference
; 24–27 October 2007; Madrid, Spain; 2007.
27. Trottier B, Mills A, Cahn P, Clotet B, Grinsztejn B, Towner W, et al
. Durable efficacy and safety of etravirine (ETR; TMC125) in treatment-experienced, HIV-1-infected patients: pooled week 96 results from the phase III DUET-1 and DUET-2 trials
[abstract #P148]. 18th Annual Canadian Conference on HIV/AIDS Research
; 23–26 April 2009; Vancouver, British Columbia, Canada; 2009.
28. De Meyer S, Vangeneugden T, Van Baelen B, De Paepe E, Van Marck H, Picchio G, et al
. Resistance profile of darunavir: combined 24-week results from the POWER trials. AIDS Res Hum Retroviruses 2008; 24:379–388.
29. Vingerhoets J, Peeters M, Azijn H, Tambuyzer L, Hoogstoel A, Nijs S, et al
. An update of the list of NNRTI mutations associated with decreased virologic response to etravirine (ETR): multivariate analyses on the pooled DUET-1 and DUET-2 clinical trial data [abstract 24]. Antivir Ther 2008; 13(Suppl 3):A26.
30. Hachiya A, Kodama EN, Sarafianos SG, Schuckmann MM, Sakagami Y, Matsuoka M, et al
. Amino acid mutation N348I in the connection subdomain of human immunodeficiency virus type 1 reverse transcriptase confers multiclass resistance to nucleoside and nonnucleoside reverse transcriptase inhibitors. J Virology 2008; 82:3261–3270.
31. Picchio G, Vingerhoets J, Staes M, Tambuyzer L, Bacheler L, Pattery T, de Béthune MP. Prevalence of TMC125 resistance-associated mutations in a large panel of clinical isolates
[abstract #866]. 15th Conference on Retroviruses and Opportunistic Infections
; 3–6 February 2008; Boston, Massachusetts, USA; 2008.
32. Tural C, Ruiz L, Holtzer C, Schapiro J, Viciana P, González J, et al
. Clinical utility of HIV-1 genotyping and expert advice: the Havana trial. AIDS 2002; 16:209–218.
33. Cingolani A, Antinori A, Rizzo MG, Murri R, Ammassari A, Baldini F, et al
. Usefulness of monitoring HIV drug resistance and adherence in individuals failing highly active antiretroviral therapy: a randomized study (ARGENTA). AIDS 2002; 16:369–379.
34. Coakley E, Chappey C, Benhamida J, Napolitano L, Tambuyzer L, Vingerhoets J, et al
. Defining the upper and lower clinical cutoffs for etravirine in the PhenoSense HIV assay
[abstract #687]. 16th Conference on Retroviruses and Opportunistic Infections
; 8–11 February 2009; Montreal, Quebec, Canada; 2009.
35. Bannister WP, Ruiz L, Cozzi-Lepri A, Mocroft A, Kirk O, Staszewski S, et al
. Comparison of genotypic resistance profiles and virological response between patients starting nevirapine and efavirenz in EuroSIDA. AIDS 2008; 22:367–376.
36. Picchio G, Vingerhoets J, Parkin N, Azijn H, de Bethune M-P. Nucleoside-associated mutations cause hypersusceptibility to etravirine (ETR)
[abstract #23]. 17th International HIV Drug Resistance Workshop
; 10–14 June 2008; Sitges, Spain; 2008.
37. Marcelin AG, Flandre P, Descamps D, Morand-Joubert L, Charpentier C, Izopet J, et al
. Factors associated with virological response to etravirine in non nucleoside reverse transcriptase inhibitor experienced HIV-1 infected patients. Antimicrob Agents Chemother 2010; 54:72–77.
38. Benhamida J, Chappey C, Coakley E, Parkin NT. HIV-1 genotype algorithms for prediction of etravirine susceptibility: novel mutations and weighting factors identified through correlations to phenotype [abstract #130]. Antivir Ther 2008; 13(Suppl 3):A142.
40. Cotte L, Trabaud MA, Tardy JC, Brochier C, Gilibert RP, Miailhes P, et al
. Prediction of the virological response to etravirine in clinical practice: comparison of three genotype algorithms. J Med Virol 2009; 81:672–677.
41. Vingerhoets J, Azijn A, Tambuyzer L, Dierynck I, De Meyer S, Rimsky L, et al
. Activity of etravirine on different HIV-1 subtypes: Week 48 data of the pooled DUET trials and in vitro susceptibility in treatment-naive patients
[abstract #P189]. Ninth International Congress on Drug Therapy in HIV Infection
; 9–13 November 2008; Glasgow, UK; 2008.
© 2010 Lippincott Williams & Wilkins, Inc.