In this study, we observed a VR (defined as a decrease of at least 1.5 log10 copies/ml and/or HIV-1 RNA at <50 copies/ml at month 2) in 82% of the patients studied. The following NNRTI mutations were associated with a decreased VR to ETR: Y181V, V179I, V106I, Y188L, and E138A. The K103N mutation had no effect and was even associated with a better VR to ETR. In the multivariate analysis, some variables were retained as independently positively associated with the VR, including the number of new drugs used in combination with ETR for the first time and the K103N mutation, in addition to others negatively associated with the VR, such as mutations Y181V and E138A.
The VR was defined here as a decrease of at least 1.5 log
10 copies/ml and/or HIV-1 RNA at <50 copies/ml at month 2, which is a criterion currently used in such studies (
7,
14-
16). Month 2 has been chosen as an endpoint because the impact of mutations on the response needs to be measured early after initiation of a new agent to avoid possible interference of drug discontinuation, dropout, toxicity, and adherence to the regimen in the VR.
From the 17 ETR RAMs involved in the weighted genotypic score, only 3 of them were associated in this study with a decreased VR to ETR in univariate analysis. A certain number of mutations had a very low frequency in our study, limiting the power to detect their potential effect on the virological response. For example, the following 8 mutations were found in <5% of patients: Y181I/V, K101P, M230L, G190S, K103H, and V179D/T. Of note, nine mutations (K101N, K103T, V179F, Y188C, G190C, F227C, Y318F, N348I, and N348T) were not found in our sample. The Y181V has been previously described to have a high weight (3.0) and both V106I and E138A a medium weight (1.5) on VR to ETR (
28). Two mutations not previously included in the 17 ETR RAMs have been identified in this study: V179I and Y188L. In vitro selection experiments identified mutations selected by ETR that included known NNRTI-associated mutations L100I, Y181C, G190E, M230L, and Y318F and the novel mutations V179I and V179F (
21,
26). Thus, it is understandable that V179I has been associated with a decreased VR to ETR since it can be selected in vitro by ETR. Moreover, it has been shown in the DUET trials that V179I can emerge upon virological failure in patients on an ETR-containing regimen (
24). Y188L is a NNRTI resistance mutation already known that can be selected in case of EFV failure, and viruses with this mutation in the RT gene show reduced in vitro susceptibility to EFV and NVP (
3). Moreover, Y188L was previously described in a study to have a potential role in ETR resistance (
4).
K103N is the most commonly reported resistance mutation for NNRTIs arising from clinical use of this drug class (
12). A number of structures of the RT K103N mutant have been reported (
5,
11,
20). An early suggestion of how this mutation can give rise to resistance was based on the crystal structure of unliganded HIV-1 RT into which the K103N change was modeled. This work showed that a hydrogen bond could be made between the hydroxyl group of the Tyr188 side chain in the unliganded down position and the Asn103 amide group. Such an interaction would have the effect of stabilizing the apo-RT conformation and hence create an energy barrier to binding NNRTIs, thereby giving a reduction in potency. Modeling of ETR entry into the NNRTI using targeted molecular dynamics has led to the suggestion that hydrogen bonding from a cyano substituent may assist in breaking the Asn103-Tyr188 H-bond, thereby explaining the potency of this compound against the K103N mutant RT (
22). In the DUET trials, it has been shown that K103N was not associated with a decreased VR (
27). Our results confirm that the K103N mutation has no effect on VR to ETR and further suggest it may be associated with a better VR when present. Phenotypic studies have shown that ETR retains in vitro efficacy against K103N mutants, but there is no evidence of hypersusceptibility. The positive impact of the K103N mutation on ETR VR should be further investigated, especially the type of mutations associated with K103N that could explain a hypersensitivity to ETR, such as NRTI mutations (
18).
It is noteworthy that certain baseline characteristics of our patients were similar to those of the DUET trial populations, such as the number of antiretroviral drugs received before starting the ETR regimen; the numbers of NRTIs, NNRTIs, and PIs received in the past; the CD4 cell count; and the plasma viral load. However, we did not identify strictly the same mutations impacting the ETR VR. Indeed, this may be due to the use of different patient populations, resistance methodologies, and statistical approaches. These different approaches are nonetheless interesting because this allows identification of mutations in a population that would not be present in another population, thus contributing to enlarging the knowledge of mutations implicated in resistance to a drug.
Our results confirm that previous exposure to NVP was associated with a poorer VR to ETR that is consistent with other studies showing that NVP rather than EFV was associated with an increased number of ETR mutations (
9,
19). Thus, use of EFV may be less likely to lead to ETR resistance than that of NVP. In addition, it has been shown that the duration of initial NNRTI exposure was associated with an increasing likelihood of developing in vitro ETR resistance (
9). Therefore, the choice and treatment duration of the initial NNRTI may affect the subsequent ETR response. Finally, the number of active drugs associated with ETR seems to be a factor strongly impacting the VR to ETR, as it was independently associated with the VR. Prior publications have already indicated the number of new agents to be a common predictor of success (
2,
29). In our study, in patients receiving 3 active drugs in their regimen comprising ETR, 95% of patients were responders. This result, combined with the fact that the use of a boosted PI in combination with ETR was not associated with a better VR, suggests that the most important condition is the number of active drugs in the ETR regimen regardless of the therapeutic class. However, as the virological and clinical efficacy of ETR has been demonstrated in combination with boosted darunavir in the DUET trials, the use of ETR in combination with a boosted PI should be prescribed when possible.
In conclusion, in this population of NNRTI-experienced patients, ETR was associated with high response rates. Factors associated with a better VR to ETR were the number of new drugs (among RAL, DRV, or T20) used for the first time in combination with ETR and the presence of K103N. The use of a boosted PI in combination with ETR was not associated with a better VR, suggesting that the most important factor is the use of active drugs regardless of the therapeutic class. Mutations Y181V and E138A were independently associated with poor VR, whereas no effect on VR was observed with Y181C. The positive impact of K103N mutation on ETR VR should be further investigated, especially the type of mutations associated with K103N that could explain a hypersensitivity to ETR, such as NRTI mutations.