Pre-existent NRTI and NNRTI resistance impacts on maintenance of virological suppression in HIV-1-infected patients who switch to a tenofovir/emtricitabine/rilpivirine single-tablet regimen

Introduction

NNRTIs played an important role for several years in the first-line treatment of HIV-1 infection—before the recent introduction of integrase inhibitors (INIs) in the antiretroviral armamentarium—and remain the first option in low-income countries.¹

Indeed, despite some concerns about the low genetic barrier to resistance development and central nervous system toxicity,^2,3 NNRTIs offer a high efficacy combined with the advantage that they can be administered as part of a single-tablet regimen (STR) in many patients.⁴ Moreover, the introduction of the second-generation NNRTI rilpivirine has brought about important improvements in treating patients as a first-line regimen or a switching regimen for simplification/convenience reasons.^5–7

Rilpivirine administered as an STR in combination with tenofovir and emtricitabine (TDF/FTC/RPV) is currently recommended for first-line treatment of patients with plasma HIV-1 RNA <100 000 copies/mL without NRTI and NNRTI resistance mutations.^8–10 TDF/FTC/RPV as an STR also represents an attractive option for virologically suppressed treatment-experienced patients. In this regard, only one randomized clinical trial and few studies from clinical practice have evaluated this option.^6,11,12 In particular, the SPIRIT trial demonstrated that switching virologically suppressed patients from a regimen containing a ritonavir-boosted PI (PI/r) plus two NRTIs to TDF/FTC/RPV ensured the maintenance of virological suppression (VS).⁶ However, findings from clinical practice showed that the presence of resistance to NRTIs and/or NNRTIs or previous virological failures with these drug classes before TDF/FTC/RPV switching are associated with a higher propensity of experiencing virological failure.¹²

Another aspect that should be considered in treatment simplification strategy treatment is the role of pre-therapy viraemia before starting first-line treatment. Indeed, recent findings have demonstrated that a high pre-therapy viraemia before first-line treatment is associated with a higher rate of virological rebound (VR) after the achievement of VS.^13,14 However, the role of high pre-therapy viraemia on VR in patients switching therapy with undetectable plasma HIV-1 RNA (<50 copies/mL) is still unknown.

Based on these considerations, the aim of the present study was to evaluate, in clinical practice, the impact of pre-existent resistance (pRes) and pre-therapy viraemia on the maintenance of VS in drug-treated HIV-1-infected patients with viraemia <50 copies/mL switching to a TDF/FTC/RPV STR, for whom at least one previous genotypic resistance test (GRT) was available.

Materials and methods

Patients

Virologically suppressed HIV-1-infected patients switching to a TDF/FTC/RPV STR in several clinical centres across north and central Italy were retrospectively selected on the basis of the following criteria: (i) age ≥18 years; (ii) plasma HIV-1 RNA <50 copies/mL at the time of starting therapy; (iii) at least one viraemia measurement available after starting TDF/FTC/RPV; (iv) CD4 cell count available before therapy switching; (v) complete information about therapeutic history; and (vi) availability of at least one plasma protease/reverse transcriptase GRT before switching. Patients were censored at the last viraemia measurement available before TDF/FTC/RPV discontinuation or treatment interruption. The time of virological undetectability before starting TDF/FTC/RPV was defined as the period where patients remained persistently with viraemia undetectable (plasma HIV-1 RNA <50 copies/mL), or experiencing sporadic viral blips (defined as a plasma HIV-1 RNA value ranging between 50 and 999 copies/mL preceded and followed by another value below the assay limit of 50 copies/mL).¹⁵

Ethics

Approval by an ethics committee was deemed unnecessary for all patients evaluated for diagnostic purpose according to Italian law (articles 6 and 9, legislative decree 211/2003). This study was conducted on anonymous samples (legislative decree 196/2003) and in accordance with the principles of the Declaration of Helsinki. All information, including virological, clinical and therapy data, was recorded in an anonymous database.

HIV-RNA quantification

Plasma viraemia was determined using three different assays (according to the locally available tests): the Roche Cobas CA/CTM v. 2.0 (Mannheim, Germany), the Abbott RealTime HIV-1 (Chicago, IL, USA) and the bDNA version 3.0 (Bayer Corporation, Diagnostics Division, Tarrytown, NY, USA) as previously described.¹³

Genotyping

Viral RNA was extracted from 1 mL plasma sample and retrotranscribed/amplified as previously described.^16,17 Sequencing of the entire PR and the first 240/335 amino acids of the RT open reading frame was performed using commercially available kits (ViroSeq HIV-1 Genotyping System, Abbott Molecular, Des Plains, IL, USA; Trugene-HIV-1 Genotyping-Kit, Bayer HealthCare LLC, Tarrytown, NY, USA) or home-made systems, as previously described.^16,17

Resistance evaluation

pRes was evaluated by exploring all plasma GRTs available before starting TDF/FTC/RPV. Patients were stratified in three groups, according to the following characteristics: (i) presence of concomitant resistance to both NRTIs and NNRTIs; (ii) presence of resistance to NRTIs or NNRTIs; and (iii) absence of reverse transcriptase inhibitor (RTI) resistance. Pre-existent resistance-associated mutations (pRAMs) to NRTIs, NNRTIs and PIs panelled by the International Antiviral Society in 2015¹⁸ and Stanford HIV Drug resistance database 2015 (http://hivdb.stanford.edu/pages/download/resistanceMutations_handout.pdf) were evaluated.

Pre-existent genotypic susceptibility score (pGSS) (according to HIVDB v. 7.0.1) at the time of starting therapy was evaluated by considering all plasma GRTs available before starting TDF/FTC/RPV. Patients were stratified into three groups, according to the following pGSS category: (i) virus with full or intermediate resistance to both tenofovir/emtricitabine and rilpivirine; (ii) virus with full or intermediate resistance to tenofovir/emtricitabine or rilpivirine; and (iii) virus fully susceptible to TDF/FTC/RPV.

The prevalence of each pRAM and/or combination of pRAMs considered in all plasma GRTs was compared between patients who maintained VS and those who experienced VR.

Finally, in patients who experienced VR and for whom a subsequent GRT was available, resistance at failure was evaluated according to the resistance lists described above.

Statistical analysis

All the analyses were performed using the statistical open source environment R (v. 3.2.3).

(i) Survival analysis

Kaplan–Meier curves were used to estimate the time and probability of experiencing VR. VR was defined as either a plasma HIV-1 RNA value >50 copies/mL at two consecutive visits or a single plasma HIV-1 RNA value >1000 copies/mL, as previously described.¹⁹ Kaplan–Meier curves were compared by using the Peto and Peto modification of the Gehan–Wilcoxon test.²⁰

Cox regression analysis was performed, by evaluating the proportional hazards assumption, to estimate the predictive impact of pRes and of pGSS on VR by using as confounders the following variables: age, gender, B subtype, risk factor, duration of VS and presence of at least one viral load blip before TDF/FTC/RPV switching, treatment at TDF/FTC/RPV switching, cumulative experience with NNRTIs, baseline and nadir CD4 cell count, number of previous regimens before TDF/FTC/RPV switching.

For a sub-group of patients, beyond the above-mentioned variables, pre-therapy viraemia as drug naive (defined as the last plasma HIV-1 RNA value before starting first-line antiretroviral therapy) was also considered. The following pre-therapy levels were considered: <100 000, 100 000–500 000 and >500 000 copies/mL.

The characteristics of these sub-groups of patients were compared with the characteristics of the overall population, to avoid any potential bias due to missing values in the survival analysis.

(ii) Comparison of pRAMs according to VR

χ² or Fisher’s exact tests were used, when appropriate, to evaluate potential differences in the prevalence of pRAMs between patients who maintained virological undetectability and those who experienced VR.

Results

Patients’ characteristics before the TDF/FTC/RPV switch

Overall, 309 patients were included in the analysis. Baseline characteristics are summarized in Table 1. Before switching, patients had already been under VS for a median (IQR) of 21 (6–50) months. During this undetectability period, 71 (23%) patients experienced at least one viral blip. According to all available GRTs, 81.6% of patients did not show any RTI pRAM before the TDF/FTC/RPV switching. Among the 57 (18.4%) patients with previous RTI resistance, 18 showed resistance to both NRTIs and NNRTIs, while 39 patients showed resistance to only one of these two classes. By evaluating resistance according to pGSS, the majority of patients (86.4%) harboured a virus fully susceptible to TDF/FTC/RPV before switching, whereas 31 (10%) harboured a virus with full/intermediate resistance to tenofovir/emtricitabine or rilpivirine, and 11 (3.6%) patients harboured a virus with full/intermediate resistance to all the three drugs contained in the regimen (Table 1).

Survival analysis for the evaluation of VR

By 72 weeks after the TDF/FTC/RPV switch, the overall probability of VR was low (11.3%), with a median (IQR) viral load at rebound of 2.6 (2.3–3.7) log₁₀ copies/mL. Patients having pRes to both NRTIs and NNRTIs showed a significantly higher probability of experiencing VR compared with patients with pRes to only one of these classes or without RTI pRes (concomitant NRTI and NNRTI pRes: 39.2%; NRTI or NNRTI pRes: 11.5%; no RTI pRes: 9.4%, P < 0.0001; Figure 1a).

Figure 1

Kaplan–Meier estimates of the probability of experiencing VR. Kaplan–Meier estimates at 72 weeks were performed by stratifying patients according to pre-existent resistance (a) and genotypic susceptibility (b), evaluated by considering all plasma GRTs available before TDF/FTC/RPV switching. Patients were followed until the last viraemia measurement available under TDF/FTC/RPV. P values were calculated by using the Peto and Peto modification of the Gehan–Wilcoxon test. TDF, tenofovir; FTC, emtricitabine; RPV, rilpivirine; GRTs, genotypic resistance tests; VR, virological rebound.

Open in new tabDownload slide

Cox models confirmed the impact of concomitant NRTI/NNRTI pRes on VR. In particular, by multivariable analysis, patients with concomitant NRTI/NNRTI pRes had a higher adjusted hazard ratio (AHR) of experiencing VR compared with those without RTI pRes [AHR (95% CI) = 5.80 (1.50–22.38), P = 0.011; Table 2].

Based on the pGSS, patients having a virus with full/intermediate resistance to both tenofovir/emtricitabine and rilpivirine showed a significantly higher probability of experiencing VR compared with those having a virus with full/intermediate resistance to tenofovir/emtricitabine or rilpivirine, and those having a fully susceptible virus (36.4% versus 17.8% versus 9.7%, P < 0.001; Figure 1b). Cox models confirmed this finding. In particular, by multivariable analysis, patients harbouring a virus with full/intermediate resistance to both tenofovir/emtricitabine and rilpivirine had a higher AHR compared with patients harbouring a virus fully susceptible to TDF/FTC/RPV [AHR (95% CI) = 5.04 (1.17–21.63), P = 0.030; Table 2].

Impact of pre-therapy viraemia as drug naive on VR

Pre-therapy viraemia was available for 229 (74.1%) patients. No significant statistical differences in demographic, viro-immunological and pRes characteristics were found by comparing the subset of 229 patients and the overall population (data not shown). Among these 229 patients, 101 (44.1%), 80 (34.9%) and 48 (21%) had pre-therapy viraemia of <100 000, 100 000–500 000 and >500 000 copies/mL, respectively.

Kaplan–Meier estimates showed that by 72 weeks after TDF/FTC/RPV switching, the probability of VR was significantly higher in patients with pre-therapy viraemia >500 000 copies/mL compared with those with pre-therapy viraemia ranging between 100 000 and 500 000 copies/mL and those with pre-therapy viraemia <100 000 copies/mL (16% versus 9.3% versus 4.8%, respectively, P = 0.009, by Gehan–Wilcoxon test).

Cox models confirmed this finding. In particular, by multivariable analysis, patients with pre-therapy viraemia >500 000 copies/mL had a significantly higher AHR of experiencing VR compared with those with pre-therapy viraemia <100 000 copies/mL, regardless of the resistance indicator (pRes or pGSS) used in the adjustments [AHR (95% CI), model with pRes: 5.69 (1.31–24.74), P = 0.020; AHR (95% CI), model with pGSS: 4.32 (1.02–18.33), P = 0.047; Table S1, available as Supplementary data at JAC Online].

Prevalence of pre-existent RTI resistance before the TDF/FTC/RPV switch according to virological response

An overview of patients with RTI pRAMs before the TDF/FTC/RPV switch is reported in Table 3. As indicated in bold in Table 3, the combination of at least one NRTI pRAM [including M184V/I and/or thymidine analogue mutations (TAMs) and/or other NRTI mutations] and at least one NNRTI pRAM (including K103N and/or rilpivirine RAMs and/or other NNRTI RAMs) was more frequently observed among patients experiencing VR (7/29, 24.1%) compared with those who maintained VS [11/280 (3.9%), P < 0.001].

By contrast, the proportion of patients with pRAMs for only one drug class (NRTI or NNRTI; underlined in Table 3) was 10.3% (3/29 patients) among patients experiencing VR and 12.9% (36/280 patients) among patients maintaining virological success (P = 1.000).

Evaluation of resistance detected after VR under a TDF/FTC/RPV STR

Among the 29 patients experiencing VR, 13 had an available GRT after rebound. Of these, 7/13 (53.8%) showed resistance to at least one drug in the regimen (Table 4). In particular, five patients with pRAMs accumulated additional mutations, one patient with pRAMs maintained only E138G mutation and one patient without any previous resistance acquired novel RAMs.

Discussion

In the present study, we evaluated the role in clinical practice of pre-existent resistance and pre-therapy viraemia on the maintenance of VS in patients switching to a TDF/FTC/RPV STR with viraemia <50 copies/mL. By 72 weeks after switching regimen, patients maintained a very high rate of VS (89%), confirming, at longer observation time, the efficacy data already available.^{11,12,21–23} Based on the availability of the complete information about previous resistance, we found that among the few patients who experienced VR, the presence of pRes (evaluated by both class resistance and pGSS) plays an important role in losing virological control.

So far, only a few studies (both randomized and observational) have evaluated pRes as a predictor of virological failure. The findings of these studies can be considered controversial because they are often based on incomplete resistance data retrieved only for subgroups of patients.^6,12,22,23

In the SPIRIT trial, the rate of VS in patients with pRes remained high; in particular, patients with single mutants had no virological failure through week 48. Moreover, the majority of patients with K103N (about 95%) maintained undetectability.⁶ However, it should be taken into account that resistance analysis was underestimated because patients with pRes to tenofovir, emtricitabine and/or rilpivirine were excluded, and only some patients with K103N or having RAMs detected at baseline GRTs from proviral DNA were analysed.⁶

In the present study, by analysing the pattern of pre-existent mutations, we found that the combination of several NNRTI (including K103N and rilpivirine RAMs) and NRTI mutations (including TAMs and M184I/V) showed a higher prevalence in patients who experienced VR compared with those who maintained VS, whereas, pRes to only one drug class was not associated with VR.

The presence of K103N may cause uncertainty in clinicians in the selection of treatment containing second-generation NNRTIs (such as rilpivirine and etravirine). Indeed, this mutation is a marker of previous failure with efavirenz or nevirapine²⁴ and might be a marker of potential archived resistance to NNRTIs.

In our population, among patients who experienced VR, we found that the pre-existent K103N was always associated with NNRTI and/or NRTI RAMs, present in most cases in a complex pattern. In patients maintaining VS, however, K103N was usually found only together with other NRTI mutations or alone. A recent study confirmed this finding, demonstrating that the previous presence of the single K103N does not affect virological response in patients switching to a TDF/FTC/RPV STR.²⁵

Similarly, in patients experiencing VR we never found the pre-existent mutation M184V alone, but always in combination with other NRTI and/or NNRTI mutations. This finding is apparently in contrast with that described by Gazaignes et al.,¹² who showed an association between the previous presence of M184I/V and the risk of losing virological control under a TDF/FTC/RPV STR. This apparent discrepancy can be explained by the fact that in the study by Gazaignes et al.¹² resistance information was available only for a sub-group of patients and the patterns of pre-existent mutations were not extensively explored.¹² It should be taken into account that, even though M184V confers a high level of resistance to emtricitabine, it confers hypersusceptibility to tenofovir.²⁶ Thus, the role of this mutation in the TDF/FTC/RPV response might be difficult to explain without a phenotypic evaluation.

Another important finding observed in our study is the negative role played by very high pre-therapy viraemia before first-line treatment in the maintenance of VS. We found that the probability of VR was significantly higher (16%) for patients who started their first-line treatment with a viraemia >500 000 copies/mL compared with those with lower viraemia levels (<10%, P = 0.009).

As far as we know, the present study is the first to describe the impact of pre-therapy viraemia on therapy switches. Our finding of the adverse effect of high pre-therapy viraemia on the TDF/FTC/RPV response might be useful to distinguish patients with a potential higher probability of virological failure after switching therapy.

Our study might have some limitations. Firstly, our observations about the role of specific pRAMs on virological response might be only descriptive because of the low number of VR events (only 29 out 309) found in our population. Certainly, this finding reflects good clinical practice, where clinicians decide to simplify treatment only for selected patients (e.g. switching from a successful first-line treatment, or without previous resistance and/or virological failures). Furthermore, we cannot extrapolate any data about the selection of resistance under TDF/FTC/RPV because only 13 patients were tested for resistance development after VR. Further studies with a larger number of patients might better elucidate the role of the specific pRAMs involved in rilpivirine response and the resistance evolution at virological failure.

Moreover, even though we found an effect of high pre-therapy viraemia in drug-naive patients on TDF/FTC/RPV response, we cannot estimate whether this phenomenon is related to the size of the viral reservoir. Further studies evaluating the impact of viral reservoir size (e.g. HIV-DNA level) on virological response might solve this issue.

In conclusion, the present paper shows that switching to TDF/FTC/RPV is effective for properly selected HIV-1-suppressed patients. Patients with previous single rilpivirine RAMs or RAMs to only one RTI class have a low risk of VR, while major concerns arise only regarding patients with previous patterns of NRTI and NNRTI RAMs.

An accurate evaluation of treatment history, previous virological trends and historical resistance might prevent a negative response due to archived resistance and/or a high previous exposure to viral replication.

Acknowledgements

The manuscript was in part presented at the Fifteenth European AIDS Conference, Barcelona, Spain, 2015 (Abstract PE9/6) and at the Fourteenth European Meeting on HIV & Hepatitis—Treatment Strategies & Antiviral Drug Resistance, Rome, Italy, 2016 (Abstract O_06).

We gratefully thank C. Alteri, A. Biddittu, M. Bruni, L. Carioti, L. Fabeni, S. Giannella, A. Giannetti, T. Guenci, D. Pizzi, M. Romani, F. Stazi and V. Serafini for sequencing and data management, together with all the members of the Resistance Group of the National Institute for Infectious Disease ‘Lazzaro Spallanzani’-IRCCS: R. Acinapura, A. Ammassari, A. Antinori, R. Bellagamba, G. Berno, E. Boumis, S. Carta, G. De Carli, P. De Longis, G. D’Offizi, F. Forbici, L. Fabeni, V. Fedele, F. M. Fusco, S. Galati, M. L. Giancola, E. Girardi, C. Gori, R. Libertone, L. Lo Iacono, G. Liuzzi, P. Lorenzini, S. Mosti, V. Neri, E. Nicastri, N. Orchi, C. F. Perno, N. Petrosillo, C. Pinnetti, D. Pizzi, P. Scognamiglio, C. Tommasi, U. Visco-Comandini and M. Zaccarelli, and the Resistance Group of University Hospital Tor Vergata: C. Alteri, M. Andreoni, D. Armenia, A. Bertoli, F. Ceccherini-Silberstein, C. Cerva, L. Dori, E. Gentilotti, S. Giannella, S. Gini, T. Guenci, D. Leoni, G. Maffongelli, V. Malagnino, A. Ricciardi, R. Salpini, M. M. Santoro, L. Sarmati, V. Serafini, V. Svicher, F. Stazi, E. Teti and M. Viscione.

Funding

This work was financially supported by the European Commission Framework 7 Programme (CHAIN, the Collaborative HIV and Anti-HIV Drug Resistance Network, Integrated Project no. 223131), the Italian Ministry of Health (Progetto Ricerca Finalizzata, grant no. RF-2009-1539999), the Italian Ministry of Education, University and Research (MIUR) (Bandiera InterOmics Protocollo PB05 1°) and an unrestricted grant from AVIRALIA foundation.

Transparency declarations

None to declare.

References

Author notes