SEARCH

SEARCH BY CITATION

Keywords:

  • nonnucleoside reverse transcriptase inhibitor mutations;
  • ongoing viraemia;
  • Poisson regression;
  • rate of accumulation

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Background

Virological failure of first-generation nonnucleoside reverse transcriptase inhibitors (NNRTIs) can compromise the efficacy of etravirine as a result of the accumulation of NNRTI resistance mutations. How quickly NNRTI resistance accumulates in patients with a delayed switch from nevirapine or efavirenz despite virological failure, when these drugs are used as a component of combination antiretroviral therapy (cART), remains unclear.

Methods

The rate of NNRTI resistance accumulation was estimated in patients in EuroSIDA with at least two available genotypic resistance tests (GRTs), provided that (1) the date of the first GRT (t0) was after the date of the first virological failure (VF) of an NNRTI, and (2) patients were receiving an NNRTI and HIV RNA was >500 HIV-1 RNA copies/mL in all measurements between GRTs.

Results

A total of 227 patients were included in the study, contributing 467 GRT pairs. At baseline-t0, a median of 3 months after VF, 66% of patients had at least one NNRTI mutation: 103N (34%), 181C (22%) and 190A (20%) were the most common mutations. Overall, 180 additional NNRTI mutations were found to have accumulated over 295 years [1 new/1.6 years; 95% confidence interval (CI) 1.5–1.8]. The rate of accumulation was faster in the first 6 months from VF (1 new/1.1 years), and slower in patients exposed to nevirapine vs. those receiving efavirenz [relative risk (RR) 0.66; 95% CI 0.46–0.95; P=0.03].

Conclusions

There is an initial phase of rapid accumulation of NNRTI mutations close to the time of VF followed by a phase of slower accumulation. We predict that it should take approximately one year of exposure to a virologically failing first-generation NNRTI-based cART regimen to reduce etravirine activity from fully susceptible to intermediate resistant, and possibly longer in patients kept on a failing nevirapine-containing regimen.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Global access to antiretroviral drugs has increased dramatically in recent years [1], and concerns regarding the development of drug resistance remain in both resource-rich and resource-limited settings [2,3]. In resource-limited settings, NNRTIs are a fixed component of first-line combination antiretroviral therapy (cART) [3], but HIV-infected populations typically have little access to virological monitoring and/or genotypic resistance testing, which is likely to result in the accumulation of NNRTI resistance. An improved access to NNRTI drugs for preventing mother-to-child transmission has further complicated this issue. A previous analysis of patients in EuroSIDA focused on the estimation of the rate of accumulation of thymidine analogue mutations (TAMs) in patients kept on zidovudine or stavudine despite a viral load of >500 HIV-1 RNA copies/mL [4,5]. NNRTI resistance accumulation could compromise the efficacy of second-generation NNRTIs (e.g. etravirine [6]) if they ever become available in these settings. Indeed, etravirine has already been used in some resource-limited settings as a component of second-line regimens in patients who could not tolerate protease inhibitors (PIs) [7]. Data on etravirine resistance in patients already exposed to first-generation NNRTIs show that, among 17 mutations in the reverse transcriptase gene, at least three must be present simultaneously in order to reduce etravirine activity, although just two mutations can greatly decrease susceptibility in some cases [7–9]. In addition, this activity is likely to diminish to zero as NNRTI-associated resistance mutations further accumulate. Our analysis is based on data for patients enrolled in clinics in Europe. However, while there are differences in the prevalence of HIV subtypes, some infections and in access to health care between resource-rich and resource-limited settings, there is otherwise generally little evidence of differences between these settings in the damage caused by HIV or the effect of ART [10–12]. Also, many of the patients included in studies in the UK, for example, are originally from Africa, and substantive differences in the response to ART have not been observed between these patients and those with origins in the UK [13]. Thus, the aim of this study was to improve our understanding of the rate of NNRTI resistance accumulation under selection pressure from nevirapine or efavirenz in the presence of a detectable viral load, in order to improve predictions of the activity and potential benefits of subsequent use of etravirine in both resource-rich and resource-limited countries.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Patients

The EuroSIDA study is a prospective, observational, open cohort study of 16 599 HIV-1-infected patients in 102 centres across 31 European countries, Israel and Argentina. The study is described in detail at http://www.cphiv.dk and by Kirk et al. [14]. EuroSIDA requests plasma samples from patients to be collected prospectively every 6 months and stored in a central repository. Patients were included if stored plasmas samples at the time points needed for this analysis were available for them. Retrospective genotypic testing was carried out on these samples. In EuroSIDA, HIV-1 RNA is isolated from patient blood plasma using the QIAamp kit (Qiagen, Barcelona, Spain) and sequence analysis of the HIV-1 reverse transcriptase (RT) and protease (PR) reading frames is performed using the Trugene HIV-1 genotyping Kit (Siemens Healthcare, Barcelona, Spain) and the OpenGene DNA Sequencing System (Bayer, Barcelona, Spain) according to the manufacturer's recommendations. Mutations are identified by comparison against a reference sequence of the subtype B isolate HXB2. Sequences are regularly submitted to GenBank at the time of analysis. Each EuroSIDA participating site has obtained local Institutional Review Board (IRB) approval for contribution to the study.

In this analysis, we included patients who experienced virological failure while receiving an NNRTI-containing regimen [with virological failure defined as occurring at (1) the time of the first viral load >500 HIV-1 RNA copies/mL ≥6 months after starting the NNRTI while still receiving an NNRTI, or (2) the first detection of an International AIDS Society (IAS)-USA NNRTI-associated mutation (see Table S1 for a complete list), whichever occurred earlier] and for whom at least two genotypic resistance tests (GRTs) while still on NNRTI were available after the estimated date of failure. GRTs performed before the estimated date of virological failure were used to estimate the prevalence of NNRTI transmitted resistance. Viral load had to be >500 HIV-1 RNA copies/mL in all measurements between the date of failure and the first GRT and between all subsequent GRTs (including the actual date of the GRT). Data were analysed as pairs of genotypes, and patients with j GRTs (j≥2) contributed j – 1 pairs (e.g. a patient with two eligible genotype tests contributed one pair, a patient with three eligible genotype tests contributed two pairs, etc.). The time of the very first GRT for an individual patient was defined as baseline-t0, while the time of the first GRT in a genotype pair was defined as t0 (so baseline-t0 and t0 coincide for the very first genotype in a patient) and the time of the second GRT in a pair was defined as t1. For this analysis, we also insisted that patients had to be receiving an NNRTI-containing regimen at all times between GRTs in a pair, but no restrictions were imposed on the other drugs (Fig. 1 illustrates a virtual patient who was kept on a nevirapine-containing regimen). Furthermore, to be sure that patients had experienced failure with resistance, we included only those harbouring a virus predicted by the Rega interpretation system (IS) to have reduced susceptibility to at least one of the drugs (not necessarily the NNRTI) received at the first GRT; versions 8.0.1 of the Rega IS for the drugs currently in use in clinical practice and 6.4.1 for the remaining drugs (nonboosted PIs, etc.) were used to predict the number of active drugs in the ART regimen at the time of each GRT [15].

image

Figure 1.  A virtual patient on a nevirapine (NVP)-based failing regimen contributing two genotypic resistance test (GRT) pairs. 3TC, lamivudine; FTC, emtricitabine; LPV/r, lopinavir boosted with ritonavir; TDF, tenofovir; ZDV, zidovudine.

Download figure to PowerPoint

Statistical analysis

Patients' characteristics at t0 were described and average (mean or median) changes in laboratory markers from t0 to t1 were evaluated using simple regression and multilevel modelling, accounting for nonindependence of observations (with similar results).

NNRTI-associated mutations were those currently listed in the IAS-USA report as of December 2009 [16]. We assumed that NNRTI-associated mutations identified at t0 were still present in a patient's body at t1, even if they were not actually identified by the GRT at t1.

The rate of NNRTI resistance accumulation was calculated as number of NNRTI mutations detected at t1 that had not been detected at t0 divided by the time between t0 and t1 [and expressed as a rate per person-years of follow-up (PYFU) with a viral load>500 HIV-1 RNA copies/mL while receiving an NNRTI]. A multivariable Poisson regression model was used to identify independent predictors of both NNRTI resistance accumulation and IAS etravirine-specific mutations. All factors known or thought potentially to be associated with the risk of accumulation of resistance were included in a final multivariable model showing mutually adjusted relative rates (RRs). The full list of predictors included in the multivariable model is shown in Table 3 below. In order to adjust the estimate of the parameters variance to account for the fact that a patient could contribute more than one pair of genotypes, a generalized estimating equation (GEE) model with first-order autoregressive working correlation structures was fitted (but results were robust to the choice of this working matrix) using PROC GENMOD in sas [17,18]. In addition, as there was evidence that the rate of accumulation was dependent on the time difference between GRTs in a pair mainly because of a higher prevalence of short time differences close to the time of virological failure (supporting information, Table S2), the model was further adjusted for categorical covariates indicating the time difference in months between GRTs (1 month difference was chosen as the reference group) as well as geographical variation (one group for each country included). A sensitivity analysis was performed after including only the first GRT pair per patient.

Table 3.  Adjusted relative rates (RRs) of accumulation of resistance mutations after fitting a Poisson regression model [all genotypic resistance test (GRT) pairs; n=467]
CharacteristicsAccumulation of NNRTI mutationsAccumulation of etravirine mutations
Adjusted RR (95% CI)P-valueAdjusted RR (95% CI)P-value
  1. Data are adjusted also for the time difference between GRTs in a pair within a patient and geographical region.

  2. CI, confidence interval; IDU, injecting drug use; NRTI, nucleoside reverse transcriptase inhibitor; NNRTI, nonnucleoside reverse transcriptase inhibitor; PI, protease inhibitor; PI/r, protease inhibitor boosted with ritonavir; t0, time of first genotypic test in a pair; VL, viral load.

Age (years)
Per 10 years older0.99 (0.82, 1.21)0.960.63 (0.45, 0.90)0.01
Gender
 Male1.00 1.00 
 Female0.65 (0.35, 1.24)0.190.89 (0.33, 2.44)0.82
Mode of HIV transmission
 Homosexual contact1.00 1.00 
 IDU1.43 (0.82, 2.48)0.210.44 (0.14, 1.34)0.15
 Heterosexual contact1.09 (0.67, 1.78)0.730.82 (0.36, 1.87)0.64
 Other/unknown0.98 (0.58, 1.64)0.931.02 (0.43, 2.42)0.96
Calendar year of t0
 Per more recent year0.80 (0.69, 0.93)0.0040.79 (0.62, 1.00)0.05
Viral load at t0 (log10 copies/mL)
 Per log10 higher0.80 (0.64, 1.01)0.060.59 (0.38, 0.92)0.02
CD4 count at t0 (cells/μL)
 Per 100 cells/μL higher1.04 (0.93, 1.16)0.520.92 (0.76, 1.12)0.41
Time from last VL≤50 copies/mL or start of NNRTI to t0
 Per year longer0.76 (0.64, 0.91)0.0030.86 (0.62, 1.18)0.34
HIV subtype
 B1.00 1.00 
 Non-B1.57 (0.82, 3.01)0.170.85 (0.30, 2.42)0.77
NNRTI in regimen
 Efavirenz1.00 1.00 
 Nevirapine0.66 (0.46, 0.95)0.030.56 (0.31, 1.02)0.06
No. of drugs failed before t0
 Per one additional0.97 (0.88, 1.07)0.541.04 (0.89, 1.22)0.63
No. of drugs in regimen at t0
 Per one additional1.07 (0.83, 1.38)0.590.64 (0.37, 1.11)0.11
Predicted activity of NNRTI in regimen (Rega)
 Non-active1.00 1.00 
 Half active0.53 (0.07, 4.37)0.560.98 (0.11, 8.46)0.99
 Fully active3.26 (2.27, 4.68)<.0011.41 (0.46, 4.31)0.54
No. of active drugs in addition to NNRTI in regimen (Rega)
 Per one additional drug1.13 (0.90, 1.42)0.280.92 (0.60, 1.40)0.68
Drug classes in regimen in addition to NNRTI
 NRTI only1.00 1.00 
 NRTI and PI0.85 (0.52, 1.38)0.511.11 (0.52, 2.35)0.79
 NRTI and PI/r1.03 (0.52, 2.06)0.932.15 (0.53, 8.69)0.28
 PI/r only0.96 (0.36, 2.57)0.94--

We also simulated a hypothetical situation in which all patients included in the study, at the end of a prolonged period of unsuppressed viraemia while receiving an NNRTI, would be switched to an etravirine-containing regimen which, as a result of the accumulation of NNRTI mutations over t0–t1, would have a certain predicted diminished activity at t1. The Rega IS was again used to derive the predicted susceptibility at both t0 and t1. The difference in etravirine predicted activity between t0 and t1 was calculated, averaged, standardized per time between t0 and t1, and used as a measure of the decrease in susceptibility to etravirine caused by the accumulation of NNRTI resistance.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Patients' characteristics

A total of 227 patients were included in the study, who remained on a virologically failing NNRTI-based regimen and contributed 467 pairs of GRTs, with the following distribution: 124 patients contributed one pair, 55 contributed two pairs, 25 contributed three pairs, nine contributed four pairs and 14 contributed more than four pairs. The breakdown of these contributions is given in Table 1a, which also shows the main characteristics of the target population. Only six of the 35 female patients included (17%) had a history of pregnancy prior to baseline-t0. Two hundred and eighty-eight patients with at least one GRT pair were excluded because there was no evidence that they experienced virological failure because of resistance (supporting information, Table S3).

Table 1.  Characteristics of (a) the study population (n=227) at the time of the first genotypic test (baseline-t0) and (b) the genotype pairs (n=467) at the time of the first test in a pair (t0)
Characteristic 
  • ART, antiretroviral therapy; IDU, injecting drug use; IQR, interquartile range; NRTI, nucleoside reverse transcriptase inhibitor; NNRTI, nonnucleoside reverse transcriptase inhibitor; PI, protease inhibitor; PI/r, protease inhibitor boosted with ritonavir; t0, time of first genotypic test in a pair; VF, virological failure, VL, viral load.

  • *

    Measured only in 48 patients with viral load pre-ART available.

  • Only pairs failing nevirapine.

  • Only pairs failing efavirenz.

(a)
NNRTI in pair [n (%)]
 Efavirenz104 (45.8)
 Nevirapine123 (54.2)
Gender [n (%)]
 Female35 (15.4)
Age (years) [median (range)]40 (20, 71)
Mode of HIV transmission [n (%)]
 Homosexual contact117 (51.5)
 Heterosexual contact35 (15.4)
 IDU30 (13.2)
 Other/unknown45 (19.8)
Viral load (log10 copies/mL) [median (IQR)]4.14 (3.41, 4.72)
CD4 count (cells/μL) [median (IQR)]264 (143, 383)
HIV subtype [n (%)]
 Non-B24 (10.6)
Started the NNRTI from ART-naïve [n (%)]14 (6.2)
Months from VF of NNRTI [n (%)] [median (IQR)]3 (0, 18)
Geographical region [n (%)]
 Belgium11 (4.8)
 Czech Republic/Poland/Romania31 (13.7)
 Estonia/Lithuania2 (0.9)
 France28 (12.3)
 Germany41 (18.1)
 Greece1 (0.4)
 Spain/Portugal24 (10.6)
 Italy12 (5.3)
 Denmark/Sweden53 (23.3)
 Switzerland11 (4.8)
 United Kingdom12 (5.3)
Number of contributing pairs [n (%)]
 1124 (54.6)
 255 (24.2)
 325 (11.0)
 49 (4.0)
>414 (6.2)
(b)
NNRTI in pair [n (%)]
Efavirenz255 (54.6)
Nevirapine212 (45.4)
Time from last VL≤50 copies/mL or start of NNRTI to t0 (months) [median (IQR)]16 (7, 32)
Calendar year of t0 [median (range)]2000 (1997, 2005)
Extent of viral suppression below pre-ART levels (log10 copies/mL) [median (range)]*0.40 (−2.26, 3.30)
Viral load at t0 (log10 copies/mL) [median (IQR)]4.18 (3.45, 4.77)
CD4 count at t0 (cells/μL) [median (IQR)]222 (130, 367)
Time between pair of tests (months) [median (range)]6 (1, 74)
Duration of exposure to nevirapine from NNRTI failure to t0 (months) [median (IQR)]2 (0, 13)
Duration of exposure to efavirenz from NNRTI failure to t0 (months) [median (IQR)]5 (0, 18)
Number of drugs received at t0 [median (IQR)]4 (3, 4)
Drug classes in addition to NNRTI at t0 [n (%)] 
 NRTI only189 (40.5)
 NRTI and PI181 (38.8)
 NRTI and PI/r84 (18.0)
 PI/r only13 (2.8)

Laboratory markers

At t0, the median viral load of the patients was 4.18 log10 copies/mL [interquartile range (IQR) 3.45–4.77 log 10 copies/mL] and the median CD4 count was 222 cells/μL (IQR 130–367 cells/μL). In the 48 patients with a viral load measurement before the initiation of ART, the median viral load suppression below this value at t0 was 0.40 log10 copies/mL (range –2.26 to 3.30 log10 copies/mL; Table 1b), suggesting that HIV was somewhat suppressed compared with its maximum level of replication.

Over the intervals t0–t1 (with a median of 6 months between tests and a median number of two viral load values over this time period), the viral load was observed to be stable {mean change+0.17 [standard deviation (SD) 1.83] logs10 copies/mL per year; P=0.12} and a small increase in CD4 count was found [mean change+21 (SD 312) cells/μL per year; P=0.15]; the changes in these variables were not significantly different from zero. The corresponding figures for 178 patients who received an NNRTI-based regimen without a PI were +0.29 (SD 1.52) copies/mL per year (P=0.01) for viral load and +53 (SD 353) cells/μL per year (P=0.04) for CD4 cell count. There was no difference in the median time between GRTs between patients receiving nevirapine (median 6 months; IQR 3–9 months) and those receiving efavirenz (median 6 months; IQR 3–8.5 months; Wilcoxon test, P=0.73).

Antiretroviral drugs

For 212 pairs the patient was on a nevirapine-containing regimen, and for 255 pairs the patient was on an efavirenz-containing regimen (Table 1b). Only six patients underwent a switch in NNRTI between t0 and t1: four patients switched from nevirapine to efavirenz and two patients did the opposite and were classified according to their NNRTI exposure at t0. At the first GRT in a pair, the median number of drugs in the regimen was 4 (IQR 3–4) and the most frequently used NRTIs at t0 together with the NNRTI were lamivudine (56%), stavudine (49%) and didanosine (36%). At t0, two NRTIs plus either nevirapine or efavirenz were used in 189 (41%) of the pairs while the remaining pairs were on combinations including PIs (Table 1b). The frequency of use of other antiretrovirals besides nevirapine/efavirenz at t1 was similar to that observed at t0 (data not shown), suggesting that these patients had in most cases been kept on the same drugs over t0–t1 despite virological failure.

HIV drug resistance

The median number of NNRTI mutations detected at baseline-t0 was 2 (range 0–8) and the majority of patients (66%) had at least one NNRTI mutation (supporting information, Table S4). For only 36 of the GRT pairs (8%) were no NNRTI mutations detected at both GRTs. In 2% of the patients included in the study, NNRTI mutations were detected at the GRT performed prior to the estimated date of virological failure. Table 2a shows the prevalence of patients with at least one IAS NNRTI mutation, the distribution of individual IAS NNRTI mutations detected in major virus populations at t0 and the estimated proportions at t1. Table 2a also shows the total number of NNRTI mutations (overall and stratified by specific NNRTI drug) at t0 and t1, and the estimate of the rate of accumulation of NNRTI resistance over the observation period.

Table 2.  Nonnucleoside reverse transcriptase inhibitor (NNRTI) resistance accumulation. (a) Number (%) of pairs in which an NNRTI mutation was detected at t0 and t1 and crude rate of accumulation of single mutations; the total number of NNRTI mutations is also shown; (b) predicted susceptibility to NNRTI in a failing regimen at t0 and t1 and estimated accumulation of NNRTI mutations according to susceptibility to the drugs received at t0; (c) mean reduction in Rega-predicted susceptibility to NNRTI overall and stratified by NNRTI use (all patients); (d) mean reduction in Rega-predicted susceptibility to NNRTI overall and stratified by NNRTI use (only patients with first GRT <3 months after the date of virological failure)
(a)
Mutationt0t1New mutations/PYFU*Crude rate of accumulation/100 PYFU (95% CI)
Patients with specific mutation [n (%)]
 90I33 (7.1)37 (7.9)4/2741.5 (0.40–3.7)
 98G60 (12.8)67 (14.3)7/2652.6 (1.1–5.4)
 100I32 (6.9)44 (9.4)12/2774.3 (2.3–7.4)
 101E31 (6.6)39 (8.4)8/2673.0 (1.3–5.8)
 101H5 (1.1)6 (1.3)1/2860.3 (0.0–1.9)
 101P24 (5.1)28 (6.0)4/2891.4 (0.4–3.5)
 103N234 (50.1)277 (59.3)43/15627.6 (20.7–35.3)
 106A36 (7.7)37 (7.9)1/2730.3 (0.0–2.0)
 106I40 (8.6)44 (9.4)4/2781.4 (0.4–3.6)
 106M20 (4.3)21 (4.5)1/2900.3 (0.0–1.9)
 108I82 (17.6)99 (21.2)17/2516.7 (4.0–10.6)
 138A5 (1.1)6 (1.3)1/2930.3 (0.0–18.9)
 179D11 (2.4)12 (2.6)1/2880.3 (0.0–19.1)
 179F0 (0.0)1 (0.2)1/2950.3 (0.0–18.7)
 179T0 (0.0)0 (0.0)0/2950.0 (0.0–12.4)
 181C173 (37.0)197 (42.2)24/19612.2 (8.0–17.7)
 181I25 (5.4)31 (6.6)6/2832.1 (0.8–4.6)
 181V0 (0.0)1 (0.2)1/2950.3 (0.0–18.7)
 188C0 (0.0)1 (0.2)1/2950.3 (0.0–18.7)
 188H13 (2.8)17 (3.6)4/2871.4 (0.4–3.5)
 188L16 (3.4)23 (4.9)7/2862.4 (1.0–5.0)
 190S18 (3.9)23 (4.9)5/2841.8 (0.6–4.1)
 190A151 (32.3)170 (36.4)19/2029.4 (5.8–14.3)
 225H7 (1.5)12 (2.6)5/2901.7 (0.6–4.0)
 230L6 (1.3)9 (1.9)3/2921.0 (0.2–3.0)
Any IAS NNRTI mutations392 (83.9)431 (92.3)39/4979.6 (65.6–89.8)
Number of IAS NNRTI mutations10221202180/29561.0 (55.2–66.6)
Number of IAS nevirapine mutations782917135/29545.8 (40.0–51.6)
Number of IAS efavirenz mutations758897139/29547.1 (41.3–53.0)
Number of IAS etravirine mutations614715101/29534.2 (28.8–40.0)
(b)
Rega-predicted susceptibilityt0t1New NNRTI mutationsPYFUCrude rate of NNRTI mutation accumulation/100 PYFU (95% CI)
NNRTI activity
 Non-active378 (80.9%)422 (90.4%)9323639.4 (33.1–45.9)
 Half active3 (0.6%)3 (0.6%)1425.0 (0.6–80.6)
 Fully active86 (18.4%)42 (9.0%)8655156.4 (127–189.4)
Number of active drugs in addition to NNRTI used at t0
 0208 (44.5%)231 (49.5%)5810654.7 (44.8–64.4)
 0.5–1.5225 (48.2%)205 (43.9%)10517061.8 (54.0–69.1)
 ≥234 (7.3%)31 (6.6%)171989.5 (66.9–98.7)
(c)
NNRTIMean (SD) Rega-predicted susceptibility 
t0t1Mean (SD) reduction in predicted susceptibility/year
All regimens received (n=467)
 Nevirapine0.187 (0.39)0.093 (0.29)0.402 (1.74)
 Efavirenz0.185 (0.39)0.093 (0.29)0.400 (1.74)
 Etravirine0.692 (0.33)0.622 (0.33)0.282 (1.07)
Nevirapine-based regimens (n=212)
 Nevirapine0.184 (0.39)0.102 (0.30)0.421 (1.84)
 Efavirenz0.180 (0.39)0.102 (0.30)0.418 (1.84)
 Etravirine0.731 (0.28)0.665 (0.29)0.320 (1.23)
Efavirenz-based regimens (n=255)
 Nevirapine0.191 (0.39)0.083 (0.27)0.378 (1.61)
 Efavirenz0.191 (0.39)0.083 (0.27)0.378 (1.61)
 Etravirine0.644 (0.37)0.571 (0.37)0.237 (0.84)
(d)
NNRTIMean (SD) Rega-predicted susceptibility (patients with t0 less than 3 months from VF) 
t0t1Mean (SD) reduction in predicted susceptibility/year
  1. CI, confidence interval; IAS, International AIDS Society; PYFU, person-years of follow-up; SD, standard deviation; VF, virological failure.

All regimens received (n=165)
 Nevirapine0.385 (0.48)0.148 (0.35)1.084 (2.79)
 Efavirenz0.385 (0.48)0.148 (0.35)1.084 (2.79)
 Etravirine0.852 (0.24)0.730 (0.27)0.487 (1.27)
Nevirapine-based regimens (n=89)
 Nevirapine0.326 (0.47)0.124 (0.33)1.152 (2.98)
 Efavirenz0.326 (0.47)0.124 (0.33)1.152 (2.98)
 Etravirine0.809 (0.26)0.685 (0.28)0.574 (1.41)
Efavirenz-based regimens (n=76)
 Nevirapine0.454 (0.49)0.178 (0.37)1.005 (2.56)
 Efavirenz0.454 (0.49)0.178 (0.37)1.005 (2.56)
 Etravirine0.901 (0.20)0.783 (0.26)0.385 (1.08)

Rate of accumulation of NNRTI-associated resistance

The highest rate of accumulation was observed for mutations 103N (27.6 new mutations per 100 years; 95% CI 20.7–35.3), 181C (12.2/100 years; 95% CI 8.0–17.7) 190A (9.4/100 years; 95% CI 5.8–14.3) and 108I (6.7/100 years; 95% CI 4.0–10.6). Other mutations such as 98G, 100I, 101E, 181I and 188L were also accumulated, although at the lower rate of 0.2–0.4/100 years. The number of pairs for which there was at least one NNRTI mutation that was detected at t1 but not at t0 was 39/49 PYFU, giving a rate of accumulation of at least 0.79 new NNRTI mutations/year (95% CI 0.66–0.90; Table 2a). Overall, 180 IAS NNRTI resistance mutations were accumulated over 295 PYFU (average rate of 0.61 per year; 95% CI 0.55–0.67), while the rate of accumulation of NNRTI drug-specific mutations was somewhat slower, at 0.46/year, and that of etravirine mutations was a little lower compared with nevirapine or efavirenz mutations.

However, there was a large variability in rates according to the time difference between GRTs in a pair, shorter time differences being associated with higher rates (supporting information, Tables S2), and the extent of treatment exposure prior to failing the NNRTI regimen. This was mainly explained by time since virological failure, as there was a higher prevalence of shorter time differences if t0 was closer to the date of virological failure. An initial phase of rapid accumulation followed by phases of slower accumulation were identified: 0.90/year (95% CI 0.84–0.95) for GRT pairs with a t0 within 6 months of the date of virological failure, 0.43/year (95% CI 0.32–0.56) for the period 7–18 months after failure and 0.24/year (95% CI 0.15–0.34) for the period >18 months after failure (supporting information, Table S2). The overall estimated rate was slower when the analysis was restricted to 14 participants who had failed the NNRTI regimen that they started when they were ART-naïve: four new NNRTI mutations over 18 PYFU (rate 0.22/year; 95% CI 0.06–0.57). In contrast, when only the first GRT pair per patient was used, the rate was higher than the average estimate at 1.02/year (95% CI 0.85–0.12; supporting information, Table S4).

Table 2b shows that the rate of accumulation was higher in patients with a virus predicted at t0 to be susceptible to the NNRTI used, at 1.56/year (95% CI 1.27–1.89; 86 mutations over 55 PYFU), compared with those with a virus predicted to be resistant, for whom the rate was 0.39/year (95% CI 0.33–0.46; 93 mutations over 236 PYFU). Despite the slower accumulation of etravirine-specific mutations, overall the predicted etravirine activity showed the largest drop, decreasing from 0.69 (meaning that the activity of etravirine was already reduced by a third at t0) to 0.62, resulting in an absolute mean change of 0.28/year (Table 2c). This drop was even more dramatic when we restricted the analysis to GRT pairs started within 3 months of virological failure (0.49/year when starting from almost fully susceptible; Table 2d). On the basis of these estimates and assuming a piecewise linear model, we predict that it should take approximately 1.0 year (calculated as 0.5/0.49) of exposure to a virologically failing regimen including nevirapine or efavirenz to reduce etravirine activity from fully susceptible to intermediate resistant [and a further 1.8 years (0.50/0.28) to reach zero activity]. As a consequence of rapid accumulation of classic NNRTI resistance upon failure, both nevirapine and efavirenz had lost almost all their activity at t0, even when the analysis was restricted to 165 pairs in which t0 was within 3 months of the date of failure (Table 2c and d).

Predictors of accumulation of NNRTI-associated resistance

In the Poisson regression analysis, independent predictors of a slower accumulation of NNRTI mutations were a more recent calendar year of t0 (RR 0.80; 95% CI 0.69–0.93; P=0.004; Table 3), a longer interval from the time of last virological suppression on the NNRTI (RR 0.76; 95% CI 0.64–0.91; P=0.003) and receiving nevirapine instead of efavirenz (RR 0.66; 95% CI 0.46–0.95; P=0.03). Patients receiving a fully active NNRTI accumulated mutations much more rapidly than those with a virus that was already resistant to their NNRTI (RR 3.26; 95% CI 2.27–4.68; P<0.001). The results for the accumulation of etravirine-specific mutations were similar, although the analysis had lower power (Table 3).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Our analysis indicated that, in patients who were kept on NNRTI-based virologically failing regimens, there was an initial phase of rapid acquisition of new NNRTI mutations (one new NNRTI mutation/year over the first 6 months) followed by a phase in which rates of accumulation were 0.4/year and lower. The estimated average rate was at least 3-fold higher than the rate of accumulation of TAM previously estimated in this cohort [4]. Some mutations such as 103N (for efavirenz) and 181C (for nevirapine), which tend to appear earlier in the clinical course of failure, appeared to accumulate at a higher rate than other mutations. This is consistent with other data and with the biological hypothesis that significant NNRTI resistance is typically achieved early in the course of virological failure and no fitness-compensatory mutations are later required [19–21].

On average, the rate of accumulation of etravirine-specific mutations was somewhat lower, at one new mutation per 3 years. Using the Rega IS and assuming a linear rate of loss of susceptibility within each phase, we predicted that, from being fully active against the virus, etravirine is likely to become intermediate resistant over a time span of one year and to become completely inactive after a further 1.8 years. Note that, although the prediction of loss of etravirine susceptibility over time has been extrapolated using a piecewise linear assumption, this does not mean that we assumed that per each accumulated mutation the etravirine genotypic susceptibility score (GSS) was expected to decrease linearly. In fact, according to the Rega IS, each NNRTI mutation has a specific weight and a variable impact on the etravirine GSS [15].

At baseline-t0, after a median of 3 months from the time of first virological failure on an NNRTI, an appreciable amount of NNRTI-associated resistance could already be detected: 66% of patients had at least one NNRTI mutation, with an average of two NNRTI mutations. Of note, there could be a number of reasons for the lack of a resistance test closer to the date of virological failure, but this seems to reflect routine clinical practice in Europe and elsewhere [22–24]. It has been argued that a key factor in preventing resistance accumulation is an early treatment switch guided by virological monitoring and resistance testing [25]. Our analysis is in agreement with this view, as it shows a strong association between both the time from virological failure to t0 and the time from the last viral load ≤50  HIV-1 RNA copies/mL on the NNRTI to t0 and the subsequent rate of resistance accumulation.

While there is evidence that, at the time of first failure of the NNRTI, susceptibility to etravirine is more severely compromised in those who experience failure on nevirapine than in those who experience failure on efavirenz, possibly because of the early appearance of 181C [26–29], our data suggest that the subsequent accumulation of etravirine-associated mutations is slower in those who are kept on nevirapine. These findings potentially have clinical implications for decisions regarding which patients may experience a greater benefit from starting etravirine after prolonged exposure to NNRTI-based failing regimens. However, our interpretation relies on the predictions of currently available IS which are known to be imperfect. It is possible that the estimates may have varied if an alternative system (e.g. Stanford-HIVDB) had been used [30].

Two studies performed in the USA showed a rate of NNRTI accumulation very similar to ours (approximately 0.35 new NNRTI mutations/year) [31,32]. Two more recent analyses of patients with HIV clade C showed a high level of NNRTI resistance at the failure of their first ART regimen [33,34]. In one of these analyses, at the detection of viraemia, five (71%) of seven tested patients had NNRTI resistance mutations; this number increased to eight (89%) of nine patients by 6 months, 11 (78%) of 14 patients by 12 months, and 15 (94%) of 16 patients by 18 months, perhaps suggesting a higher rate of accumulation in the population mainly infected with C subtype viruses [34]. However, the difference in virus subtype is likely not to be the only difference between this cohort and that of EuroSIDA.

Some limitations of this analysis should be discussed. First, in the absence of adherence data, in order to exclude patients who might have been completely nonadherent, we restricted the analysis to those for whom there was evidence of resistance to at least one of the drugs used at t0. Secondly, it is not possible from our data to establish the most likely reason that patients in EuroSIDA were kept on virologically failing regimens (reasons may have included waiting for the results of a genotypic test, a lack of available options, and patients' choice) so selection bias cannot be ruled out. Further, because standard genotyping can only detect mutations that are well represented in major populations, we cannot rule out the possibility that mutations defined in our analysis as ‘newly detected at t1’ could already have been present at t0 but not detectable in the majority virus, resulting in a possible overestimate of the true rate of NNRTI accumulation. Data obtained from ultra-deep sequencing are not yet available for patients in EuroSIDA. Also, not all participants were tested prior to failing the NNRTI regimen and therefore we could have underestimated the proportion of resistance detected at failure which was caused by transmission of resistant variants. Lastly, our results may only apply to patients with little initial resistance to etravirine but with extensive resistance to nevirapine, efavirenz and other drugs. Indeed, only a minority of participants (6%) were followed up after failing an NNRTI regimen that was started when they were ART-naïve.

In conclusion, our data suggest that, in the setting of patients who are kept on NNRTI-based, virologically failing regimens, the rate of accumulation of NNRTI mutations is 0.8 mutations/year on average (>3-fold faster than the rate at which TAMs accumulate) and even faster in the first 6 months after failure. Patients who experienced virological failure with NNRTI resistance and who have a history of long exposure to nevirapine might gain greater benefits from switching to etravirine than those with long previous exposure to efavirenz.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Funding: Primary support for EuroSIDA is provided by the European Commission BIOMED 1 (CT94-1637), BIOMED 2 (CT97-2713), 5th Framework (QLK2-2000-00773), 6th Framework (LSHP-CT-2006-018632) and 7th Framework (FP7/2007-2013, EuroCoord n° 260694) programmes. Current support also includes unrestricted grants from Gilead, Pfizer, Bristol-Myers Squibb and Merck and Co. The participation of centres in Switzerland was supported by The Swiss National Science Foundation (Grant 108787).

Conflicts of interest: None of the authors has any financial or personal relationships with people or organizations that could inappropriately influence this work, although most members of the group have, at some stage in the past, received funding from a variety of pharmaceutical companies for research, travel, speaking engagements or consultancies.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • 1
    UNAIDS 2009. Report on the global AIDS epidemic. Available at http://data.unaids.org/pub/Report/2009/JC1700_Epi_Update_2009_en.pdf (accessed 11 March 2011).
  • 2
    BHIVA Guidelines for the Treatment of HIV-1 infected adults with antiretroviral therapy (2008). Available at http://www.bhiva.org/files/file1030835.pdf (accessed 11 March 2011).
  • 3
    World Health Organization (WHO). Antiretroviral therapy for HIV infection in adults and adolescents: rapid advice 2009. Available at http://www.who.int/hiv/pub/arv/rapid_advice_art.pdf (accessed 11 March 2011).
  • 4
    Cozzi-Lepri A, PhillipsAN, Martinez-PicadoJet al. for the EuroSIDA Study Group. Rate of accumulation of thymidine analogue mutations in patients continuing to receive virologically failing regimens containing zidovudine or stavudine: implications for antiretroviral therapy programs in resource-limited settings. J Infect Dis2009; 200: 687697.
  • 5
    Stevens WS. Slow accumulation of HIV resistance mutations: Implications for resource-limited settings? J Infect Dis 2009; 200: 670672.
  • 6
    Ghosn J, ChaixML, DelaugerreC.HIV-1 resistance to first- and second-generation non-nucleoside reverse transcriptase inhibitors. AIDS Rev2009; 11: 165173.
  • 7
    Madruga JV, CahnP, GrinsztejnBet al. for the DUET-1 study group. 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: 2938.
  • 8
    Lazzarin A, CampbellT, ClotetBet al. for the DUET-2 study group. 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: 3948.
  • 9
    Katlama C, HaubrichR, LalezariJet al. for the DUET-1, DUET-2 study groups. Efficacy and safety of etravirine in treatment-experienced, HIV-1 patients: pooled 48 week analysis of two randomized, controlled trials. AIDS 2009; 23: 22892300.
  • 10
    Braitstein P, ZalaC, YipBet al. Immunologic response to antiretroviral therapy in hepatitis C virus-coinfected adults in a population-based HIV/AIDS treatment program. J Infect Dis2006; 193: 259268.
  • 11
    Bachani D, GargR, RewariBBet al. Two-year treatment outcomes of patients enrolled in India's national first-line antiretroviral therapy programme. Natl Med J India2010; 23: 712.
  • 12
    Stringer JS, ZuluI, LevyJet al. Rapid scale-up of antiretroviral therapy at primary care sites in Zambia: feasibility and early outcomes. JAMA 2006; 296: 782793.
  • 13
    Geretti AM, HarrisonL, GreenHet al. for the UK Collaborative Group on HIV Drug Resistance. Effect of HIV-1 subtype on virologic and immunologic response to starting highly active antiretroviral therapy. Clin Infect Dis2009; 48: 12961305.
  • 14
    Lundgren JD, PhillipsAN, VellaSet al. for the EuroSIDA Study Group. Regional differences in use of antiretroviral agents and primary prophylaxis in 3122 European HIV-infected patients. J Acquir Immune Defic Syndr Hum Retrovirol1997; 16: 153160.
  • 15
    Van Laethem K, GerettiAM, CamachoR, VandammeA-M.Algorithm for the use of genotypic HIV-1 resistance data. Rega v8.0.2©, Leuven, 16 June 2009. Available at http://www.rega.kuleuven.be/cev (accessed 11 March 2011).
  • 16
    Johnson VA, Brun-VézinetF, ClotetBet al. Update of the drug resistance mutations in HIV-1: December 2010. Top HIV Med2010; 18: 156163.
  • 17
    Statistical Analysis Software Version 9, Cary, NC. Available at http://www.sas.com (accessed 11 March 2011).
  • 18
    Zeger SL, LiangKY.Longitudinal data analysis for discrete and continuous outcomes. Biometrics1986; 42: 121130.
  • 19
    Bangsberg DR, AcostaEP, GuptaRet al. Adherence-resistance relationships for protease and non-nucleoside reverse transcriptase inhibitors explained by virological fitness. AIDS2006; 20: 223231.
  • 20
    Bacheler LT, AntonED, KudishPet al. Human immunodeficiency virus type 1 mutations selected in patients failing efavirenz combination therapy. Antimicrob Agents Chemother2000; 44: 24752484.
  • 21
    Beerenwinkel N, DrtonM.A mutagenetic tree hidden Markov model for longitudinal clonal HIV sequence data. Biostatistics2007; 8: 5371.
  • 22
    Lee KJ, DunnD, GilsonRet al. Treatment switches after viral rebound in HIV-infected adults starting antiretroviral therapy: multicentre cohort study. AIDS 2008; 22: 19431950.
  • 23
    Hart E, CurtisH, WilkinsE, JohnsonM.National review of first treatment change after starting highly active antiretroviral therapy in antiretroviral-naive patients. HIV Med2007; 8: 186191.
  • 24
    Eyawo O, FernandesKA, BrandsonEKet al. Suboptimal use of HIV drug resistance testing in a universal health-care setting. AIDS Care2011; 23: 4251.
  • 25
    Gupta RK, HillA, SawyerAWet al. Virological monitoring and resistance to first-line highly active antiretroviral therapy in adults infected with HIV-1 treated under WHO guidelines: a systematic review and meta-analysis. Lancet Infect Dis 2009; 9: 409417.
  • 26
    MacArthur RD, Huppler HullsiekK, PengGet al. for the CPCRA 058 Study Team, the Terry Beirn Community Programs for Clinical Research on AIDS (CPCRA) and the International Network for Strategic Initiative in Global HIV Trials (INSIGHT). Failing therapy with efavirenz results in significantly fewer mutations limiting to etravirine than failing therapy with nevirapine: on-treatment analyses from the CPCRA FIRST study. XVII International HIV Drug Resistance Workshop. Sitges, Spain, June 2008 [Abstract 129].
  • 27
    Marcelin AG, FlandreP, DescampsDet al. for the ANRS AC11 Resistance Study Group. Factors associated with virological response to etravirine in nonnucleoside reverse transcriptase inhibitor-experienced HIV-1-infected patients. Antimicrob Agents Chemother2010; 54: 7277.
  • 28
    Lapadula G, CalabresiA, CastelnuovoFet al. Prevalence and risk factors for etravirine resistance among patients failing on non-nucleoside reverse transcriptase inhibitors. Antivir Ther2008; 13: 601605.
  • 29
    Poveda E, AntaL, BlancoJLet al. for the Resistance Platform of the Spanish AIDS Research Network (ResRIS). Etravirine resistance associated mutations in HIV-infected patients failing efavirenz or nevirapine in the Spanish antiretroviral resistance database. AIDS2010; 24: 469471.
  • 30
    Scherrer AU, HasseB, von WylVet al. Prevalence of etravirine mutations and impact on response to treatment in routine clinical care: the Swiss HIV Cohort Study (SHCS). HIV Med 2009; 10: 647656.
  • 31
    Kantor R, ShaferRW, FollansbeeSet al. Evolution of resistance to drugs in HIV-1-infected patients failing antiretroviral therapy. AIDS2004; 18: 15031511.
  • 32
    Napravnik S, EdwardsD, StewartP, StalzerB, MattesonE, EronJJJr.HIV-1 drug resistance evolution among patients on potent combination antiretroviral therapy with detectable viremia. J Acquir Immune Defic Syndr2005; 40: 3440.
  • 33
    Orrell C, WalenskyRP, LosinaE, PittJ, FreedbergKA, WoodR.HIV type-1 clade C resistance genotypes in treatment-naive patients and after first virological failure in a large community antiretroviral therapy programme. Antivir Ther2009; 14: 523531.
  • 34
    Hoffmann CJ, CharalambousS, SimJet al. Viremia, resuppression, and time to resistance in human immunodeficiency virus (HIV) subtype C during first-line antiretroviral therapy in South Africa. Clin Infect Dis2009; 49: 19281935.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
hiv_943_sm_suppmatt_crx.docx43KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.