Susceptibility to Etravirine of HIV Type 1 Subtype C Isolates from Nevirapine/Efavirenz-Experienced Patients: Comparative Interpretation of ANRS and STANFORD Algorithms

Abstract

We analyzed subtype C HIV-1 isolates from patients at failure of a regimen including nevirapine or efavirenz for their susceptibility or resistance to etravirine according to the ANRS and STANFORD algorithms. Statistical analysis showed a consensus that more than 45% of these viral strains are potentially resistant to etravirine.

Introduction

T he efficacy of first-generation nonnucleoside reverse transcriptase inhibitors (NNRTIs) such as nevirapine (NVP) and efavirenz (EFV) is limited by the low genetic barrier to resistance. Etravirine (ETV) is a more recent NNRTI with a higher genetic barrier to viral resistance (in vitro) and offers the potential to achieve antiviral activity even in patients exhibiting resistance to first-generation NNRTIs. In vitro studies and DUET therapeutic trials have identified resistance-associated mutations (RAMs) correlating with decreased virological response to ETV.^1,2 Numerous RAMs have been identified and introduced into subtype B resistance algorithms, particularly those from ANRS and STANFORD.

The triple antiretroviral (ARV) combination d4T/AZT-3TC-NVP/EFV is the least expensive first-line combination and is widely used in the southern hemisphere. Such is the case in India, where it is used as first line in the public and private sectors. The aim of the present study was to analyze the susceptibility to ETV of HIV-1 subtype C isolates from Indian patients at failure for a first-line regimen consisting of d4T/AZT-3TC-NVP/EFV and where EFV replaces NVP if the patient is coinfected with tuberculosis.

Materials and Methods

Eighty patients from Mumbai with 1 to 8 years of exposure to NVP/EFV in a d4T/AZT-3TC-NVP/EFV regimen volunteered for resistance testing, which was carried out free of cost. After informed written consent was obtained, blood was collected in EDTA tubes and plasma was separated and used for viral load (VL) quantitation before being frozen at −80°C for other purposes. VL was determined using the Cobas Taqman assay (Roche) and values ranged from 10,000 to 5,817,977 copies/ml.

For genotypic analysis, plasmas were thawed, homogenized, and spotted on Whatman 903 filter paper cards with three spots of 50 μl. The samples were allowed to dry at room temperature for one night with no contact of the two faces of the spot, then placed in plastic bags with desiccants. They were stored at room temperature for 2 to 7 days before dispatch to Bordeaux, France. The spots were cut with scissors and eluted in 220 μl/spot of a buffer containing phosphate-buffered saline (PBS), Tween 20, and fetal calf serum. The samples were then agitated at 4°C for 1 h before being vortexed. Of the final elution 140 μl was used for RNA extraction using the QIAamp Viral RNA Mini Kit (Qiagen). At that time, the extraction and amplification sensitivity of the assay was around 10,000 copies/ml.

The RNA was used in reverse transcription polymerase chain reaction of viral reverse transcriptase (RT) using two sets of primers in a GeneAmp PCR System 9700 (Applied Biosystems, Foster City, CA) thermal cycler. The outer and inner primers for RT have been described previously.³ The fragments obtained were sequenced on both strands using the CEQ DTCS Quick Start kit on a Beckman CEQ 2000 DNA automated sequencer in the virology department of the University Hospital of Bordeaux. The derived nucleotide sequences of the RT region were aligned by the Clustal W 1.74 alignment program with known reference strains of M and N pooled from the HIV-1 databank. Final genetic trees were inferred using the neighbor-joining method for matrix distances calculated after gapstipping of alignments with Kimura two-parameter algorithms. The resistance mutations of RT were noted and interpreted according to the French ANRS (v19) and STANFORD (v6) algorithms using the SmartGene IDNS service module.

The GenBank accession numbers for RT sequences of viral isolates 425 to 451 are GU108151 to 108177 and those of isolates 452 to 504 are JF895621 to JF895673.

We compared the level of agreement between algorithms using Cohen's kappa coefficient with 95% confidence limits. Statistical analysis was performed with SAS 9.1.3 (SAS Institute Inc., Cary, NC) software. We selected the combination with the highest kappa coefficient.

For the Agence Nationale de Recherches sur le SIDA (ANRS) algorithm, we interpreted susceptible as Susceptible only and resistance as either Intermediate resistance or Resistance.

For the STANFORD algorithm, we interpreted susceptible as either Susceptible, Potential low resistance, or Low resistance and resistance as either Intermediate resistance or High resistance.

Results

All RT sequences but one (A1 subtype) clustered with subtype C; the analysis focused on the 79 isolates of subtype C. Resistance mutations to NNRTIs of the first line and considered as RAMs to ETV are presented in Tables 1 and 2 according to the ANRS and STANFORD algorithms. Globally, if we consider RAMs common to both algorithms, the list with decreasing frequencies above 15% is Y181C, G190A, K101E, and A98G; K103N, which is considered by STANFORD but not by ANRS, had a prevalence above 30%; V179I and V90I, which are considered by ANRS, had percentages of 24.1% and 17.7%, respectively. H221Y, which is a RAM for ANRS when it is associated with Y181C, had a percentage of 20.3% when alone and 12.7% when in the Y181C/H221Y association. We also carried out a statistical analysis of the different responses to ETV provided by the algorithms with different combinations of these responses in order to find the combination with the highest kappa coefficient. Table 3 presents the ANRS/STANFORD comparison with the combination exhibiting the highest kappa coefficient (0.50). Globally, resistance was observed for 46% and susceptibility for 54%.

Table 1.

Resistance-Associated Mutations to Etravirine in the Viral Isolates of the Study According to ANRS and STANFORD Algorithms

N. Bordeaux	Synthesis of interpretations	List of mutations according to ANRS	List of mutations according to STANFORD
425	RI	90I_179I_181C+221Y	103N_181C
426	SP	179I	188L
427	SS	/	103N
428	SI	90I_181C	181C
429	RR	98G_101E_179I_181I_190A_221Y	98G_101E_106M_181I_190A
430	SP	/	188L
431	RI	181C+221Y	103N_181C
432	SI	181C	181C
433	SS	/	103N
434	SI	98G_101P	98G_101P_103N_225H
435	II	181C_138Q	103N_138Q_181C
436	SI	101P_190A	101P_190A
437	RI	90I_101E_179I_190A	101E_106M_190A
438	SP	179I_190A	190A
439	SP	90I	103N_238T
440	SP	90I	103N_238T
441	RR	190A_181C+221Y	181C_190A
442	RI	181V_221Y	103N_181V
443	SI	181C	103N_181C
444	SP	/	188L
445	II	90I_179I_181C	181C
446	SI	181C	106M_181C
447	SL	/	103N_106M_227H
448	SI	90I_181C	103N_181C
449	RR	101E_179I_181C_190A	101E_181C_190A
450	SP	221Y	188L
451	IL	98G_138A	98G_103N_138A_238T
452	SS	/	103N
453	SI	179I_181C	181C
454	RI	90I_181C+221Y	181C
455	RR	181V_190A	181V_190A
456	SI	181C	181C
457	RL	98G_101E_179I_190S	98G_101E_190S
458	RL	90I_98G_101E_179I_190A	98G_101E_190A
459	SS	/	/
460	SL	98G_221Y	98G_103N_106A
462	SI	98G_181C	98G_181C
463	SL	221Y	188L_227L
464	RR	98G_101E_181C_190A	98G_101E_181C_190A
465	RR	101E_179T_181C_190A	101E_181C_190A
466	IL	101E_179I_190A	101E_190A_238T
467	RI	90I_181C+221Y	181C
468	IR	98G_181C_190A	98G_181C_190A
469	SL	190A	103N_190A
470	SS	/	/
471	IL	98G_101E_190A	98G_101E_190A
472	SI	181C	181C
473	SP	190A	103N_190A
474	RI	138A_179T_181C+221Y	103N_138A_181C
475	II	101E_179I_190A	101E_188L_190A
476	RR	101E_181C+221Y_230L	101E_106M_181C_230L
477	IR	101H_181C_190A	101H_181C_190A
478	II	101E_138Q_190S	101E_138Q_190S
479	RR	98G_101E_181C_190A	98G_101E_181C_190A
480	RR	98G_179I_181C_190A	98G_106M_181C_190A_227L
481	SS	90I	103N
482	II	101E_179I_190A	101E_188L_190A
483	SL	179I_190A	106M_190A
484	II	101P_179I_190A_221Y	101P_103S_190A
485	SS	/	/
486	II	101E_138Q_190S	101E_138Q_190S
487	IL	98G_179I_190A	98G_190A
488	SL	/	103N_225H
489	SL	101E_179I	101E_190C
490	RI	98G_101E_179I_190A	98G_101E_190A_238N
491	SL	/	103N_225H_238T
492	SS	/	103N
493	RR	179T_190A_181C+221Y	181C_190A
494	SL	90I_101E	101E_103N
495	SS	90I	103N
496	SL	190A	106M_190A_227L
497	IL	98G_101H_190A	98G_101H_190A
498	SL	190A	103N_190A
499	SL	/	103N_225H
500	SI	179D_230L	106M_179D_230L
501	SL	101E_190S	101E_190S
502	SL	190A	106M_190A
503	RR	101E_190A_181C+221Y	101E_181C_190A
504	RI	90I_181C+221Y	101N_181C

The second column is devoted to interpretation by ANRS (first letter) and STANFORD (second letter). ANRS: S, susceptible; I, possible resistance; R, resistance. STANFORD: S, susceptible; P, potential low-level resistance; L, low-level resistance; I, intermediate resistance; R, high-level resistance.

Table 2.

Percentages of Resistance-Associated Mutations to Etravirine in the Viral Isolates of the Study According to ANRS and Stanford Algorithms

		90I	98G	98S	100I	101E	101H	101I	101N	101P	101R	103N	103R
ANRS	n=79	17.7%	19.0%		0.0%	25.3%	2.5%	0.0%		3.8%	0.0%
STANFORD	n=79		17.7%		0.0%	24.1%	2.5%		1.3%	3.8%		30.4%
		103S	106A	106M	106I	108I	138A	138Q	179D	179E	179F	179I	179L
ANRS	n=79				0.0%		2.5%	3.8%	1.3%		0.0%	24.1%	0.0%
STANFORD	n=79	2.5%	1.3%	12.7%			2.5%	3.8%	1.3%	0.0%	0.0%
		179M	179T	181C	181I	181V	188L	188H	188C	190A	190C	190E	190S
ANRS	n=79	0.0%	3.8%	35.4%^a	1.3%	2.5%				38.0%			5.0%
STANFORD	n=79			38.0%	1.3%	2.5%	8.9%	0.0%	0.0%	36.7%	1.3%	0.0%	3.8%
		221Y	225H	227C	227L	230L	238N	238T
ANRS	n=79	20.3%^a				2.5%
STANFORD	n=79		5.1%	0.0%	5.1%	2.5%	1.3%	6.3%

ANRS: 181C+221Y, 12.77.

Resistance-associated mutations (RAMs) as considered by the two algorithms. A blank position is RAM not considered by the algorithm.

Table 3.

Comparative ANRS/STANFORD Analysis with Coefficient Kappa=0.50 IC 95% (0.28–0.71)

	STANFORD algorithm
	Susceptible		Resistant		Total
ANRS algorithm	n	%	n	%	n	%
Susceptible	31	82	12	29	43	54
Resistant	7	18	29	71	36	46
Total	38	100	41	100	79	100

Discussion

We first ruled out the possibility that RAMs to ETV as considered by the two algorithms could occur as a natural polymorphism in subtype C viruses from naive patients by using personal unpublished data and those from a previous study.³ At that time we had noted very rare substitutions at positions 90 and 179. In both cases, double populations V90VI and V179VI were observed. These data are in agreement with the results published by Maiga et al.⁴ and highlight the very low prevalence of subtype C substitutions in positions related to potential resistance to ETV. They are also in accordance with the international list of drug resistance mutations for surveillance of transmitted HIV-1 drug resistance that was updated in 2009.⁵ The prevalence of natural G190A, Y181C, and K103N in subtype C is so low (below 0.2%) that these substitutions may be considered as nonpolymorphic.

The main question in this study was the potential use of ETV at the end of a first-line HAART including NVP and/or EFV. Clearly, comparison between the algorithms analyzed here is quite complicated. ANRS considers full resistance with four mutations including V90I, A98G, L100I, K101E/H/I/P/R, V106I, V179D/F/I/L/M/T, Y181C/I, G190A/S, M230L, Y181V alone, or Y181C+H221Y. Possible resistance is associated with three mutations or E138A/G/K/Q/R. In the STANFORD algorithm, the major mutations to ETV with a high weight are L100I, K101P, V179F, Y181C/I/V, G190E, F227C, and M230L.

The major finding of the statistical analysis is that more than 45% of the HIV-1 subtype C isolates from these NVP/EFV-experienced patients are not susceptible to ETV. This result, which is based on the analysis of two algorithms, supports other data recently published using the VIRCO algorithm.⁶ Moreover, although we did not investigate the minority variants bearing resistance mutations to NNRTIs, we postulate that the potential resistance to second-line ETV is quite common, as already suggested.⁷ Our results are also in concordance with those obtained in Thailand ⁸ showing that ETV induced a suboptimal virological response compared to a protease inhibitor (PI) in patients infected with non-C viruses at failure for a first line with NRTIs plus an NNRTI. More recently, the same group from Thailand used the STANFORD algorithm and Monogram weighted score to show that 60% of the CRF01_AE isolates from NVP/EFV-experienced patients are resistant to ETV.⁹

In conclusion, in patients infected with subtype C HIV-1 who are declared at failure for a regimen including NVP/EFV according to clinical/immunological WHO criteria and whose viral isolates exhibit numerous mutations to NVP/EFV, the use of ETV as a second-line therapy instead of a PI should be excluded since more than 45% of the cases will not benefit from it.

Footnotes

Acknowledgment

This work was funded by UMR CNRS 5234.

Author Disclosure Statement

No competing financial interests exist.

References

Vingerhoets

, Tambuyzer

, Azijn

et al. Resistance profile of etravirine: Combined analysis of baseline genotypic and phenotypic data from the randomized, controlled Phase III clinical studies. AIDS, 2010; 24:503–514.

Marcelin

, Flandre

, Descamps

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.

Deshpande

, Recordon-Pinson

, Deshmukh

et al. Molecular characterization of HIV type 1 isolates from untreated patients of Mumbai (Bombay), India, and detection of rare resistance mutations. AIDS Res Hum Retroviruses, 2004; 20:1032–1035.

Maiga

, Descamps

, Morand-Joubert

et al. Resistance-associated mutations to etravirine (TMC-125) in antiretroviral-naive patients infected with non-B HIV-1 subtypes. Antimicrob Agents Chemother, 2010; 54:728–733.

Bennett

, Camacho

, Otelea

et al. Drug resistance mutations for surveillance of transmitted HIV-1 drug-resistance: 2009 update. PLoS One, 2009; 4:e4724.

Neogi

, Shet

, Shamsundar

, Ekstrand

. Selection of nonnucleoside reverse transcriptase inhibitor-associated mutations in HIV-1subtype C: Evidence of etravirine cross-resistance. AIDS, 2011; 25:1123–1126.

Varghese

, Shahriar

, Rhee

et al. Minority variants associated with transmitted and acquired HIV-1 non-nucleoside reverse transcriptase inhibitor resistance: Implications for the use of second-generation non-nucleoside reverse transcriptase inhibitors. J Acquir Immune Defic Syndr, 2009; 52:309–315.

Ruxrungtham

, Pedro

, Latiff

et al. Impact of reverse transcriptase resistance on the efficacy of TMC125 (etravirine) with two nucleoside reverse transcriptase inhibitors in protease inhibitor-naive, non-nucleoside reverse transcriptase inhibitor-experienced patients: Study TMC125-C227. HIV Med, 2008; 9:883–896.

Bunupuradah

, Ananworanich

, Chetchotisakd

et al. Etravirine and rilpivirine resistance in HIV-1 subtype CRF01_AE-infected adults failing non-nucleoside reverse transcriptase inhibitor-based regimens. Antivir Ther, 2011; 16,7:1113–1121.