Decreased Phenotypic Susceptibility to Etravirine in Patients with Predicted Genotypic Sensitivity

Background A sensitive, phenotypic reverse transcriptase (RT)-based drug susceptibility assay for the detection of etravirine (ETR) resistance in patient isolates was developed and compared with the results from direct sequencing and ultra-deep pyrosequencing (UDPS). Methods Samples were obtained from 15 patients with antiretroviral therapy (ART) failure and from five non-nucleoside reverse transcriptase inhibitor (NNRTI)-naïve patients of whom four were infected by an NNRTI-resistant strain (transmitted drug resistance, TDR). In five patients, two consecutive samples (a and b) were taken for follow up of the virological response. HIV-1 RT was purified and drug susceptibility (IC50) to ETR was estimated. Direct sequencing was performed in all samples and UDPS in samples from nine patients. Results Increased IC50 to ETR was found in samples from 13 patients where direct sequencing predicted resistance in only four. UDPS identified additional (N = 11) NNRTI resistance associated mutations (RAMs) in six of nine tested patients. During early failure, IC50 increases were observed in three of six patients without any ETR-RAMs detected by direct sequencing. In further two patients, who stopped NNRTI before sampling, increased IC50 values were found shortly after, despite absence of ETR-RAMs. In two patients who had stopped NNRTI for >1 year, a concordance between phenotype and genotypes was found. Two patients with TDR had increased IC50 despite no ETR-RAMs were detected by direct sequencing. UDPS revealed additional ETR-RAMs in four patients with a discrepancy between phenotype and direct sequencing. Conclusions The RT-based phenotypic assay showed decreased ETR susceptibility in patients where direct sequencing predicted ETR-sensitive virus. This increased phenotypic sensitivity was to a large extent supported by UDPS and treatment history. Our method could be valuable for further studies on the phenotypic kinetics of NNRTI resistance. The clinical relevance remains to be studied in larger patient-populations.


Introduction
The first generation non-nucleoside reverse transcriptase inhibitors (NNRTIs) nevirapine (NVP) or efavirenz (EFV) are frequently used in combination with two nucleoside reverse transcriptase inhibitors (NRTIs) as first-line antiretroviral therapy (ART) in both high-and low-middle income countries. The second generation NNRTI etravirine (ETR) is approved for treatment-experienced HIV-1 infected adults with resistance to NVP and/or EFV [1][2][3], although access to ETR in low-middle income countries (LMIC) is limited. ETR can be used in such patients due to its limited cross-resistance pattern relative to first generation of NNRTI. However, the pattern of resistanceassociated mutations (RAMs) has to be considered and several constellations of mutations cause indeed cross-resistance [4,5]. A large number of ETR RAMs have been revealed [6], including the major mutations L100I, K101E/P, E138A/G/K/Q, Y181C/I/ V, Y188L, G190A/S/E, and M230L.
The genotypic resistance tests (GRT) used in clinical care detect RAMs that are present in .20% of the viral population [7][8][9]. More sensitive techniques, such as allele-specific PCR (ASPCR) [10][11][12] and ultra-deep sequencing [13,14], are not yet used in the clinical routine. It has however been reported that minor resistant quasispecies may influence the outcome of ART [15][16][17]. One alternative to GRT is phenotypic tests which have several disadvantages in the clinical setting. However, new methods for reverse transcriptase (RT) extraction and sensitive RT assays have made analyses of drug susceptibility profiles of retroviral RTs recovered directly from plasma virions feasible [18][19][20]. Drug susceptibility testing on RT offers advantages compared to traditional phenotypic tests, such as short turnover time, lower cost, a need of only serology laboratory equipment, and they might therefore be useful in LMIC. This method has successfully been used to describe drug resistance to NVP [21] and also recently to ETR [22]. In contrast to genotypic assays, no complex interpretations are required.
The aim of the study was to thoroughly evaluate clinical samples in our newly adapted RT-based assay for assessment of resistance to ETR and to compare the obtained phenotype with the genotype obtained with standard direct sequencing. Also, we performed ultra-deep pyrosequencing (UDPS) to identify any minor sequence variants in the HIV-1 RT gene of cases where there was a discrepancy between the major genotype and the RT phenotype.

Patients
Altogether, 25 plasma EDTA samples of 20 HIV-1 infected patients were retrospectively included from the HIV cohort at Department of Infectious Diseases, Karolinska University Hospital, Stockholm, Sweden (Table 1). Of these, 15 treatmentexperienced patients were randomly selected among subjects with ART failure. All of these patients had been treated or were treated with nevirapine or efavirenz. In patients 3, 5, 8, 9 and 11 (Table 1), who did not respond to treatment two consecutive samples (a and b) were drawn for follow up of the virological response. For eight samples (no 3b, 5a, 5b, 8a, 8b, 10, 11a, 18), the failing regimen contained an NNRTI. For 12 samples (no 1, 3a, 6, 7, 9a, 9b, 11b, 12,14,15,17,20), the NNRTI had been stopped earlier and the failing regimen contained now antiretroviral drugs from other categories (Table 1). Four treatment naïve patients (no 2, 13, 16,19) were included since they had been infected with NNRTIresistant strains. GRT had been performed within the clinical care by routine direct sequencing on samples from all of these individuals and one or more NNRTI mutations had been found in all of them, except in patient 7 (Table 2). One NNRTI-naïve patient (no 4), who had a K103R mutation, was chosen as a negative control since this mutation is reported not to affect ETR susceptibility. All the work presented in the manuscript was conducted in accordance with the Declaration of Helsinki. Ethical permission was obtained from the Regional Ethical Committee at Karolinska Institutet, Dnr. 2005/1167-31/3, Dnr. 2005/772-31/ 4 and written informed consent was obtained from each individual.

Phenotypic RT-based analysis of resistance to ETR
Lysates containing intraviral RT were recovered from plasma using ExaVir Load, version 3 (Cavidi, Uppsala, Sweden) [18]. Briefly, virions from up to 1 ml plasma were bound to an ion exchange gel, washed, and the RT was extracted by lysing the purified virions. The RT was used to catalyse the incorporation of bromodeoxyuridine triphosphate (BrdUTP) into DNA via a poly-rA (ribodenylic acid template) plate. The amount of BrdU monophosphate (BrdUMP) incorporated into DNA was detected with an alkaline phosphatase (Ap) conjugated anti-BrdU monoclonal antibody [20]. The Ap substrate para-nitrophenyl-diphosphate was used for colorimetric detection. The susceptibility of the RT activity to ETR was measured using ExaVir Drug susceptibility assay, as previously described [22]. Briefly, this assay measures the ability of RT to catalyse the above reaction in the presence of a 2.5 fold step dilution of ETR. The RT activity at each drug concentration was recalculated into percentage of RT activity in the absence of drug. The IC 50 values were calculated using the median effect equation [23]. The analysis was performed blindly in relation to the genotype results and to type of treatment. We have earlier described low IC 50 values (mean 6SD: 2.561.0 mM) for RT activity in the presence of ETR in treatment-naïve patients and for reference recombinant RTs [22]. Also, the reproducibility analysis of the assay showed an inter-assay variation (CVs) of 9.4 and 11.1% and that the IC 50 values for ETR (mean 6 SD: 1.660.03 mM and 3.160.04) were not influenced by the RT amount within the 40-828 fg/ml RT range. In the present study, two reference recombinant RTs, BH10-wild-type and its mutant form L100I, were analysed together with patient samples in each assay. The results of the phenotypic tests were evaluated independently by two persons (E.A, C.FR.K).

Scoring of ETR susceptibility
Three genotypic scoring systems were used to predict ETR susceptibility (Stanford University, Monogram Weighted Score, Tibotec Weighted Genotype Score). The Monogram and Tibotec systems use a score calculated from a list of mutations, each with their own weight factor [24]. The Stanford scores were obtained using http://hivdbstanford.edu/cgi-bin/Marvel.cgi. The correlation between Monogram/Tibotec/Stanford scorings and IC 50 determined by our phenotypic RT assay was assessed by Spearman's rank test using software in GraphPad Prism (San Diego, California, USA).

Ultra deep HIV-1 sequencing
HIV-1 RNA was extracted from 1 mL of plasma from nine samples (QIAamp Viral RNA Mini Kit, QIAGEN, Valencia, CA) after centrifugation at 9,0006g for 1 h. The 39 end of the reverse transcriptase region was reverse transcribed and amplified using SuperScript III and Platinum Taq High Fidelity (Life Technologies, Paisley, UK). Primer used were Forward primer (HXB2: 2508-2528) 59-GGA AGA AAT CTG TTG ACT CAG-39 and Reverse primer (HXB2: 3441-3459) 59-GAA GCA GAG CTA GAA CTG G-39. Cycling conditions were: 30 min at 52uC for reverse transcription;2 min at 94uC for initial denaturation; 20 cycles of 2 min at 94uC, 30 sec at 50uC and 1 min 30 sec at 68uC; and a final polymerisation step of 5 min at 68uC. Amplicon libraries were generated from first-round PCR products. These amplicons incorporated adapters A and B, and also identifiers used in parallel sample sequencing needed for bidirectional 454 sequencing. Nested PCR conditions were a first denaturation step of 2 min at 94uC; 20 cycles of 2 min at 94uC, 30 sec at 50uC and 45 sec at 68uC; and a final polymerisation step of 3 min at 68uC using Platinum Taq High Fidelity (Life Technologies, Paisley, UK). A set of 5 overlapped amplicons were designed to cover the protein region previously amplified. Amplicon 1: Forward (2541 -Table 1. Characteristics of 20 HIV-1 infected patients included in the study.

Ultra-deep pyrosequencing (UDPS)
UDPS was performed on nine samples (1, 2, 5b, 10, 11a, 13,14,19,20) for which there seemed to be a discrepancy between the major genotype and the RT phenotype (Table 3). The median nucleotide coverage per sample ranged from 1499 to 2506 sequences. Relative to UDPS, direct sequencing detected all mutations corresponding to .20% but no mutations ,20% of the viral population. Altogether eleven mutations were detected by UDPS, but not by direct sequencing, ranging from 0.54% to 19.56%.
In three samples (2, 5b, 20), identical mutations were found with the two sequencing techniques. Thus, the UDPS did not reveal any further minor variants that could explain the increased IC 50 values of 100 mM, 71.4 mM and 7.2 mM, respectively.

Correlation between RT-based phenotypic sensitivity and genotype scoring based on direct sequencing
There was a significant correlation between the Stanford scoring system and the phenotypic assay results (r = 0.70 p, 0.0001) assessed by Spearman's rank test. The Monogram and Tibotec scoring systems also correlated with the IC 50 values (r = 0.65 p,0.0005; r = 0.64 p,0.0005, respectively).
In patients 6 and 14, a clear decreased phenotypic sensitivity against ETR (6: .100 mM; 14: 28.8 mM) was found despite that . 1 year and .3 years, respectively, had passed since the cessation of an NNRTI. Genotyping further supported this finding for patient 6 (direct sequencing: V90I + L100I + K103N) and patient 14 (UDPS: In four patients, several years had passed since their NNRTI treatment was terminated (7, 11b, 15: .5.5 years; 9a: 3.5 years). In patient 12, 18 weeks had passed. No phenotypic resistance was found in these patients although non-ETR RAMs persisted (11b: V179I; 12 and 15: K103N; 9a: A98S + K103N + V108I + K238T). However, a second sample (9b) of patient 9 drawn four months later showed a slightly decreased phenotypic sensitivity (9.2 mM) with the same mutational pattern with a concomitant increase in plasma viral load. UDPS could not be performed due to lack of plasma.
In four samples, discordance was seen between the RT-based phenotype and the genotype. In patient 5, two samples were drawn with three months interval during failure with EFVcontaining regimen. An increasing IC 50 of 45.8 mM and 71.4 mM, respectively, was seen despite that direct sequencing as well as UDPS showed only mutations (K101I/R + V106G) which are not known to be ETR-associated. Patient 10 exhibited an increased IC 50 (13.6 mM), but the identified mutations (direct sequencing: K103N + V108I; UDPS: K103N + V108I + V179I) predicted an ETR sensitive virus. In patient 11 a sample drawn early after failure showed a slightly increased IC 50 (7.3 mM), while direct sequencing showed K103N + V179I. However UDPS identified additionally G190A giving support for that resistance was developing.

Effects of ETR on patients with transmitted NNRTI resistance
Four patients (no 2, 13,16,19), who had been infected with an NNRTI resistant strain, were also analysed. Patients 2 and 13 had a strongly increased IC 50 values (.100 mM; 68.8 mM) despite that the sequence analysis predicted only a slightly decreased sensitivity (2: K103N + E138Q, K238T; 13: K103N + G190A) ( Table 2). The remaining two samples had the K103N only, and a low 1.7 mM (16) and a slightly increased 6.4 mM (19) IC 50 value, respectively, was found. Patient 4 who had a K103R mutation was also sensitive with the phenotypic assay.

Discussion
In high-income countries, there has been a considerable decline of HIV-1 drug resistance during the last decade [25,26]. In contrast, transmitted and acquired HIV drug resistance is increasing in low-income countries [27,28], especially resistance to the first generation NNRTIs, NVP and EFV [29][30][31]. The second generation NNRTI, etravirine (ETR), is approved for treatment-experienced patients and cross-resistance to NVP and EFV is considered to be limited. We have earlier shown that a simple and affordable RT-based phenotypic resistance test can be used for detection of ETR resistance, including cross-resistance to other NNRTIs in clinical samples. In this study, we further expanded the evaluation of the method in a larger clinical cohort including patients with past or ongoing failure on the first generation NNRTIs by comparing the results with those of direct sequencing and UDPS.
We found that most of strains from patients with ongoing or past failure to a first generation NNRTI had various degrees of decreased ETR susceptibility by our phenotypic test, despite that the mutational pattern based on direct sequencing did not always predict ETR resistance. For our assay, we previously demonstrated low IC 50 values (mean 6SD: 2.561.0 mM) in treatment-naïve patients, a good reproducibility and an independence on the amount of plasma HIV RNA for the outcome of the test [22]. It can be argued that to describe a more precise laboratory cut-off for decreased susceptibility an extensive evaluation has to be done and that the assay may overestimate the presence of decreased susceptibility to ETR. However, in the present experiments ETR gained expected changes in IC 50 values for all recombinant HIV-1 RTs equivalent to previous reports [5,20], one NNRTI naïve patient (no 4) had a low IC 50 value, and the test could plausibly describe the lack of impact of K103N on the IC 50 value of ETR in the majority of samples [2,5]. Thus, it is clear that the method can discriminate between mutants with or without ETRrelevant mutations.
One possible explanation for the discrepancy between the phenotypic and the genotypic test results obtained by direct sequencing is resistance in minor viral variants. Therefore, UDPS was used in nine samples in which a discrepancy was found. There was a concordance between the direct sequencing and the UDPS for mutations consisting of .20% of the viral population. Also, eleven additional RAMs were found by UDPS, in all cases ,20% of the viral population, which is well in line with earlier results on the detection levels of direct sequencing [32][33][34][35]. Importantly, in four patients ETR RAMs were found in the minor quasispecies. In at least three of them, the proportion of mutated virus was so high that it is possible that these mutations may have contributed to the phenotypic resistance explaining the discrepancy with the genotype obtained by direct sequencing. Thus, it is possible that our phenotypic method is more sensitive than direct sequencing in identifying minor quasispecies with RAMs in the RT. The degree and pattern of an increased sensitivity as well as the clinical relevance remains to be determined.
Also, the clinical treatment history was concordant with our phenotypic results, which corresponded to a true decreased susceptibility for ETR. Thus, during early ART failure and before the NNRTI was stopped, an increase of the IC 50 was seen in four of eight samples despite that a sensitive virus was predicted by the genotype. Also, in two patients who stopped NNRTI some weeks before the sampling, an increased IC 50 but not ETR-resistance mutations was found. For subjects who had stopped the 1 st generation NNRTI-containing regimen for one or more years, a good concordance between the methods was seen. In addition, in two subjects (7 and 16) with very poor adherence, no phenotypic resistance was identified and only K103N in one of them. These patient histories and the UDPS comparison indicate that our phenotypic method may detect resistance to ETR despite that the direct sequencing predicts a sensitive phenotype.
The cause of the discrepancies between the phenotype and the genotype results is not known presently. The mutations predictive for cross-resistance to ETR-resistance have been mainly identified in clinical studies using direct sequencing [36] and it cannot be excluded that not-yet described mutations exist which may influence the ETR sensitivity. The effects of the defined mutations may also vary in different genetic environments. It seems less likely that the replicative capacity of the virus influenced the results although the ratio of relative RT activity to infectious titer has been reported to group according to virus tropism [37]. Thus, there was no correlation between the viral load in plasma, the CD4 T-cell counts and the IC 50 of the phenotypic test (data not shown). Also, we have previously shown that the RT amount can vary within a range of 40-828 fg RT/ml without affecting the determination and reproducibility of IC 50 values [22]. It should be remembered that our phenotypic method uses lysates originating from intact virions. In contrast, the genotypic and phenotypic methods based on recombinant viruses use HIV RNA which partly may represent defective virus [37,38]. Actually, it has been reported that endogenous RT activity, not p24 content or viral RNA load, is the best surrogate measure of infectious HIV-1 titer in both cell-free supernatants and viruses purified on sucrose cushions [37]. It can therefore be speculated that the viral population which is studied with our phenotypic assay is more representative for the ongoing viral replication.
The clinical utility of the phenotypic method remains to be established. It is clear that there is a strong correlation between the results of our phenotypic method and the predicted antiviral scores according to three genotypic scoring systems. A high IC 50 was found in all samples with key-mutations, Y181C and L100I, which are associated to ETR resistance. However, in a substantial number of samples there was a phenotypically decreased susceptibility but no known ETR-RAMs. The clinical relevance of these findings and clinical cut off of our method should be studied on larger patient-populations. However, in view of the high potency of newer anti-HIV drugs and the low failure rate of todays ART, such a study is likely to demand samples from a larger international study.
In the present perspective, it is believed that the RT-based phenotypic method for detection of ETR resistance in plasma is possible it could be used as an alternative for studying resistance to first and second generation NNRTI. While waiting for studies on its clinical usefulness, it could also be useful for studies on the kinetics of NNRTI resistance during ART failure.