CCR5 antagonists are a relatively new class of antiretroviral drugs for the treatment of HIV-1. Currently, maraviroc (MVC) (ViiV Healthcare) is the only clinically approved CCR5 antagonist.1 HIV-1 cellular entry is dependent on interactions of two host cell receptors, the CD4 receptor in conjunction with 1 of 2 chemokine coreceptors, CCR5 or CXCR4. Viral populations may consist of CCR5-using virus, “R5”; CXCR4-using virus, “X4”; virus able to use both CCR5 and CXCR4 or a mixture of both CCR5-using and CXCR4-using virus, “D/M”.2,3 R5 viruses are responsible for the majority of new infections, whereas the presence of X4 viruses, usually in mixture with R5 virus, is generally associated with late-stage disease and poor prognosis.4,5 As the name suggests, CCR5 antagonists target CCR5-using virus exclusively by binding to the CCR5 coreceptor thus inhibiting the CCR5-gp120 interaction.6,7 As such, it is highly recommended that a tropism screening assay be performed before the initiation of a CCR5 antagonist as CCR5 antagonists are ineffective against X4 or D/M variants. The Monogram Biosciences Trofile Assay and Enhanced Sensitivity Trofile Assay (ESTA) have been the most widely used phenotypic tropism assays.8–10
The MOTIVATE trials demonstrated that MVC, when used with an optimized background therapy, can effectively reduce plasma viral load in treatment-experienced patients infected with R5 virus.11,12 The MERIT trial was a multicenter, randomized, double-blind comparative trial in drug-naive individuals consisting of three arms: MVC once daily, MVC twice daily and EFV, all in combination with Combivir [zidovudine (ZDV)/lamivudine (3TC)].13 Only patients with R5 virus determined by the original phenotypic Trofile assay (OTA) were enrolled in the trial. At week 16, the Data Safety Monitoring Board discontinued the MVC once-daily study arm due to failure to meet prespecified noninferiority criteria of MVC once daily versus EFV.13 In the primary 48-week analysis, MVC twice daily was noninferior to the EFV arm at the <400 copies per milliliter endpoint but not at the <50 copies per milliliter endpoint (65% vs. 69%) using a threshold of −10%. However, when retrospectively rescreened using the ESTA assay, approximately, 15% of patients were shown to have non-R5 virus, and the noninferiority margin was met for the coprimary endpoint of <50 copies per milliliter, suggesting the enrollment of non-R5 patients in MERIT who were unlikely to respond to MVC.14
Genotypic methods have also been developed to determine tropism to reduce the costs and practical limitations of phenotypic tropism testing. Numerous genetic determinants of tropism have been identified primarily within the V3-loop of the gp120 protein encoded by the HIV-1 env gene.15,16 From RNA extracted from plasma samples, the V3-loop can be sequenced using population-based or “deep” sequencing technologies. Resultant sequences can then be interpreted using a bioinformatics algorithm, such as geno2pheno[coreceptor] (g2P) or the position-specific scoring matrix, from which tropism may be inferred.17,18 Previously, our group retrospectively analyzed the clinical response to MVC in treatment-experienced patients of the MOTIVATE trial, where patient samples were rescreened using a population-based sequencing method.19,20
Here we retrospectively analyzed MVC treatment response in treatment-naive individuals from the MERIT trial where tropism was predicted using a population-based V3 genotypic tropism assay.
Treatment-naive patients enrolled in the MERIT trial receiving at least 1 dose of study medication, with screening tropism results available using the original Trofile, the enhanced Trofile and population sequencing assays (n = 876) were included in this analysis. Patient viral loads were ≥2000 copies per milliliter for purposes of successfully performing the Trofile assay. Of the original screening population, only those using the CCR5 coreceptor exclusively, as determined by the OTA, were included in the MERIT trial.13,14 Patients were enrolled in a ratio of 1:1:1 to the three treatment arms: EFV (600 mg taken once daily), MVC once daily (300 mg once daily), or MVC twice daily (300 mg twice daily). All patients were also assigned a fixed-dose background therapy of Combivir (ZDV/3TC) taken twice daily. Enrollment in MERIT required patients to harbor no genotypic resistance mutations to the trial drugs, including EFV, 3TC, and ZDV.13,14
The nucleic acid from stored frozen plasma samples taken during the screening visit of patients enrolled in the MERIT trial was extracted using the standard procedures for the NucliSENS easyMAG (bioMérieux, St Laurent, Canada). Staff was blinded to virological outcome data and ESTA results. The V3 loop of gp120 from the HIV-1 env gene was amplified in triplicate from the RNA extracts using nested reverse transcriptase–polymerase chain reaction (RT-PCR). Reverse transcription and first-round polymerase chain reactions (PCRs) were combined into a single step with the SuperScript III One-Step RT-PCR system with Platinum Taq High Fidelity enzyme (Invitrogen, Burlington, Canada). One-step RT-PCR was performed using forward primer SQV3F1 (5′ GAG CCA ATT CCC ATA CAT TAT TGT 3′) and reverse primer CO602 (5′ GCC CAT AGT GCT TCC TGC TGC TCC CAA GAA CC 3′). Expand high-fidelity enzyme (Roche, Laval, Canada) with forward primer SQV3F2 (5′ TGT GCC CCA GCT GGT TTT GCG AT 3′) and reverse primer CD4R (5′ TAT AAT TCA CTT CTC CAA TTG TCC 3′) were used in the second-round PCR reaction. The resultant PCR amplicons were sequenced in both the 5′ and 3′ directions using the ABI 3730 automated sequencer (Applied Biosystems, Burlington, Canada) using BigDye Terminator (Applied Biosystems). The sequencing reactions employed forward primer V3O2F (5′ AAT GTC AGY ACA GTA CAA TGT ACA C 3′) and reverse primer SQV3R1 (5′ GAA AAA TTC CCT TCC ACA ATT AAA 3′).
Population-based sequencing methods were used to generate a consensus sequence of the V3-loop (∼105bp) for each replicate. Fully automated base calling was performed using custom software “RECall”.21,22 Mixtures were determined if the area under the secondary peak exceeded 12.5% of the area under the primary peak. Sequences were processed with the clonal geno2pheno (g2P) algorithm with no additional parameters added. The previously optimized false-positive rate (FPR) cutoff of 5.75% was used, where samples below this FPR value were inferred as being non-R5. This clinical cutoff was based on analyses of the combined MOTIVATE/A4001029 trials.23 Each sample was amplified and sequenced in triplicate, and the sequence with the lowest g2P FPR value was used to assign the viral tropism. Of the 885 samples attempted, we were unable to sequence 9 (1%); 4 patients included in the analyses were amplified successfully in duplicate, of which 3 were R5 and 1 was non-R5. To investigate the effect of different cutoff points, sequences were also processed using 3 additional FPR cutoff points, 3.5%, 10%, and 20%, and the effect of triplicate analyses were explored (see Supplemental Digital Content, http://links.lww.com/QAI/A343, showing the results of each).
Outcome Measures and Predictor Variables
The predefined study endpoints included the percentage of the population able to suppress their viral load below detectable levels at week 48 [<50 copies/mL, where missing data were considered failure (M = F)]. A secondary analysis investigated the median change in plasma viral load from baseline at all time points. In the event of missing plasma viral load data, the last observation was carried forward (LOCF). A total of 244 patients were missing plasma viral load data at week 48. A survival analysis was performed based upon the time elapsed between baseline and the first sample with non-R5 virus determined using the OTA. A comparative analysis of the rate of concordance and discordance in tropism calls between the V3 genotypic and ESTA methods was also performed. The treatment response, change in plasma viral load and percent of patients with viral load <50 copies per milliliter associated with each was analyzed to week 48.
The MVC once-daily arm served as a second independent investigation of the V3 genotype assay used here. This trial arm was discontinued in January 2006 due to failure to achieve predetermined noninferiority criteria of MVC once daily.13 At this point, the once-daily arm was unblinded and patients were given the option to switch to the MVC twice-daily arm. In this study, a censored analysis consisting of only those patients who remained on MVC once daily for the duration of the study was conducted as well as an uncensored analysis consisting of all recipients of MVC once daily, including those who went on to receive MVC twice daily. Note that not all MVC once-daily samples were rescreened using ESTA, therefore, these comparisons could not be made for this arm.
Upon rescreening the MERIT EFV and MVC twice-daily populations (n = 705) using a population-based V3 genotype assay, 58 patients (8%) were identified as having non-R5 virus based on their g2P score despite having been screened as R5 by the OTA. Baseline characteristics were broadly comparable between the R5 and non-R5 groups and treatment arms (Table 1). R5 and non-R5 genotype categories were used in the analyses, including the percentage of the population able to suppress their viral load to undetectable levels at 48 weeks (<50 HIV RNA copies/mL) after the start of treatment (Table 2). Patients screened as having R5 virus by V3 genotype had similar responses at all weeks in the MVC-twice daily arm and EFV arm (Fig. 1A), whereas those screened with non-R5 virus in the MVC twice-daily arm did not (Fig. 1B). We performed the same analysis at three additional commonly used FPR cutoff points (3.5%, 10%, and 20%) which demonstrated an improved response to MVC twice daily in patients screened as having non-R5 virus when the FPR cutoff is increased, suggesting decreased sensitivity in detecting X4 virus as the FPR cutoff increases (see Figure, Supplemental Digital Content 1, http://links.lww.com/QAI/A343). At the predefined primary endpoint of 48 weeks, MVC twice daily was noninferior when compared with that of EFV in the same group of individuals (67% vs. 68% with plasma viral load <50 RNA copies/mL) (Fig. 1A; Table 2).
Median plasma viral load decline in the MVC arms was roughly 1 log10 copies per milliliter less in the population identified as having non-R5 virus by genotype when compared with the R5 population. For example, when screened as R5 by the V3 genotypic assay, patients assigned MVC twice daily had a reduction in their plasma HIV RNA viral load of approximately 3 log10 copies per milliliter after 16 weeks on treatment similar to the EFV arm (Fig. 1C). In comparison, a viral load reduction of just over 2 log10 copies per milliliter was observed after 16 weeks on MVC for patients screened as having non-R5 virus by V3 genotype (Fig. 1D). The median viral load in the non-R5 group on EFV decreased roughly 3 log10 copies per milliliter (Fig. 1D). The analysis was repeated using three additional g2P FPR cutoff points (see Figure, Supplemental Digital Content 2, http://links.lww.com/QAI/A343).
To further investigate the ability of V3 genotype to determine tropism and predict clinical outcome, patients initially assigned to the MVC once-daily arm were also investigated regardless of whether or not they subsequently switched to a MVC twice-daily regimen. Patients assigned to the MVC once-daily arm had greater and more sustained suppression of viremia if rescreened as having R5 virus (Fig. 1E) than if rescreened as having non-R5 virus (Fig. 1F). Similar patterns were observed in a censored analysis, which included only those patients who remained on the once-daily regimen, though the number of patients becomes very small beyond week 16 (data not shown). As with MVC twice daily, the analysis was repeated using three additional g2P FPR cutoff points (see Figure, Supplemental Digital Content 3, http://links.lww.com/QAI/A343).
Subsequent Changes in OTA Results
A longitudinal analysis was performed using the time to a change of tropism from R5 to non-R5 as determined by the OTA to investigate whether changes in on-therapy tropism were predicted by screening V3 genotype. Patients screened as having non-R5 virus by genotype had a greater likelihood of changing tropism than patients with R5 virus whether initially assigned to the MVC twice-daily (Fig. 2A) or MVC once-daily (Fig. 2B) arms. After two weeks on MVC twice daily, 10% of patients screened as having non-R5 virus by genotype had shifted to non-R5 by Trofile. By week 48, 35% of patients screened as having non-R5 virus by genotype had shifted to non-R5 by Trofile, compared with the shift occurring in only 10% of subjects screened as R5 by V3 genotype. To further investigate the use of different FPR cutoff points, we analyzed the data in terms of the time to change tropism from R5 to non-R5 at 3.5%, 10%, and 20%, again showing decreased sensitivity as the FPR cutoff increases (see Figure, Supplemental Digital Content 4, http://links.lww.com/QAI/A343).
Comparison With ESTA
The MVC twice-daily baseline samples were also rescreened using ESTA. There were 283 concordant R5 calls (81%) and 9 concordant non-R5 calls (3%) between ESTA and V3 genotype. Of the 59 discordant cases, there were 39 (11%) where ESTA identified non-R5 virus but V3 genotype identified R5-using virus, while the converse occurred in 20 cases (6%). Where both assays determined an R5 viral population, patients on MVC twice daily had a viral load reduction of approximately 3 log10 copies per milliliter compared with approximately 2 log10 copies per milliliter where both assays identified a non-R5 population (Fig. 3A). Of interest, in the instance of discordance between the two assays, patients showed an intermediate response after the start of MVC twice daily, with a viral load reduction of roughly 2.5 log10 copies per milliliter regardless of the direction of discordance (Fig. 3A). Similar results were obtained if concordance and discordance between the two assays were examined in terms of the percentage of patients with undetectable plasma viral loads (Fig. 3B). Patients identified as having non-R5 virus by both assays respond poorly to MVC twice daily with an average of only 15% achieving an undetectable viral load at week 48. In the instance of concordant R5 or where assay results were discordant, roughly 60%–70% of patients achieved undetectable viral loads by week 16 and sustained these to week 48. From this data, it is difficult to determine whether one assay outperformed the other in the case of discordance. Using triplicate testing resulted in the interpretation of 29 patients as having non-R5 virus, compared with only 20 if only single amplification testing was used (see Figure, Supplemental Digital Content 5, http://links.lww.com/QAI/A343).
Assay Performance in Different HIV Clades
The MERIT trial was performed at multiple centers worldwide, enabling the comparison of assay performance in B and non-B clades of HIV. Most non-B clade virus was clade C, originating in South Africa.13,24 Regardless of the screening assay used, patients with clade B virus had a greater response to MVC twice daily than non-B virus. At week 48, the percentage of the population whose virus was suppressed below 50 copies per milliliter was comparable between those rescreened as R5 by ESTA (74%) and V3 genotype (70%) who were clade B (n = 176; 194) and also similar between ESTA and V3 genotype among those with clade C virus (63% and 62%, respectively) rescreened as R5 (n = 105; 96).24 Fewer than 36 patients were infected with HIV other than clades B or C, resulting in limited power to distinguish differences in other subtypes.
This study was designed to investigate the ability of a V3 genotypic tropism assay to predict the treatment response to CCR5 antagonists in the MERIT trial which compared MVC versus EFV in treatment-naive individuals determined to have exclusively R5 virus at screening by the OTA. Within this population, V3 genotyping was able to identify patients who were less likely to respond well to MVC treatment and more likely to change tropism as defined by the OTA after exposure to MVC, but not to EFV, similar to ESTA.
The MERIT trial has particular advantages when comparing tropism screening assays in comparison to studies without treatment outcomes as endpoints. The trial was a large, worldwide, multicenter, double-blind, randomized controlled trial. Patients enrolled were treatment naive and all received the same nucleoside background therapy, reducing confounding factors. The EFV control arm was particularly useful in highlighting the difference in response for those rescreened non-R5. In addition, the MVC treatment arms (twice daily and once daily) provided two independent populations for further investigation of the genotypic method in treatment response prediction.
Numerous studies have compared V3 genotypic prediction methods and algorithms with Trofile and ESTA, the most widely used tropism assays. Initially, Trofile was found to have an apparently greater sensitivity in identifying non-R5 variants.25 In 2007, however, Skrabal et al26 compared the prediction of viral tropism using 2 phenotypic methods, including OTA, and a genotypic method consisting of population-based sequencing paired with a bioinformatics algorithm based on a support vector machine. In this study, they were able to identify 86.5% concordance in tropism call between the OTA and the genotypic method used.26
Few of these studies, however, have been able to directly investigate the relative ability of genotypic algorithms to determine tropism using actual treatment outcomes in well-powered studies as the gold standard. The V3 genotypic tropism assay described here was also employed to rescreen the MOTIVATE population, where, despite discordant results genotype was broadly comparable to the OTA in predicting treatment response after the start of MVC.20 It is interesting to note that a considerable proportion of patients achieve virological success despite having non-R5 virus by either genotype or ESTA. This is likely due to the activity of the nucleoside reverse transcriptase inhibitor background coupled with additional partial suppression of R5 variants by MVC in what are likely mixed tropic viral populations.
It is not clear why MVC response rates were higher in patients infected with clade B versus clade C HIV, but there exist a very large number of possible confounders including differences in race, gender, and regional differences in health care management. From the viewpoint of a screening assay, however, there was no obvious evidence of clade-specific decreases in performance relative to the ESTA assay between clade B and clade C.
There are a number of inherent limitations of this study, including the retrospective nature of rescreening and the resulting fact that only patients with R5 results by the OTA could be examined for treatment outcomes. However, a similar analysis was performed with the MOTIVATE trial populations where non-R5 patients by OTA were enrolled in a sister safety study, A4001029.20,27 The study was also restricted to patient screening samples. Follow-up data would have allowed for ongoing comparison in tropism assays as the study progressed and viral populations changed. In addition, the majority of patients enrolled in the study were infected with HIV-1 clades B or C, leaving only limited power to investigate other non-B clades. It should also be noted that the Combivir nucleoside reverse transcriptase inhibitor background used at the time that the MERIT trial was designed is much less commonly used now. A further limitation of this study is the sequencing technology employed. Population-based sequencing, generally, is only able to identify minority species that exist at a prevalence of approximately >20%. Therefore, samples were amplified in triplicate to increase the sensitivity for detecting non-R5 virus. The 5.75% g2P FPR cutoff point used here was optimized using data from the MOTIVATE trial and may not be ideal as the study populations of MOTIVATE and MERIT were different (treatment experienced vs. treatment naive, respectively). Although the MERIT population may not be ideal for determining an optimal FPR cutoff point (both because it is a smaller population and has been completely prescreened, unlike MOTIVATE/A4001029), the data are consistent with the suggestion that cutoffs of 5.75 to 10% FPR combined with triplicate sequence analysis are quite reasonable values for clinical interpretation of population-based V3 sequencing data (see Supplemental Digital Content, http://links.lww.com/QAI/A343). Similar analyses using “deep” sequencing, with a detection limit for minority species of approximately 2%, yield similar results and showed only a minor incremental benefit of this increased sensitivity.28
In summary, when patients enrolled in the MERIT trial were retrospectively rescreened using a population-based V3 genotypic tropism assay, MVC was found to be noninferior to EFV in the population identified as R5 by genotype, but not in the non-R5 populations. These results confirmed earlier results obtained when the same population was rescreened using ESTA. As well, the performance characteristics of ESTA and genotype are broadly similar. When population-sequencing results were discordant with ESTA, neither assay outperformed the other in terms of treatment response. These results have also been obtained using “deep” sequencing technology; with the reduced cost and broader worldwide availability, population-based genotyping is an attractive option as a routine screening tool for tropism assessment.
1. MacArthur RD, Novak RM. Maraviroc: the first of a new class of antiretroviral agents. Clin Infect Dis. 2008;47:236–241.
2. Berger EA, Murphy PM, Farber JM. Chemokine receptors as HIV-1 coreceptors: roles in viral entry, tropism and disease. Ann Rev Immunol. 1999;17:657–700.
3. Low AJ, McGovern RA, Harrigan PR. Trofile HIV co-receptor usage assay. Expert Opin Med Diagn. 2009;3:181–191.
4. Connor RI, Sheridan KE, Ceradini D, et al.. Change in coreceptor use correlates with disease progression in HIV-1 infected individuals. J Exp Med. 1997;185:621–628.
5. Xiao L, Rudolph DL, Owen SM, et al.. Adaptation to promiscuous usage of CC and CXC-chemokine coreceptors in vivo correlates with HIV-1 disease progression. AIDS. 1998;12:F137–F143.
6. Dorr P, Westby M, Dobbs S, et al.. Maraviroc (UK-427, 857), a potent, orally bioavailable, and selective small-molecule inhibitor of chemokine receptor CCR5 with broad-spectrum anti-human immunodeficiency virus type 1 activity. Antimicrob Agents Chemother. 2005;49:4721–4732.
7. Soriano V, Geretti AM, Perno CF, et al.. Optimal use of maraviroc in clinical practice. AIDS. 2008;22:2231–2240.
8. Whitcomb JM, Huang W, Fransen S, et al.. Development and characterization of a novel single-cycle recombinant-virus assay to determine human immunodeficiency virus type 1 coreceptor tropism. Antimicrob Agents Chemother. 2007;51:566–575.
9. Vandekerckhove L, Verhofstede C, Vogelaers D. Maraviroc: integration of a new antiretroviral drug class into clinical practice. J Antimicrob Chemother. 2008;61:1187–1190.
10. Reeves JD, Coakley E, Petropoulos CJ, et al.. An enhanced-sensitivity trofileTM HIV coreceptor tropism assay for selecting patients for therapy with entry inhibitors targeting CCR5: a review of analytical and clinical studies. J Viral Entry. 2009;3:94–102.
11. Gulick RM, Lalezari J, Goodrich J, et al.. Maraviroc for previously treated patients with R5 HIV-1 infection. N Engl J Med. 2008;359:1429–1441.
12. Fätkenheuer G, Nelson M, Lazzarin A, et al.. Subgroup analyses of maraviroc in previously treated R5 HIV-1 infection. N Engl J Med. 2008;359:1442–1455.
13. Cooper DA, Heera J, Goodrich J, et al.. Maraviroc versus efavirenz, both in combination with zidovudine-lamivudine, for the treatment of antiretroviral-naïve subjects with CCR5-tropic HIV-1 infection. J Infect Dis. 2010;201:803–813.
14. Saag M, Heera J, Goodrich J, et al.. Reanalysis of the MERIT study with the Enhanced TrofileTM Assay (MERIT-ES) [abstract]. Presented at: 48th ICAAC/IDSA, 46th Annual Meeting; October 25–28, 2008; Washington, DC. H-1232a.
15. Cann AJ, Churcher MJ, Boyd M, et al.. The region of the envelope gene of human immunodeficiency virus type 1 response for determination of cell tropism. J Virol. 1992;66:305–309.
16. Fouchier RAM, Greonink M, Kootstra NA, et al.. Phenotype-associated sequence variation in the third variable domain of the human immunodeficiency virus type 1 gp120 molecule. J Virol. 1992;66:3183–3187.
17. Lengauer T, Sander O, Sierra S, et al.. Bioinformatics prediction of HIV coreceptor usage. Nat Biotechnol. 2007;25:1407–1410.
18. Jensen MA, Li FS, van't Wout AB, et al.. Improved coreceptor usage prediction and genotypic monitoring of R5-to-X4 transition by motif analysis of human immunodeficiency virus type 1 env V3 loop sequences. J Virol. 2003;77:13376–13388.
19. Harrigan PR, McGovern R, Dong W, et al.. Screening for HIV tropism using population-based V3 genotypic analysis: a retrospective virological outcome analysis using stored plasma screening samples from MOTIVATE-1. Presented at: 5th IAS Conference on HIV Pathogenesis, Treatment and Prevention; July 19–22, 2009; Cape Town, South Africa. WELBA101.
20. McGovern RA, Thielen A, Mo T, et al.. Population-based V3 genotypic tropism assay: a retrospective analysis using screening samples from the A4001029 and MOTIVATE studies. AIDS. 2010;24:2517–2525.
21. Dong W, Wynhoven B, Mo T, et al.. Performance of RE_Call basecalling software for high-throughput HIV drug resistance basecalling using in-house methods [abstract]. Presented at: 11th Annual Canadian Conference on HIV/AIDS Research; April 25–28, 2002; Winnipeg, Canada. 138P.
22. Brooks JI, Woods CK, Merks H, et al.. Evaluation of an automated sequence analysis tool to standardize HIV genotyping results [abstract]. Presented at: 18th Annual Canadian Conference on HIV/AIDS Research; April 23–26, 2009; Vancouver, Canada. O061.
23. McGovern RA, Dong W, Mo T, et al.. Optimization of clinically relevant cutpoints for the determination of HIV co-receptor usage to predict maraviroc responses in treatment experienced (TE) patients using population V3 genotyping [abstract]. Presented at: 12th European AIDS Conference; November 11–14, 2009; Cologne, Germany. PE3.4/8.
24. Portsmouth S, Chapman D, Lewis M, et al.. Virologic outcome by V3 loop genotypic population sequencing and 454 ‘deep sequencing’ in clade B and non-B virus in MERIT at 48 and 96 weeks [abstract]. Presented at: XVIII International AIDS Conference; July 18–23, 2010; Vienna, Austria. TUPE0134.
25. Low AJ, Dong W, Chan D, et al.. Current V3 genotyping algorithms are inadequate for predicting X4 co-receptor usage in clinical isolates. AIDS. 2007;21:F17–F24.
26. Skrabal K, Low AJ, Dong W, et al.. Determining human immunodeficiency virus coreceptor use in a clinical setting: degree of correlation between two phenotypic assays and bioinformatic model. J Clin Microbiol. 2007;45:279–284.
27. Saag M, Goodrich J, Fatkenheuer G, et al.. A double-blind, placebo-controlled trial of maraviroc in treatment-experienced patients infected with non-R5 HIV-1. J Infect Dis. 2009;199:1638–1647.
28. Swenson LC, Dong W, Mo T, et al.. Large-scale application of “Deep” sequencing using 454 technology to HIV tropism screening [abstract]. Presented at: 17th Conference on Retroviruses and Opportunistic Infections; February 16–19, 2010; San Francisco, CA. M-268.