 |
|
|
| |
Quantification of HIV Tropism by "Deep" Sequencing Shows a Broad Distribution of Prevalence of X4 Variants in Clinical Samples that Is Associated with Virological Outcome
|
| |
| |
Reported by Jules Levin
CROI 2009 Montreal Feb 8-12
L Swenson1, W Dong1, T Mo1, C Woods1, A Thielen2, M Jensen3, C Glascock1, J Montaner1,4, and Richard Harrigan*1,4
1BC Ctr for Excellence in HIV/AIDS, Vancouver, Canada; 2Max-Planck-Inst for Informatics, Saarbrucken, Germany; 3Fortinbras Res, Buford, GA, US; and 4Univ of British Columbia, Vancouver, Canada
Background: "Deep" sequencing (Roche GS-FLX) detects minority HIV variants in clinical samples. Here, we performed both deep and standard sequence analyses of the HIV V3 region from individuals entering a clinical trial of maraviroc, all of whom had evidence of X4/DM virus and would not be expected to show a virological response.
Methods: Screening samples from the Pfizer A4001029 study were polymerase chain reaction (PCR) amplified in triplicate (n = 202). Tropism phenotype of all samples was X4 or DM by the standard Monogram Trofile assay as defined by the study entry criteria. Conventional (n = 153) and "deep" sequencing using the GS-FLX was performed using a "barcoding" approach, allowing the simultaneous analysis of 48 samples in both directions (n = 202) in a blinded manner. V3 genotypes were interpreted using the PSSM and/or geno2pheno algorithms.
Results: An average of >4000 V3 sequences were obtained from each sample. All samples had detectable levels of X4 HIV by deep sequencing, with 4% of patients having ≦1% inferred X4 by PSSM, 14% having 1 to 10% X4, 50% having 10 to 90% X4, and 31% of patients having more than 90% X4. Tight correlations were generally observed between the forward and reverse sequencing directions, with <4% coefficient of variation.. The percentage of X4 determined was generally similar whether the PSSM or geno2pheno interpretations were used, (72% of samples fell within 4% of each other), despite some large outliers. Standard sequence analysis failed to detect X4 in 28% of samples. This was primarily driven by the low prevalence of X4-samples which standard sequence detected as having X4 virus had a much higher proportion of X4 (median 78% X4; IQR 38 to 99% by "deep" analysis) compared to those missed by standard sequencing (median 9% X4; IQR 2.5 to 21%); p <<0.05. Preliminary analysis of viral load reductions in the twice-daily maraviroc arm showed greater response for those with <10% X4 by deep sequencing/PSSM (-1.8, -2.2, and -2.6 mean log changes at weeks 2, 4, and 8, respectively) compared to either those with >10% X4 or to those who received placebo. All of these showed <-1..75 log reductions at these points.
Conclusions: Deep sequence analyses detect and quantify low prevalence of X4 HIV within clinical isolates that are not detected by standard sequence analysis. A low prevalence of X4 was associated with improved virological response to maraviroc, even where standard Trofile indicated DM or X4 virus.
|
| |
|
|
|
|
|
