Correlation between genotypic (V3 population sequencing) and phenotypic (Trofile ES) methods of characterizing co-receptor usage of HIV-1 from 200 treatment-naïve HIV patients screened for Study A4001078

Assessment of HIV-1 co-receptor usage is essential to identify patients who are likely to respond to maraviroc (MVC)-containing regimens. Co-receptor usage of plasma virus from all treatment-naïve patients screened for a MVC clinical trial was assessed using phenotypic and genotypic methodologies to evaluate concordance between testing methods and to assess the quantity of CXCR4-using (non-R5) virus in samples giving discordant results. Co-receptor usage was prospectively measured using the enhanced sensitivity Trofile assay (Trofile ES) to screen patients for enrollment in Study A4001078. Population and deep sequencing methodologies were utilized retrospectively to analyze all screening samples, with co-receptor usage determined using the geno2pheno algorithm. Concordance between methods was explored using descriptive statistics. The quantity of non-R5 virus in all samples was measured using deep sequencing. Trofile ES and matched genotype results were obtained for 199screening samples. Concordance of Trofile ES with population genotyping (5.75% false-positive rate [FPR]) and deep sequencing (3.5% FPR; 2% non-R5 threshold) was 91.7% and 89.6%, respectively. Population genotype data were available for all samples with non-reportable Trofile ES results; the distribution of co-receptor usage in this set was consistent with that in the overall population: 75% (12/16) R5 and 25% (4/16) non-R5. The majority of samples contained non-R5 plasma HIV-1 RNA estimated at either <1 log(10) (62.0%) or ⩾4 log(10) (30.5%) copies/mL; the absolute amount of detectable non-R5 virus remained stable between screening and baseline visits. Samples originally classified as non-R5 by Trofile ES but R5 by population sequencing had a relatively low absolute amount of non-R5 virus (mean 2.1%, median 0.1%). The determination of co-receptor usage using either Trofile ES or genotyping methodologies showed similar frequencies of R5 and non-R5 virus in this treatment-naïve study population. For both concordant and discordant samples, population sequencing appropriately identified R5 samples with low levels of non-R5-using virus.