Determining human immunodeficiency virus coreceptor use in a clinical setting: degree of correlation between two phenotypic assays and a bioinformatic model

Abstract

Two recombinant phenotypic assays for human immunodeficiency virus (HIV) coreceptor usage and an HIV envelope genotypic predictor were employed on a set of clinically derived HIV type 1 (HIV-1) samples in order to evaluate the concordance between measures. Previously genotyped HIV-1 samples derived from antiretroviral-naïve individuals were tested for coreceptor usage using two independent phenotyping methods. Phenotypes were determined by validated recombinant assays that incorporate either an approximately 2,500-bp ("Trofile" assay) or an approximately 900-bp (TRT assay) fragment of the HIV envelope gp120. Population-based HIV envelope V3 loop sequences ( approximately 105 bp) were derived by automated sequence analysis. Genotypic coreceptor predictions were performed using a support vector machine model trained on a separate genotype-Trofile phenotype data set. HIV coreceptor usage was obtained from both phenotypic assays for 74 samples, with an overall 85.1% concordance. There was no evidence of a difference in sensitivity between the two phenotypic assays. A bioinformatic algorithm based on a support vector machine using HIV V3 genotype data was able to achieve 86.5% and 79.7% concordance with the Trofile and TRT assays, respectively, approaching the degree of agreement between the two phenotype assays. In most cases, the phenotype assays and the bioinformatic approach gave similar results. However, in cases where there were differences in the tropism results, it was not clear which of the assays was "correct." X4 (CXCR4-using) minority species in clinically derived samples likely complicate the interpretation of both phenotypic and genotypic assessments of HIV tropism.

FIG. 1.

Receiver operating characteristic curve of the Trofile HIV coreceptor assay versus the TRT coreceptor assay. The gray line uses the Trofile coreceptor results as the reference while varying the cutoff for the X4 OD_Sample/X4 OD_Ctrl data obtained from the TRT assay. The black line uses the TRT coreceptor results as the reference while varying the cutoff for the X4 RLU data obtained from the Trofile assay.

FIG. 2.

SVM-predicted probabilities of CXCR4 coreceptor usage. Each position on the x axis represents a sample from a single patient. Sequences containing amino acid mixtures were expanded into all possible permutations per sample, and each permutation was assigned a score based on its probability of harboring X4-containing virus (where 0 is low probability and 1 is high probability). Sample permutations were tested and scored individually, and the probability of being X4 is indicated on the y axis. Each individual sample permutation is represented by a single symbol, generating a series of vertically aligned symbols for each clinically derived sample. The samples were grouped along the x axis, labeled according to the results obtained from both phenotypic assays as “R5” or “R5/X4,” and ordered by “Trofile assay/TRT assay.” The solid triangles represent samples phenotyped as R5/X4 by both assays, open squares represent samples phenotyped as R5/X4 by the Trofile assay and as R5 by the TRT assay, solid diamonds represent samples phenotyped as R5 by the Trofile assay and as R5/X4 by the TRT assay, and open circles represent samples phenotyped as R5 by both assays. The dashed horizontal line represents the SVM cutoff, above which samples are classified as R5/X4.

FIG. 3.

Box plot summary of the predicted probability of CXCR4 coreceptor usage. Mixtures were separated into all possible permutations per sample. The samples were grouped and labeled according to the results obtained from both phenotypic assays as “R5” or “R5/X4” and ordered by “Trofile assay/TRT assay.” The maximum score observed for each isolate was used to create the box plot. The boxes represent the interquartile range, and the solid line inside each box represents the median of the maxima of the group of samples. The box plot whiskers represent the range of the data in each category, and outliers are represented as open circles.

FIG. 4.

Concordances between phenotypic coreceptor assays and an SVM bioinformatic predictor. The black bars represent the concordances between the Trofile coreceptor assay (TF) and SVM, and the gray bars represent concordances between the TRT coreceptor assay and SVM. Percent discordances are represented by the white bars.