Correlation between genotypic predictions based on V3 sequences and phenotypic determination of HIV-1 tropism

Genotypic prediction of human immunodeficiency virus type 1 CRF02-AG tropism

We assessed the performance of genotypic algorithms for predicting the tropism of human immunodeficiency virus type 1 coreceptor usage in 52 patients infected with the CRF02-AG subtype. The combined criteria of the 11/25 and net charge rules accurately detected CXCR4-using CRF02-AG viruses, whereas the Geno2pheno tool lacked sensitivity and the position-specific scoring matrix (PSSM) tool WebPSSM lacked specificity.

Current tests to evaluate HIV-1 coreceptor tropism

PURPOSE OF REVIEW

HIV-1 entry into target cells is a complex multistage process involving the envelope glycoprotein, primary cellular receptor CD4, and at least two main cellular coreceptors, CCR5 and CXCR4. The identification of the HIV-1 coreceptors led to the rapid development of several drug candidates that selectively block this interaction, that is, CCR5 or CXCR4 antagonists. Here, we review different methodologies used to determine the ability of the virus to use one or both coreceptors and their potential role in managing HIV-infected individuals treated with these novel drugs.

RECENT FINDINGS

Most commercially available HIV-1 tropism assays are cell-based (phenotypic) tests, which use different methodologies to generate env-recombinant viruses and distinct detection systems. On the other hand, a large effort is being devoted to develop more robust bioinformatic (genotypic) tools that may expedite HIV-1 tropism assays without compromising their accuracy. The main goal, however, continues to be to improve the sensitivity to detect minor CXCR4-tropic variants within the in-vivo HIV-1 quasispecies.

SUMMARY

An accurate determination, and perhaps quantification, of HIV-1 coreceptor usage is necessary for the successful management of HIV-infected individuals in the new era of entry inhibitors. Further studies, aimed to the development of novel methodologies, are essential for the success of this new class of drugs.

Design and validation of new genotypic tools for easy and reliable estimation of HIV tropism before using CCR5 antagonists

BACKGROUND

Genotypic tools may allow easier and less expensive estimation of HIV tropism before prescription of CCR5 antagonists compared with the Trofile assay (Monogram Biosciences, South San Francisco, CA, USA).

METHODS

Paired genotypic and Trofile results were compared in plasma samples derived from the maraviroc expanded access programme (EAP) in Europe. A new genotypic approach was built to improve the sensitivity to detect X4 variants based on an optimization of the webPSSM algorithm. Then, the new tool was validated in specimens from patients included in the ALLEGRO trial, a multicentre study conducted in Spain to assess the prevalence of R5 variants in treatment-experienced HIV patients.

RESULTS

A total of 266 specimens from the maraviroc EAP were tested. Overall geno/pheno concordance was above 72%. A high specificity was generally seen for the detection of X4 variants using genotypic tools (ranging from 58% to 95%), while sensitivity was low (ranging from 31% to 76%). The PSSM score was then optimized to enhance the sensitivity to detect X4 variants changing the original threshold for R5 categorization. The new PSSM algorithms, PSSM(X4R5-8) and PSSM(SINSI-6.4), considered as X4 all V3 scoring values above -8 or -6.4, respectively, increasing the sensitivity to detect X4 variants up to 80%. The new algorithms were then validated in 148 specimens derived from patients included in the ALLEGRO trial. The sensitivity/specificity to detect X4 variants was 93%/69% for PSSM(X4R5-8) and 93%/70% for PSSM(SINSI-6.4).

CONCLUSIONS

PSSM(X4R5-8) and PSSM(SINSI-6.4) may confidently assist therapeutic decisions for using CCR5 antagonists in HIV patients, providing an easier and rapid estimation of tropism in clinical samples.

HIV-1 coreceptor usage prediction without multiple alignments: an application of string kernels

BACKGROUND

Human immunodeficiency virus type 1 (HIV-1) infects cells by means of ligand-receptor interactions. This lentivirus uses the CD4 receptor in conjunction with a chemokine coreceptor, either CXCR4 or CCR5, to enter a target cell. HIV-1 is characterized by high sequence variability. Nonetheless, within this extensive variability, certain features must be conserved to define functions and phenotypes. The determination of coreceptor usage of HIV-1, from its protein envelope sequence, falls into a well-studied machine learning problem known as classification. The support vector machine (SVM), with string kernels, has proven to be very efficient for dealing with a wide class of classification problems ranging from text categorization to protein homology detection. In this paper, we investigate how the SVM can predict HIV-1 coreceptor usage when it is equipped with an appropriate string kernel.

RESULTS

Three string kernels were compared. Accuracies of 96.35% (CCR5) 94.80% (CXCR4) and 95.15% (CCR5 and CXCR4) were achieved with the SVM equipped with the distant segments kernel on a test set of 1425 examples with a classifier built on a training set of 1425 examples. Our datasets are built with Los Alamos National Laboratory HIV Databases sequences. A web server is available at http://genome.ulaval.ca/hiv-dskernel.

CONCLUSION

We examined string kernels that have been used successfully for protein homology detection and propose a new one that we call the distant segments kernel. We also show how to extract the most relevant features for HIV-1 coreceptor usage. The SVM with the distant segments kernel is currently the best method described.