In order to reach WHOs goal of HCV elimination it has been suggested to prioritize populations that are actively propagating infection, given e.g. several outbreaks of HCV among HIV-infected men who have sex with men in the last decade. Genotyping the infecting virus is part of routine care to guide antiviral treatment, but commercial assays have been shown to occasionally report inaccurate results. In this study, we use next-generation sequencing (NGS) to increase the discriminatory power and evaluate the accuracy of genotyping in routine HCV care.
From the University Hospitals of Leuven, Belgium, 64 samples from patients with HIV/HCV co-infection were selected and matched with 86 samples from HCV mono-infected patients to exhibit a similar genotype distribution based on determinations with the VERSANT HCV Genotype Assay. For the co-infected patients, 30.4%, 66.1%, and 46.4% reported intravenous drug use, same-sex practices, and being born outside of Belgium, respectively. HCV genomes were generated using the veSeq-HCV protocol and an in-house optimized bioinformatics pipeline. Concordance between geno- and subtypes designated by VERSANT and the Hepatitis C Virus Phylogenetic Typing Tool v2.4 using the generated consensus sequence was determined.
When considering only the 87 samples with an associated VERSANT genotyping record and >90% of the coding region of HCV sequenced to a depth >100, the genotype distribution following NGS was: genotype 1: 72% (42: 1b, 21: 1a), genotype 4: 15% (8: 4d, 2: 4k, 1 each: 4q, 4c, 4r), genotype 3: 9% (3: 3a) and genotype 2: 3% (1 each: 2a, 2c, 2i). Despite not all samples passing quality control thresholds, 112 samples had both a genotype determined by VERSANT and the phylogenetic typing tool. Of these, 78% had identical subtypes using VERSANT and NGS, 20% had a genotype specified into one of its constituent subtypes, one sample had a different subtype (VERSANT: 1b, NGS: 1a) and one had a different genotype (VERSANT: 1a, NGS: 4d). Based on near full-genome coverage by contigs of different genotypes generated de novo, 5 samples showed signs of mixed infection not indicated by VERSANT.
While the applied sequencing strategy requires further optimisation to reliably classify all geno- and subtypes across a broad viral load range, a good overall concordance was found with the genotype determined by VERSANT. The higher resolution of NGS proves capable of resolving specific subtypes and detecting cases of potential mixed infections.