|Home | About | Journals | Submit | Contact Us | Français|
Before starting chronic hepatitis C treatment, the viral genotype/subtype has to be accurately determined and potentially coupled with drug resistance testing. Due to the high genetic variability of the hepatitis C virus, this can be a demanding task that can potentially be streamlined by viral whole-genome sequencing using next-generation sequencing as demonstrated by an article in this issue of the Journal of Clinical Microbiology by E. Thomson, C. L. C. Ip, A. Badhan, M. T. Christiansen, W. Adamson, et al. (J Clin Microbiol. 54:2455–2469, 2016, http://dx.doi.org/10.1128/JCM.00330-16).
Hepatitis C virus (HCV) is well-known for its high genetic variability, with seven distinct major HCV genotypes and nearly 70 officially assigned subtypes (1). For this reason, even though revolutionary progress has recently been made with direct-acting antivirals (DAAs) against HCV with high efficacy rates and an improved safety profile, there is still no universal antiviral treatment available that works equally effectively against all HCV genotypes (and especially subtypes) and across all patient populations (2, 3). Recent approval by the U.S. Food and Drug Administration (on 28 June 2016) of the first all-oral, pan-genotypic, single-tablet regimen for treating HCV-infected adults with and without cirrhosis based on a combination of sofosbuvir (NS5B inhibitor) and velpatasvir (pan-genotypic NS5A inhibitor) is an important step forward, but only time will tell whether the excellent results obtained in the clinical trials across all six major HCV genotypes and their subtypes (4, 5) will be reproduced in real-life settings.
Accurate determination of the HCV genotype and selected HCV subtypes before treatment initiation has been mandatory for more than 2 decades in order to select the most appropriate treatment regimen and to decide on its duration and the use of ribavirin, and it is here to stay for at least the next few years (6). A few sequencing-based and several non-sequencing-based HCV genotyping assays are commercially available and are used worldwide for routine determination of the HCV genotype and selected subtypes. The non-sequencing-based HCV genotyping assays are mainly founded on reverse hybridization or real-time PCR. Because both assay groups belong to DNA probe-based technologies, they occasionally fail to provide unambiguous results, particularly when the genetic diversity of the target is high, as in the case of HCV. Even with the last versions of commercial non-sequencing-based HCV genotyping assays, the HCV genotype/subtype fails to be assigned in 5 to 10% of patients (7, 8). The failure to obtain an unambiguous genotyping result by probe-based assays is frustrating, but it can be overcome by the use of a supplementary test (9). However, more concerning are cases of genotype misclassification because these often remain unnoticed until the clinical consequences of misclassification are evident, for example, after treatment failure with an HCV genotype 1-specific DAA combination. Recent studies have shown a substantial number of HCV genotype/subtype misclassifications, occurring in 2% to more than 50% of patients, depending on the genotyping assay and population studies (9,–12). However, it should be stressed that the great majority of misclassifications account for erroneous assignment of the HCV genotype 1 subtype, occasionally with important clinical consequences (9,–12).
When a probe-based HCV genotyping assay fails to determine the HCV genotype (or HCV subtype) and if capacities allow, laboratories often turn to population Sanger sequencing of a carefully selected, usually relatively short part of the HCV genome using commercial sequencing-based assays or in-house sequencing protocols (10). Although population sequencing is considered more accurate than probe-based assays, in certain circumstances, population sequencing can even be inferior (10). The main disadvantage of using population sequencing for HCV genotyping is that it can determine only the most prevalent genomic variant(s) and thus underperforms in the case of simultaneous infection with different HCV genotypes/subtypes, colloquially known as mixed infection. The impact of mixed HCV infections on treatment response to current DAA combinations is still to be resolved. However, if treatment is tailored for the dominant HCV genotype (e.g., with a genotype 1-specific DAA combination) in a patient with undetected baseline mixed infection and the treatment fails due to the emergence of another preexisting minority HCV genotype (e.g., genotype 3) that was not identified in a pretreatment sample, apart from therapeutic failure, the entire situation would mistakenly be interpreted as reinfection. If the mixed infection had been identified beforehand with the use of a more-sensitive genotyping method, treatment could have been tailored accordingly (e.g., using a pan-genotypic DAA combination) and treatment failure consequently avoided (13). In addition, because only a relatively short part of the HCV genome is usually targeted, sequenced, and analyzed for genotyping purposes, recombinant forms of HCV can be overlooked and mistaken for a genotype of the sequenced region. CRF1_2k/1b is currently the most prevalent HCV recombinant form. It has been reported in several countries worldwide—the countries of the former Soviet Union, Cyprus, France, the Netherlands, Belgium, and the United States (14,–19), and its prevalence is probably underestimated. Despite the fact that in comparison to human immunodeficiency virus (HIV), not many recombinant HCV forms have been identified so far, their reliable and timely recognition is crucial for selecting the optimal treatment regimen. Recently, a sustained viral response of only 27% was achieved in patients that were initially misidentified as infected with HCV genotype 2 and therefore treated with sofosbuvir/ribavirin for 12 or 16 weeks but were later identified as infected with the CRF1_2k/1b recombinant form (19). Poor sustained viral response could be avoided with prolongation of treatment or an alternative regimen if the recombinant form were detected before the treatment initiation (19).
All-oral treatment options available in 2016 for patients chronically infected with HCV include various combinations of three DAA classes: NS3/4A protease inhibitors, NS5A inhibitors, and NS5B nucleotide analogue inhibitors or nonnucleoside inhibitors, with or without ribavirin (2, 3). Each of these DAA classes has its own characteristics regarding drug resistance (2). Nucleotide analogue NS5B inhibitors are currently the DAA class with the highest barrier to resistance, and treatment failure is rarely seen even in monotherapy with these drugs (20). Resistant variants that carry resistance-associated substitutions (RASs) with reduced susceptibility in the NS5B region that do occasionally emerge are usually not fit and quickly disappear after treatment discontinuation (20, 21). On the other hand, resistant variants carrying RASs in the NS5A and NS3 regions often preexist at baseline and have been associated with treatment failure in specific groups of patients (6). RASs that emerge during treatment in the NS5A region are also the ones that tend to persist after treatment failure for several years, possibly indefinitely, and could potentially influence retreatment results with less-potent DAA combinations (6). However, a recent study that analyzed data from 2,144 patients with HCV genotype 1a or 1b infection who received a ledipasvir/sofosbuvir combination (with or without ribavirin) and with 16% overall prevalence of detectable baseline RASs in NS5A showed that baseline RASs in NS5A have a minimal impact on patients' response to this powerful therapeutic combination (21). The study also showed that, when NS5A RASs do have an impact, they could largely be overcome by extending treatment duration or through treatment intensification (21).
Although evidence is growing to support HCV drug resistance testing in certain situations, consensus is still lacking regarding its clinical significance and cutoffs for RASs present either at baseline or before retreatment (6). Recent data have shown that the presence of more than 15% of NS5A-resistant variants in a patient's quasispecies population at baseline affects the chances for sustained viral response, especially in specific groups of patients, such as those with genotype 1a or 3 infection and cirrhosis and/or prior nonresponders to pegylated interferon (IFN)-based treatment (6). Interestingly, the proposed 15% sensitivity cutoff for detecting clinically important minority resistant variants when using next-generation sequencing (NGS) is on the order of magnitude of population Sanger sequencing. Due to the absence of commercial, standardized, and externally validated assays for reliable detection of resistant variants and lack of consensus on the list of clinically relevant RASs and interpretation and reporting of HCV resistance data, systematic and broad use of resistance testing before treatment with DAAs is not recommended at present (6). One current exception is the NS3 protease inhibitor simeprevir: until this DAA becomes obsolete, the Q80K mutation, which can confer resistance to simeprevir and is widely distributed among HCV genotype 1a strains, should be assessed before starting a simepravir-containing regimen. However, resistance testing could be beneficial in many other instances, especially to guide retreatment decisions, but in these cases, testing should be performed only in adequately equipped laboratories with longstanding experience in resistance testing, interpretation, and reporting of RASs.
Population Sanger sequencing is still the most frequently used approach for HCV resistance testing, with amplification of the targeted region with reverse transcription and PCR using primers that are genotype specific and even subtype specific. Due to the high genetic diversity of HCV and/or frequent low HCV RNA viral load present in the sample, PCR amplification commonly results in suboptimal quantities of PCR products for downstream Sanger sequencing, and nested PCR has to be performed. All of this requires the design and purchase of numerous primers for PCRs and primers for sequencing in order to obtain a bidirectional nucleotide sequence of the targeted HCV genome segment, followed by laborious optimization of each PCR. Because the desire to detect all RASs present in three DAA target genome regions (NS3, NS5A, and NS5B) already requires sequencing of one-third of the HCV genome, the possibility of sequencing the entire HCV genome instead appears to be an attractive alternative. Two approaches for obtaining HCV whole-genome sequence (WGS) are usually used: population sequencing or NGS. When striving for applicability in routine clinical practice, it soon becomes clear that population Sanger sequencing is too labor-intensive and time-consuming. Several approaches to attempting HCV WGS using NGS have been made: (i) preamplification of large overlapping fragments with specific primers, (ii) preamplification in single long-range PCR with specific primers, (iii) generation of HCV replicon transcripts coupled with single-primer isothermal amplification, and (iv) direct sequencing of RNA extracts (for samples with a high HCV RNA viral load only) (6, 22,–24). This is usually followed by de novo assembly of obtained reads (22, 23). All of the strategies mentioned for obtaining HCV WGS using NGS have some important limitations; they are considered costly, too laborious, and technically challenging for many laboratories, and the overall success rate of obtaining a reliable consensus sequence is still not adequate in many instances (22). Thus, NGS is generally perceived as not affordable and not yet ready to be adopted in routine clinical practice. In this issue of the Journal of Clinical Microbiology, Emma Thomson and Camilla L. C. Ip and colleagues compared the ability of three different NGS approaches to generate high-quality and deep WGS of HCV and evaluated their utility for diagnostics and clinical assessment (25). The study compared three high-throughput NGS methods: (i) an unselected HCV RNA metagenomic approach, (ii) preenrichment of HCV RNA by probe capture, and (iii) HCV preamplification by PCR. The metrics of sequence coverage and depth, quasispecies diversity and detection of RASs, mixed HCV genotype(s), and coinfections with other viruses were compared using a panel of samples with different HCV RNA viral loads, HCV genotypes, and mixed HCV genotypes/subtypes. Surprisingly, each NGS method generated near-complete HCV genome sequences from more than 90% of samples. Enrichment methods and PCR preamplification generated greater sequence depth and were more effective for samples with low HCV RNA viral load. All NGS methodologies accurately identified mixed-HCV-genotype infections. The consensus WGSs generated by different NGS methods were generally concordant, and the majority of RASs were consistently detected. However, methods differed in their ability to detect minor populations of RASs and coinfections with other human viruses, such as pegiviruses. The authors showed that the method of preenrichment of HCV RNA by probe capture has great potential to be used routinely in clinical practice. In this approach, HCV sequences are targeted for capture from metagenomic sequencing libraries using panels of oligonucleotide probes, but at the potential expense of missing very divergent HCV sequences. The preenrichment approach was highly effective for samples with low viral loads, generating deep coverage along the HCV genome, and is considered suitable for high-throughput screening. The metagenomic approach, although less attractive for routine use in daily clinical practice, remains interesting, because the libraries generated may be probed for additional pathogens that may contribute to disease development, and a rich data set for future pathogen discovery research is provided. The third approach studied, PCR preamplification, remains too laborious, but it may still have a role in samples with very low HCV RNA viral loads.
In conclusion, in this issue of the Journal of Clinical Microbiology, United Kingdom researchers have undoubtedly shown that NGS provides a rapid method for generating the whole HCV genome to accurately and simultaneously determine HCV genotypes/subtypes, RASs, and quasispecies diversity and to allow comprehensive viral strain analysis. In other words, NGS has great potential to become the first diagnostic one-stop shop for hepatitis C. As in every shop, customers are interested in both quality and price. Surprisingly, and in contrast to the general perception, in the authors' hands, NGS appears to be not only a rapid and accurate approach for generating whole HCV genomes but also an affordable one: sequencing by any of the methods evaluated in the study can be achieved at a cost of approximately £120/sample ($158/sample), which is a price comparable to that of existing clinical genotyping assays.
Although current NGS solutions are becoming increasingly accessible and affordable, there are still some important limitations to be considered. Standard NGS technologies produce reads that are too short to enable the detection of the physical linkages of individual single nucleotide variants across the haplotype (quasispecies) of each viral strain present. To overcome this issue, new sequencing technologies are currently emerging, such as single-molecule sequencing technologies, also known as third-generation sequencing. Third-generation sequencing has already been successfully piloted for both HCV (26, 27) and HIV-1 (28). In these pilot studies, one of the third-generation sequencing solutions, the PacBio RS II platform from Pacific Biosciences (Menlo Park, CA), provided very long, single-molecule reads allowing the full length of each targeted HCV or HIV-1 molecule to be continuously sequenced in one contiguous pass so that the linkage information between every individual mutation on the same HCV or HIV-1 molecule is well characterized (26,–28). Using the PacBio RS II platform and an appropriate (and complex) bioinformatic analysis strategy, Bull et al. obtained sequence reads greater than 9 kb that covered nearly the full-length HCV amplicon in a single read and enabled analysis of near-full-length quasispecies (26), and Bergfors et al. detected low levels of HCV NS5A-resistant variants in treatment-naive patients infected with genotypes 1a and 3a (27). As the sequence read technology continues to improve, the near-full-length sequence data will improve analyses across an array of virological interests (28), streamlining HCV diagnostics and addressing and hopefully resolving many important open questions in HCV and HIV fields.
I am grateful to Maja Lunar for her generous help with reference searches and manuscript draft preparation and to Katja Seme for critical comments.
The views expressed in this Commentary do not necessarily reflect the views of the journal or of ASM.
For the article discussed, see doi:10.1128/JCM.00330-16.