Hepatitis C virus deep sequencing for sub-genotype identification in mixed infections: A real-life experience.

BACKGROUND: The effectiveness of the new generation of hepatitis C treatments named direct-acting antiviral agents (DAAs) depends on the genotype, subtype, and resistance-associated substitutions present in individual patients. The aim of this study was to evaluate a massive sequencing platform for the analysis of genotypes and subtypes of hepatitis C virus (HCV) in order to optimize therapy.
METHODS: A total of 84 patients with hepatitis C were analyzed. The routine genotyping methodology for HCV used at the study institution (Versant HCV Assay, LiPA) was compared with a deep sequencing platform (454/GS-Junior and Illumina MiSeq).
RESULTS: The mean viral load in these HCV patients was 6.89×106±7.02×105. Viral genotypes analyzed by LiPA were distributed as follows: 26% genotype 1a (22/84), 55% genotype 1b (46/84), 1% genotype 1 (1/84), 2.5% genotype 3 (2/84), 6% genotype 3a (5/84), 6% genotype 4a/c/d (5/84). When analyzed by deep sequencing, the samples were distributed as follows: 27% genotype 1a (23/84), 56% genotype 1b (47/84), 8% genotype 3a (7/84), 5% genotype 4d (4/84), 2.5% genotype 4f (2/84). Six of the 84 patients (7%) were infected with more than one subtype. Among these, 33% (2/6) failed DAA-based triple therapy.
CONCLUSIONS: The detection of mixed infection could explain some treatment failures. Accurate determination of viral genotypes and subtypes would allow optimal patient management and improve the effectiveness of DAA therapy.

Introduction

An estimated 170 million individuals worldwide are chronically infected with hepatitis C virus (HCV), and according to the US Centers for Disease Control and Prevention, approximately 10% to 20% of them will develop severe liver disease (https://wwwnc.cdc.gov/travel/yellowbook/2018/infectious-diseases-related-to-travel/hepatitis-c). The complex genetic variability of HCV has been classified into four hierarchical strata: genotype, subtype, isolate, and quasispecies (Farci and Purcell, 2000). A consensus proposal for HCV classification was initially presented in 2005 (Simmonds et al., 2005) and updated in 2014 (Smith et al., 2014). To date, seven HCV genotypes have been confirmed, with whole-genome nucleotide sequences differing by >30%, and each can be further subdivided into related subtypes (67 confirmed), with nucleotide sequence divergence of between 15% and 30% (Smith et al., 2014).

Genotype identification has long been used in clinical practice. The different genotypes and subtypes of HCV show different responses to antiviral therapy with both the traditional treatment of pegylated interferon α/ribavirin and the currently available HCV direct-acting antiviral (DAA) drugs (Legrand-Abravanel et al., 2009, Poveda et al., 2014). Therefore, accurate genotyping and subtyping of HCV is needed for the selection of the appropriate treatment regimen. The quasispecies provide a large pool of genetic variants that can adapt to new selection pressures such as the host’s cell defence and antiviral treatments, resulting in chronic infection and HCV drug resistance (Tarr et al., 2015). However, subtype identification has been used mainly in epidemiological studies. Both in vitro studies and clinical trials with different classes of DAA (NS3 protease inhibitors, NS5A inhibitors, and nucleos(t)ide and non-nucleos(t)ide NS5B polymerase inhibitors), given with pegylated interferon α/ribavirin or in interferon-free combinations, have shown lower response rates for HCV genotype 1a than for HCV genotype 1b (Poordad et al., 2014).

Currently available genotyping methods based on reverse hybridization with subtype-specific primers and probes targeting the 5′ untranscribed region (UTR) and core regions (Versant HCV Genotype 2.0 system; Siemens), as well as real-time PCR assays based on the 5′ UTR and NS5B sequencing (Abbott HCV Genotype II assay), accurately differentiate major HCV genotypes in the majority of cases and are widely used because of their technical simplicity. However, these assays have not been designed to confidently identify mixed infections, and their ability to accurately discriminate HCV subtypes other than 1a and 1b is very limited (Chevaliez et al., 2009, Avó et al., 2013, González et al., 2013).

A subtyping assay based on phylogenetic analysis of reads obtained by deep sequencing of an NS5B fragment has recently been developed (Quer et al., 2015). The aim of this study was to evaluate the effectiveness of this massive sequencing methodology for the accurate classification of HCV in the real-life setting, in order to optimize antiviral therapy.

Methods

All patients participating in this study signed an informed consent agreement. Serum samples from 84 HCV-infected patients were used, independent of clinical information. All specimens were collected and processed as recommended and stored at −80 °C. The study was approved by the Ethics Committee of Valme University Hospital. This study was conducted in accordance with the provisions of the Declaration of Helsinki and Good Clinical Practice guidelines.

HCV RNA was extracted from 650 μl of plasma or serum by automated RNA extraction using a total nucleic acid isolation (TNAI) kit (AmpliPrep System; Roche Diagnostics, Barcelona, Spain), or by manual RNA extraction (150 μl) using a QIAamp Viral RNA Mini Kit (Qiagen, Hilden, Germany).

Routine HCV genotyping was performed using the HCV Genotype 2.0 assay (LiPA HCV v.2.0; Siemens Healthcare Diagnostics, Eragny, France). This system is a line probe assay based on reverse hybridization of biotinylated products targeting the 5′ UTR and a fragment of the core region, which are hybridized to immobilized oligonucleotide probes in nitrocellulose strips that are specific for the 5′-UTR core regions of the six genotypes.

The NS5B target for HCV subtyping was a 339-nucleotide (nt) region spanning nt 8280 to nt 8618, according to data for isolate H77 available in GenBank (accession number AF009606). Primer and PCR condition data have been deposited under patent number EP13382278. Further details of the PCR methods (primers sequences, etc.) have been reported previously by Quer et al. (2015).

Deep sequencing

After nested PCR amplification, the final product is flanked by universal M13 sequences at both ends. The last PCR amplification is performed in a 96-well plate previously preloaded with a lyophilized universal primer pair consisting of an M13 universal primer plus MID plus oligonucleotide A or B (454 sequencing) (M13-MID_A/B primer; TIB Molbiol, Berlin, Germany). Each patient sample requires a different MID sequence. The PCR product of this amplification is then automatically purified using Agencourt AMPure XP magnetic beads (Beckman Coulter, Brea, CA, USA) and quantified using a fluorometric titration Quant-iT PicoGreen double-stranded DNA (dsDNA) kit (Life Technologies, Eugene, OR, USA) in an Infinity fluorometer (Tecan, Männedorf, Switzerland), with equimolecular pooling and a final enrichment step using an automated REMe STARlet Hamilton workstation (Hamilton Robotics, Reno, NV, USA).

Massive sequencing was performed using the 454/GS-Junior platform (Roche, Branford, CT, USA) and titanium chemistry (GS-Junior Titanium Sequencing kit), which enables sequencing of 400–500-nt fragments. An average of 1250 sequences (reads) was obtained (minimum 200, maximum 5240).

The FASTA file from the GS-Junior was demultiplexed to obtain a FASTA file for each sample and strand. Sequences not covering the full amplicon (positions 8280 to 8618) or showing more than 2 Ns or 3 gaps were discarded. The threshold of 90% identity was selected on the basis of the minimum observed distance between pairs of reference sequences of different subtypes.

Statistical analysis

Rates and proportions were analyzed by binomial test and Fisher’s exact test. A p-value of <0.05 was considered statistically significant.

Results

Patients attending the outpatient clinic of Valme University Hospital (January 2015–May 2016) with a chronic active HCV infection were selected and classified according to the HCV genotype determined using the HCV Genotype 2.0 assay (LiPA HCV v.2.0). All patient data were coded to maintain anonymity. Serum samples obtained from patients before treatment were used to determine viral load and genotypes. The mean viral load was 6.89 × 106 ± 7.02 × 105.

A total of 84 patients were included; 51 were male and 33 were female. Their average age at baseline was 53.8 ± 8.3 years. Thirty-five patients (42%) received triple therapy for HCV (pegylated interferon + ribavirin + telaprevir/boceprevir). Forty-nine patients (58%) received sofosbuvir-based therapy. Seventeen (14%) patients did not respond to antiviral therapy: 15 did not respond to triple therapy and two did not respond to sofosbuvir-based therapy.

The LiPA assay failed (no data after analysis) for three of the 84 samples tested (3.5%), while the ultra-deep sequencing system failed for one sample (1%), due to degraded RNA. Table 1 shows the major differences found in the identification of genotypes on comparison of the routine method with the deep sequencing technology: the HCV genotype was not properly identified for eight patients (10%) when using the routine method. The 454 technology has recently been discontinued by Roche. Therefore, the samples of a subset of 24 patients were re-analyzed: the same analysis was performed with the Illumina MiSeq platform. The same results (in terms of high-resolution genotyping) were found for this group of patients (n = 24) when compared to the data obtained with the 454/GS-Junior platform.

Table 1

Comparison of results obtained with the routine method (patients with indeterminate genotypes) and deep sequencing genotyping.

Patient code	Viral load	Versant HCV genotyping	Deep sequencing subtyping (GS-Junior)
327-20	1 430 000	4*	4d
327-25	246 000	4*	4d
68648	562 000	3*	3a
364956	6 040 000	1a	1b**
7217	548 000	1*	1a + 1b
327-61	1 620 000	4*	4d
72-69060	656 000	1a	1a + 1b
57946	59 900	1b	1a**

HCV, hepatitis C virus.

Indeterminate sub-genotype.

Routine method failed in HCV genotyping.

The HCV genotype distributions for the study patients according to the genotyping method used – routine (LiPA) and deep sequencing – are shown in Figure 1. The main difference was found for patients infected with HCV genotypes 3 and 4, which could not be classified into subtypes using the routine method. For two more samples, the subtype could not be identified using LiPA. This problem was resolved when using the deep sequencing method.

Figure 1

HCV genotype distribution in patients according to the genotyping method. The LiPA method failed in the genotyping of three of the 84 patients. Deep sequencing failed for one patient. *Mixed infections were detected, with a majority of genotype 1b. **Two mixed infections 1b + 3a. See Table 2 for details.

Table 2

Mixed infections detected in HCV patients using the deep sequencing method.

Patient code	Viral load	Genotypes detected
7217	548 000	1a (2%) + 1b (98%)a
72-69060	656 000	1a (1%) + 1b (99%)
62205	3 670 000	1a (1%) + 1b (99%)
696561	2 900 000	1a (1%) + 3a (99%)b
707400	583 000	1b (99%) + 3a (1%)
331235	6 892 395	1a (3%) + 1b (97%)

HCV, hepatitis C virus.

This patient failed treatment with sofosbuvir + ledipasvir.

This patient failed treatment with sofosbuvir + daclatasvir.

A typical deep sequencing analysis output file from the analysis of 20 samples is given in the Supplementary Material (Figure S1).

The deep sequencing method was able to identify patients with HCV mixed infections (more than one HCV genotype) (Table 2). Massive sequencing analysis revealed that six of the 84 patients (7%) were infected with more than one subtype simultaneously. Four of them (4/6, 67%) were infected with subtypes 1a and 1b, while the other two (2/6, 33%) were infected with subtypes 1b and 3a.

Interestingly, two of the six patients with mixed infection (33%) failed DAA-based triple anti-HCV therapy. These two samples were analyzed using population Sanger sequencing and no resistance-associated substitutions for DAA inhibitors were detected (F. García, personal communication). Mixed genotype HCV infections cannot be detected using routine methods. Deep sequencing is the only method available that can identify these.

Discussion

Some DAAs used to treat HCV infections act in a genotype-specific manner (McNaughton et al., 2014, Bagaglio et al., 2016, Patiño-Galindo et al., 2016). The potential outcome of DAA treatment regimes used on mixed HCV infections, i.e. concurrent infections with more than one HCV genotype, has not been considered in the clinical guidelines. This is probably because commercially available genotyping methods are only capable of identifying the dominant genotype present within a mixed infection sample, leaving minor genotypes undetected.

In this study, it was demonstrated that the routine method (LiPA) failed to accurately subtype seven out of 84 infections (8.5%). This is not a minor issue from the clinical view point, since this information is useful for clinical diagnosis, guidance of clinical treatment, patient management, and HCV epidemiology studies (Tong et al., 2015, Nieto-Aponte et al., 2017).

An added benefit of deep sequencing techniques is the possibility of detecting mixed infections. In this study, six mixed infections were detected in 84 samples (7%) (Table 2). Other studies have found mixed HCV infections at rates from 5% to 25.3% (Pham et al., 2010, Thomson et al., 2011). Moreover, DAA treatment in patients with mixed infections may be associated with the occurrence of genotype switching, whereby a previously undetected minority variant drug-resistant genotype expands to replace the successfully treated majority variant genotype (Quer et al., 2016). Further work is required to assess the impact of minority variant strains on patients treated with DAA therapy.

In the present study, most of the patients received triple therapy with boceprevir or telaprevir. Two of the patients who were found to have mixed infections failed therapy with sofosbuvir + ledipasvir and sofosbuvir + daclatasvir (Table 2). Using population Sanger sequencing, no resistance-associated substitutions relevant to DAA antivirals were found. Genotype diversity also directly affects the outcome of infected patients, due to genotype-specific differences in response to treatment and disease severity (EASL, 2015). Hence, several more effective and better tolerated DAAs are now in clinical development (Wendt et al., 2014). Very recently, it was shown that NS5A resistance-associated substitutions in patients with genotype 1a HCV had a negative impact on the treatment outcome (Zeuzem et al., 2017). The present authors are currently developing a deep sequencing protocol to detect HCV resistance mutations in the NS3, NS5A, and NS5B regions.

In conclusion, subtyping analysis based on deep sequencing is much more reliable and useful than the systems currently used in routine clinical practice. However, deep sequencing methods are still expensive compared to routine methods, and a cost-effectiveness analysis is required before the wide implementation of this technique in clinical practice. The accurate determination of viral genotypes and subtypes would allow optimal patient management and the most effective therapy. Moreover, the detection of mixed infections could explain the failure of some therapies.

Ethical approval and consent to participate

This study was approved by the Ethics Committee of Valme University Hospital. This study was conducted in accordance with the provisions of the Declaration of Helsinki and Good Clinical Practice guidelines.

Availability of data and material

All data generated or analysed during this study are included in this published article.

Funding

This work was supported by grants from the Andalusian Government (#PI0892-2012) and Instituto de Salud Carlos III (PI14/01349, PI16/00337) co-financed by the European Regional Development Fund (ERDF).

Conflict of interest

The authors declare that they have no competing interests.