Comparison of Sanger sequencing for hepatitis C virus genotyping with a commercial line probe assay in a tertiary hospital

Due to its high rate of mutation, hepatitis C virus (HCV) forms viral quasispecies, classified based on the highly variable regions in the envelope protein and nonstructural 5B protein (NS5B). As a result of this large genetic variability, seven genotypes and more than 67 subtypes have been identified [1]. The global repartition of genotypes varies across countries but genotype 1 is the most common with more than 45% worldwide, followed by genotype 3 [2]. In Belgium, a recent publication showed the same distribution of HCV genotypes with 53.6% of genotype 1 (31.6% subtype 1b and 19.7% subtype 1a) followed by genotype 3 with 22.0% [3].

HCV genotyping is part of the evaluation of newly diagnosed patients and has always been important in guiding treatment [4]. In the last EASL recommendations for the treatment of hepatitis C [5], identification of HCV genotype remains an essential consideration. Since the development of pan-genotype direct acting agents (DAA), correct identification of the genotype is important to guide the choice of the DAA combination and the duration of the treatment. Indeed, several studies have shown that a misclassification of the HCV genotype can lead to therapeutic failure. Starace et al. showed that 14.9% of DAA failure are related to a genotyping error [6] and Di Maio et al. report that 6/197 (3%) of DAA-failing patients and in particular 4/7 non responders to a DAA INF-free regimen were impacted on a wrong genotype assignment [7].

In addition to its impact on the choice of treatment, genotyping is also important for epidemiological information and the development of cost-effectiveness national treatment strategies [8].

Considering that genotype is a key element in the management of patients infected with hepatitis c virus, the determination of the genotype is recommended but in recent guidelines of the European Association of the Study of the Liver (EASL) published in 2018 [5], no recommendation was made on the assay to use. It is only mentioned that “the most widely used method is based on reverse hybridization with the line probe assay” and that “a kit based on deep sequencing will soon be available” [5]. Guidelines also mentioned that genotype assays should target the 5′ untranslated region (5’UTR) and another region for identification of the genotype 1 subtype (1a or 1b), generally the core or the NS5B region [5]. In the EASL 2018 guidelines [5], it is recommended that genotypes 1(G1), 2 (G2), 3 (G3), 4 (G4), 5 (G5) and 6 (G6) and the subtypes a and b of genotype 1 (G1a and G1b) should be distinguished. Therefore, we focused this work only on subtyping G1, no analysis was done for the subtype of the other genotypes.

As mentioned above, the technique most frequently used to genotype HCV is reverse transcriptase PCR (RT-PCR), with VERSANT HCV Genotype 2.0 Line Probe Assay (VERSANT LiPA; Healthcare SIEMENS, Munich, Germany), the other most widely used commercial genotyping assays is the Abbott RealTime HCV Genotype II (Abbott GTII; Abbott, Illinois, USA) [8].

The two commercial assays detect genotypes 1 to 6 and subtypes 1a and 1b. The Abbott GTII assay targets the 5’UTR region for classification of HCV genotypes and the NS5B region for detection of the genotype 1 subtypes. The VERSANT LiPA assay targets the 5’UTR region for the genotype and the core region for subtype 1a and 1b.

These commercial assays have shown some limitations that must be kept in mind. The first limitation is the percentage of non-subtypable G1. Some publications reported that Genotype 1 subtypes were not identified by the Abbott GTII assay in 3.7 to 15.9% of cases (with a mean of 8.5%) [9‐11] and in 2.2 to 7.4% of cases with the VERSANT LiPA assay [10]. Other limitations are that the commercial assays are not always able to differentiate between G6 and G1b [10], coinfection [12] or recombination of HCV genotypes [13]. In addition, an undetermined result is obtained for some samples. In our experience, over a 2-year period (2016–2017) and 678 genotype analyses performed with the VERSANT LiPA assay in the laboratory at the Cliniques Universitaires Saint-Luc, a tertiary hospital, out of a total of 276 G1b samples, 6% (17/276) could not be distinguished from genotype 6. In addition 3.5% (24/678) of analyzed samples gave an undetermined genotype during the same period. Noppornpanth et al. reported similar results in their study [14].

Because the VERSANT LiPA was unable to provide an accurate genotype/subtype for a significant number of samples, we decided to develop an in-house method with the goal of accurately identifying the genotype for all tested samples. As a National Centre of reference for hepatitis in Belgium, we decided to develop a homebrewed sequencing assay for 5’UTR and core regions starting from VERSANT LiPA amplicons. To our knowledge, this is the first description of the sequencing of VERSANT LiPA amplicons that might be useful for many laboratories worldwide using the VERSANT LiPA to genotype samples with undetermined results. We also decided to develop in-house sequencing assay for NS5B region because it is well recognized that NS5B is the most discriminant region for HCV genotyping.

Following the arrival of the newer approved anti-HCV pangenotypic DAAs, one might think that genotype determination is of little use, however, according to the last EASL recommendations [5] for the Treatment of Hepatitis C published in 2018, determining the HCV genotype, including subtypes 1a and 1b, remains useful when choosing a treatment and its duration. Assays that are recommended for genotyping are those that can discriminate genotyping/subtyping with amplification of the 5’UTR region plus one other region, usually the NS5B or core region. Commercial assays perform these goals with a globally good performance [18]. The two most frequently used commercial assays are the VERSANT LiPA and GII Abbott assays. Regardless of the assay used, discriminating between genotype 6 and genotype/subtype 1a and 1b remains a challenge. In approximately 5% of samples, this distinction cannot be made [10]. For the GII Abbott assay, more than 90% of genotypes and subtypes would be correctly identified [10, 11, 18, 19] and Mallory et al. [11] recommended to confirm samples with coinfected genotypes or an unkown G1 subtype with another method, representing 5.5% of samples included in their cohort. The VERSANT LiPA failed to distinguish the subtype of genotype 1b in 17 samples (6%) of 276 samples over a 2-year period in our hospital. During the same period, 24 (3.5%) of 678 genotypes performed could not be determined. These percentages are concordant with the literature [14, 19‐21].

In this study, we investigated the accuracy of the VERSANT LiPA assay for genotyping HCV with sequencing of three different regions of the HCV: the 5’UTR, the core region and the NS5B region. Sequencing of the NS5B region was performed on EDTA plasma samples while sequencing of the 5’UTR and core regions was performed from VERSANT LiPA amplicons. One hundred samples were included and all were pre-genotyped by the VERSANT LiPA assay. Sixty-four VERSANT LiPA amplicons were sequenced for 5’UTR and core regions. Seventy-seven EDTA plasma samples were sequenced for the NS5B region. We showed that the amplification rate was the lowest for NS5B, only 69 samples out of 77 (89,6%) were correctly amplified, this low rate of amplification is probably explained by the fact that most of the samples included in this study were complicated samples. In addition, matrices used have undergone higher freeze-thaw cycles than a routine cohort. It is well known that the NS5B region is less preserved than the Core or 5’UTR region but in contrast, it allows the most correct HCV genotype determination [22].

The HCV genotypic distribution of the samples included in the study was not representative of the HCV genotypic prevalence in Belgium; indeed the samples included were preselected to have a mix of the different genotypes found in our country.

When taking the VERSANT LiPA assay as a reference, we showed that the global accuracy for genotyping was high if undetermined genotypes were not taken into account; globally, accuracy was greater than 95%. If we take the sequencing of the NS5B region as reference, this global accuracy fell. However, sequencing enabled discrimination of all the samples (27/100) that were not identified by the VERSANT LiPA assay. Genotype/subtype 1a, 1b and 6 were accurately identified in more than 95% of samples with sequenced Core and NS5B regions, but this percentage was lower for 5’UTR region sequencing with accurate identification in only 69% of samples. These values are concordant with those reported in the literature [10, 11, 14, 18‐21]. Indeed, it is well established that sequencing of the 5’UTR region of HCV is not sufficient to distinguish between genotype/subtypes 1a, 1b and 6 [9].

Few studies have shown that the genotyping challenge is higher for the subtyping of G1a and 1b [23‐25]. Knowing that G1a and G1b are associated with different rates of resistance-associated variants and different responses to DAAs, correct identification of G1 subtype remains important and may have implications for responses rate to DAAs [25, 26].

Based on the literature and our own data, we would be very cautious in interpreting HCV genotyping result if the Abbott test is used in the first line and we would not hesitate to do the NS5B sequencing assay as a confirmation test. When using the VERSANT LiPA assay, G1 subtyping without core region information should be confirmed by a sequencing method. All undetermined results must also be sequenced. HCV genotyping results that would be confirmed represent approximately 5% in our cohort. Taking in account the cost of these confirmation tests, we report here an assay that uses VERSANT LiPA amplicons already available allowing a significant saving of costs, workload, set up of a RT-PCR steps and would not require extra plasma samples.

We would therefore recommend the sequencing of the core region when the VERSANT LiPA amplicons are available and the sequencing of the NS5B region starting from plasma sample. As shown in the analysis of phylogenetic trees, the sequencing of the NS5B region allows for better discrimination of HCV genotyping and subtyping, the sequencing of the core region gives results close to those of NS5B sequencing, the 5′ UTR region is the least recommended of the 3 regions due to low discrimination. Homemade or commercial Sanger sequencing or Next Generation sequencing assays should be performed at least in specialized laboratories such as National Reference Centers.

The high genetic variability of HCV makes it a challenge to correctly determine genotype and subtypes using commercial assays. For undetermined samples, supplementary testing is required. Even if no G5 and G7 were included in our study, we describe new and original methods of sequencing 5’UTR and Core regions to confirm HCV genotypes not discriminated by a commercial assay, in particular by using amplicons already obtained by the VERSANT HCV Genotype 2.0 Line Probe Assay. This method thus saves costs and workload of RT-PCR steps if a confirmation assay is needed and might be of usefulness for overcoming the problem of undetermined results in many laboratories worldwide performing the VERSANT LiPA assay.