Real-World Outcomes of Direct-Acting Antiviral Treatment and Retreatment in United Kingdom–Based Patients Infected With Hepatitis C Virus Genotypes/Subtypes Endemic in Africa

Abstract Background Chronic hepatitis C virus (HCV) infection affects 71 million individuals, mostly residing in low- and middle-income countries (LMICs). Direct-acting antivirals (DAAs) give high rates of sustained virological response (SVR) in high-income countries where a restricted range of HCV genotypes/subtypes circulate. Methods We studied United Kingdom–resident patients born in Africa to examine DAA effectiveness in LMICs where there is far greater breadth of HCV genotypes/subtypes. Viral genome sequences were determined from 233 patients. Results Full-length viral genomic sequences for 26 known subtypes and 5 previously unidentified isolates covering 5 HCV genotypes were determined. From 149 patients who received DAA treatment/retreatment, the overall SVR was 93%. Treatment failure was associated primarily with 2 subtypes, gt1l and gt4r, using sofosbuvir/ledipasvir. These subtypes contain natural resistance-associated variants that likely contribute to poor efficacy with this drug combination. Treatment failure was also significantly associated with hepatocellular carcinoma. Conclusions DAA combinations give high SVR rates despite the high HCV diversity across the African continent except for subtypes gt1l and gt4r, which respond poorly to sofosbuvir/ledipasvir. These subtypes are widely distributed across Western, Central, and Eastern Africa. Thus, in circumstances where accurate genotyping is absent, ledipasvir and its generic compounds should not be considered as a recommended treatment option.

Direct-acting antivirals (DAAs) capable of clearing chronic hepatitis C virus (HCV) infection from 90%-95% of treated individuals are a cornerstone of the World Health Organization (WHO) strategy to eliminate viral hepatitis as a public health concern by 2030 [1]. Consequently, many high-income countries (HICs) have implemented action plans for controlling infection and transmission to achieve HCV eradication. By contrast, low-and middle-income countries (LMICs) face obstacles and challenges limiting their ability to implement similar strategies [2]. Moreover, clinical trials and data on DAA efficacy are almost exclusively derived from studies in HICs where circulating HCV genotypes (gt)/subtypes are more restricted than those in LMICs [3][4][5][6]. In particular, most subtypes of HCV gt4, gt5, and gt6 have been largely neglected, yet they represent about 15% of chronic HCV infections [7].
We recently reported a substantial gap in our knowledge of HCV genomic sequences circulating in LMICs across large geographic areas [8]. In addition, we and others have identified HCV gt1 and gt4 subtypes in infected patients of African origin that do not respond to DAA therapy as well as subtypes typically transmitted in HICs; treatment failure using a sofosbuvir (SOF)/ledipasvir (LDV) combination has been particularly evident with HCV subtypes gt1l [9,10] and gt4r [10][11][12][13][14]. Reduced DAA efficacy for these subtypes is frequently related to polymorphisms in DAA target proteins (herein defined as resistance-associated variants 2022;226:995-1004 [RAVs]), which can give inherent resistance to certain DAA combinations.
In this study, we address the issue of treatment outcomes in response to DAA therapy in HCV-infected individuals originating from Africa but residing in the United Kingdom (UK). We also aimed to increase the number of available complete HCV genomic sequences derived from African patients, as there is a lack of such genetic data in public databases.

Study Design and Patients
HCV-infected patients were enrolled into the HCV Research UK cohort at clinical sites in the UK [15]. All patients whose country of birth was in Africa were included in this study. Further details of the HCV Research UK clinical database are given in the Supplementary Methods. The study conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the institution's human research committee. Ethics approval for HCV Research UK was given by National Research Ethics Service (NRES) Committee East Midlands-Derby 1 (Research Ethics Committee reference 11/ EM/0314).

Next-Generation Sequencing by Metagenomics and Target Enrichment
The next-generation sequencing (NGS) methods for determining HCV viral sequences by metagenomics and target enrichment have been published previously [16] and are described in the Supplementary Methods.

HCV Sequence and Variant Analysis
Next-generation sequencing data quality was assessed using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/), and low-quality bases (Phred score <30 or read length <50 nucleotides) were trimmed using Trim Galore! script (https://www.bioinformatics.babraham.ac.uk/projects/ trim_galore/). Cleaned reads were submitted for de novo assembly using SPAdes [17]. Where de novo assembly failed to generate a full-length contig, reads were mapped to the closest reference genome. Reference genomes were selected by a kmerbased approach using unique kmers from all HCV genotypes. Reads were mapped to the reference genome (or to de novo contigs) using Tanoti read mapper (https://bioinformatics. cvr.ac.uk/software/tanoti/). Any minor variants associated with DAA resistance were identified using the GLUE software package samReporter with default settings [18,19].

Statistical Analysis
Statistical analyses were performed using Stata version 10 software. Continuous variables were reported as median (and interquartile range) and analyzed by the Mann-Whitney U test for univariate analysis. Categorical univariate analysis was performed using Fisher exact test. Multivariate analysis was carried out using logistic regression. Variables with a P value of <.1 in univariate analysis were included in multivariate analysis.

Demographics of Subjects Originating From Africa in the Cohort
Within the HCV Research UK cohort [15], 319 patients were born in 32 countries across Africa. Analysis of this African group revealed that 66.5% were male (Table 1) with an age range of 24-88 years, averaging 59 years. The most frequently cited probable sources of infection were "born abroad" (40%) and "blood/blood products" (17%) ( Table 1). "Born abroad" was included as it was significantly associated as a risk factor for infection in a previous study [20]. Only 10% of individuals probably acquired infection through injection drug use, considerably lower than that for white UK nationals from the HCV Research UK cohort (n = 5360/8419 [64%]). Moreover, 44 cases were cited as "other" (14% of the cohort) as a probable source of infection; within this category, there was a range of possible sources of infection, but none were reported as having acquired infection in the UK (Supplementary Table 1). Indeed, for 11 individuals, country of birth was recorded specifically as the likely location for acquiring infection (Table 1 and Supplementary  Table 1).
b For 2 individuals, country of birth was recorded as the likely location for acquiring infection.
We compared the genotypes/subtypes identified by NGS analysis with those recorded by the clinical site that used commercial kits (such as the INNO-LiPA test [Innogenetics]). We obtained genotype and subtype data for both commercial assays and NGS for 133 and 60 samples, respectively (genotype and subtype). HCV genotype was identical in 124 samples (93%), but the percentage of concordant HCV subtypes was much lower (n = 41/60 samples [68%]). The majority of mismatches corresponded to subtypes that are not typically identified or differentiated in commercial kits (data not shown).

Outcomes of DAA Therapy
We evaluated treatment outcomes for all DAA drug combinations used to treat the cohort, yielding data on 149 patients. This group included 12 patients who had received prior DAA therapy on 1 (n = 11) or 2 (n = 1) occasions. Hence, there was a total of 162 treatment episodes recorded ( Table 2). Two patients were lost to follow-up or died before treatment outcome was known but were included in the overall analysis.
From univariate analysis, presence of HCC (P = .006) was significantly associated with DAA failure; a higher failure rate was noted also for those with decompensated liver disease (P = .061; Supplementary  IFN] and DAA based), treatment-naive patients were significantly more likely to respond compared with treatment-experienced individuals (Supplementary Table 5). Stratifying the various treatment regimens according to either IFN or DAA targeted at HCV proteins (NS3, NS5A, and NS5B) revealed that prior exposure to antivirals against NS5A and NS5B was significantly associated with subsequent treatment failure (Supplementary Table 5).
To identify the basis of this finding, we examined the viral sequence data for the group that received DAA therapy. In total, viral sequences were available for 131 patients who had received 144 treatment episodes. The most striking observation from analyzing the different DAA combinations for these patients  was low SVR rates for gt1l and gt4r with NS5A/NS5B antiviral combinations (0% and 44%, respectively; Table 3). Statistical analysis of both HCV subtypes combined showed that a low SVR was highly significant in both univariate and multivariate analysis compared to other subtypes (Supplementary Table 5). By comparison, high SVR rates were achieved for other subtypes that are common across Africa but found far less frequently in HCV-infected populations in other continents.

DAA-Based Therapy and Retreatment Outcomes
In terms of virological outcomes (ie, excluding patients either lost to follow-up or who died before outcome was known), there were 24 treatment episodes with DAA that led to treatment failure ( Table 2) in 19 patients. Final records for these patients showed that 9 were successfully retreated. Four patients who were retreated did not achieve SVR and, therefore, 10 patients remained viremic (Table 4). Five patients were retreated with the same drug combination and only 2 (both infected with gt3a) achieved SVR. The other 3 patients were infected with either gt1l (n = 2) or gt4r (n = 1). Six of 7 patients who received retreatment with more potent DAA combinations, either glecaprevir (GLE)/ pibrentasvir (PIB) or SOF/velpatasvir (VEL)/voxilaprevir (VOX), achieved SVR; the 1 retreatment failure in this category was infected with gt4r and received SOF/ VEL/VOX (Table 4).

Lower Response to DAA Therapy
Given the frequency of gt1l and gt4r treatment failure, we examined all individuals in the cohort infected with these HCV subtypes for both their viral sequences and outcomes from antiviral therapy. Seven patients were infected with gt1l (Supplementary Table 4), 6 of whom received DAA therapy (Table 5). Twentysix patients were infected with gt4r (Supplementary Table 4), 12 of whom received at least 1 episode of DAA therapy (Table 5).
Combining the data for gt1l and gt4r indicated that there were 13 instances of DAA treatment failure, 10 of which resulted from SOF/LDV treatment; the remaining treatment failures arose from use of SOF/daclatasvir (DCV) (n = 2) for a gt1l-infected individual and SOF/VEL/VOX for retreatment in a gt4r infection (Table 4). Previously, we have shown that sequences from these subtypes can share a common motif in NS5A, M 28 R 30 M 31 , that could give resistance to NS5A inhibitors and thereby reduce effectiveness of DAA therapy [10]. From examining the NS5A coding region for all gt1l and gt4r viral sequences in the cohort, we found that gt1l had 2 patterns at positions 28, 30, and 31; methionine was invariant at positions 28 and 31, while position 30 encoded either arginine or glutamine ( Table 5). The same positions in gt4r had more complex amino acid combinations with 6 distinct patterns of residues identified in the 11 patients who received DAA treatment (Table 5). Position 30 encoded an arginine residue that was invariant, while position 31 encoded predominantly leucine with methionine present at a lower frequency. The highest variability was observed at position 28 with methionine found at highest frequency, but 4 other amino acids were also encoded at this position (Table 5).

DISCUSSION
Our study had 2 major objectives; first, to analyze DAA treatment outcomes for individuals originating from countries across Africa, and second, to determine the HCV sequences circulating in the cohort. Given the sparsity of viral sequences at RAV positions and real-world outcomes of DAA therapy from African individuals, our study addresses a critical gap in such information from LMICs [8].
The cohort totaled 319 individuals, originating from 32 African countries. This gives a broad geographical spread with larger groups coming from Egypt (Northern Africa), Nigeria (Western Africa), DRC/Republic of Congo (Central Africa), Somalia (Eastern Africa), and South Africa (Southern Africa). For most individuals, HCV infection was likely acquired in their country of origin given their more advanced stage of liver disease (Table 1 and Supplementary Table 2) and the  diversity of genotypes and subtypes represented in our analysis (Supplementary Tables 3, 4, and 6), as reported previously for African individuals [22][23][24][25]28]. In addition, some individuals reported likely transmission in their country of origin (Supplementary Table 1). We cannot exclude the possibility that a proportion of infections occurred in the UK or in countries from which individuals did not originate. In the UK, HCV is predominantly found in people who inject drugs (PWID), and the genotypes observed are almost exclusively gt1 (53%) and gt3 (41%; Supplementary Table 6). By contrast, African individuals had much lower occurrence of gt3 infection (8%) but far higher prevalence of gt4 infection (47% compared to 0.5% in the UK PWID population; Supplementary Table 6). Moreover, the subtypes identified in the UK PWID population are typically gt1a, gt1b, and gt3a (data not shown), but there is greater diversity of gt1 subtypes in the African group (Supplementary Tables 3  and 4) and there were no cases of gt3h in the UK PWID cohort. We did attempt to distinguish gt1a, gt1b, and gt3a strains in African individuals compared to UK-based infections, but phylogenetic analysis did not discriminate between possible UK and African strains (data not shown). Thus, it is highly probable that for the African group, HCV infection occurred in their country of origin. Far larger studies would be required to determine whether the higher prevalence of liver disease was associated with any virological factors given the infrequency of many of the subtypes outside Africa. Almost half of the cohort was treated with 9 different DAA regimens including NS3 protease + NS5A inhibitors, NS5A + NS5B inhibitors, and drug combinations against all 3 viral targets. The most frequently prescribed DAA therapies were SOFbased regimens (104/162 [64%]). SOF was occasionally used as mono-DAA therapy with ribavirin (RBV) but was mostly prescribed in dual combination with NS5A inhibitors with or without RBV (LDV, VEL, or DCV) or triple combination (VEL/ VOX) ( Table 2). The overall SVR for SOF-based treatment was 79% (82/104). This relatively low SVR rate was primarily a consequence of relapse with SOF/LDV in gt1l-and gt4r-infected patients ( Table 2). These findings confirm previous reports from ourselves and others on the lower SVR rates achieved with SOF/ LDV therapy for both of these subtypes [9][10][11][12][13][14]. By contrast, 8 of 9 patients (89%) infected with either gt1l or gt4r achieved SVR following treatment with either NS3/NS5A combinations or triple SOF/VEL/VOX treatment.

Table 4. Characteristics of Patients With Unsuccessful Direct-Acting Antiviral Treatment Episodes and Outcomes of Retreatment
Aside from gt1l and gt4r, the SVR for other gt1 and gt4 subtypes was 97%. This includes gt4d, another unusual subtype associated with treatment failure with ombitasvir (OBV)/ paritaprevir (PTV)/ritonavir (RTV) in a clinical trial [29]. There were 4 gt4d-infected patients in our study, all of whom achieved SVR with OBV/PTV/RTV (n = 1) and SOF-based therapy (n = 3). Thus, it is likely that there are few gt1 and gt4 subtypes, other than gt1l and gt4r, circulating in Africa that would not give high SVR rates for the various DAA therapies. Nonetheless, exceptions could arise such as the unassigned gt1 strain from a Nigerian patient who had a T28 + S30 + N93 amino acid combination in NS5A and was a responder-relapser to grazoprevir/ elbasvir treatment (patient 2 in Table 4). Moreover, the gt1ainfected case, who was classified as a nonresponder to GLE/PIB, had a natural M31 RAV in NS5A that appears infrequently in gt1a sequences but could arise more frequently in certain LMIC populations that have not been extensively analyzed. Thus, continued surveys of viral sequences in Africa and outcomes from therapy would be beneficial.
Thirteen patients in our cohort who were not successfully cured by initial DAA therapy received retreatment, with 9 achieving SVR. Aside from 1 individual who relapsed on retreatment, the more potent DAA combinations containing GLE/PIB and SOF/VEL/VOX achieved cure in 6 patients. SOF/ VEL/VOX retreatment was unsuccessful in 1 individual with gt4r (Table 4) who had received previous unsuccessful SOF/ LDV therapy. The reason for retreatment failure with SOF/VEL/  VOX is not known since there were no serial samples available following SOF/LDV treatment. In addition, the patient did not have underlying severe liver disease. We also did not find any potential resistance-associated substitutions (RASs) in the NS3 protein that could explain lack of efficacy for VOX. It is possible that initial SOF treatment could have led to emergence of an S282T RAS in NS5B, which would yield high levels of resistance. From previous reports, S282T appears to emerge at higher frequency with gt4r than other HCV genotypes and subtypes [10,13]. This RAS did emerge transiently in 1 patient in our study (patient 9 in Table 4) following initial SOF/LDV treatment. We also found a V321I variant [12] in the NS5B sequences of 6 gt4r-infected patients who received SOF treatment; 3 patients achieved SVR whereas the remainder relapsed from therapy. Notably, in the French study that documented successful retreatment outcomes in 7 patients infected with gt4r, again GLE/PIB and SOF/VEL/VOX were used in most retreatment regimens [13]. Overall, studies from our group and others agree that SOF/LDV therapy is suboptimal and should not be recommended in regions with potentially significant prevalence of gt1l and gt4r infection. This recommendation is reinforced by a recent report using subgenomic replicons containing gt1l and gt4r NS5A sequences, which shows that variants at positions 30 and 31 identified here increase resistance to LDV [30]. It is worth noting that African [14], UK [9,10], and French studies [11][12][13] have now highlighted difficulties with treating patients of African origin infected with these subtypes using SOF/LDV therapy. Therefore, treating this patient group with a SOF/LDV combination in high-income settings should be avoided unless robust HCV genotype/subtype assays are available. In resource-limited settings where there is considerable diversity of HCV genotypes and subtypes that could include gt1l and gt4r, more potent pangenotypic DAA combinations should prove effective as either initial or rescue therapy, thereby obviating the need for sophisticated assays for viral genotyping and subtyping. During the review of this manuscript, the most recent European Association for the Study of the Liver (EASL) guidelines excluded SOF/LDV as recommended therapy, but do highlight the need for additional treatment outcome data for unusual HCV subtypes such as gt1l and gt4r with more potent DAA combinations [31].
In conclusion, our study describes HCV subtypes and viral sequences circulating in countries for which there are very limited data, and complements our recent report describing gt4 subtypes in Uganda and DRC as well as novel gt7 strains [22]. Crucially, we show that gt4r is not a "rare" subtype across large geographical regions in Africa. Indeed, excluding Egypt where gt4a dominates [26,32], gt4r is the most prevalent gt4 subtype in our study and therefore may be highly prevalent in certain African countries. Moreover, the extent of gt1l infection in Nigeria and other unassigned gt1 strains has not been fully evaluated. As highlighted in EASL guidelines, it would be prudent to continue to record outcome data to determine the optimum regimens for use in regions where such subtypes circulate.