- Email: [email protected]
Next-Generation Sequencing Analysis of NS5A and NS5B Minor Resistance-Associated Variants in Patients with Hcv Genotype 3 Infection who Failed Treatment with Daclatasvir plus Sofosbuvir
POSTER PRESENTATIONS statins, diuretics, or rifaximin did not affect SOF/VEL PK in Phase 2/3 studies. Conclusions: SOF/VEL exhibits a favorable DDI profile allowing use with various drugs commonly used by HCV-infected patients. FRI-169 STEADY-STATE PHARMACOKINETICS OF SOFOSBUVIR AND VELPATASVIR IN HCV-INFECTED SUBJECTS WITHOUT CIRRHOSIS, WITH COMPENSATED CIRRHOSIS, OR WITH DECOMPENSATED CIRRHOSIS IN THE PHASE 3 ASTRAL STUDIES E. Mogalian1, Di An2, Y. Zhu2, Y. Maruca3, C. Casey3, J. McNally4, J. Ling5, A. Mathias1. 1Clinical Pharmacology; 2Biometrics; 3Clinical Operations; 4 Clinical Research; 5Bioanalytical, Gilead Sciences, Inc., Foster City, United States E-mail: [email protected] Background and Aims: The Phase 3 ASTRAL studies evaluated a fixed-dose combination of sofosbuvir (SOF), a nucleotide analog HCV NS5B inhibitor, and velpatasvir (VEL), a pangenotypic HCV NS5A inhibitor, for the treatment of chronic HCV infection. This analysis describes the steady-state pharmacokinetics (PK) of SOF, SOF metabolites, and VEL in HCV-infected patients without cirrhosis, with compensated cirrhosis, and with decompensated cirrhosis. Methods: Patients enrolled in the ASTRAL-1 and ASTRAL-4 studies who consented to participate in a PK substudy had serial plasma sample drawn over a period of 24 hours at the Week 2 or Week 4 ontreatment visit. Study drug administration times were recorded for the week preceding serial PK draws. Plasma concentration of SOF, SOF metabolites, and VEL were quantified by LC/MS/MS. ASTRAL-1 patients received SOF/VEL 400/100 mg once-daily and ASTRAL-4 patients were randomized to receive SOF/VEL 400/100 mg oncedaily ± ribavirin (RBV). Results: Sixty-nine subjects without cirrhosis or with compensated cirrhosis in ASTRAL-1 had evaluable serial PK data. Thirty-three subjects with decompensated cirrhosis in ASTRAL-4 had evaluable serial PK data; 27 receiving SOF/VEL and 6 receiving SOF/VEL + RBV. Subjects with decompensated cirrhosis resulted in VEL AUCtau that was 7% (with RBV) to 14% (no RBV) lower than subjects without cirrhosis and those with compensated cirrhosis. Velpatasvir Cmax was modestly lower (27% to 41%) and Ctau was modestly higher (41% to 54%) in subjects with decompensated cirrhosis compared to those without cirrhosis and those with compensated cirrhosis. Sofosbuvir AUCtau and Cmax were higher (90% to 106% and 30 to 37%, respectively) in subjects with decompensated cirrhosis. GS–331007 ( primary circulating SOF metabolite) exposures were similar in subjects regardless of hepatic function. The differences in SOF and VEL PK were consistent with those observed in dedicated Phase 1 studies evaluating the PK of SOF and VEL in HCV-uninfected subjects with mild, moderate, or severe hepatic impairment compared to healthy subjects. Conclusions: Modest differences in the PK of SOF and VEL were observed in HCV-infected subjects with decompensated cirrhosis versus those without cirrhosis or with compensated cirrhosis. Results are consistent with previously observed SOF and VEL PK in populations with hepatic impairment and are not expected to be clinically meaningful. FRI-170 STABLE INCIDENCE OF HEPATITIS C VIRUS INFECTION AMONG PEOPLE WITH A HISTORY OF INJECTING DRUG USE IN AN AUSTRALIAN PRISON SETTING, 2005–2014: THE HITS-P STUDY E.B. Cunningham1, F. Luciani2, B. Hajarizadeh1, B. Betz-Stablein2, N.A. Bretana2, S. Teutsch2, G.J. Dore1, J. Grebely1, A.R. Lloyd2. 1Kirby Institute – UNSW; 2Inflammation and Infection Research Centre, School of Medical Sciences, Sydney, Australia E-mail: [email protected] Background and Aims: High rates of hepatitis C virus (HCV) transmission have been reported in Australian prisons. S614
Understanding the trends in incidence and factors associated with infection in this setting is crucial for the development of HCV prevention and treatment programs. This study investigated trends in HCV incidence and factors associated with HCV infection among a prisoner cohort of participants in New South Wales (NSW), Australia. Methods: Data were available from the Hepatitis C Incidence and Transmission Study in prisons (HITS-p) from 2005–2014. Temporal trends in HCV incidence were evaluated. Factors associated with time to HCV seroconversion were assessed using Cox proportional hazards regression.
Results: Among 590 participants enrolled, 328 were eligible for inclusion in this analysis (at least one follow-up visit, lifetime history of injecting drugs, and both HCV antibody and HCV RNA negative at baseline). The mean age was 28 years, 72% (n = 236) were male, 63% (n = 207) reported injecting drug use during follow-up, 24% (n = 80) reported Aboriginal identity, and 36% (n = 119) reported syringe sharing during follow-up. Over a total of 913 person-years of follow up, 101 HCV infections were observed with a median time to infection of 538 days (interquartile range: 228-1012 days). The overall HCV incidence was 11.1 per 100 person years (95% CI: 9.1–13.4 cases per 100 person-years), with a relatively stable trend over the study period (Figure). In the total population, time to incident HCV infection was independently associated with injecting daily or more [adjusted hazard ratio (aHR): 3.4; 95% CI: 1.9–6.2] and being on opioid substitution therapy (OST) (aHR: 2.1; 95% CI: 1.4–3.1). Among people who reported injecting drug use during follow-up, time to incident HCV infection was independently associated with injecting heroin (aHR: 2.5; 95% CI: 1.3–4.8) and being on OST (aHR: 1.6; 95% CI: 1.0–2.6). Conclusions: HCV incidence in Australian prisons has remained consistently high over the last decade, demonstrating an association with injecting drug use, and illustrating no protective effect of OST. These findings highlight the need for both improved harm reduction strategies and evaluation of interferon-free HCV treatment as prevention in prisons. FRI-171 NEXT-GENERATION SEQUENCING ANALYSIS OF NS5A AND NS5B MINOR RESISTANCE-ASSOCIATED VARIANTS IN PATIENTS WITH HCV GENOTYPE 3 INFECTION WHO FAILED TREATMENT WITH DACLATASVIR PLUS SOFOSBUVIR F. McPhee1, D. Hernandez1, N. Zhou1. 1Bristol-Myers Squibb, Wallingford, United States E-mail: [email protected] Background and Aims: The ALLY-3 phase 3 study assessed the efficacy of 12 weeks of daclatasvir ( pangenotypic NS5A inhibitor) + sofosbuvir (NS5B nucleotide inhibitor) in patients with HCV GT 3 infection. The majority (89%) of these patients achieved a sustained virologic response. Of the 17 patients who failed treatment, emergent
Journal of Hepatology 2016 vol. 64 | S425–S630
POSTER PRESENTATIONS variants in the HCV NS5A region at L31I, S62L, and Y93H were observed. Here, the presence of minor variants in the HCV NS5A and NS5B regions at baseline and at failure were assessed in the 17 patients who failed treatment in the ALLY-3 study using nextgeneration sequencing. Methods: Plasma samples were analysed using next-generation sequencing (Illumina technology; sensitivity ≥1% [DDL, Netherlands]). The HCV NS5A (amino acids 28, 30, 31, and 93) and NS5B (amino acids 159, 282, and 321) regions were surveyed and the results compared with those obtained using direct sequencing (sensitivity cut-off ∼20%). Results: Using next-generation sequencing, minor variants in the HCV NS5A region at M28, A30, or Y93 were observed in three patients. M28V (1.3%) or A30V (1.6%) were no longer detected at relapse in two patients with these respective variants at baseline, while the minor variant Y93H (2%) observed in one patient was enriched (99.4%) at failure. No minor variants associated with resistance to sofosbuvir were observed, including the two patients who had also failed prior treatment with sofosbuvir. At treatment failure, the only detected signature variants in NS5Awere A30K (one patient), L31I (one patient) and Y93H (15 patients; eight were emergent). All of these variants were also detected by direct sequencing. One patient had emergent NS5B S282T together with emergent NS5A Y93H at treatment failure; neither variant was detected at baseline or by post-treatment Week 12 using next-generation sequencing. Conclusions: In general, minor variants in NS5A (at 28, 30, 31, or 93) and NS5B (at 159, 282, or 321) were not detected at baseline by next-generation sequencing in patients with emergent variants associated with resistance to daclatasvir or sofosbuvir. In trying to determine which patients with HCV GT 3 infection would respond to treatment with 12 weeks of daclatasvir + sofosbuvir, next-generation sequencing delivered similar results to those obtained using direct sequencing. From these data, the presence of minor variants in NS5A or NS5B at baseline does not appear to predict response to treatment with daclatasvir + sofosbuvir. FRI-172 NEXT GENERATION SEQUENCING FOR HCV GENOTYPING AND OPTIONAL IDENTIFICATION OF RESISTANCE-ASSOCIATED VARIANTS E. Rakhmanaliev1, Z. Rui1, W. Huang1, K.S. Poon2, W.C. Cui2, J.K. Mui2, E. Koay2, E. Passomsub3, W. Chantratita3, G. Michel1. 1Vela Diagnostics Pte. Ltd.; 2Department of Laboratory Medicine, Molecular Diagnosis Centre, National University Hospital, Singapore; 3Department of Pathology, Faculty of Medicine, Ramathibodi Hospital Mahidol University, Bangkok, Thailand E-mail: [email protected] Background and Aims: Despite the advent of a number of direct acting antiviral drugs interferon treatment is still a valid therapy option, requiring HCV genotype (GT) determination before initiation of therapy. Moreover, preliminary data from several studies indicate that genotype testing, in particular GT3, might be useful for decision making for some novel drug regimens. Therefore accurate detection of HCV GTs is critical and can have significant impact on outcome of therapy. The objective of this study was to compare a widely used line probe-based test (VERSANT HCV Genotype 2.0 LiPA) and a newly developed automated next generation sequencing (NGS)-based integrated workflow, comprised of an upfront robotic liquid handling system and kits for RNA extraction and library preparation (Sentosa SQ HCV Genotyping Assay), instruments and kits for template preparation and Ion Torrent based deep sequencing, as well as bioinformatics analysis and reporting software. The latter gives full data reports on GTs 1a and 1b including known Resistance Associated Variants (RAVs). Methods: A total of 136 RAVs are included in the analysis across GTs and can be retrieved by the user. However, the system does not make
direct treatment recommendations, which are left to the respective investigator. This study included a cohort of 346 patients with chronic HCV, eligible for therapy. Archived serum and plasma samples across all 6 HCV GTs were tested on both platforms. Results: For 47/346 (13.6%) samples GT results by VERSANT were “indeterminate”. In 19/299 (6.4%) of the samples, discordant results between the two methods were obtained. All discordant and indeterminate samples were subjected to Sanger sequencing. The ability to correctly determine HCV genotypes was 93.7% (95%CI: 90.3–95.9%) for VERSANT and 100% (95%CI: 98.7–100%) for Sentosa SQ HCV Genotyping Assay. Among the 19 discordant samples, 10 GT6 were incorrectly classified as GT1b by line probing, 6 GT3 as GT4, 2 GT3 as GT6, and 1 GT1c as GT1a. GT distribution among the 47 samples indeterminate by VERSANT was: 5 GT1a, 1 GT2, 19 GT3, 1 GT4, 20 GT6 and 1 mixed infection (GTs 2 and 3). Clinical sensitivity aggregated was 86.4% (95%CI: 82.4–89.6%) for VERSANT and 100% (95%CI: 98.9–100%) for Sentosa SQ HCV Genotyping Assay. Conclusions: Our data indicate that NGS in combination with thorough bioinformatics data analysis is highly accurate for determination of HCV genotypes and identification of RAVs, both representing essential (GTs) and potentially important (RAVs) information for HCV treatment approaches. FRI-173 HCV ERADICATION DOES NOT IMPACT GUT DYSBIOSIS OR SYSTEMIC INFLAMMATION IN CIRRHOTIC PATIENTS J.S.S. Bajaj1, R. Sterling1, P. Hylemon1, D. Nixon1, M. Fuchs1, D. Heuman1, M. SIkaroodi2, P. Gillevet2. 1Virginia Commonwealth University, Richmond; 2George Mason University, Manassas, United States E-mail: [email protected] Background and Aims: Cirrhotic patients could have several reasons for their inherently harmful systemic pro-inflammatory milieu. This could be partly related to the underlying disease activity or gut dysbiosis. With the ease of HCV eradication, the impact of this sustained virological response (SVR) on the gut microbiota and systemic inflammation in HCV cirrhotic patients is important. Define the impact of HCV eradication on gut microbiota composition and systemic inflammatory milieu in cirrhosis. Methods: Outpatients with cirrhosis with only HCV as the cirrhosis etiology underwent stool and serum collection and a 3-day dietary recall. HCVRNA, genotype and SVR status were evaluated. We excluded subjects with concomitant cirrhosis etiologies, HIV coinfection and with decompensated cirrhosis (use of lactulose, antibiotics, rifaximin, probiotics). Stool microbiota composition using multi-tagged sequencing and serum inflammatory cytokines (IL-1b & TNF using ELISA) and endotoxemia were studied between HCV cirrhotics who had achieved an SVR compared to those who were still viremic. Analysis of microbiota was performed using UNIFRAC and metastats. Results: We included 105 compensated HCV cirrhotics, of which 21 had reached SVR 12 ± 3 months prior to enrollment. There was no difference in age (56 vs 57 years) or MELD score (11 vs. 12) between patients with and without SVR. Viremic patients had a median viral load of 0.9 million copies/mL and were mostly genotype 1 (77%). There was no difference in daily caloric intake (2891 ± 439 SVR vs. 2942 ± 750 no SVR). Microbiota: On UNIFRAC there was no significant differences in gut microbial composition ( p = 0.3) between groups. No changes in specific phyla relative abundance were observed (Firmicutes 45% vs 48%, Bacteroidetes 34% vs 31%, Proteobacteria 5% vs 5% in SVR/no SVR groups). When specific comparisons between families were performed (Figure), no changes in relative abundances of potentially pathogenic families (Enterobacteriaceae) or potentially beneficial ones (Clostridiales XIV,Lachnospiraceae, Ruminococcaceae) were seen.
Journal of Hepatology 2016 vol. 64 | S425–S630
S615
Understanding the trends in incidence and factors associated with infection in this setting is crucial for the development of HCV prevention and treatment programs. This study investigated trends in HCV incidence and factors associated with HCV infection among a prisoner cohort of participants in New South Wales (NSW), Australia. Methods: Data were available from the Hepatitis C Incidence and Transmission Study in prisons (HITS-p) from 2005–2014. Temporal trends in HCV incidence were evaluated. Factors associated with time to HCV seroconversion were assessed using Cox proportional hazards regression.
Results: Among 590 participants enrolled, 328 were eligible for inclusion in this analysis (at least one follow-up visit, lifetime history of injecting drugs, and both HCV antibody and HCV RNA negative at baseline). The mean age was 28 years, 72% (n = 236) were male, 63% (n = 207) reported injecting drug use during follow-up, 24% (n = 80) reported Aboriginal identity, and 36% (n = 119) reported syringe sharing during follow-up. Over a total of 913 person-years of follow up, 101 HCV infections were observed with a median time to infection of 538 days (interquartile range: 228-1012 days). The overall HCV incidence was 11.1 per 100 person years (95% CI: 9.1–13.4 cases per 100 person-years), with a relatively stable trend over the study period (Figure). In the total population, time to incident HCV infection was independently associated with injecting daily or more [adjusted hazard ratio (aHR): 3.4; 95% CI: 1.9–6.2] and being on opioid substitution therapy (OST) (aHR: 2.1; 95% CI: 1.4–3.1). Among people who reported injecting drug use during follow-up, time to incident HCV infection was independently associated with injecting heroin (aHR: 2.5; 95% CI: 1.3–4.8) and being on OST (aHR: 1.6; 95% CI: 1.0–2.6). Conclusions: HCV incidence in Australian prisons has remained consistently high over the last decade, demonstrating an association with injecting drug use, and illustrating no protective effect of OST. These findings highlight the need for both improved harm reduction strategies and evaluation of interferon-free HCV treatment as prevention in prisons. FRI-171 NEXT-GENERATION SEQUENCING ANALYSIS OF NS5A AND NS5B MINOR RESISTANCE-ASSOCIATED VARIANTS IN PATIENTS WITH HCV GENOTYPE 3 INFECTION WHO FAILED TREATMENT WITH DACLATASVIR PLUS SOFOSBUVIR F. McPhee1, D. Hernandez1, N. Zhou1. 1Bristol-Myers Squibb, Wallingford, United States E-mail: [email protected] Background and Aims: The ALLY-3 phase 3 study assessed the efficacy of 12 weeks of daclatasvir ( pangenotypic NS5A inhibitor) + sofosbuvir (NS5B nucleotide inhibitor) in patients with HCV GT 3 infection. The majority (89%) of these patients achieved a sustained virologic response. Of the 17 patients who failed treatment, emergent
Journal of Hepatology 2016 vol. 64 | S425–S630
POSTER PRESENTATIONS variants in the HCV NS5A region at L31I, S62L, and Y93H were observed. Here, the presence of minor variants in the HCV NS5A and NS5B regions at baseline and at failure were assessed in the 17 patients who failed treatment in the ALLY-3 study using nextgeneration sequencing. Methods: Plasma samples were analysed using next-generation sequencing (Illumina technology; sensitivity ≥1% [DDL, Netherlands]). The HCV NS5A (amino acids 28, 30, 31, and 93) and NS5B (amino acids 159, 282, and 321) regions were surveyed and the results compared with those obtained using direct sequencing (sensitivity cut-off ∼20%). Results: Using next-generation sequencing, minor variants in the HCV NS5A region at M28, A30, or Y93 were observed in three patients. M28V (1.3%) or A30V (1.6%) were no longer detected at relapse in two patients with these respective variants at baseline, while the minor variant Y93H (2%) observed in one patient was enriched (99.4%) at failure. No minor variants associated with resistance to sofosbuvir were observed, including the two patients who had also failed prior treatment with sofosbuvir. At treatment failure, the only detected signature variants in NS5Awere A30K (one patient), L31I (one patient) and Y93H (15 patients; eight were emergent). All of these variants were also detected by direct sequencing. One patient had emergent NS5B S282T together with emergent NS5A Y93H at treatment failure; neither variant was detected at baseline or by post-treatment Week 12 using next-generation sequencing. Conclusions: In general, minor variants in NS5A (at 28, 30, 31, or 93) and NS5B (at 159, 282, or 321) were not detected at baseline by next-generation sequencing in patients with emergent variants associated with resistance to daclatasvir or sofosbuvir. In trying to determine which patients with HCV GT 3 infection would respond to treatment with 12 weeks of daclatasvir + sofosbuvir, next-generation sequencing delivered similar results to those obtained using direct sequencing. From these data, the presence of minor variants in NS5A or NS5B at baseline does not appear to predict response to treatment with daclatasvir + sofosbuvir. FRI-172 NEXT GENERATION SEQUENCING FOR HCV GENOTYPING AND OPTIONAL IDENTIFICATION OF RESISTANCE-ASSOCIATED VARIANTS E. Rakhmanaliev1, Z. Rui1, W. Huang1, K.S. Poon2, W.C. Cui2, J.K. Mui2, E. Koay2, E. Passomsub3, W. Chantratita3, G. Michel1. 1Vela Diagnostics Pte. Ltd.; 2Department of Laboratory Medicine, Molecular Diagnosis Centre, National University Hospital, Singapore; 3Department of Pathology, Faculty of Medicine, Ramathibodi Hospital Mahidol University, Bangkok, Thailand E-mail: [email protected] Background and Aims: Despite the advent of a number of direct acting antiviral drugs interferon treatment is still a valid therapy option, requiring HCV genotype (GT) determination before initiation of therapy. Moreover, preliminary data from several studies indicate that genotype testing, in particular GT3, might be useful for decision making for some novel drug regimens. Therefore accurate detection of HCV GTs is critical and can have significant impact on outcome of therapy. The objective of this study was to compare a widely used line probe-based test (VERSANT HCV Genotype 2.0 LiPA) and a newly developed automated next generation sequencing (NGS)-based integrated workflow, comprised of an upfront robotic liquid handling system and kits for RNA extraction and library preparation (Sentosa SQ HCV Genotyping Assay), instruments and kits for template preparation and Ion Torrent based deep sequencing, as well as bioinformatics analysis and reporting software. The latter gives full data reports on GTs 1a and 1b including known Resistance Associated Variants (RAVs). Methods: A total of 136 RAVs are included in the analysis across GTs and can be retrieved by the user. However, the system does not make
direct treatment recommendations, which are left to the respective investigator. This study included a cohort of 346 patients with chronic HCV, eligible for therapy. Archived serum and plasma samples across all 6 HCV GTs were tested on both platforms. Results: For 47/346 (13.6%) samples GT results by VERSANT were “indeterminate”. In 19/299 (6.4%) of the samples, discordant results between the two methods were obtained. All discordant and indeterminate samples were subjected to Sanger sequencing. The ability to correctly determine HCV genotypes was 93.7% (95%CI: 90.3–95.9%) for VERSANT and 100% (95%CI: 98.7–100%) for Sentosa SQ HCV Genotyping Assay. Among the 19 discordant samples, 10 GT6 were incorrectly classified as GT1b by line probing, 6 GT3 as GT4, 2 GT3 as GT6, and 1 GT1c as GT1a. GT distribution among the 47 samples indeterminate by VERSANT was: 5 GT1a, 1 GT2, 19 GT3, 1 GT4, 20 GT6 and 1 mixed infection (GTs 2 and 3). Clinical sensitivity aggregated was 86.4% (95%CI: 82.4–89.6%) for VERSANT and 100% (95%CI: 98.9–100%) for Sentosa SQ HCV Genotyping Assay. Conclusions: Our data indicate that NGS in combination with thorough bioinformatics data analysis is highly accurate for determination of HCV genotypes and identification of RAVs, both representing essential (GTs) and potentially important (RAVs) information for HCV treatment approaches. FRI-173 HCV ERADICATION DOES NOT IMPACT GUT DYSBIOSIS OR SYSTEMIC INFLAMMATION IN CIRRHOTIC PATIENTS J.S.S. Bajaj1, R. Sterling1, P. Hylemon1, D. Nixon1, M. Fuchs1, D. Heuman1, M. SIkaroodi2, P. Gillevet2. 1Virginia Commonwealth University, Richmond; 2George Mason University, Manassas, United States E-mail: [email protected] Background and Aims: Cirrhotic patients could have several reasons for their inherently harmful systemic pro-inflammatory milieu. This could be partly related to the underlying disease activity or gut dysbiosis. With the ease of HCV eradication, the impact of this sustained virological response (SVR) on the gut microbiota and systemic inflammation in HCV cirrhotic patients is important. Define the impact of HCV eradication on gut microbiota composition and systemic inflammatory milieu in cirrhosis. Methods: Outpatients with cirrhosis with only HCV as the cirrhosis etiology underwent stool and serum collection and a 3-day dietary recall. HCVRNA, genotype and SVR status were evaluated. We excluded subjects with concomitant cirrhosis etiologies, HIV coinfection and with decompensated cirrhosis (use of lactulose, antibiotics, rifaximin, probiotics). Stool microbiota composition using multi-tagged sequencing and serum inflammatory cytokines (IL-1b & TNF using ELISA) and endotoxemia were studied between HCV cirrhotics who had achieved an SVR compared to those who were still viremic. Analysis of microbiota was performed using UNIFRAC and metastats. Results: We included 105 compensated HCV cirrhotics, of which 21 had reached SVR 12 ± 3 months prior to enrollment. There was no difference in age (56 vs 57 years) or MELD score (11 vs. 12) between patients with and without SVR. Viremic patients had a median viral load of 0.9 million copies/mL and were mostly genotype 1 (77%). There was no difference in daily caloric intake (2891 ± 439 SVR vs. 2942 ± 750 no SVR). Microbiota: On UNIFRAC there was no significant differences in gut microbial composition ( p = 0.3) between groups. No changes in specific phyla relative abundance were observed (Firmicutes 45% vs 48%, Bacteroidetes 34% vs 31%, Proteobacteria 5% vs 5% in SVR/no SVR groups). When specific comparisons between families were performed (Figure), no changes in relative abundances of potentially pathogenic families (Enterobacteriaceae) or potentially beneficial ones (Clostridiales XIV,Lachnospiraceae, Ruminococcaceae) were seen.
Journal of Hepatology 2016 vol. 64 | S425–S630
S615