Improvement of Gut Diversity and Composition After Direct-Acting Antivirals in Hepatitis C Virus–Infected Patients With or Without Human Immunodeficiency Virus Coinfection

Abstract

Background

The influence of direct-acting antivirals (DAAs) on the composition of gut microbiota in hepatitis C virus (HCV)–infected patients with or without human immunodeficiency virus (HIV) is unclear.

Methods

We enrolled 62 patients with HCV monoinfection and 24 patients with HCV/HIV coinfection receiving elbasvir-grazoprevir from a clinical trial. Fecal specimens collected before treatment and 12 weeks after treatment were analyzed using amplicon-based 16S ribosomal RNA sequencing.

Results

Sustained virological response rates in the monoinfection and coinfection groups were similar (98.4% vs 95.8%). Pretreatment bacterial communities in the patient groups were less diverse and distinct from those of healthy controls. Compared with HCV-monoinfected patients, HCV/HIV-coinfected individuals showed comparable microbial alpha diversity but decreased Firmicutes-Bacteroidetes ratios. The improvement of microbial dysbiosis was observed in responders achieving sustained virological response across fibrosis stages but was not found in nonresponders. Responders with a low degree of fibrosis exhibited a recovery in alpha diversity to levels comparable to those in healthy controls. Reciprocal alterations of increased beneficial bacteria and reduced pathogenic bacteria were also observed in responders.

Conclusions

This study indicates a short-term effect of direct-acting antivirals in restoration of microbial dysbiosis. The favorable changes in gut microbiota profiles after viral eradication might contribute toward the reduction of HCV-related complications among infected individuals.

An estimated 70 million people globally are chronically infected with hepatitis C virus (HCV), and cirrhosis and hepatocellular carcinoma will eventually develop in a significant proportion of these individuals [1]. The risk of liver-related complications in patients with chronic HCV infection is associated with several factors, including older age at infection, heavy alcohol intake, obesity, gut microbiota, and human immunodeficiency virus (HIV) coinfection [2, 3]. Indeed, HCV/HIV-coinfected individuals are at higher risk for cirrhosis and complications than those without HIV infection [4]. The emerging of highly effective direct-acting antivirals (DAAs) has dramatically changed the standard of care in HCV infection, as sustained virological response (SVR) could be achieved in >90% [5]. HCV eradication is associated with decreased risks of hepatic and extrahepatic complications, as well as improvement in health-related quality of life [5].

Alteration of gut microbiota diversity and composition (so-called dysbiosis) and liver diseases are closely linked owing to the cross talk of the gut-liver axis [3, 6]. In this context, increased intestinal permeability and dysbiosis leading to translocation of bacterial components could involve immune activation and increased liver injury [7, 8]. Thus, gut dysbiosis is not only associated with progressive liver disease but is also connected to complications of chronic liver diseases [3, 9]. Among HCV-infected individuals, gut dysbiosis was persistently observed in mild hepatitis and showed greater alteration in association with disease progression [10, 11]. At present, data regarding the impact of HCV therapy on microbial restoration have generated conflicting results and are restricted in Western populations [12–14], whose genetics and ethnic backgrounds, as well as dietary habits and lifestyles differ, from those of Asian populations. In addition, altered gut microbiota diversity and composition after DAA therapy in HCV/HIV-coinfected patients is yet to be determined.

In the current study, we aimed to compare the diversity and composition of gut microbiota in Thai patients with HCV monoinfection or HCV/HIV coinfection. Moreover, we investigated the short-term impact of DAA therapy on the microbial communities among these patients. Our results indicated that the microbial patterns of HCV-monoinfected and HCV/HIV-coinfected patients differed significantly from those of healthy controls. Interestingly, DAA therapy significantly ameliorated gut dysbiosis in patients achieving SVR, particularly among individuals who had early fibrosis stages at baseline.

MATERIALS AND METHODS

Patients

Patients with HCV monoinfection and HCV/HIV coinfection treated with an elbasvir-grazoprevir (EBR/GZR) combination (clinicaltrials.gov; NCT03037151) at the King Chulalongkorn Memorial Hospital, Bangkok, Thailand, between August 2018 and April 2019, were recruited. Inclusion criteria were anti-HCV positivity for >6 months, detectable HCV RNA levels >10 000 IU/mL, and infection with HCV genotype 1. For HCV/HIV-coinfected patients, everyone received antiretroviral therapy (ART) with undetectable plasma HIV-RNA levels at enrollment. Treatment-naive patients were assigned to treatment with EBR/GZR for 12 weeks, while those experienced with pegylated interferon and ribavirin (RBV) were treated with EBR/GZR and RBV for 16 weeks. Patients were followed up for 12 weeks after therapy to assess SVR at 12 weeks (defined by HCV RNA level <12 IU/mL).

Pretreatment liver stiffness (LS) was measured with magnetic resonance (MR) elastography, using a Philips Ingenia MR imaging system at 3.0 T (Philips Healthcare) [15]. LS was measured in 3 sections, with an average value expressed in kilopascals. The cutoff values of LS for significant fibrosis (≥F2), advanced fibrosis (≥F3), and cirrhosis (F4) were 3.2, 4.0, and 4.6 kPa, respectively [16]. In this study, patients were further divided into 2 groups, including no or mild fibrosis (F0–1) and significant fibrosis to cirrhosis (F2–4).

Exclusion criteria were hepatitis B virus coinfection, concomitant liver diseases (including nonalcoholic fatty liver disease and alcohol liver disease), previous DAA treatment, decompensated cirrhosis, and hepatocellular carcinoma. Patients who received antibiotics, prebiotics, probiotics, or proton pump inhibitors within 2 weeks before enrollment and throughout the study period were also excluded. In this study, healthy individuals without any liver or systemic diseases were included as controls.

All participants had provided written informed consent as approved by the Institutional Review Board of the Faculty of Medicine, Chulalongkorn University (no. 378/61). The study protocol was conducted in compliance with the Helsinki Declaration and Good Clinical Practice guidelines.

Fecal Sample Collection, Microbial DNA Extraction, and Amplicon-Based 16S Ribosomal RNA Sequencing

Stool specimen collection followed the standard operating procedures for the International Human Microbiome Standard procedure [17]. Fresh stool specimens were self-collected by participants at home using DNA/RNA Shiel Fecal Collection Tube. The sample were stored at 80ºC until analysis.

Total bacterial 16s ribosomal RNA (rRNA) from feces were extracted using a bead-beating method with a Quick-DNA Fecal/Soil Microbe Midiprep Kit (Zymo Research), according to the manufacturer’s protocol. The quality and quantity of genomic DNA were evaluated with a Qubit fluorometer (Invitrogen), and the integrity of the DNA was performed by agarose gel electrophoresis.

The samples were processed and analyzed with the ZymoBIOMICS Targeted Sequencing Service for Microbiome Analysis, using the Quick-16 NGS Library Prep Kit (Zymo Research). Briefly, the hypervariable V3–V4 region of the 16S rRNA gene was amplified by custom-designed by Zymo Research to provide the best coverage of the 16S gene while maintaining high sensitivity. The ZymoBIOMICS Microbial Community DNA Standard (Zymo Research) was used as a positive control for each targeted library preparation. Negative controls (ie, blank extraction control and blank library preparation control) were included to assess the level of bioburden carried by the wet-laboratory process. The final library was sequenced on an Illumina MiSeq sequencer with a v3 reagent kit (600 cycles). The sequencing was performed with 10% PhiX spike-in.

Bioinformatics Analysis

In this analysis, 16S rRNA unique amplicon sequence variants (ASVs) were inferred from raw reads using the DADA2 pipeline [18]. Chimeric sequences were removed with the DADA2 pipeline. Taxonomy assignment was performed using Uclust from Qiime version 1.9.1 with the Zymo Research Database, a 16S database that is internally designed and curated, as reference. Composition visualization and analyses of alpha diversity, including Chao1, Shannon, and Simpson indices, and beta diversity, including weighted and unweighted UNIFRAC analyses and Bray-Curtis dissimilarity method were performed with Qiime version 1.9.1 [19]. If applicable, taxonomy with significant abundance among different groups were identified by means of linear discriminative analysis effect size (LEfSe) [20], using default settings. The principal coordinate analysis (PCoA) plots were performed with in-house scripts.

Microbial Community Functions Prediction

Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) was evaluated to predict functional profiles of microbial communities [21] using the Kyoto Encyclopedia of Genes and Genomes Orthology database [22] as a reference. Welch’s test was used to confirm significant difference of functional pathways between pretreatment and posttreatment groups.

Statistical Analysis

Statistical analyses were performed using SPSS software (version 22.0.0; SPSS). Categorical data were analyzed using χ2 tests and 1-way analysis of variance (ANOVA). The comparison of Specific relative abundances of bacterial genera between groups were compared using the Kruskal-Wallis test. The nonparametric Wilcoxon signed rank test was used to compare paired data samples, and the Mann-Whitney test to compare unpaired groups. Benjamini-Hochberg false discovery rate correction was used to adjust for multiple testing (adjusted P< .05).

RESULTS

Baseline Clinical Characteristics of Patients

A total of 62 patients with HCV monoinfection, 24 patients with HCV/HIV coinfection, and 20 healthy controls were included (Table 1). Healthy controls and HCV-monoinfected patients were significantly older than the coinfection group, whereas there was no difference in sex and body mass index among groups. Compared with healthy individuals, the patient groups had higher aspartate aminotransferase and alanine aminotransferase levels. There was no difference between the patient groups regarding other biochemical parameters, HCV RNA levels, and LS. The sustained virological response rate of the overall cohort at 12 weeks was 97.7% (84 of 86), while the rates were comparable between the monoinfection and coinfection groups (98.4% [61 of 62] vs 95.8% [23 of 24], respectively; P = .48). Significant adverse effects of RBV were not observed and none of the patients discontinued the therapy.

Table 1.

Baseline Characteristics of Healthy Controls and Hepatitis C Virus–Infected Patients

Characteristic	Healthy Controls (n = 20)	HCV Monoinfection (n = 62)	HCV/HIV Coinfection (n = 24)	P Value
Sex, no. (%)
Male	11 (55.0)	46 (74.2	18 (75.0)	.23
Female	9 (45.0)	16 (25.8)	6 (25.0)
Age, mean (SD), y	47.8 (9.0)	50.3 (10.8)	44.2 (7.5)	.04a
Mode of HIV transmission, no. (%)	…	-	MSM, 3 (12.5); IVDU, 15 (62.5); HS, 6 (25.0)	…
BMI, mean (SD)b	23.4 (2.5)	24.4 (3.6)	22.2 (4.0)	.06
Diabetes, no. (%)		10 (16.1)	3 (12.5)	.67
Hypertension, no. (%)		15 (24.2)	5 (20.8)	.74
Dyslipidemia, no. (%)		17 (27.4)	4 (16.7)	.30
Laboratory values, mean (SD)
Hemoglobin, g/dL	…	14.1 (1.2)	14.3 (2.2)	.88
WBC count, ×103/μL	…	5.9 (1.8)	5.9 (1.6)	.36
Platelet count, × 109/L	…	195.8 (72.6)	203.0 (78.5)	<.001a
Creatinine, mg/dL	…	1.6 (2.3)	1.0 (0.2)	.07
eGFR, mL/min/1.73 m2	…	91.1 (30.6)	94.4 (15.5)	.69
AST, IU/L	21.7 (6.4)	53.6 (37.4)	43.7 (17.2)	.001a
ALT, IU/L	22.8 (8.0)	63.8 (52.3)	50.9 (25.0)	.002a
Log10 HCV RNA, IU/mL	…	6.2 (0.7)	6.3 (0.7)	.53
CD4+ T-cell count, cells/μL	…	…	455.9 (151.2)	…
CD8+ T-cell count, cells/μL	…	…	646.2 (285.1)	…
CD4+/CD8+ T-cell ratio	…	…	0.5 (0.5)	…
HIV viral suppression (<50 copies/mL), no. (%)	…	…	24 (100)	…
Fibrosis stage by MRE, no. (%)
F0–1		29 (46.8)	16 (66.7)
F2–4		33 (53.2)	8 (33.3)	.10
Previous PEG-IFN/RBV therapy, no. (%)		15 (24.2)	6 (25.0)	.94
SVR12, no. (%)		61 (98.4)	23 (95.8)	.48

Abbreviations: ALT, alanine aminotransferase; AST; aspartate transaminase; BMI, body mass index; eGFR, estimated glomerular filtration rate; HCV, hepatitis C virus; HIV, human immunodeficiency virus; HS, heterosexual; IVDU, intravenous drug user; MRE, magnetic resonance elastography; MSM, men who have sex with men; PEG-IFN, pegylated interferon; RBV, ribavirin; SD, standard deviation; SVR, sustained virological response; WBC, white blood cell.

^aSignificant at P < .05.

^bBMI calculated was calculated as weight in kilograms divided by height in meters squared.

Baseline Gut Alpha and Beta Diversities

After quality filtering, denoising, read merging and chimera removal, an average of 40 220.50 ASVs per sample were obtained (Supplementary Table 1). The sequencing depths were examined by plotting the rarefaction curve from the observed species. All samples reached plateaus, suggesting that sequencing depth was sufficient to detect all the genera within each sample (Supplementary Figure 1).

We compared baseline microbial alpha diversity (Chao1, Shannon, and Simpson indices) between groups. Regarding richness of gut microbiota, Chao1 indices in the monoinfection and coinfection groups (95% CI lower-upper, 104.8 [98.2–111.5] and 113.1 [99.2–127.0], respectively) were lower than in the healthy controls (142.9 [130.5–155.3]; P < .001 and P = .002 for comparisons with monoinfection and coinfection groups, respectively). However, no difference in Chao1 indices between the patient groups was observed (P = .26) (Figure 1A). Regarding fibrosis, there was no difference in Chao1 indices between and F34 to F3–4, but significant differences (P < .001) between F012 and F34 (Supplementary Figure 2A and 2B).

Figure 1.

A–C, Baseline alpha diversity in healthy controls, patients with hepatitis C virus (HCV) monoinfection, and patients with HCV/human immunodeficiency virus (HIV) coinfection (HCV/HIV), as shown by Chao1 (A), Shannon (B), and Simpson (C) indices. D, Principal coordinate analysis (PCoA) plots of beta diversity calculated using Bray-Curtis dissimilarity method (Bray Method); samples from healthy controls (green) clustered separately from the HCV (blue) and HCV/HIV (red) groups before treatment (wk0).

For richness and evenness of taxa, our results showed that patients with HCV monoinfection had significantly lower Shannon indices than healthy controls (P = .02), while a similar trend was observed in patients with HCV/HIV coinfection but did not reach statistical significance (P = .17) (Figure 1B). There was no difference between groups in terms of Simpson index (Figure 1C). The Chao1 index was negatively correlated with aspartate aminotransferase (Spearman r = −0.222; P = .02) and alanine aminotransferase (−0.253; P = .008) (Supplementary Figure 3) but not with other parameters.

Regarding the beta diversity of bacteria composition, PCoA based on Bray-Curtis dissimilarity index was performed. The result demonstrated that the clustering of healthy controls and the patient groups was remarkably separated into 2 distinct clusters (Figure 1D), which was confirmed by permutational multivariate ANOVA (P = .001).

Baseline Bacterial Composition and Differential Abundance

Gut microbiota in all fecal samples contained 15 phyla, 27 orders, 67 families, and 208 genera. To determine whether difference in bacterial composition was established at baseline, we compared the most abundance in the monoinfection and coinfection groups with findings in healthy controls.

Comparison of the most abundant ASVs (≥1%) at phylum, family, and genus levels identified in the patients and healthy controls are shown in Figure 2A–2C. The gut microbial composition was dominated by the phyla Firmicutes (median [interquartile range], (45.3% [35.5–57.2%]), Bacteroidetes (44.1% [30.7–54.9%]), and Proteobacteria (5.2% [3.1–8.9%]). The abundance of Bacteroidetes-dominant phyla did not differ significantly among groups. The relative abundances of Firmicutes and Actinobacteria were lower in patients with HCV/HIV coinfection than in healthy controls (Mann-Whitney test, adjusted P < .05), while the abundance of Proteobacteria was higher in patients with HCV/HIV coinfection than in patients with HCV monoinfection or healthy controls (adjusted P < .05). Together, a significantly lower Firmicutes-Bacteroidetes ratio was found in HCV/HIV-coinfected patients compared with the monoinfection group (adjusted P = .02) and healthy controls (adjusted P = .008) (Supplementary Figure 4A).

Figure 2.

Composition bar plots of the top-most relative abundance used to classify baseline gut microbiota composition for healthy controls, the hepatitis C virus (HCV) group, and the HCV/ human immunodeficiency virus (HIV) group. A, Phylum level. B, Family level. C, Genus level. Relative abundances in the right panels are shown as medians with interquartile ranges. *Adjusted P < .05.

At the family level, the median abundances of Lachnospiraceae (16.9% [12.9–24.1%]), Bacteroidaceae (14.5% [7.6–32.1%]), Ruminococaceae (13.3% [8.0–23.3%]), Prevotellaceae (10.8% [1.1–33.1%]), Acidaminococcaceae (3.3% [1.2–5.5%]), Alcaligenaceae (1.9% [1.3–2.8%]), Porphyromonadaceae (1.8% [0.8–2.7%]), and Rikenellaceae (1.4% [0.3–2.7%]) were found among healthy controls and the patient groups. Differences in Bacteroidaceae, Provotellaceae and Ruminococaceae between groups were not observed. However, relative abundances of Lachnospiraceae and Coriobacteriaceae were significant lower in the patient groups, while Enterobacteriaceae was significantly less abundant in healthy controls (adjusted P < .05).

At the genus level, the abundances of dominant genera, including Bacteroides (14.5% [24]), Faecalibacterium (5.9% [6]), Prevotella (2.4% [16]), Blautia (2.3% [2]), Lachnoclostridium (2.3% [3]), Roseburia (1.7% [2]), Phascolarctobacterium (1.7% [3]), Sutterella (1.6% [2]), and Alistipes (1.1% [2]), were comparable between healthy controls and the patient groups. Compared with healthy controls, the patient groups had lower abundances of Lachnospira, Blautia, Subdoligranulum, and Ruminococcus but higher abundance of Alloprevotella (adjusted P < .05). However, no significant differences in Pretovotella-Bacteroides ratios were observed among groups (Supplementary Figure 4B).

Effect of HCV Eradication with DAAs on Gut Biodiversity

We further explored whether alpha and beta diversities and microbial community could be altered after DAA therapy. In this context, 81 of 86 individuals (94.2%) (58 with HCV monoinfection and 23 with HCV/HIV coinfection) had available posttreatment fecal specimens and were included in subsequent analyses. Changes in Chao1, Shannon, and Simpson indices from baseline in relation to treatment response are shown in Figure 3 and Supplementary Figure 5.

Figure 3.

Alpha diversity changes in patients from before treatment (Pre) to 12 weeks after treatment (Post). A, Overall cohort. B, Subgroup analysis of hepatitis C virus (HCV) infection versus HCV/human immunodeficiency virus (HIV) coinfection. C, Sustained virological response (SVR) versus non-SVR. D, Patients with fibrosis stage F0–1 versus stage F2–4.

Overall, our results showed significant difference in Chao1 and Shannon indices between healthy controls and the pretreatment group (P < .001 for Chao1 and P = .04 for Shannon index). However, we did not observe any difference in Chao1, Shannon, and Simpson indices between healthy controls and the posttreatment group. We then performed a paired Wilcoxon signed rank test (2 sided) and found that Chao1 indices were significantly increased in posttreatment compared with pretreatment groups (median [interquartile range], 139.9 [131.2–148.5] vs 107.2 [101.1–113.2], respectively; P < .001). There was no difference between healthy controls and the posttreatment group (P = .86) (Figure 3A). There were no differences in Shannon or Simpson indices between pretreatment and posttreatment groups (Supplementary Figure 5).

In subgroup analysis, the improvement of Chao1 diversity after treatment was observed in both groups of patients (P < .001) (Figure 3B). Regarding treatment outcome, SVR was achieved in 79 patients (97.5%) (responders), while 2 (2.5%) were nonresponders. After treatment, only responders had significantly higher Chao1 indices than before treatment, while there was no difference in any biodiversity among nonresponders (Figure 3C). Among responders, patients with stage F01 fibrosis had more improvement in Chao1 index after treatment than those with stage F2–4 (P = .01) (Figure 3D). A similar trend was observed with the Shannon index (P = .04) (Supplementary Figure 6A), but no significant difference was identified with the Simpson index (Supplementary Figure 6B).

Regarding beta diversity based on PCoA plot, the significant differences in clustering between the pretreatment and posttreatment groups was clearly observed (permutational multivariate ANOVA, P = .001) (Figure 4) These results indicated that overall gut microbiota composition was recovered and clustered with healthy controls after therapy. Data from heat map clustering of the top key genera are available in Supplementary Figure 7 and Supplementary Table 2.

Figure 4.

Principal coordinate analysis (PCoA) was calculated using Bray-Curtis dissimilarity method (Bray Method) before treatment (wk0) (blue dots for the hepatitis C virus (HCV) group (blue dots) and the HCV/human immunodeficiency virus (HIV) group (purple dots), and after treatment (sustained virological response at 12 weeks [SVR12]) for the HCV group (green dots), the HCV/HIV group (red dots), and the healthy controls (orange dots).

Effect of HCV Eradication with DAAs on Bacterial Composition

Detailed taxonomic information about major relative abundances at phylum, family, and genus levels in the pretreatment and posttreatment groups is available in Supplementary Figure 8. To investigate the alteration of specific gut microbiota composition after DAA therapy, the relative abundance between before and after treatment at the genus level was compared using paired Wilcoxon rank sum tests. Overall, the top relative abundances of Bacteroides, Prevotella, Faecalibacterium, Lachinoclostridium, Escherichia Shigella, Phascolarctobacterium, Sutterella, Roseburia, Blautia, Alistipes, Alloprevotella, Bifidobacterium, Megamonas, and Fusobacterium genera remained stable between the pretreatment and posttreatment groups (Supplementary Figure 8 and Supplementary Table 3).

However, there was a significant increase in beneficial bacteria, including Parabacteroides (P = .047) and Subdoligranulum (P = .02), while the relative abundance of pathogenic bacteria (Eubacterium) was significantly decreased (P < .001) (Figure 5A). Subgroup analysis regarding HIV status demonstrated that there was a significant depletion in Eubacterium after treatment among HCV-monoinfected patients (P < .001), while a significant decrease in Eubacterium (P < .001) and an increase in beneficial bacteria, including Lachnospira and Subdoligranulum, were detected in HCV/HIV-coinfected individuals (Figure 5B).

Figure 5.

Composition bar plot of relative abundance in genera that exhibited significant differences between before treatment (Pre) and after treatment (Post). A, Overall cohort. B, Subgroup analysis of the hepatitis C virus (HCV) and HCV/human immunodeficiency virus (HIV) groups. C, Only patients achieving sustained virological response at 12 weeks (SVR12). D, Patients with fibrosis stage F0–1 or F2–4 (data analyzed using paired Wilcoxon rank sum tests).

After stratification by treatment outcome, changes in gut microbiota composition after treatment were found only in responders. For instance, enrichment of Parabacteroides and Subdoligranulum and reduction in Eubacterium were observed in responders (P = .04, P = .01, and P < .001, respectively) (Figure 5C). Among responders, patients with fibrosis stage F01 had significantly increases in beneficial Phascolarctobacterium (P = .03), whereas a significant increase in Subdoligranulum was observed in patients with fibrosis stage F2–4 (P = .01). Moreover, there was a significant decrease in Eubacterium in both fibrosis groups (P < .001) (Figure 5D).

Distribution of Operational Taxonomic Units Before and After Treatment

To confirm and identify gut microbial alteration associated with DAAs at the taxonomic level, we determined the differences in microbial clade between before and after treatment, using LEfSe analysis. Cladogram plots (linear discriminative analysis score, ≥3.0) with relative abundance levels were applied to identify biomarkers accounting for the greatest differentially abundant taxa enriched in gut microbiota from the pretreatment or posttreatment groups (Figure 6A). At the genus level, the LEfSe analysis indicated that Akkermansia, Collinsella, Subdoligranulum, Victivallis, and Parabacteroides were enriched in the posttreatment group, while Eubacterium and Romboutsia were enriched in the pretreatment group. Detailed biomarkers ordered by effect size between groups are shown in Figure 6B.

Figure 6.

A, Linear discriminative analysis (LDA) effect size (LEfSe) cladogram using LDA score ≥3.0 identifies taxa whose distributions differ significantly between before and after treatment. B, Effect size plot using LDA score ≥3.0 between before and after treatment.

Effect of Treatment on the Change of Functional Analysis

PICRUSt analysis was used to demonstrate the alteration of functional pathways after therapy (Supplementary Figure 9A). At baseline, 9 pathways were up-regulated, including enriched drug metabolism, xenobiotics biodegradation, polyketide sugar unit biosynthesis, and cyanoamino acid metabolism. After therapy, 11 pathways, such as RNA degradation, amino acid biosynthesis, protein export, riboflavin metabolism, and pantothenate/coenzyme A biosynthesis, were up-regulated (Supplementary Figure 9B).

DISCUSSION

The role of gut microbiota in chronic liver disease has received considerable attention [9]. In the current study, we analyzed the diversity and composition of gut microbiota in HCV-infected patients with or without HIV coinfection. We clearly demonstrated that pretreatment bacterial communities among infected individuals were less diverse and distinct from those of healthy controls. In the longitudinal study, successful DAA therapy was associated with the improvement of microbial dysbiosis, in addition to the enrichment of some beneficial bacteria and the reduction of pathogenic bacteria. Collectively, our results suggest that gut dysbiosis is associated with the pathogenesis of chronic HCV infection, which might be alleviated after viral eradication by DAAs.

Previous studies demonstrated that changes in gut microbiota composition were distinctly observed in HCV-infected patients compared with healthy subjects at phylum, genus, and species levels [10, 12, 13, 23]. In agreement with aforementioned reports, we observed higher microbial diversity in fecal samples of healthy controls compared with the patient groups. Indeed, decreased bacterial diversity has been consistently observed in cirrhotic and noncirrhotic patients [11, 13, 14, 24]. Moreover, the main gastrointestinal commensals families Lachnospiraceae and Ruminococcaceae were found to be decreased in HCV-monoinfected and HCV/HIV-coinfected patients, while the family Enterobacteriaceae increased significantly in the patient groups [10, 12, 24], which was in line with our findings.

A possible explanation of enriched Enterobacteriaceae is decreasing primary bile acid in the intestine, leading to overgrowth of potentially pathogenic bacteria associated with inflammatory conditions [25, 26]. Remarkable, gut dysbiosis might occur in early stages [10, 11], as bacterial translocation was observed not only in the cirrhotic state but also in the absence of significant fibrosis [27, 28], indicating that altered gut microbiota exists during persistent HCV infection, independently of fibrosis stages.

Chronic HIV infection is found to be associated with altered gut microbiota, which could be partially restored after effective ART [29]. However, direct comparison of gut microbiota in HCV-infected individuals with or without HIV infection is not yet well characterized. Our results showed that HCV/HIV-coinfected patients receiving ART seemed to have alpha diversity (richness) comparable to that found in HCV monoinfection. Notably, a Prevotella-rich/Bacteroides-poor type, which was predominantly found in men who have sex with men (MSM [29], was not detected in our study. In fact, the majority of HCV/HIV-coinfected patients in this study were not men who have sex with men, similar to a cohort dominated by heterosexual transmission [30]. Moreover, gut dysbiosis depleted Firmicutes and enriched Bacteroidetes were mostly found among HCV/HIV-coinfected individuals, in line with an earlier report [23]. Indeed, the Firmicutes phylum includes some commensal bacteria that potentially induce regulatory immunity, while the Bacteroidetes phylum is composed of bacteria with known proinflammatory properties [31].

In addition, the abundance of certain genera, such as Alloprevotella, was significantly higher in HCV/HIV-coinfected patients than in healthy controls, although the difference compared with the monoinfection group did not reach statistical significance This finding was similar to previous data demonstrating Alloprevotella enrichment in HIV-infected individuals [32]. It seems that HCV/HIV-coinfected patients might not have significant alteration in alpha diversity but exhibit gut microbiota composition dissimilar to findings in HCV-monoinfected individuals.

Currently, it is unclear whether DAA therapy is capable of restoring microbial dysbiosis, because previous studies with relatively small sample sizes showed inconsistent results. An Italian study indicated that gut microbiota diversity and composition were significantly improved among 12 cirrhotic patients achieving SVR after various DAA regimens [13]. Conversely, in a cross-sectional cohort in the United States persistent gut dysbiosis was seen in cirrhotic patients or and without SVR after therapy with pegylated interferon and RBV therapy [12]. In a Spanish report including 22 noncirrhotic patients treated with DAAs, partial restoration of alpha diversity was demonstrated only in individuals with mild fibrosis but not in those with significant fibrosis [14].

Unlike previous studies, our study recruited more patients across different stages of liver fibrosis, e treated with the same DAA regimen. Moreover, this is the first study investigating the effect of DAAs on gut microbiota in patients with HCV/HIV coinfection. Interestingly, we observed significant restoration of alpha diversity in patients with HCV monoinfection or HCV/HIV coinfection in whom SVR was achieved. In addition, a recovery in alpha diversity to a level comparable to that in healthy controls was detected in patients with a low degree of fibrosis. Remarkably, beta diversity representing the structure of gut microbiota in responders showed a shift away from untreated individuals and displayed a similar community structure, approaching findings in healthy controls. Together, these findings indicated improvement in microbial balance and gut health after successful therapy. Indeed, these data were in line with reports that HCV cure in HCV/HIV-coinfected patients resulted in rapid and significant declines in immune activation and microbial translocation markers, regardless of fibrosis stage [33, 34].

Notably, also observed was a reciprocal change in beneficial bacteria, such as Parabacteroides, Lachnospira, Subdoligranulum, and Phascolarctobacterium, and pathogenic bacteria, including Eubacterium, after successful therapy, findings comparable to previous data demonstrating a decrease in Eubacterium after DAA therapy in cirrhotic patients [13]. In this respect, Eubacterium is associated with gastrointestinal diseases [35], while enriched Lachnospiraceae and Ruminococcaceae improve duodenal mucosal diversity and antimicrobial peptide expression [36]. We also found increasing Subdoligranulum, Lachnospira, and Phascolarctobacterium, which produce valuable short-chain fatty acids [37–40], implicating the depletion of short-chain fatty acid–producing taxa in untreated patients. The improvement of these bacteria was particularly detected in responders with low-grade fibrosis, while such a finding was not observed in nonresponders. Unlike findings in aforementioned longitudinal studies [13, 14], including both responders and nonresponders was very informative, because that would ensure that gut microbiota modulation was associated with treatment outcome.

According to current international recommendations, HCV therapy should be considered without delay in patients with significant fibrosis or cirrhosis [41]. In the era of highly effective interferon-free regimens, however, treating HCV-infected patients at early stages also appears to be cost-effective [42, 43], because successful DAA therapy could result in decreasing hepatic and extrahepatic complications, such as diabetes mellitus and chronic kidney disease [5, 44]. In this report, we extend clinically significant evidence that early initiating treatment not only prevents liver disease progression over time but could also improve gut dysbiosis approaching healthy controls. Whether gut microbiota restoration is linked to reducing long-term HCV-related complications needs further longitudinal investigations.

The current study had some limitations, as we included the relatively small number of coinfected individuals. In addition, patients’ dietary records were inadequate, particularly among participants who received RBV that might alter eating habits. Moreover, pretreatment liver biopsy was not performed. Instead, we measured LS using MR elastography, currently is the most reliable noninvasive method for assessing liver fibrosis [45]. This study also had a short follow-up period; thus, the long-term impact of EBR/GZR in the same cohort is currently under investigation. Finally, although the relationships between gut microbiota and chronic liver disease are now well accepted, further investigations to move beyond association studies to mechanistic studies are required [46]. In summary, our results indicate the essential role of gut dysbiosis in chronic HCV infection and underscore the use of DAA therapy in any infected individuals, regardless of the severity of fibrosis, to potentially improve microbial balance and gut health.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. We thank Surachate Siripongsakun and Pantajaree Hiranrat (Sonographer School, Faculty of Health Science Technology, Chulabhorn Royal Academy, Bangkok, Thailand) for technical support with magnetic resonance elastography measurement. We also thank Sawannee Sutheeworapong for comments and valuable discussion on this manuscript.

Financial support. This work was supported by the Ratchadapiseksompotch Fund, Faculty of Medicine, Chulalongkorn University (grant RA62-033), the Thailand Research Fund (senior research scholarship RTA6280004), the Center of Excellence in Hepatitis and Liver Cancer, Chulalongkorn University, the National Research Council of Thailand (NRCT)-Research Grants for Talented Young Researchers to N.C., NRCT5-RGJ63001-007 and Second Century Fund (C2F), Chulalongkorn University to T.J., and The National Institute of General Medical Sciences, National Institutes of Health (grant P20GM125503] to I. N.

Potential conflict of interests. Y. T. is currently conducting research funded by Janssen Pharmaceutical and has received an honorarium from Gilead Sciences. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.