Gut Microbiome Composition in Lean Patients with NASH Is Associated with Liver Damage Independent of Caloric Intake: a Prospective Pilot Study*

Nutrition, Metabolism & Cardiovascular Diseases (2018) 28, 369e384

Available online at www.sciencedirect.com

Nutrition, Metabolism & Cardiovascular Diseases

journal homepage: www.elsevier.com/locate/nmcd

Gut microbiome composition in lean patients with NASH is associated with liver damage independent of caloric intake: A prospective pilot study*

S.M.B. Duarte a, J.T. Stefano a, L. Miele b,**, F.R. Ponziani b, M. Souza-Basqueira c, L.S.R.R. Okada a, F.G. de Barros Costa a, K. Toda a, D.F.C. Mazo a, E.C. Sabino c, F.J. Carrilho a, A. Gasbarrini b, C.P. Oliveira a,* a Divisao de Gastroenterologia e Hepatologia, Departamento de Gastroenterologia (LIM-07), Faculdade de Medicina FMUSP, Universidade de Sao Paulo, Sao Paulo, SP, Brazil b Division of Internal Medicine, Gastroenterology and Hepatology, Area Gastroenterologica, Fondazione Policlinico Universitario Agostino Gemelli Università Cattolica del Sacro Cuore, Rome, Italy c Departamento de Doenças Infecciosas e Instituto de Medicina Tropical, Faculdade de Medicina FMUSP, Universidade de Sao Paulo, Sao Paulo, SP, Brazil

Received 13 March 2017; received in revised form 9 October 2017; accepted 13 October 2017 Handling editor: A. Siani Available online 26 October 2017

KEYWORDS Abstract Background and Aim: The aim of the study was to compare the gut microbiomes from Gut microbiome; obese and lean patients with or without NASH to outline phenotypic differences. NASH; Methods and Results: We performed a cross-sectional pilot study comprising biopsy-proven Lean; NASH patients grouped according to BMI. Microbiome DNA was extracted from stool samples, Obese; and PCR amplification was performed using primers for the V4 region of the 16S rRNA gene. Overweight The amplicons were sequenced using the Ion PGM Torrent platform, and data were analyzed using QIIME software. Macronutrient consumption was analyzed by a 7-day food record. Liver fibrosis F2 was associated with increased abundance of Lactobacilli (p Z 0.0007). NASH patients showed differences in Faecalibacterium, Ruminococcus, Lactobacillus and Bifidobacterium abundance compared with the control group. Lean NASH patients had a 3-fold lower abundance of Faecalibacterium and Ruminococcus (p Z 0.004), obese NASH patients were enriched in Lacto- bacilli (p Z 0.002), and overweight NASH patients had reduced Bifidobacterium (p Z 0.018). Moreover, lean NASH patients showed a deficiency in Lactobacillus compared with overweight and obese NASH patients. This group also appeared similar to the control group with regard to gut microbiome alpha diversity. Although there were qualitative differences between lean NASH and overweight/obese NASH, they were not statistically significant (p Z 0.618). The study limitations included a small sample size, a food questionnaire that collected only qualitative and

Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; CSS, cumulative sum scaling; DNA, deoxyribonucleic acid; GGT, gamma glutamyltransferase; HC, healthy controls; HCC, hepatocellular carcinoma; HE, hematoxylin-eosin; HFD, high-fat diet; LPS, lipopolysaccharides; NAS, NASH Clinical Research Network Scoring System; NAFLD, Nonalcoholic fatty liver disease; NASH, Nonalcoholic steatohepatitis; PCoA, Principal Coordinates Analysis; PCR, polymerase chain reaction; rRNA, ribosomal RNAs; SCFAs, short chain fatty acids. * Financial support: This work is supported by Fundação de Amparo à Pesquisa do Estado de São Paulo grant 2013/06828-0. * Corresponding author. Divisao de Gastroenterologia e Hepatologia, Departamento de Gastroenterologia (LIM-07), Faculdade de Medicina FMUSP, Universidade de Sao Paulo, Av. Dr. Enéas de Carvalho Aguiar no 255, Instituto Central, # 9159, 05403-000, Sao Paulo, SP, Brazil. Fax: þ55 11 2661 7830. ** Corresponding author. Division of Internal Medicine, Gastroenterology and Hepatology, Area Gastroenterologica, Fondazione Policlinico Universitario Agostino Gemelli Università Cattolica del Sacro Cuore, 8 Largo Gemelli, 00168 Rome, Italy. E-mail addresses: [email protected] (L. Miele), [email protected] (C.P. Oliveira). https://doi.org/10.1016/j.numecd.2017.10.014 0939-4753/ª 2017 The Italian Society of Diabetology, the Italian Society for the Study of Atherosclerosis, the Italian Society of Human Nutrition, and the Department of Clinical Medicine and Surgery, Federico II University. Published by Elsevier B.V. All rights reserved. 370 S.M.B. Duarte et al.

semi-quantitative data, and variations in group gender composition that may influence differences in FXR signaling, bile acids metabolism and the composition of gut microbiota. Conclusion: Our preliminary finding of a different pathogenetic process in lean NASH patients needs to be confirmed by larger studies, including those with patient populations stratified by sex and dietary habits. ª 2017 The Italian Society of Diabetology, the Italian Society for the Study of Atherosclerosis, the Italian Society of Human Nutrition, and the Department of Clinical Medicine and Surgery, Feder- ico II University. Published by Elsevier B.V. All rights reserved.

Introduction Methods

Nonalcoholic fatty liver disease (NAFLD) encompasses a This is a cross-sectional study comprising patients large spectrum of diseases from simple steatosis to with biopsy-proven NASH enrolled consecutively at the nonalcoholic steatohepatitis (NASH), and can evolve to- NAFLD outpatient clinic (A2MG700) in the Clinical ward cirrhosis and hepatocellular carcinoma (HCC) [1]. Gastroenterology Department of the Hospital das Clínicas Currently, NAFLD is the most common disease in the world da Faculdade de Medicina da Universidade de São Paulo [2] and an important cause of HCC [3]. (HC-FMUSP). Sedentary lifestyle, inappropriate food intake with high Inclusion criteria for this study were the following: fat and fructose consumption, as well as obesity, metabolic adult men and women aged 18e75 years of any racial disorders, hormonal status and genetic background, have demographics; presence of NASH confirmed by liver his- been described as the primary “culprits” of NAFLD [4e6]. tology; an absence of the following: alcohol daily con- Despite the knowledge about NAFLD pathogenesis, the sumption >20 g ethanol for women and >30 g for men, reason why some patients develop NASH with fibrosis and drug addiction, schistosomiasis, hepatitis B or C infection, progress to advanced liver disease is still unclear. Recently, detectable serum autoantibodies (antinuclear, anti-smooth intestinal dysbiosis has been reported to have an impor- muscle, anti-mitochondria, anti-liver kidney microsome tant role in metabolic disorders, such as obesity, metabolic type 1) with titers 1/160; altered serum ceruloplasmin syndrome, diabetes and cardiovascular diseases [7e12]. levels and hemochromatosis. Studies in both humans and animal models have Exclusion criteria for this study were the following: demonstrated that the gut microbiome is an important pregnancy and/or lactation; use of antibiotics, probiotics, factor in energy storage and contributes to the develop- prebiotics, and laxatives in the last month preceding the ment of NAFLD [13,14], helping to explain the observed collection of fecal material; or refusal to participate in the phenotypic differences related to this disease. Studies have study. identified that patients with NAFLD show fewer pro- For all study participants, physical examinations, portions of Bacteroidetes and higher proportions of Pre- medical histories, anthropometric and body composition votella and Porphyromas than healthy patients [15,16].An assessments, and laboratory tests were obtained at the increase in Lactobacillus, Escherichia and Streptococcus time of enrollment. Biochemical investigations included abundance, as well as a decrease in Ruminococcaceae and fasting glucose, total cholesterol and fractions, tri- in Faecalibacterium prausnitzii, have also been described in glycerides, alanine aminotransferase (ALT), aspartate NAFLD patients [17]. aminotransferase (AST), and gamma glutamyltransferase Boursier et al. [18] demonstrated an independent (GGT). Blood for testing was collected for each participant associations between Bacteroides abundance and NASH, as after a 12-h overnight fast and evaluated at the time of well as between Ruminococcus abundance and fibrosis the liver biopsy. Diabetes, dyslipidemia and hypertension stage F2. However, most studies that included patients were diagnosed according to international guidelines with NAFLD had several limitations: heterogeneous pop- [21]. Anthropometric variables represented by weight, ulations (adults versus children), incomplete liver biopsy height, and body mass index (BMI) were determined by data, and varied fecal microbiota characterization methods stadiometer (Biospace model BSM370, GangnamGu, (quantitative polymerase chain reaction (PCR) vs. pyrose- Seoul, Korea). According to the BMI value, participants quencing). Nevertheless, these studies showed that while were then assigned to one of three groups: lean (BMI NAFLD is commonly associated with overweight and <25 kg/m2), overweight (BMI 25e29.9 kg/m2) and obese obesity, a percentage of these patients are of normal (BMI 30 kg/m2). weight [19,20]. The control group consisted of lean healthy volunteers Thus, the link between dysbiosis and NAFLD severity, as with normal BMI and metabolically healthy obese in- well as the role of gut microbiota alterations in lean NAFLD dividuals recruited through internal disclosure at the patients, remains poorly documented. The aim of our study hospital. In addition to fulfilling the same inclusion/ was to compare the gut microbiome from lean NASH patients exclusion criteria as patients with NASH, these participants versus overweight/obese patients with or without NASH. had (a) normal liver exams, (b) no diabetes or chronic Microbiota in lean, overweight, and obese patients with NASH 371 gastrointestinal symptoms, and (c) no evidence of fatty The ManneWhitney test, the KruskaleWallis test and liver at abdominal ultrasound examination. the chi-square test were used to highlight any difference in The study was approved by the Ethics Committee for sex, age, NAS score, fibrosis, nutrient intake, and preva- Research Project Analysis, CAPPesq, at the Clinical Hospital lence of hypertension, diabetes, or dyslipidemia between of the Medical School of University of São Paulo. All par- NASH patients and controls. For any statistical significance ticipants signed informed consent before enrollment. found with the KruskaleWallis test, the Dunn test with Bonferroni adjustment was applied to verify relevant dif- Nutritional analysis ferences between variable levels. Statistical analysis was conducted using the R statistics Assessment of food intake program version 3.1.2. All statistical tests were two-sided, All enrolled participants received instructions to record and differences were considered significant for p-values their daily dietary intake for 7 days, including a weekend below 0.05. (R Core Team. R: A language and environment day. Macronutrients (protein, carbohydrates and lipids for statistical computing. R Foundation for Statistical including saturated, monounsaturated and poly- Computing, Vienna, Austria. URL http://www.R-project. unsaturated fatty acids, well as cholesterol), total energy org/2014.) and dietary fiber intakes were calculated with the Avanutri Biostatistical analysis of metagenomic data is reported 4.0 software (Avanutri, Rio de Janeiro, Brazil). in the Supplementary materials.

Histological evaluation Results All study participants had a record of a liver biopsy within 6 months prior to enrollment, as reported according to the We prospectively recruited 13 biopsy-proven NASH pa- care routine of NAFLD outpatient protocol (A2MG700) of tients and 10 matched controls, which were stratified ac- the Clinical Gastroenterology Department at the Hospital cording BMI (Table 1). The female gender was more das Clínicas da Faculdade de Medicina da Universidade de prevalent among obese participants, while the male São Paulo (HC-FMUSP). gender was more prevalent among the lean and over- Liver fragments were fixed in 10% formalin saline and weight NASH participants (p Z 0.018). subjected to hematoxylin-eosin (HE) stain, as well as to Metabolic complications were present in 8 (61.5%) of Masson and Perls stain for pigments. Histological param- the 13 NASH patients, with diabetes as the most common. eters were revised according to the NASH Clinical Research While there was a slightly higher prevalence of hyper- Network Scoring System (NAS score) [22]. tension in the obese NASH group, no significant difference in the frequency of single complications was observed Analysis of fecal microbiota among the three NASH groups. The gut microbiota composition of patients and con- Fecal samples collection, DNA extraction and trols was heterogeneous, with no relevant clustering at the pyrosequencing ordination plot according to BMI, even after stratifying For preparation of DNA extraction, 0.25 g of each fecal groups for the presence of NASH (PERMANOVA p Z 0.691; sample was analyzed with the Power Soil DNA Isolation PCoA, Fig. 1). Kit (Mobio Laboratories, Carlsbad, CA) according to Comparing the fecal bacteria of obese, overweight and manufacturer instructions. lean NASH patients with that of obese and lean controls, After seven days of food recall, fecal samples were no significant difference was found at the phylum level collected, mixed with RNA, and then stored at 20 C. (Table 2). After a maximum of 4 h, the samples were separated into However, analyzing the composition of bacterial genera, aliquots of 200 mg each and stored at 80 C. NASH patients showed marked differences in Faecali- Extraction and purification of microbial DNA from the bacterium, Ruminococcus, Lactobacillus and Bifidobacterium fecal samples were performed at the FMUSP using the DNA abundance compared to the control group (Table 3 and Qiamp DNA Stool Mini Kit (Qiagen, USA). PCR amplification Fig. 2). In particular, lean NASH patients had a 3-fold lower wasthenconducted using primers for the V4region of the 16S abundance of Faecalibacterium and Ruminococcus (logFC rRNA gene. Finally, bacterial DNA amplicons were sequenced 3.625, p Z 0.004, adj. p Z 0.05 and logFC 3.782, using the Ion PGM Torrent platform (Carlsbad, CA e USA). p Z 0.004, adj. p Z 0.05, respectively). By comparing lean Reads obtained after pyrosequencing were processed NASH patients with lean healthy individuals and with through the QIIME pipeline [23] and assigned to taxo- obese participants, this difference was confirmed. nomic units. Fecal microbiota of obese NASH patients was enriched in Lactobacilli (logFC 5.459, p Z 0.002, adj. p Z 0.048), Statistical analysis while Bifidobacterium was reduced in overweight NASH For the analysis of non-metagenomic data, non-parametric patients (logFC 4.494, p Z 0.018, adj. p Z 0.041). Minor statistics were used. differences were also observed in Roseburia and Lachno- Continuous variables were expressed as median and spira abundance. range, while categorical variables were expressed as fre- Ruminococcus abundance was lower in lean NASH pa- quencies and percentages. tients than their overweight and obese counterparts (lean 372 S.M.B. Duarte et al.

Table 1 Study participant characteristics. Values are expressed as median (range) and frequencies (percentages). Statistically signiﬁcant comparisons are highlighted in bold.

Variable Lean NASH (4) Overweight Obese NASH (4) Lean Obese p-value p-value NASH (5) Controls (5) Controls (5) (NASH (overall) patients) Age 48 53 43 30 44 (34e53) 0.715 0.595 (24e53) (27e65) (27e55) (26e50) Sex (male) 3 (75) 5 (100) 0 3 (60) 1 (20) 0.006 0.018 Diabetes 3 (75) 2 (50) 3 (75) eee0.336 mellitus Hypertension 1 (25) 1 (20) 3 (75) eee0.084 Dyslipidemia 3 (75) 1 (20) 2 (50) eee0.322 NAS score 5 (5e6) 5 (5e7) 6 (6e7) eee0.084 Fibrosis stage 3(1e4) 2 (0e4) 3 (3e4) ee0.412 (Kleiner) Fibrosis 2 3 (75) 3 (60) 4 (100) ALT (U/L) 66.5 37.5 65.8 56.6 154.8 177 13.5 0.7 18.3 4.3 0.75 0.035 AST (U/L) e 42.4 30.3 76 54.4 14 2.6 15 1.7 0.2 0.037 GGT (U/L) 181 223 120 73.2 402 320 15 422 14.4 0.32 0.022 Glucose 118.5 20.5 140.4 89 145 60.5 77.3 6.5 89.3 13 0.07 0.06 (mg/dL) Cholesterol 146.5 43 162 105 175 30.3 154 5.7 207.2 21 0.56 0.3 (mg/dL) Triglycerides 119 45.3 130 80.3 93.5 39 80 17 123.3 58 0.73 0.85 (mg/dL) LDL (mg/dL) 83.5 60.1 87.4 76 130 10.5 90 15.5 129.7 10.5 0.83 0.32 HDL (mg/dL) e 76 21.2 65.8 11.4 48 12.7 52.7 5.3 0.2 0.33 Carbohydrate 52.85 (38e58.1) 52.95 (50e61) 56.3 (44.7e61.8) 51.2 (32.2e54) 56 (43.2e62) 0.840 0.740 intake (%) Protein intake 4.46 (13.6e24) 5.77 (13.2e25.8) 2.89 (14e19.7) 6.22 (17e32.8) 7.22 (13e27.3) 0.385 0.788 (%) Lipid intake (%) 28.35 (21.7e40.3) 23.05 (14e36.8) 24 (20.5e41.3) 28.8 (27e35) 25.3 (25e29.9) 0.618 0.604 Fiber intake (%) 9.65 (7.1e16.5) 12.3 (8.7e21) 6.8 (3.3e11.6) 12 (9.9e15.4) 9.2 (1.7e17.5) 0.242 0.381 Caloric intake 1541.5 (970e2196) 1160 (961e3014) 1200 (1152e1803) 1802 (1527e2018) 1894 (1120e2010) 0.781 0.552 (kcal)

NASH vs obese NASH: logFC 3.340, p Z 0.034, adj. bacterial diversity (p Z 0.05). Obese and overweight pa- p Z 0.415; lean NASH vs overweight NASH: logFC 3.643, tients with NASH had higher alpha diversity values p Z 0.015, adj. p Z 0.337; Table 3 and Fig. 2). Lean NASH compared to controls (p Z 0.05). No statistically signifi- patients were also deficient in Lactobacilli (lean NASH vs cant differences were observed among lean, overweight obese NASH: logFC 4.031, p Z 0.0037, adj. p Z 0.415). and obese NASH patients. The analysis of fecal microbiome alpha diversity also According to the 7-day food registry, qualitative differ- highlighted substantial differences among groups (Fig. 3). ences in dietary habits were observed among the study In particular, lean NASH patients appeared more similar to groups. In the lean NASH group, the consumption of sim- the control group (p Z 0.265) and clearly different from ple carbohydrates (i.e., foods produced from white flour or overweight and obese NASH patients who had a higher refined sugar) was greater than the consumption of complex ones (e.g., whole foods, fresh fruits and vegetables). Moreover, this group’s intake of animal proteins such as meats, milk and derivatives was greater than the intake of vegetal proteins. Medications used by the patients in this group were N-acetylcysteine and metformin supplemen- tation. The amount of exercise per week of these patients did not exceed 90 min per week. In contrast, the lean control group reported a low consumption of simple carbohydrates, consuming complex carbohydrates during at least two meals per day. Protein consumption for this group was well-varied between source protein and plant- derived protein such as beans, lentils and seeds (chia, quinoa and gergelin). None of the participants in this group were on continuous medication. Physical activity did not exceed 90 min per week. In the obese/overweight Figure 1 Principal coordinates analysis (PCoA) on weighted UniFrac distance. Patient stratification was performed according to BMI and NASH groups, we observed a high consumption of pro- presence/absence of NASH. cessed and super processed foods such as soft drinks, irboai en vregt n bs ainswt NASH with patients obese and overweight, lean, in Microbiota

Table 2 Differential abundance analysis of bacterial phyla between lean, overweight and obese NASH patients and controls. Changes in abundance are reported as log2 fold changes (logFC). The increase or decrease in bacterial abundance is expressed by a positive or negative logFC and proportional to the logFC absolute value. Statistical signiﬁcance is expressed as p-value and p-value adjusted for multiple comparisons.

Phylum Lean NASH p value Adjusted Lean NASH vs. p value Adjusted Lean NASH vs. p value Adjusted Lean NASH vs. p value Adjusted vs Controls p value Lean Controls p value Obese Controls p value Obese NASH p value (logFC) (logFC) (logFC) (logFC) Bacteroidetes 0.9 0.392 0.908 0.171 0.880 0.893 0.074 0.948 0.965 1942 0.109 0.436 Actinobacteria 0.291 0.791 0.908 544 0.666 0.893 0.055 0.965 0.965 0.043 0.972 0.972 Firmicutes 0.132 0.865 0.908 0.417 0.648 0.893 0.291 0.753 0.965 0.408 0.644 0.972 Proteobacteria 0.075 0.908 0.908 0.104 0.893 0.893 0.124 0.874 0.965 0.136 0.854 0.972 Phylum Overweight p value Adjusted Overweight p value Adjusted Overweight p value Adjusted Overweight p value Adjusted NASH vs p value NASH vs. p value NASH vs. p value NASH vs. Lean p value Controls Lean Controls Obese Controls NASH (logFC) (logFC) (logFC) (logFC) Firmicutes 1226 0.089 0.359 1362 0.106 0.427 1062 0.225 0.581 1353 0.131 0.524 Actinobacteria 0.388 0.692 0.908 0.186 0.868 0.875 0.631 0.596 0.795 0.687 0.570 0.760 Bacteroidetes 0.142 0.878 0.908 0.691 0.5 0.875 1148 0.290 0.581 1073 0.328 0.657 Proteobacteria 0.068 0.908 0.908 0.11 0.875 0.875 0.017 0.981 0.981 0.142 0.849 0.849 Phylum Obese NASH p value Adjusted Obese NASH vs. p value Adjusted Obese NASH vs. p value Adjusted Obese NASH vs. p value Adjusted vs Controls p value Lean Controls p value Obese Controls p value Overweight p value (logFC) (logFC) (logFC) NASH (logFC)

FirmicutesBacteroidetes 0.2751.04 0.7280.329 0.927 0.4170.171 0.6480.880 0.893 0.1172.01 0.8990.093 0.9880.373 0.9500.899 0.2730.434 0.731 Actinobacteria 0.334 0.764 0.927 0.544 0.666 0.893 0.099 0.938 0.988 0.722 0.548 0.731 Proteobacteria 0.06 0.927 0.927 0.104 0.893 0.893 0.011 0.988 0.988 0.007 0.991 0.991 373 Table 3 Differential abundance analysis of bacterial genera between lean, overweight and obese NASH patients and controls. Changes in abundance are reported as log2 fold changes (logFC). The 374 increase or decrease in bacterial abundance is expressed by a positive or negative logFC and proportional to the logFC absolute value. Statistical signiﬁcance is expressed as p-value and p-value adjusted for multiple comparisons. Signiﬁcant variables are highlighted in bold.

Phylum/Class/Order/Family/ Lean NASH vs p value Adjusted Lean NASH vs. p value Adjusted Lean NASH vs. p value Adjusted Lean NASH vs. p value Adjusted Genus Controls (logFC) p value Lean Controls p value Obese Controls p value Obese NASH p value (logFC) (logFC) (logFC) Firmicutes; Clostridia; L3625 0.004* 0.047** L4278 0.006* 0.107 L3.72 0.015* 0.17 1814 0.211 0.743 Clostridiales; Ruminococcaceae; Faecalibacterium Firmicutes; Clostridia; L3782 0.004* 0.048** L4625 0.009* 0.107 L4.7 0.003* 0.07 L3340 0.034* 0.415 Clostridiales; Ruminococcaceae; Ruminococcus Firmicutes; Clostridia; 1391 0.183 0.993 1875 0.166 0.734 2045 0.109 0.602 1636 0.198 0.743 Clostridiales; Lachnospiraceae; Roseburia Firmicutes; Erysipelotrichi; 1.85 0.346 0.993 1808 0.514 0.844 1445 0.573 0.849 0.389 0.863 0.959 Erysipelotrichales; Erysipelotrichaceae; Eubacterium Proteobacteria; 1303 0.411 0.993 0.457 0.838 0.970 1528 0.458 0.849 1872 0.306 0.743 Deltaproteobacteria; Desulfovibrionales; Desulfovibrionaceae; Bilophila Firmicutes; Clostridia; 1032 0.314 0.993 1.17 0.362 0.844 2229 0.07 0.517 0.988 0.426 0.743 Clostridiales; Lachnospiraceae; Coprococcus Firmicutes; Clostridia; 1038 0.315 0.993 1368 0.308 0.844 1492 0.237 0.849 1556 0.217 0.743 Clostridiales; Lachnospiraceae; Blautia Firmicutes; Clostridia; 1047 0.426 0.993 2187 0.229 0.841 1371 0.413 0.849 1278 0.399 0.743 Clostridiales; Veillonellaceae; Phascolarctobacterium Firmicutes; Bacilli; 1427 0.356 0.993 2917 0.146 0.734 1.06 0.575 0.849 L4031 0.037* 0.415 Lactobacillales; Lactobacillaceae; 0.589 0.606 0.993 0.576 0.717 0.876 1095 0.459 0.849 0.029 0.982 0.982 Lactobacillus Firmicutes; Clostridia; Clostridiales; Ruminococcaceae; Firmicutes; Clostridia; 0.514 0.614 0.993 0.137 0.921 0.970 1.14 0.382 0.849 1413 0.236 0.743 al. et Duarte S.M.B. Oscillospira Clostridiales; Lachnospiraceae; Lachnospira Bacteroidetes; Bacteroidia; 0.241 0.85 0.993 1308 0.43 0.844 1037 0.498 0.849 1188 0.439 0.743 Bacteroidales; [Odoribacteraceae]; Butyricimonas (continued on next page) (continued ) NASH with patients obese and overweight, lean, in Microbiota Bacteroidetes; Bacteroidia; 0.012 0.993 0.993 2095 0.272 0.844 1684 0.332 0.849 0.525 0.757 0.926 Bacteroidales; [Odoribacteraceae]; Odoribacter Firmicutes; Erysipelotrichi; 0.099 0.935 0.993 0.694 0.685 0.876 0.54 0.733 0.878 0.504 0.725 0.926 Erysipelotrichales; Erysipelotrichaceae; Holdemania Bacteroidetes; Bacteroidia; 1152 0.584 0.993 1129 0.575 0.844 0.285 0.911 0.966 0.406 0.872 0.959 Bacteroidales; Prevotellaceae; Prevotella Actinobacteria; Actinobacteria; 0.487 0.677 0.993 1315 0.442 0.844 0.797 0.618 0.849 2217 0.126 0.743 Bifidobacteriales; Bifidobacteriaceae; Bifidobacterium Bacteroidetes; Bacteroidia; 0.392 0.737 0.993 0.056 0.970 0.970 0.623 0.665 0.861 0.669 0.637 0.877 Bacteroidales; Porphyromonadaceae; Parabacteroides Firmicutes; Clostridia; 0.240 0.792 0.993 0.089 0.939 0.970 0.869 0.431 0.849 0.559 0.615 0.877 Clostridiales; Lachnospiraceae; Dorea Firmicutes; Clostridia; 0.087 0.929 0.993 0.8 0.536 0.844 0.117 0.922 0.966 0.102 0.932 0.977 Clostridiales; Lachnospiraceae; Ruminococcus Firmicutes; Bacilli; 0.059 0.966 0.993 2959 0.091 0.668 0.067 0.966 0.966 1574 0.363 0.743 Lactobacillales; Streptococcaceae; Streptococcus Bacteroidetes; Bacteroidia; 0.037 0.978 0.993 1023 0.575 0.844 0.942 0.581 0.849 1092 0.524 0.824 Bacteroidales; Bacteroidaceae; Bacteroides Proteobacteria; 0.022 0.985 0.993 0.869 0.575 0.844 0.444 0.758 0.878 1295 0.377 0.743 Betaproteobacteria; Burkholderiales; Alcaligenaceae; Sutterella p value Phylum/Class/Order/Family/ Overweight NASH vs p value Adjusted Overweight p value Adjusted Overweight p value Adjusted Overweight p value Adjusted Genus Controls (logFC) p value NASH vs. p value NASH vs. p value NASH vs. Lean Lean Controls Obese Controls NASH (logFC) Actinobacteria; Actinobacteria; L4494 0.001* 0.041**(logFC)L3284 0.048* 0.697(logFC)2765 0.074 0.814 1973 0.210 0.763 Bifidobacteriales; Bifidobacteriaceae; Bifidobacterium Bacteroidetes; Bacteroidia; 2516 0.148 0.915 1067 0.484 0.918 2713 0.06 0.814 0.706 0.708 0.764 Bacteroidales; [Odoribacteraceae]; Odoribacter Firmicutes; Clostridia; Clostridiales; 1432 0.160 0.915 0.308 0.778 0.918 1586 0.134 0.987 0.310 0.808 0.808 375 Lachnospiraceae; Lachnospira (continued on next page) 376 Table 3 (continued ) Phylum/Class/Order/Family/ Overweight NASH vs p value Adjusted Overweight p value Adjusted Overweight p value Adjusted Overweight p value Adjusted Genus Controls (logFC) p value NASH vs. p value NASH vs. p value NASH vs. Lean p value Lean Controls Obese Controls NASH (logFC) (logFC) (logFC) Bacteroidetes; Bacteroidia; 1616 0.207 0.915 2493 0.063 0.697 0.147 0.903 0.991 1356 0.401 0.763 Bacteroidales; [Odoribacteraceae]; Butyricimonas Firmicutes; Clostridia; Clostridiales; 2494 0.202 0.915 2618 0.235 0.918 2254 0.279 0.991 0.839 0.729 0.764 Eubacteriaceae; Pseudoramibacter_Eubacterium Firmicutes; Clostridia; Clostridiales; 1292 0.270 0.991 0.770 0.539 0.918 0.94 0.424 0.991 1106 0.366 0.763 Lachnospiraceae; Roseburia Firmicutes; Clostridia; Clostridiales; 1151 0.387 0.993 1882 0.203 0.918 1324 0.335 0.991 2391 0.099 0.763 Ruminococcaceae; Faecalibacterium Bacteroidetes; Bacteroidia; 1818 0.442 0.993 2589 0.332 0.918 1746 0.47 0.991 1467 0.556 0.764 Bacteroidales; Prevotellaceae; Prevotella Firmicutes; Clostridia; Clostridiales; 0.829 0.470 0.993 0.295 0.805 0.918 0.763 0.498 0.991 1475 0.219 0.763 Lachnospiraceae; Coprococcus Firmicutes; Clostridia; Clostridiales; 0.940 0.503 0.993 0.982 0.504 0.918 1057 0.443 0.991 3643 0.015* 0.337 Ruminococcaceae; Ruminococcus Firmicutes; Clostridia; Clostridiales; 0.753 0.514 0.993 0.448 0.720 0.918 0.572 0.625 0.991 0.921 0.451 0.763 Lachnospiraceae; Blautia Firmicutes; Bacilli; Lactobacillales; 0.798 0.644 0.993 1.5 0.42 0.918 0.357 0.841 0.991 1402 0.444 0.763 Lactobacillaceae; Lactobacillus Proteobacteria; Betaproteobacteria; 0.619 0.646 0.993 0.673 0.644 0.918 1098 0.422 0.991 1538 0.280 0.763 Burkholderiales; Alcaligenaceae; Sutterella Firmicutes; Clostridia; Clostridiales; 0.340 0.761 0.993 0.505 0.685 0.918 0.012 0.991 0.991 1135 0.424 0.763 Ruminococcaceae; Oscillospira 0.013 0.993 0.993 0.273 0.877 0.918 0.798 0.623 0.991 0.797 0.689 0.764 Proteobacteria; Deltaproteobacteria; Desulfovibrionales; Bacteroidetes;Desulfovibrionaceae; Bacteroidia; Bilophila 0.544 0.679 0.993 0.661 0.648 0.918 0.019 0.988 0.991 0.610 0.665 0.764 Bacteroidales; Porphyromonadaceae; Parabacteroides Firmicutes; Clostridia; Clostridiales; 0.207 0.872 0.993 0.695 0.630 0.918 0.120 0.929 0.991 1418 0.374 0.763 Veillonellaceae; Phascolarctobacterium al. et Duarte S.M.B. Firmicutes; Erysipelotrichi; 0.162 0.893 0.993 0.092 0.945 0.918 0.246 0.847 0.991 0.767 0.613 0.764 Erysipelotrichales; Erysipelotrichaceae; Holdemania Firmicutes; Bacilli; Lactobacillales; 0.308 0.846 0.993 0.914 0.568 0.918 1977 0.193 0.991 2019 0.227 0.763 Streptococcaceae; Streptococcus (continued on next page) irboai en vregt n bs ainswt NASH with patients obese and overweight, lean, in Microbiota (continued ) Bacteroidetes; Bacteroidia; 0.104 0.947 0.993 0.678 0.692 0.918 0.759 0.636 0.991 1700 0.309 0.763 Bacteroidales; Bacteroidaceae; Bacteroides Firmicutes; Clostridia; Clostridiales; 0.064 0.954 0.993 0.208 0.863 0.918 0.474 0.676 0.991 0.585 0.622 0.764 Lachnospiraceae; Ruminococcus Firmicutes; Clostridia; Clostridiales; 0.057 0.955 0.993 0.648 0.558 0.918 0.131 0.898 0.991 0.744 0.494 0.764 Lachnospiraceae; Dorea

Phylum/Class/Order/Family/ Obese NASH vs P value Adjusted Obese P value Adjusted Obese P value Adjusted Obese p value Adjusted Genus Controls (logFC) p value NASH vs. Lean p value NASH vs. Obese p value NASH vs. p value Controls (logFC) Controls (logFC) Overweight NASH (logFC) Firmicutes; Bacilli; 5459 0.002* 0.048** 6293 0.002* 0.044* 4436 0.027* 0.199 4660 0.027* 0.306 Lactobacillales; Lactobacillaceae; Lactobacillus Firmicutes; Clostridia; L3028 0.009* 0.101 L2943 0.025* 0.277 L3112 0.018* 0.199 1735 0.208 0.923 Clostridiales; Lachnospiraceae; Roseburia Firmicutes; Clostridia; L1927 0.044* 0.325 1285 0.231 0.702 L2564 0.024* 0.199 0.495 0.670 0.923 Clostridiales; Lachnospiraceae; Lachnospira Firmicutes; Clostridia; Clostridiales; 2325 0.061 0.340 2739 0.06 0.3 1922 0.190 0.804 2117 0.164 0.923 Veillonellaceae; Phascolarctobacterium Firmicutes; Clostridia; Clostridiales; 1811 0.153 0.621 2092 0.157 0.579 1535 0.295 0.834 0.659 0.671 0.923 Ruminococcaceae; Faecalibacterium Actinobacteria; Actinobacteria; 1730 0.169 0.621 1473 0.36 0.794 1991 0.219 0.804 6225 0.0003* 0.008** Bifidobacteriales; Bifidobacteriaceae; Firmicutes;Bifidobacterium Clostridia; Clostridiales; 2248 0.212 0.669 2429 0.255 0.702 2066 0.341 0.834 0.246 0.912 0.923 Eubacteriaceae; Pseudoramibacter_Eubacterium Bacteroidetes; Bacteroidia; 1429 0.246 0.677 2483 0.068 0.3 0.137 0.916 0.958 0.187 0.901 0.923 Bacteroidales; [Odoribacteraceae]; Butyricimonas Firmicutes; Bacilli; Lactobacillales; 1633 0.279 0.684 3083 0.062 0.3 0.191 0.904 0.958 1324 0.479 0.923 Streptococcaceae; Streptococcus Proteobacteria; Betaproteobacteria; 1272 0.319 0.703 1059 0.467 0.826 1484 0.310 0.834 0.653 0.679 0.923 Burkholderiales; Alcaligenaceae; Sutterella Bacteroidetes; Bacteroidia; 1130 0.450 0.858 1.09 0.526 0.826 1.17 0.495 0.958 1234 0.508 0.923 Bacteroidales; Bacteroidaceae; Bacteroides (continued on next page) 377 378 Table 3 (continued ) Phylum/Class/Order/Family/ Obese NASH vs P value Adjusted Obese P value Adjusted Obese P value Adjusted Obese p value Adjusted Genus Controls (logFC) p value NASH vs. Lean p value NASH vs. Obese p value NASH vs. p value Controls (logFC) Controls (logFC) Overweight NASH (logFC) Firmicutes; Clostridia; Clostridiales; 0.559 0.589 0.858 0.301 0.803 0.841 0.819 0.507 0.958 0.900 0.491 0.923 Ruminococcaceae; Oscillospira Proteobacteria; 0.568 0.696 0.858 1.16 0.503 0.826 0.090 0.958 0.958 0.582 0.748 0.923 Deltaproteobacteria; Desulfovibrionales; Desulfovibrionaceae; Bilophila Bacteroidetes; Bacteroidia; 0.538 0.692 0.858 1531 0.307 0.752 2249 0.13 0.716 1617 0.342 0.923 Bacteroidales; [Odoribacteraceae]; Odoribacter Firmicutes; Clostridia; Clostridiales; 0.518 0.633 0.858 0.58 0.642 0.827 0.456 0.715 0.958 1272 0.350 0.923 Lachnospiraceae; Blautia Firmicutes; Erysipelotrichi; 0.404 0.719 0.858 0.482 0.714 0.827 0.328 0.806 0.958 0.241 0.864 0.923 Erysipelotrichales; Erysipelotrichaceae; Holdemania Bacteroidetes; Bacteroidia; 0.745 0.738 0.858 1219 0.647 0.827 0.376 0.883 0.958 1072 0.696 0.923 Bacteroidales; Prevotellaceae; Prevotella Firmicutes; Clostridia; Clostridiales; 0.442 0.736 0.858 0.406 0.781 0.841 0.482 0.741 0.958 0.498 0.761 0.923 Ruminococcaceae; Ruminococcus Firmicutes; Clostridia; Clostridiales; 0.318 0.741 0.858 0.709 0.521 0.826 0.070 0.949 0.958 0.261 0.862 0.923 Lachnospiraceae; Dorea Bacteroidetes; Bacteroidia; 0.277 0.822 0.897 0.064 0.964 0.964 0.615 0.669 0.958 0.267 0.862 0.923 Bacteroidales; Porphyromonadaceae; Parabacteroides Firmicutes; Clostridia; Clostridiales; 0.19 0.856 0.897 0.532 0.66 0.827 0.150 0.901 0.958 0.125 0.923 0.923 Lachnospiraceae; Ruminococcus Firmicutes; Clostridia; Clostridiales; 0.043 0.967 0.967 0.486 0.684 0.827 0.571 0.633 0.958 0.785 0.559 0.923 **Lachnospiraceae; Results with logFC Coprococcus1 and adjusted p-value <0.05. * Results with logFC 1 and p-value <0.05. ...Dat tal. et Duarte S.M.B. Microbiota in lean, overweight, and obese patients with NASH 379 snacks, fast food, and sugar-sweetened beverages (coffee, increase in Lactobacillus, Escherichia and Streptococcus juices) and a low consumption of complex carbohydrates abundance, as well as a decrease in Ruminococcaceae and (whole foods, fresh fruits and vegetables), which led to a in F. prausnitzii, have been found in NAFLD patients low consumption of fibers (less than 10 g per day). Protein compared with healthy controls [24e29]. consumption was very low, approximately 10% of the total Bacterial increases such as these are commonly linked daily energy value, with a higher proportion of animal to mechanisms of inflammation and toxicity, while bac- protein than vegetable-derived protein. Medications used terial decreases are associated with the production of by the patients in this group were N-acetylcysteine, los- beneficial compounds such as short chain fatty acids artan, atenolol and fluoxetine. The majority of these pa- (SCFAs). These assumptions seem to characterize gut tients did not practice any type of physical exercise. In the microbiota composition in patients with NAFLD. However, obese/overweight group without NASH, the 7-day dietary while NAFLD is commonly associated with overweight and record demonstrated a predominant consumption of obesity, a percentage of patients are of normal weight. simple carbohydrates, primarily consisting of cafeteria Data about gut microbiota alterations in lean NAFLD pa- foods and refined sugar. In contrast, complex carbohy- tients are scarce. drates were consumed in the form of fruit, one per day on Our work provides interesting findings on this issue average. The consumption of animal proteins such as compared with previous studies [16]. First, we evaluated meats, whole milk and butter was higher than the con- intestinal microbiota in biopsy-proven NASH patients ac- sumption of vegetable protein. These patients did not cording to BMI, comparing lean, overweight and obese report the use of medications, except for one patient who NASH patients to lean and obese participants without used antianxiety and antidepressant agents. None of them NASH. Second, we conducted a daily dietary intake practiced physical exercise. assessment for 7 days in all participants to explore Despite some qualitative differences, such as a higher whether the overall caloric intake or specific macronutri- dietary lipid intake for lean NASH patients than for over- ents intake has an impact on intestinal microbiota weight and obese NASH patients, these differences were composition in the studied population. Finally, this is the not statistically significant (28.35% vs. 23.05% and 24%, first study describing the fecal microbiota of biopsy- respectively, p Z 0.618; Table 1). Similar lipid intake was proven NASH patients in Latin America. seen in the control group (28.8% and 25.3% for lean and Our data suggest that the gut microbiota composition of obese healthy individuals, p Z 0.604). In addition, obese the study population was heterogeneous, with no relevant NASH patients had a slightly lower fiber intake than lean clustering at the ordination plot according to BMI, even and overweight NASH patients (6.8% vs. 9.65% and 12.3%, after stratifying groups for the presence of NASH. p Z 0.242), as well as lean and obese healthy individuals At the phylum level, no significant difference was found (12% and 9.2%, p Z 0.381). Caloric intake in lean NASH between fecal microbiota of obese, overweight and lean patients was higher than in overweight and obese NASH NASH patients and controls. It should be noted that several patients (1541 kcal vs. 1160 kcal and 1200 kcal, p Z 0.781) studies have provided discordant results in NAFLD patients and more similar to healthy controls (1802 kcal and when phyla were compared. This probably reflects differ- 1894 kcal, p Z 0.552). However, neither single macronu- ences in quantification techniques and inhomogeneous trient intake nor caloric intake was associated with specific patient characteristics [30]. differences in fecal gut microbiome composition. Analyzing the gut microbiota composition of our pop- The median NAS score was higher in obese NASH pa- ulation, we found a significantly reduced abundance of tients (6 points, range 6e7) than in lean (5 points, range Faecalibacterium and Ruminococcus in lean NASH patients. 5e6) and overweight (5 points, range 5e7) NASH patients F. prausnitzii, one of the most represented butyrate- (p Z 0.084). Three (75%) out of 4 lean NASH patients, 3 producing bacteria, accounts for more than 3.5% of the (60%) out of 5 overweight NASH patients, and all obese total bacterial population in the human gut and has shown NASH patients had fibrosis scores higher than or equal to 2 immunomodulating and anti-inflammatory properties [1]. according to Kleiner score. NAS score was not associated Butyrate, propionate, and acetate are SCFAs, which derive with a specific alteration of the gut microbiome, whereas a from bacteria degradation of dietary indigestible poly- fibrosis score higher than or equal to 2 was associated with saccharides and account for 10% of the energy harvested an increased abundance of Lactobacilli (logFC 6.11, from diet [31e33]. In lean NASH patients, a reduced Fae- p Z 0.0007, adj. p Z 0.015). calibacterium abundance may have decreased the extraction of calories from the diet and may have favored a pro- fl Discussion in ammatory environment, potentially triggering NASH development. fi The role of gut microbiota in energy storage and the Ruminococcus genus includes bene cial and pro- fl development of NAFLD has provided new insight into the in ammatory species. A decrease in the Ruminococca- pathophysiology of NAFLD and may help to explain the ceae family has been previously described in patients with observed phenotypic difference. Recent studies in both NAFLD, whereas more recently, an increase in Rumino- humans and animal models have demonstrated that gut coccus abundance has been found as an independent fi fi microbiota composition is an important factor in energy predictor of signi cant liver brosis ( F2) in NAFLD pa- fi storage and contributes to the development of NAFLD. An tients [18]. These previous results are dif cult to interpret, 380 S.M.B. Duarte et al. Microbiota in lean, overweight, and obese patients with NASH 381

Figure 3 Bacterial alpha diversity (Shannon index) of NASH patients and controls according to BMI. Obese and overweight NASH patients showed higher alpha diversity values than controls (p Z 0.05). There were no statistically significant differences among obese, overweight and lean NASH patients. and due to the lack of more specific taxonomic information For obese and overweight NASH patients, we found a in our study (e.g., species and strain characterization), only higher abundance of Lactobacillus and a lower abundance hypotheses can be put forward. Notably, many bacteria of Bifidobacterium than the control group. Aberrant belonging to the Ruminococcaceae family are butyrate composition of Bifidobacteria has been reported in several producers. This reinforces the hypothesis of a reduced disease conditions, and a reduced abundance has been energy harvesting ability and of anti-inflammatory activity described in patients with NASH, although data are scarce in gut microbiota of lean NASH patients. Conversely, the [15,18]. Lactobacillus genus comprises over 180 species reduced abundance of Ruminococcus may have exerted a with important immunological functions [35] but also protective effect in the light of its association with signif- with a wide range of metabolic activity. Indeed, Lactoba- icant fibrosis, as shown by the smaller proportion of pa- cilli produce lactic acid from the fermentation of dietary tients with significant fibrosis found in the lean NASH carbohydrates, acetate, and ethanol [36], which is involved group (75%) than the obese NASH group (100%). in liver damage in patients with NASH [30]. Interestingly,

Figure 2 Volcano plot of logarithmic fold changes (logFC) and elog10 p-values comparing bacterial abundance in NASH patients and controls. The results considered significant (logFC 1 and adjusted p-value <0.05) are colored in red, and the results of interest (logFC 1 and p-value <0.05) are colored in blue. The increase or decrease in bacterial abundance is expressed by a positive or negative value and proportional to the logFC absolute value. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 382 S.M.B. Duarte et al. lean NASH patients showed a marked deficiency in Rumi- The active role of gut microbiota in the regulation of nococcus and Lactobacillus abundance when compared to host metabolism supports the recommendations of the their obese and overweight counterparts. One possible World Health Organization and the World Cancer Research limitation of this study is that the group of obese NASH Foundation for enhancing weight loss or maintenance of a patients was composed only of female patients. This could healthier body weight through an increased consumption have influenced our findings because of the well-known of dietary fibers [42]. Indeed, organisms extract extra en- gender-dependent differences in FXR signaling, bile acids ergy from indigestible residues, such as resistant starch and metabolism, metabolic disease development, and gut dietary fiber, which are not completely hydrolyzed by host microbiota composition [34e36]. enzymes in the small intestine [43]. The main products of Moreover, our food questionnaires collected only qual- fiber fermentation produced by gut bacteria are SCFAs, itative or semi-quantitative data about the dietary habits which are absorbed across the colonic mucosa and can be of the study population. Therefore, our preliminary finding utilized for the de novo synthesis of lipids or glucose [44]. of a different pathogenetic process in lean NASH patients Interestingly, although median fiber intake of lean NASH needs to be confirmed by larger studies, including patient patients (9.65%) was similar to lean (12%) and obese populations stratified by sex and dietary habits. healthy participants (9.2%), SCFAs producing bacteria such Low gut microbiota diversity is a feature of dysbiosis as Faecalibacterium was lower in lean NASH patients, as and may affect health. Although a reduced bacterial di- previously discussed. Dietary intake of single macronutri- versity has been found in children with NAFLD [37], our ents and overall caloric intake were not associated with gut study failed to outline any alteration of microbial intra- microbiota alteration in our study, suggesting that it may individual diversity in NASH patients with different BMI. represent an independent trigger for liver disease devel- Indeed, fecal bacterial diversity of lean NASH patients was opment. However, it is important to clarify that this was the similar to the control group, but it was lower than that of major limitation of our study, since qualitative differences overweight and obese NASH patients (p Z 0.05). However, in diet between lean vs obese NASH patients did not appear further investigations are necessary to confirm a statistically significant. Thus, we cannot exclude any inter- derangement of bacterial alpha diversity in patients with action between dietary factors and the gut microbiota in NASH. the development of NASH in lean subjects. Our analysis of the overall caloric and macronutrient As a final consideration, this study also explored the intake showed no significant differences between lean, possible relationship between gut microbiome composi- overweight and obese of NASH patients nor between tion and liver histology findings in NASH patients. NASH patients and healthy participants. However, dietary Although NAS score was not associated with specific gut intake of lean NASH patients appeared more similar to that microbiome alterations, an increased abundance of Lacto- of healthy participants, characterized by a higher lipid bacilli was found in participants with a fibrosis score intake than overweight and obese NASH patients (28.35% greater than or equal to 2. This finding may be dependent vs. 23.05% and 24%, respectively), as well as lean and obese on the higher prevalence of high Lactobacilli counts among healthy participants (28.8% and 25.3%, respectively). Lean obese NASH patients, who also presented higher fibrosis NASH patients also showed a higher overall caloric intake scores. However, a possible active role of Lactobacilli in the than overweight and obese NASH patients (1541 kcal vs. progression of NASH toward significant fibrosis should not 1160 kcal and 1200 kcal, respectively), as well as lean and be underrated and needs further confirmation. obese healthy participants (1802 kcal and 1894 kcal, Finally, two other limitations of this study were its respectively). While no significant difference was cross-sectional design and small sample size; however, observed, qualitative differences were found between finding and recruiting lean NASH patients who have un- groups. According to previous studies, dietary fat and dergone liver biopsy can be difficult. Our pilot study is the cholesterol seem to interact synergistically to induce the first investigation comparing the microbiome of lean metabolic and hepatic features of NASH [38]. In animals NASH patients with that of obese NASH patients and has studies, a high-fat diet increase the Firmicutes to Bacter- provided results worthy of further study. oidetes ratio and decreases microbial diversity [39].A In conclusion, our findings suggest that lean NASH pa- recent study showed that a high-fat diet (HFD) can cause tients have a specific microbiome composition compared changes in gut microbiota structure in the pre-obesity to overweight/obese NASH patients and healthy in- state. Firmicutes, Bacteroidetes, Tenericutes and Proteobac- dividuals. A fibrosis score greater than or equal to 2 is teria were dominant microbial divisions during HFD. The associated with fecal gut microbiome composition. These relative abundance of Firmicutes was increased, but that of preliminary findings require further investigation with Bacteroidetes decreased in the HFD rats [40]. Another study integrated metabolomics approaches to clarify the role of demonstrated that in HFD-fed mice, the modulation of gut gut microbiome in the development of metabolic disorders microbiota is associated with an increased intestinal such as NASH. permeability that precedes the development of metabolic fl endotoxemia, in ammation and associated disorders. Conflict of interest Furthermore, this modulation can modify the concentra- tion of plasma lipopolysaccharides (LPS) in ob/ob mice, a The authors declare that they have no financial conflicts of mechanism involved in metabolic disorders [41]. interest with respect to this manuscript. Microbiota in lean, overweight, and obese patients with NASH 383

Authors’ contributions [12] Qin J, Li Y, Cai Z, Li S, Zhu J, Zhang F, et al. A metagenome-wide association study of gut microbiota in type 2 diabetes. Nature 2012;490:55e60. Conceived and designed the experiments: Claudia P. [13] Schnabl B, Brenner DA. Interactions between the intestinal micro- Oliveira, José T. Stefano, Ester C. Sabino. Performed the biome and liver diseases. Gastroenterology 2014;146:1513e24. experiments: Marcela de Souza, Ester C. Sabino. Collected [14] Boursier J, Diehl AM. Nonalcoholic fatty liver disease and the gut e and assembled the data: Sebastião M. B. Duarte, Lívia microbiome. Clin Liver Dis 2016;20:263 75. [15] Zhu L, Baker SS, Gill C, Liu W, Alkhouri R, Baker RD, et al. Char- Rodrigues. Analyzed and interpreted data: Luca Miele, acterization of gut microbiomes in nonalcoholic steatohepatitis Francesca R Ponziani. Recruited patients for the study: (NASH) patients: a connection between endogenous alcohol and Fernando G. B. Costa Karla Toda, Daniel F. C. Mazo. Wrote NASH. Hepatology 2013;57:601e9. [16] Betrapally NS, Gillevet PM, Bajaj JS. Changes in the intestinal the paper: Claudia P. Oliveira, Luca Miele, Francesca R. microbiome and alcoholic and nonalcoholic liver diseases: causes Ponziani, José T. Stefano. Critical revision of the manu- or effects? Gastroenterology 2016;150. 1745e1755.e1743. script. Luca Miele, Claudia P. Oliveira, Antonio Gasbarrini, [17] Machado MV, Cortez-Pinto H. Diet, microbiota, obesity, and Flair J. Carrilho. NAFLD: a dangerous quartet. Int J Mol Sci 2016;17:481. fi [18] Boursier J, Mueller O, Barret M, Machado M, Fizanne L, Araujo- All authors approved the nal version of this Perez F, et al. The severity of nonalcoholic fatty liver disease is manuscript. associated with gut dysbiosis and shift in the metabolic function of the gut microbiota. Hepatology 2016;63:764e75. [19] Sreenivasa Baba C, Alexander G, Kalyani B, Pandey R, Rastogi S, Acknowledgments Pandey A, et al. Effect of exercise and dietary modification on serum aminotransferase levels in patients with nonalcoholic steatohepatitis. J Gastroenterol Hepatol 2006;21:191e8. We thank Fundação do Estado de São Paulo (FAPESP) grant [20] Margariti E, Deutsch M, Manolakopoulos S, Papatheodoridis GV. 2013/06828-0 for supporting this research. Non-alcoholic fatty liver disease may develop in individuals with normal body mass index. Ann Gastroenterol 2012;25:45e51. [21] Neuschwander-Tetri BA, Caldwell SH. Nonalcoholic steatohepati- Appendix B. Supplementary data tis: summary of an AASLD single topic conference. Hepatology 2003;37:1202e19. [22] Bellentani S, Dalle Grave R, Suppini A, Marchesini G, Network FLI. Supplementary data related to this article can be found at Behavior therapy for nonalcoholic fatty liver disease: the need for https://doi.org/10.1016/j.numecd.2017.10.014. a multidisciplinary approach. Hepatology 2008;47:746e54. [23] Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, et al. QIIME allows analysis of high-throughput References community sequencing data. Nat Methods 2010;7:335e6. [24] Raman M, Ahmed I, Gillevet PM, Probert CS, Ratcliffe NM, Smith S, et al. Fecal microbiome and volatile organic compound metab- [1] Miquel S, Leclerc M, Martin R, Chain F, Lenoir M, Raguideau S, et al. olome in obese humans with nonalcoholic fatty liver disease. Clin fi fl Identi cation of metabolic signatures linked to anti-in ammatory Gastroenterol Hepatol 2013;11. 868e875.e861-863. effects of Faecalibacterium prausnitzii. MBio 2015;6. [25] Wong VW, Tse CH, Lam TT, Wong GL, Chim AM, Chu WC, et al. [2] Loomba R, Sanyal AJ. The global NAFLD epidemic. Nat Rev Gas- Molecular characterization of the fecal microbiota in patients with e troenterol Hepatol 2013;10:686 90. nonalcoholic steatohepatitisea longitudinal study. PLoS One 2013; [3] Wong RJ, Cheung R, Ahmed A. Nonalcoholic steatohepatitis is the 8:e62885. most rapidly growing indication for liver transplantation in pa- [26] Jiang W, Wu N, Wang X, Chi Y, Zhang Y, Qiu X, et al. Dysbiosis gut tients with hepatocellular carcinoma in the U.S. Hepatology 2014; microbiota associated with inflammation and impaired mucosal e 59:2188 95. immune function in intestine of humans with non-alcoholic fatty [4] Manti S, Romano C, Chirico V, Filippelli M, Cuppari C, Loddo I, et al. liver disease. Sci Rep 2015;5:8096. Nonalcoholic Fatty liver disease/non-alcoholic steatohepatitis in [27] Scott KP, Martin JC, Duncan SH, Flint HJ. Prebiotic stimulation of “ childhood: endocrine-metabolic mal-programming. Hepat Mon human colonic butyrate-producing bacteria and bifidobacteria, 2014;14:e17641. in vitro. FEMS Microbiol Ecol 2014;87:30e40. [5] Basaranoglu M, Basaranoglu G, Sentürk H. From fatty liver to [28] Prorok-Hamon M, Friswell MK, Alswied A, Roberts CL, Song F, fi “ brosis: a tale of second hit. World J Gastroenterol 2013;19: Flanagan PK, et al. Colonic mucosa-associated diffusely adherent e 1158 65. afaCþ Escherichia coli expressing lpfA and pks are increased in [6] Mosca A, Nobili V, De Vito R, Crudele A, Scorletti E, Villani A, et al. inflammatory bowel disease and colon cancer. Gut 2014;63: Serum uric acid concentrations and fructose consumption are 761e70. independently associated with NASH in children and adolescents. [29] Al-Jashamy K, Murad A, Zeehaida M, Rohaini M, Hasnan J. Preva- e J Hepatol 2017;66(5):1031 6. lence of colorectal cancer associated with Streptococcus bovis [7] Turnbaugh PJ, Hamady M, Yatsunenko T, Cantarel BL, Duncan A, among inflammatory bowel and chronic gastrointestinal tract Ley RE, et al. A core gut microbiome in obese and lean twins. disease patients. Asian Pac J Cancer Prev 2010;11:1765e8. e Nature 2009;457:480 4. [30] Leung C, Rivera L, Furness JB, Angus PW. The role of the gut [8] Turnbaugh P, Ley R, Mahowald M, Magrini V, Mardis E, Gordon J. microbiota in NAFLD. Nat Rev Gastroenterol Hepatol 2016;13: An obesity-associated gut microbiome with increased capacity for 412e25. e energy harvest. Nature 2006;444:1027 31. [31] Brahe LK, Astrup A, Larsen LH. Is butyrate the link between diet, ’ [9] Murphy EF, Cotter PD, Hogan A, O Sullivan O, Joyce A, Fouhy F, intestinal microbiota and obesity-related metabolic diseases? et al. Divergent metabolic outcomes arising from targeted Obes Rev 2013;14:950e9. manipulation of the gut microbiota in diet-induced obesity. Gut [32] Duncan SH, Louis P, Thomson JM, Flint HJ. The role of pH in e 2013;62:220 6. determining the species composition of the human colonic [10] Vrieze A, Van Nood E, Holleman F, Salojärvi J, Kootte RS, microbiota. Environ Microbiol 2009;11:2112e22. Bartelsman JF, et al. Transfer of intestinal microbiota from lean [33] Jumpertz R, Le DS, Turnbaugh PJ, Trinidad C, Bogardus C, Gordon JI, donors increases insulin sensitivity in individuals with metabolic et al. Energy-balance studies reveal associations between gut e syndrome. Gastroenterology 2012;143. 913 916.e917. microbes, caloric load, and nutrient absorption in humans. Am J [11] Larsen N, Vogensen FK, van den Berg FW, Nielsen DS, Clin Nutr 2011;94:58e65. Andreasen AS, Pedersen BK, et al. Gut microbiota in human adults [34] Sheng L, Jena PK, Liu HX, Kalanetra KM, Gonzalez FJ, French SW, with type 2 diabetes differs from non-diabetic adults. PLoS One et al. Gender differences in bile acids and microbiota in 2010;5:e9085. 384 S.M.B. Duarte et al.

relationship with gender dissimilarity in steatosis induced by diet overflow of dietary fat to the distal intestine. Am J Physiol Gas- and FXR inactivation. Sci Rep 2017;7:1748. trointest Liver Physiol 2012;303:G589e99. [35] Trottier J, Caron P, Straka RJ, Barbier O. Profile of serum bile acids [40] Lin H, An Y, Hao F, Wang Y, Tang H. Correlations of fecal in noncholestatic volunteers: gender-related differences in metabonomic and microbiomic changes induced by high-fat diet response to fenofibrate. Clin Pharmacol Ther 2011;90:279e86. in the pre-obesity state. Sci Rep 2016;6:21618. [36] Org E, Mehrabian M, Parks BW, Shipkova P, Liu X, Drake TA, et al. [41] Cani PD, Bibiloni R, Knauf C, Waget A, Neyrinck AM, Delzenne NM, Sex differences and hormonal effects on gut microbiota compo- et al. Changes in gut microbiota control metabolic endotoxemia- sition in mice. Gut Microbes 2016;7:313e22. induced inflammation in high-fat diet-induced obesity and dia- [37] Del Chierico F, Nobili V, Vernocchi P, Russo A, De Stefanis C, Gnani D, betes in mice. Diabetes 2008;57:1470e81. et al. Gut microbiota profiling of pediatric NAFLD and obese patients [42] Wiseman M. The second World Cancer Research Fund/American unveiled by an integrated meta-omics based approach. Hepatology Institute for Cancer Research expert report. Food, nutrition, 2017;65(2):451e64. physical activity, and the prevention of cancer: a global perspec- [38] Savard C, Tartaglione EV, Kuver R, Haigh WG, Farrell GC, tive. Proc Nutr Soc 2008;67:253e6. Subramanian S, et al. Synergistic interaction of dietary cholesterol [43] Turnbaugh PJ, Ley RE, Mahowald MA, Magrini V, Mardis ER, and dietary fat in inducing experimental steatohepatitis. Hep- Gordon JI. An obesity-associated gut microbiome with increased atology 2013;57:81e92. capacity for energy harvest. Nature 2006;444:1027e31. [39] de Wit N, Derrien M, Bosch-Vermeulen H, Oosterink E, Keshtkar S, [44] Macfarlane GT, Macfarlane S. Fermentation in the human large Duval C, et al. Saturated fat stimulates obesity and hepatic stea- intestine: its physiologic consequences and the potential contri- tosis and affects gut microbiota composition by an enhanced bution of prebiotics. J Clin Gastroenterol 2011;45. Suppl:S120-127.