The Composition Of Human Gut Microbiota In Non-alcoholic Fatty Liver Disease

Back ground: the prevalence of non-alcoholic fatty liver disease (NALFD) is increasingworldwidely, and gut flora has been proved to play an important role in the pathogenesis indevelopment of NAFLD.Methods: we selected the hospitalized participants for health check in InternationalMedical Center of the PLA General Hospital from2011September to2013February.Thediagnose of NAFLD was made according to the Guidelines for the diagnosis andmanagement of nonalcoholic fatty liver disease: update2010. The fast peripheral blood andfresh stool samples were collected and stored in-80℃refregerator within2hours forfurther tests and analyses. The genome DNA of stool samples were extracted for followingamplication of V4region of16S rDNA, and the products of V4region were thensequenced on the Illumina Miseq platform.Resusts: the current research finally collected67cases, including30NAFLD patients and37health controls. There were no statistal differences between the groups in gender, age,diastolic pressure, total cholesterol, low density lipoprotein cholesterol and aspartateaminotransferase. The body mass index (BMI) and weist-to-hip ratio (WHR) were higherin NAFLD group than those in controls (P＜0.001), and so did the levels of triglyceride (P＜0.001), alanine transaminase (P=0.004), γ-glutamyl transpeptidase (P=0.012), fastglucose (P=0.003), fast insulin (P＜0.001) and HOMA-IR (P＜0.001). Multivarite logisticregression analysis showed that HOMA-IR was the only independent factor associatedwith NAFLD.The pyrosequencing provided4460840reads in total and66579.7reads per stool sample, with an average lenth252bps per read. As a result,938optional taxonamy units (OTU)were determined by using the Uparse software with a similarity of97%as a cutoffshreshold. Alpha-diversity analysis showed that therarefaction curves of all the sampleshad a slowing tendency at the point of2000sequence, and the OTU numbers, Chao1indexand Shannon index were similar between the two groups. Principal Coordinate Analysis(PCA) showed that67samples could not be totally devided into two clusters as NAFLD orcontrol groups, and the overlaps between the two kind of the samples were obvious. Byusing UPGMA software, clustering analysis was performed to show the similarities of thegut flora composition among all the samples. The overall tendency was that the sampleswere first clustered whin the NAFLD or the control group and then further clustered withother samples or clusters.Metastats analysis showed there were statistical differences between the groups in8families. Moreover,12genus were identified with significant differences between theNAFLD and the control groups, including Porphyromonas,Odoribacter, Coprococcus,Clostridium, Blauti, Dore, Peptococcus, rc4-4, Slackia, Mitsuokella, Succinivibrio andProteus. Redundance analysis (RDA) showed the triglyceride had a smaller influence ongut flora than HOMA-IR and WHR. The Pr values of HOMA-IR and WHR were same as0.005, while the R square of HOMA-IR was higer than that of WHR (3.394vs.3.363),indicating a more important role of HOMA-IR in influencing the composition of gutmicrobiome.Conclusions:HOMA-IRwas one of the most important biomarkers related to NAFLD,implying the important role of insulin resistance in pathogenesis of the disease. All theindividuals had its unique gut flora composition. Although the overlaps existed obviously,the overall tendency was the samples were firstly clustered within the NAFLD or thecontrol groups. HOMA-IR was aolso the biomarker which related to gut flora compositionmost closely.