Abstract

Results

Clinical data of individuals in the nonobese cohort

Altogether, 126 nonobese subjects with (n=43) or without (n=83) NAFLD were enrolled in this prospective cross-sectional study. We performed anthropometric and biochemical phenotyping of the individuals (Table 1). Differences in clinical biochemical metabolic markers were compared between individuals with and without NAFLD. These markers are used to provide important information about liver function and to identify potential NAFLD patients. The liver synthetic function of all the individuals was in the normal range, as determined by albumin and International Normalized Ratio levels (data not shown). The age and gender distributions between the individuals with and without NAFLD were similar. All subjects consumed an omnivorous Chinese diet. Although all enrolled subjects were nonobese, the individuals with NAFLD had higher BMI, waist circumference, hip circumference and waist-to-hip ratio values (p=0.004, 0.0001, 0.006, and 0001, respectively). In consistent with reports by Xu et al. of nonobese NAFLD²², the individuals with NAFLD were characterized by higher levels of alanine aminotransferase (ALT) (p<0.0001), alkaline phosphatase (p=0.002), total triglycerides (p<0.0001), very low-density lipoprotein cholesterol (p<0.0001), gamma-glutamyl transpeptidase (γ-GT) (p<0.0001), and serum uric acid (p=0.016). In addition, the patients had markedly lower levels of high-density lipoprotein cholesterol (p=0.0003) compared with those of HCs. The increased levels of these markers suggest that individuals with NAFLD are featured by metabolic disturbances known to be associated with an increased risk of liver dysfunction and hepatocyte lipid accumulation and decreased β-oxidation energy production²⁴.

Table 1

Demographic, anthropometrics and laboratory data of the study cohorts.

Variables	With NAFLD (n=43)	Without NAFLD (n=83)	p-value
Sex (male/female)	43 (36/7)	83 (70/13)	—
Age (years)	47.0 (34.5–61.0)	40.5 (33.0–52.0)	0.31
Body mass index (kg/m2)	23.19 (22.19–24.22)	21.77 (20.7–23.38)	0.004
Waist circumference (cm)	89 (84–91.75)	79.5 (72.5–85)	<0.0001
Hip circumference (cm)	98 (94.75–100)	95 (91–97)	0.0058
Waist-to-hip ratio	0.90 (0.89–0.92)	0.84 (0.80–0.89)	<0.0001
Systolic blood pressure (mm Hg)	127.0 (111.5–136.0)	119 (110–129)	0.19
Diastolic blood pressure (mm Hg)	75 (70.5–82.5)	72 (67–80)	0.36
ALT (U/l)	29 (20.5–39.5)	14.5 (12–20.75)	0.0001
AST (U/l)	22 (18–25.5)	20 (18.0–21.75)	0.06
Alkaline phosphatase (g/l)	78 (63–91)	65 (54–78.75)	0.002
Total bilirubin (μmol/l)	13 (11–18)	12 (10–17)	0.182
Direct bilirubin (μmol/l)	4 (3–5)	4 (3–5)	0.95
Indirect bilirubin (μmol/l)	10 (8.5–13)	9 (6.25–12)	0.12
Total triglycerides (mmol/l)	1.83 (1.43–2.6)	1.04 (0.77–1.45)	0.0001
Total cholesterol (mmol/l)	4.9 (4.25–5.36)	4.68 (4.1–5.11)	0.12
High-density lipoprotein cholesterol (mmol/l)	1.12 (0.89–1.30)	1.4 (1.19–1.59)	0.0003
Low-density lipoprotein cholesterol (mmol/l)	2.54 (2.14–2.95)	2.50 (2.04–2.79)	0.40
Very low-density lipoprotein cholesterol (mmol/l)	1.01 (0.84–1.42)	0.7 (0.47–0.94)	<0.0001
Fasting glucose (mmol/l)	4.96 (4.74–5.47)	4.87 (4.40–5.20)	0.132
γ-Glutamyltransferase (U/l)	37 (23.5–62.5)	20 (14–28)	<0.0001
Serum uric acid (μmol/l)	383 (310–423)	340.5 (392.3–387.3)	0.016
Hemogobin (g/l)	158 (153–165)	154 (144–164)	0.07
Platelet count (×109/l)	214 (183–250)	199 (168.5–234)	0.08

Open in a separate window

Note: Data are expressed as the mean (SEM.) or median (IQR, interquartile range).

Differed pattern of fecal microbiota between individuals with and without NAFLD

Compositional differences in the microbiota of this nonobese cohort were first assessed by DGGE patterns (Supplementary Figure S1A) combined with several clustering methods. As shown in Supplementary Figure S1B, according to multidimensional scaling analysis, an unconstrained clustering method, microbiota composition differed significantly between the two groups. Moreover, PLS-DA (Fig. 1A–C and supplementary Table S1), an supervised analysis method, also allowed for the detection of differences between individuals with and without NAFLD independent of their ALT level, indicating distinctive fecal microbial communities between HC and NAFLD groups. Furthermore, PLS-DA showed that NAFLD individuals with or without normal liver enzyme levels were also associated with clearly different microbiota (Fig. 1D). The validation plot (Supplementary Figure S2) demonstrated that the PLS-DA models were valid: the Q² regression line had a negative intercept, and all of the permuted R² values to the left of the intercept were lower than the original point to the right. These findings suggested the role of gut microbiota in NAFLD and the potential of using the gut microbiota as a marker to differentiate among individuals with or without NAFLD.

Open in a separate window

Figure 1

PLS-DA score plots based on DGGE profiles distinguishing between the fecal microbial communities in nonobese individuals with and without NAFLD.

(A) NAFLD patients vs. HCs, (B) ND vs. HCs, (C) NS vs. HCs, and (D) ND vs. NS. HCs (n=55), healthy controls; ND (n=20), individuals with NAFLD; NS (n=14), NAFLD patients with elevated serum liver enzymes.

Quantification of fecal microbial changes associated with the presence of NAFLD by 454 pyrosequencing

To assess quantitative changes in the fecal microbiota associated with the presence of NAFLD, parallel pyrosequencing was performed in a selected subgroup of the nonobese cohort. The demographic and clinical information of the selected subjects and the characteristics of pyrosequencing data were listed in Supplementary Tables S2 and S3, respectively. The distribution of BMI, waist circumference, gender, smoking, alcohol consumption, and dietary habits were not statistically different between individuals with and without NAFLD in the subgroup. In agreement with the findings of DGGE profiling, decreased ecological diversities and altered fecal microbiome compositions of NAFLD patients (ND, n=8; NS, n=2) were exhibited compared with those of HCs (n=15) (Figs 2 and ​and33 and Supplementary Table S3). As shown in Fig. 2A, the Shannon diversity index (H) was significantly reduced in the NAFLD group (4.19vs. 4.71, p=0.03), whereas the Simpson’s diversity index (0.05vs. 0.02, p=0.02) was increased (Fig. 2B), suggesting decreased diversity of the IM in patients with NAFLD. Reduced richness in bacterial diversity was also observed in individuals with NAFLD by the rarefaction curve trend (Fig. 2C). Additionally, the Good’s coverage of each sample for genera was greater than 80%, demonstrating that these sequences may represent major constituents in the bacterial community. Based on the pyrosequencing data, the fecal microbial communities of individuals with or without NAFLD showed clear separation in the PLS-DA score plot (R²X=0.466, R²Y=0.76, Q²(cum)=0.239) (Fig. 2D), indicating that NAFLD was associated with the significant changes in diversity and composition of the fecal microbial community compared with those in HCs.

Open in a separate window

Figure 2

The 454 pyrosequencing analysis revealing decreased ecological diversity of the fecal microbiome in nonobese individuals with NAFLD patients (n=10) compared with individuals without NAFLD (n=15).

(A) The Shannon diversity index, (B) the Simpson diversity index, and (C) the rarefaction curve trend between patients with NAFLDs and HCs; (D) PLS-DA score plots based on the relative abundances of microbial genera of the first two components (t [1]=20.9%, t [2]=16.1%) showed the trend that the two groups were well separated.

Open in a separate window

Figure 3

The relative abundances of fecal microbiota significantly differed in individuals with and without NAFLD in the nonobese cohort.

(A) Ratios of the phyla Firmicutes and Bacteroidetes, (B) phyla Firmicutes and Bacteroidetes, (C) classes clostridia and Bacteroidia, (D) families Lachnospiraceae and Ruminococcaceae; (E) Lactobacillaceae and Peptostreptococcaceae; (F) G+ and G− bacteria; (G) The ratios of G− to G+ bacteria.

Further analysis of microbiome data showed that NAFLD was associated with changes in the fecal microbiota at the phylum, class, family and genus levels (Fig. 3, Table 2 and Supplementary Figure S3). As shown in Fig. 3A, a significant decrease in the ratio of Firmicutes to Bacteroidetes (F/B) (0.47vs. 1.4, p=0.003) was found between the groups. In addition, individuals with NAFLD had higher fecal levels of the phylum Bacteroidetes (66.0% vs. 46.4%, p=0.005) and lower levels of Firmicutes (27.5% vs. 51.9%, p=0.002) compared with HCs (Fig. 3B). Furthermore, the predominant class Bacteroidia in the phylum Bacteroidetes was enriched in patients with NAFLD (65.93% vs. 46.18%, p=0.004), whereas the class Clostridia within the phylum Firmicutes was depleted in individuals with NAFLD (27.02% vs. 50.91%, p=0.001) (Fig. 3C). Furthermore, the marked depletion of Firmicutes in NAFLD group was mostly explained by decreases in four families: Lachnospiraceae (18.67% vs. 34.26%, p=0.002), Ruminococcaceae (6.69% vs. 13.67%, p=0.018), Lactobacillaceae (0.04% vs. 0.73%, p=0.0008), and Peptostreptococcaceae (0.09% vs. 0.44%, p=0.002) (Fig. 3D,E).

Table 2

The average abundance and the variable importance in projection index of the key differentiating genera between non-obese individuals with and without NAFLD identified by the PLS-DA model based on the 454 pyrosequencing data.

NO.	Genus	Family	Phylum	VIP	With ND	Without ND	pvalue
1	Bacteroides	Bacteroidaceae	Bacteroidetes	5.9	29%±4%	20%±1.8%	0.103
2	Prevotella	Prevotellaceae	Bacteroidetes	3.8	8%±4%	3%±1%	0.789
3	Pseudobutyrivibrio	Lachnospiraceae	Firmicutes	1.9	0.45%±0.2%	2.16%±0.6%	0.02
4	Anaerotruncs	Ruminococcaceae	Firmicutes	1.9	0.19%±0.2%	0.96%±0.4%	0.004
5	Lactobacillus	Lactobacillaceae	Firmicutes	1.7	0.04%±0.03%	0.74%±0.4%	0.003
6	Roseburia	Lachnospiraceae	Firmicutes	1.6	1.48%±0.6%	2.59%±0.4%	0.01
7	Coprococcus	Lachnospiraceae	Firmicutes	1.6	0.34%±0.1%	2.4%±0.6%	0.009
8	Streptococcus	Streptococcaceae	Firmicutes	1.5	0.28%±0.01%	0.2%±0.1%	0.24
9	Ruminococcus	Ruminococcaceae	Firmicutes	1.4	1.03%±0.7%	1.59%±0.3%	0.02
10	Moryella	Lachnospiraceae	Firmicutes	1.4	0.05%±0.02%	0.35%±0.1%	0.02
11	Anaerosporabacter	Lachnospiraceae	Firmicutes	1.4	1.08%±0.1%	2.02%±0.1%	0.02
12	Oscillibacter	Ruminococcaceae	Firmicutes	1.2	0.8%±0.4%	1.4%±0.3%	0.05
13	Bifidobacterium	Bifidobacteriaceae	Actinobacteria	1.2	0.36%±0.2%	0.22%±0.1%	0.82
14	Escherichia	Enterobacteriaceae	Proteobacteria	1	0.32%±0.3%	0.1%±0.05%	0.78

Open in a separate window

Note: Data are expressed as the mean (SEM.); VIP, the variable importance in projection.

Moreover, at the genus level, 14 key microbial genera were identified as key microbes with PLS-DA. These genera were responsible for the separation between non-obese individuals with and without NAFLD (Table 2), and well-known gut bacteria such as Bacteroides, Pseudobutyrivibrio and Lactobacillus. Interestingly, differences in the fecal microbiota of individuals with NAFLD were observed only in genera belonging to Firmicutes (Table 2 and Supplementary Figure S3). For example, within the Lachnospiraceae family, Coprococcus (mean 0.34% vs. 2.4%, p=0.009), Pseudobutyrivibrio (0.45% vs. 2.16%, p=0.02), Moryella (0.05% vs. 0.35%, p=0.02), Roseburia (1.48% vs. 2.59%, p=0.01) and Anaerosporobacter (1.08% vs. 2.02%, p=0.02) were decreased in individuals with NAFLD. Within the family Ruminococcaceae, a similar decrease was observed for Anaerotruncus (0.19% vs. 0.96%, p=0.004) and Ruminococcus (1.03% vs. 1.59%, p=0.02). Notably, Lactobacillus, the main genus of the Lactobacillaceae family, was markedly reduced in both abundance (0.04% vs. 0.74%, p=0.003) and prevalence (30% vs. 82%, p=0.008) in NAFLD patients compared with HCs (Supplementary Figure S4). Furthermore, to explore the potential role of the IM in the pathogenesis of NAFLD, the relative abundances of G- bacteria and Gram-positive (G+) bacteria based on the pyrosequencing data were assessed. Importantly, in individuals with NAFLD compared with HCs, G- bacteria were markedly enriched (85.21% vs. 71.8%, p=0.009), and G+ bacteria were decreased (14.79% vs. 28.2%, p=0.01) (Fig. 3F). Additionally, the ratios of G−/G+ bacteria in individuals with NAFLD were enriched compared with those in HCs (5.72vs. 3.52, p=0.009) (Fig. 3G).

Potential correlation between intestinal dysbiosis and liver biochemical markers

We also attempted to assess the correlation between differences in microbial composition and clinical liver biochemical markers of NAFLD in the non-obese cohort. Interestingly, a clear significant correlation with the set of liver biochemistry indices in plasma was found for NAFLD-induced disturbed microbiota at the phylum and family and genus level by Spearmen’s rank correlation analysis (Fig. 4 and Supplementary Figures S5 and S6). These liver biochemistry markers are well-known clinical biochemical markers for NAFLD diagnosis^1,2,22. At the global level, significant correlations with markers of liver status, such as ALT and uric acid, were found for the two dominant phyla Firmicutes, and Bacteroidetes. In addition, a significant correlation with γ-GT, markers of liver injury, was also found for Firmicutes. Additionally, at the family level, further significant correlations were found for plasma ALT with 3 families (Lachnospiraceae, Ruminococcaceae, Lactobacillaceae), and for plasma γ-GT with Lachnospiraceae. Moreover, both plasma total triglycerides and very low-density lipoprotein cholesterol were significantly correlated with Lactobacillaceae (p<0.0001).

Open in a separate window

Figure 4

Furthermore, the correlations between the set of liver biochemistry markers and disturbed microbial genera were also revealed (Fig. 4 and Supplementary Figure S7). For instance, one metabolic marker, fasting glucose, revealed only a single microbial association with Coprococcus. Other markers had multiple correlations. For example, the increase of plasma ALT was statistically linked with seven depleted genera (Pseudobutyrivibrio, Moryella, Anaerosporobacter, Roseburia, Ruminococcus, Anaerotruncus, and Lactobacillus). These findings indicated a close interrelationship between the intestinal microbial composition and liver biochemical state, highlighting that alterations of the IM are implicated in the presence and development of NAFLD.

Discussion

To our knowledge, this is the first study assessing the gut microbiota in nonobese individuals with NAFLD compared with those without NAFLD and specifically correlating the microbial composition of individuals with clinical indices. Here, we identified the significant distinguishing fecal microbiota at all taxonomic levels in non-obese individuals with NAFLD compared with those without NAFLD; the former were characterized by a more pronounced Firmicutes-poor microbiota and marked lower overall microbial richness. Further analyses revealed that the altered microbiota associated with the presence of NAFLD significantly correlated with liver biochemical markers in the nonobese cohort.

Classifying NAFLD individuals based on BMI is crucial for deciphering the potential role of gut microbiota in NAFLD, because BMI represents a major determinant of compositional changes in microbial communities²⁵. Moreover, non-obese patients with NAFLD also represent a considerable portion of NAFLD patients. Notably, in spite of the lower BMI and lower prevalence of obesity in the Asia-Pacific region, people in this region still tend to develop NAFLD. Because both NAFLD and obesity are believed to be influenced by the gut microbiota, comparisons between non-obese individuals with and without NAFLD allowed us to further address the effects of the IM on the liver function.

In human IM, Firmicutes and Bacteroidetes are the dominant bacteria, accounting for approximately 99% of the whole microbiota^6,8. In our cohort, a significant lower abundance of Firmicutes and a higher abundance of Bacteroidetes were exhibited in individuals with NAFLD compared with those without NAFLD. This finding was partly in agreement with the findings of Zhu et al. and Schwiertz A et al., who have demonstrated lower Firmicutes and higher Bacteroidetes abundances in obese patients compared with lean healthy controls^18,26. But Zhu et al. did not find the similar difference between obese and NASH microbiome¹⁸. The novelty of our study is that we provide the first report of a direct association between lower Firmicutes and NAFLD in non-obese individuals. Interestingly, our findings were significantly different with those of a recent report of Mouzaki et al., who have shown lower Bacteroidetes levels in NASH patients with higher BMI than in patients with simple steatosis and controls with lower BMI¹⁶. These results indicated that obese NASH and nonobese NAFLD have different fecal microbiome. And the discrepancy between the findings might partly reflect the imbalance of uncontrolled effects of factors between the groups, such BMI and different detection techniques. The decreased proportion of Bacteroidetes has also been demonstrated in patients with high BMI in previously published literature in the field of obesity^8,9. Thus, the high BMI of subjects may have affected the results of the study by Zhu et al., although the linear regression, adjusting for BMI, was performed to theoretically limit the potential confounding effect of BMI. In addition, the different detection techniques also affected these results between the two study cohorts, which was evidenced by previous studies by our group¹¹ and Larsen et al.²⁷. In our study cohort, all recruited individuals were non-obese adults, which could avoid the confounding factor of BMI. It should be noted that in our non-obese cohort, our observations suggested that a slight increase in BMI, even within the normal range, in people with BMI <25kg/m² was significantly associated with an increased risk of NAFLD. However, it is unlikely that there are significant BMI-dependent variations in the IM composition within the nonobese spectrum^16,18. And in our quantification study of fecal microbial change, we only selected the nonobese individuals with the median value of BMI to ensure the similar distribution of BMI and waist circumference between two groups. In addition, unbiased metagenomics approaches were applied in this study to characterize the whole IM, including some unknown but potentially functional bacteria²⁸.

Bacterial phylotypes decrease in nonobese patients with NAFLD was mostly associated with Firmicutes. The families Lachnospiraceae and Ruminococcaceae are two predominant members of Firmicutes. These bacteria are short-chain fatty acids (SCFAs)-producing bacteria that are well known for their health-promoting functions, including the production of nutrients for the host and an energy supply for the colonic epithelium, the modulation of colonic pH²⁹, and the maintenance of host immune homeostasis³⁰. A decrease in SCFAs might result in deteriorated intestinal integrity and increased intestinal permeability, thus making decreased SCFAs a very important pathogenic factor in NAFLD³¹. In addition, these bacteria are also capable of performing 7-α dehydroxylation and deconjugation in bile acid metabolism, which have been demonstrated to play a critical role in the pathogenesis of chronic liver diseases such as inflammation, cirrhosis and hepatocellular carcinoma^12,32. Indeed, in Jone’s metagenomics study, the function bile salt hydrolases had been identified in all major bacterial phyla and archaeal species in the gut³³, supporting that IM might modulate the bile acid metabolism in NAFLD patients. In addition, it has been evidenced that IM could modulate the fecal bile acids profiles in patients with cirrhosis³² and even promote hepatocellular carcinoma development through modulating the enterohepatic circulation of BAs¹². Furthermore, we showed that fecal microbial ecological diversity was significantly reduced in patients with NAFLD. Thus far, the exact reason for decreased bacterial diversity in NAFLD remains unclear. It has been demonstrated during antibiotic treatment³⁴ and in diseases such as liver cirrhosis¹⁹ and obesity³⁵. Such alterations in the IM have recently been reported for a variety of liver diseases^19,32,36. Thus, our finding of general dysbiosis in NAFLD raises the possibility that the compositional change could result in imbalanced microbial ecology, which might itself play a role in increasing the susceptibility to liver diseases.

In this study, we also found a higher abundance of lipopolysaccharide-producing G- bacteria in non-obese individuals with NAFLD. In agreement with our study, through pyrosequencing analysis, De Minicis et al. have also revealed an increase in G- bacteria in mice with bile duct ligation fed a high-fat diet³⁷. And a significant difference in Escherichia was exhibited between obese children with and without NASH in the study by Zhu et al.¹⁸. Additionally, evidence from the literature supports a causal role for endotoxin produced by G- bacteria in NAFLD/NASH^14,38,39. Here, a significant positive correlation between G bacteria and fasting glucose levels was also found, highly suggesting the potential contribution of predictive markers for assessing the presence and development of NAFLD in non-obese individuals. It would be interesting to assess differences in the circulating levels of endotoxin between individuals with and without NAFLD in a nonobese cohort.

Notably, we first demonstrated that the health-promoting Lactobacillaceae family and its most abundant genus Lactobacillus showed marked reductions in both abundance and prevalence in patients with NAFLD compared with HCs. Probiotic bacteria such as Lactobacillus and Bifidobacterium promote beneficial effects, likely through anti-inflammatory actions and stabilization of the intestinal barrier, thereby attenuating liver pathologies⁴⁰. As concluded by many clinical studies, these probiotic bacteria reduce features of NAFLD in humans^41,42,43 and liver injury in mouse models⁴⁴. For example, some strains of Lactobacillus lead to substantial reductions in the levels of ALT in 10 patients with NASH⁴³. Consistently with the above findings, a negative correlation between Lactobacillus and a wide range of biochemical indices of NAFLD progression, including ALT, and TG, was also found, highly facilitating future efforts to better manipulate the gut microbiota for the treatment of NAFLD.

It is widely accepted that the gut microbiota is crucial for NAFLD³¹, which is considered as the hepatic manifestation of metabolic syndrome. However, to date, little is understood about the molecular host-microbiome interactions that influence host metabolic pathways. Recent studies both in adult patients with NAFLD⁴⁵ and mice¹³ supported that gut dysbiosis exerts a critical influence in the progression of NAFLD. In this study a notable finding was the clear correlation between the microbiome composition and well-known clinical indices of NAFLD progression, such as higher blood ALT, γ-GT, TG, VLDL-C, GLU and SUA, suggesting that the microbiota might be closely involved in the pathogenesis of the progression of NAFLD. For example, the blood level of ALT and γ-GT, the inflammatory indices, were significantly correlated with the distinguishing bacteria between individuals with and without NAFLD. In consistent with our findings, Boursier et al. also demonstrated that gut dysbiosis associated with NAFLD severity is accompanied by a shift in the metabolic functions of gut microbiota⁴⁵. In addition, the correlation between IM and SUA was also firstly revealed in this study. The level of SUA is elevated significantly in patients with NAFLD, which is an independent risk factor of NAFLD⁴⁶. Uric acid is the major end product of purine metabolism, and has been proposed as a natural scavenger of peroxynitrite and peroxynitrite-derived radicals^47,48. Despite the fact that it is not clear how IM regulates these opposing roles of uric acid on redox balance, our study added new clues that IM might promote steatosis to steatohepatitis in NAFLD by its metabolic pathways and products.

A considerable number of studies support the very important role of the microbiome in the pathogenesis of metabolic disturbance in NAFLD^{18,30,35,49,50}. Recently, Mouzaki et al. have reported a potential negative association between Bacteroidetes and insulin resistance when controlling for BMI¹⁸. Compared with previous reports, our study presents stronger evidence supporting these findings, because we did assess differences in the gut microbiota between nonobese individuals with and without NAFLD; additionally, our study suggests a potential correlation between alterations in fecal microbiota and metabolic disturbance in NAFLD. Future studies are required to establish the mechanistic relevance of our novel findings to the pathogenesis of NAFLD. For instance, it will be necessary to reveal the presence of gut-derived metabolites (i.e., lipopolysaccharide or SCFAs) in the peripheral or portal blood as a medium for the systemic effects of these compounds in human beings or gut microbiome biomarkers associated with the progression of NAFLD and/or NASH.

There were several limitations in this study. First, NAFLD was diagnosed by ultrasonographic methods, which could not determine the severity of NAFLD⁵¹. Nevertheless, ultrasonography is reported with reasonable accuracy and widely used for population-based studies^22,51. Here, all included subjects had no history of any disease and were recruited at the time of their annual physical examination, it was impractical and unethical to perform liver biopsy in this study. And we performed all the ultrasonographic diagnosis by a single experienced radiographer and the diagnostic performance of the ultrasonographer was evaluated by one senior ultrasonographer meanwhile. Second, this study design did not allow for the examination of insulin resistance, although limited research has found insulin resistance might be a risk factor of NAFLD among nonobese individuals²³. Future studies are necessary to investigate the correlation between insulin resistance and IM in nonobese NAFLD. Despite these limitations, our conceptually novel study provided important insights into the IM in nonobese adult patients with NAFLD and may have significant clinical importance for NAFLD manipulation.

In summary, our findings firstly showed that the non-obese patients with NAFLD were characterized by a decrease in gut microbial diversity, the depletion of Firmicutes including health-promoting bacteria such as SCFAs-producing Lachnospiraceae, 7α-dehydroxylating Ruminococcaceae and beneficial Lactobacillaceae, and the increase of potentially opportunistic pathogenic lipopolysaccharide-producing bacteria. In addition, we established a link between the altered microbiota and well-known liver biochemical indices of NAFLD progression. This novel study may pave the way for new prevention and therapeutic principles. To establish clinical feasible diagnostic and therapeutic methods, further studies utilizing the integrative approach of metabonomics and metagenomics with computational technology⁶ are required to elucidate the key microbes and its metabolic pathway associated with NAFLD.

Methods

Additional Information

How to cite this article: Wang, B. et al. Altered Fecal Microbiota Correlates with Liver Biochemistry in Nonobese Patients with Non-alcoholic Fatty Liver Disease. Sci. Rep. 6, 32002; 10.1038/srep32002 (2016).

Supplementary Material

Acknowledgments

Footnotes

References