Association Between Circulating Bile Acid Alterations and Nonalcoholic Steatohepatitis Independent of Obesity and Diabetes Mellitus

Youngae Jung; Bo Kyung Koo; Seo Young Jang; Dain Kim; Heeyeon Lee; Dong Hyeon Lee; Sae Kyung Joo; Yong Jin Jung; Jeong Hwan Park; Taekyeong Yoo; Murim Choi; Min Kyung Lee; Sang Won Kang; Mee Soo Chang; Won Kim; Geum-Sook Hwang

Disclosures

Liver International. 2021;41(12):2892-2902. 

In This Article

Results

Baseline Characteristics of the Discovery Cohort

A total of 174 subjects with biopsy-proven NAFLD and 67 no-NAFLD controls were included in the discovery cohort. Among the NAFLD subjects, 43.1% (75/174) had NASH. NAFL subjects showed higher BMI and visceral adipose tissue (VAT) area and a higher prevalence of metabolic syndrome than subjects with no-NAFLD (P = .001, P < .001, and P = .009, respectively); however, there were no differences between NAFL and NASH patients in regard to these parameters. However, the NASH group had significantly higher aspartate aminotransferase (AST), alanine aminotransferase (ALT), homeostasis model assessment of insulin resistance (HOMA-IR) and high-sensitivity C-reactive protein (hs-CRP) levels than the NAFL group (P < .001, P < .001 and P = .023 respectively). Significant fibrosis (≥F2) was found in 5.1% and 62.7% of NAFL and NASH patients respectively (P < .001; Table 1). The subjects were then divided into the nonobese (<25 kg/m2; n = 121) and obese (≥25 kg/m2; n = 120) groups according to their BMI (Table S2). Among the nonobese subjects, 46 had NAFL and 30 had NASH. There was no significant difference in BMI between nonobese NAFL patients and nonobese NASH patients (23.3 ± 1.2 kg/m2 vs 23.6 ± 1.2 kg/m2; P = .602).

Global Metabolic Profiling Based on UPLC/Q TOF-MS

Global metabolic profiling of sera collected from the 241 study subjects in the discovery cohort was performed using UPLC/Q TOF-MS. To identify significantly altered metabolites between NASH and NAFL patients, features with variable importance in projection (VIP) >1.0 in the PLS-DA model and a P-value <.05 in the Mann-Whitney test were selected among all features. A total of 18 metabolites were putatively identified using online databases, such as the Human Metabolome Database (www.hmdb.ca) and METLIN (https://metlin.scripps.edu) (Table S3). The most distinctively altered metabolites between nonobese subjects with NASH and NAFL were identified as BAs, as shown in the bar graphs (Figure 1). The levels of primary BAs, such as chenodeoxycholate (CDCA), glycocholate (GCA), glycochenodeoxycholate (GCDCA) and taurocholate (TCA), increased markedly in NASH vs NAFL nonobese subjects.

Figure 1.

Alterations in fold changes of polar metabolites according to obesity status and the histological severity of nonalcoholic fatty liver disease (NAFLD) in the discovery cohort. Fold changes were calculated using the median, and P-values were obtained from the ranked analysis of covariance (ANCOVA) test with adjustments for age and sex; Bonferroni correction was used; *P < .025, **P < .005, and ***P < .0005

Alterations in Serum BA Profiles in the Discovery Cohort

Based on global profiling analysis, BAs were identified as the most characteristic metabolite associated with the histological severity of NAFLD. However, since only a few BAs were detected through global profiling, we performed targeted profiling using UPLC/TQ-MS with a total of 225 sera samples (Table S4). In total, 12 primary/secondary BAs and/or unconjugated/conjugated BAs, such as cholate (CA), CDCA, GCA, GCDCA, TCA, taurochenodeoxycholate (TCDCA), deoxycholate (DCA), ursodeoxycholate (UDCA), glycodeoxycholate (GDCA), taurodeoxycholate (TDCA), taurolithocholate (TLCA) and tauroursodeoxycholate (TUDCA), were analysed (Table S5). Since the concentrations of TLCA were lower than the LOQ in more than 30% of all serum samples, TLCA was excluded from further analysis.

Overall, the total BA concentration increased in NASH patients compared to NAFL patients; however, it did not reach statistical significance (P = .030; Figure 2A). Subsequent stratified analysis according to obesity status showed that this trend was statistically significant only in nonobese subjects (P = .007) but not in obese subjects (Figure 2A). The ratio of unconjugated primary BA to total BA marginally increased in the nonobese NASH group compared to the nonobese NAFL group, but significantly decreased in the obese NASH group compared to the obese NAFL group (P =.009; Figure 2B and Table S6). The concentrations of unconjugated and glyco-/tauro-conjugated primary BAs increased significantly in NASH vs NAFL patients among nonobese subjects but not among obese subjects after adjustments for age, gender, HOMA-IR and PNPLA3 genotypes (Figure 2C and 2D).

Figure 2.

Alterations in primary Bile acid (BA) concentrations according to obesity status and the histological severity of NAFLD in the discovery cohort. Total BA concentration (A) and BA composition (B) in sera were compared between total, nonobese and obese subjects. Primary BA concentrations were then measured in sera collected from nonobese (C) and obese (D) subjects according to the histological severity of NAFLD. Plotted concentrations are expressed as the median and interquartile range. P-values were assessed by the ranked analysis of covariance (ANCOVA) test with adjustments for age, gender, HOMA-IR and PNPLA3 genotypes, and Bonferroni correction was applied; *P < .025 and **P < .005. BA compositions are presented as relative percentage ratios of individual BA concentrations to the total BA concentration

Association Between Serum BA Concentration and the Histological Severity of NAFLD According to Obesity Status

The correlation between serum BA concentration and liver histology, including steatosis, ballooning, lobular inflammation and fibrosis, was investigated in the discovery cohort. There was a positive correlation between conjugated primary BA and the histological severity of NAFLD in both obese and nonobese subjects except steatosis in nonobese subjects and lobular inflammation in obese subjects (Figure 3). By contrast, unconjugated primary BAs were strongly correlated with all liver histology parameters only in nonobese subjects but not in obese subjects (Figure 3).

Figure 3.

Correlations of BA concentrations with liver histology in nonobese and obese subjects in the discovery cohort. Statistical significance in Spearman's correlation heat maps is indicated with asterisks (*P < .05, **P < .01 and ***P < .001)

As primary BAs, especially tauro-conjugated BAs, were positively correlated with systemic insulin resistance and DM in nonobese subjects (Figure S1), subsequent stratified analysis according to diabetes status in nonobese subjects was performed. Significantly increased glyco-conjugated and tauro-conjugated primary BAs in NASH patients compared to NAFL patients were found among nonobese subjects irrespective of diabetes status (Figures S2A and S2B). Unconjugated primary BA also showed an increasing tendency in NASH compared to NAFL among nonobese subjects, which was statistically significant only in nonobese nondiabetic subjects (Figures S2A and S2B).

External Validation of Circulating BA Signatures According to Obesity Status and Liver Histology

To validate the characteristic alterations in serum BA concentration according to obesity status and liver histology, targeted BA analysis was also performed in the independent validation cohort including 107 biopsy-proven NAFLD subjects (70 NAFL and 37 NASH) and 29 no-NAFLD controls (Table S7). Although statistical significance was attenuated because of small sample size, the concentrations of unconjugated primary BAs showed alterations similar to those in the discovery cohort: significantly higher in NASH compared to NAFL even after adjustment for age, sex, HOMA-IR and PNPLA3 genotype (P = .045; Table S8). The significant increase in glyco-/tauro-conjugated primary BAs in nonobese NASH patients compared to NAFL patients seen in the discovery cohort was also replicated in the validation cohort (Figure 4). In addition, the correlations between serum BA concentrations and liver histology in the validation cohort also showed a pattern similar to those in the discovery cohort (Figure S3).

Figure 4.

Alterations in primary BA concentrations according to obesity status and the histological severity of NAFLD in the validation cohort. Primary BA concentrations were measured in serum samples of nonobese (A) and obese (B) subjects in the validation cohort and were assessed according to the histological severity of NAFLD. Plotted concentrations are expressed as the median and interquartile range. P-values were assessed by the ranked analysis of covariance (ANCOVA) test with adjustments for age, gender, HOMA-IR, and PNPLA3 genotypes, and Bonferroni correction was applied; *P < .025 and **P < .005

Metabolome-driven Omics Analysis Related to Hepatic and Intestinal BA Metabolism

BA metabolism is regulated by various genes related to hepatic BA synthesis, regulation, and transport. Liver tissue RNA-seq data were used to investigate whether hepatic regulatory genes of cholesterol and BA metabolism are differentially expressed between NASH patients and NAFL patients according to obesity status (Table S9). A total of 14 genes related to cholesterol and BA metabolism were analysed for differential expression using DESeq2 (Figure 5A). Expression of 3-Hydroxy-3-methylglutaryl-CoA reductase (HMGCR), Niemann-Pick C1 like 1 (NPC1L1), scavenger receptor class B type I (SCARB1) and low-density lipoprotein receptor (LDLR) genes were increased only in nonobese NASH patients. On the other hand, ATP-binding cassette subfamily G member 5/8 (ABCG5/8) genes expressions were not significantly changed in the nonobese nor in the obese NASH patients. Sterol 12α-hydroxylase (CYP8B1) gene expression was significantly upregulated in nonobese NASH patients compared to NAFL patients only in the nonobese group, and cholesterol 7α-hydroxylase (CYP7A1) gene expression was also marginally increased in nonobese NASH patients. In contrast, BATT, BACS and NTCP were significantly downregulated in NASH vs NAFL patients only in the obese group.

Figure 5.

Differential expressions of genes of liver tissues and serum fibroblast growth factor 19 levels according to BA metabolism. (A) Log2 (fold change) of differentially expressed cholesterol and BA metabolism genes from liver biopsy samples in nonalcoholic steatohepatitis (NASH) patients normalized to the expression in nonalcoholic fatty liver (NAFL) patients in the nonobese (NAFL, n = 29 and NASH, n = 21) and obese (NAFL, n = 103 and NASH, n = 92) groups. Differential expression was analysed using DESeq2, and * and *** represent false discovery rate (FDR)-adjusted P-values <.05 and <.001 respectively. (B) The levels of fibroblast growth factor 19 (FGF19) in serum samples were quantified using a human FGF19 ELISA kit in the nonobese (NAFL, n = 24 and NASH, n = 11) and obese (NAFL, n = 23 and NASH, n = 22) groups. Plotted concentrations are expressed as the mean and SD

There was no significant difference in serum FGF19 protein level between NAFL and NASH patients (Table S10; Figure 5B).

processing....