Hedgehog pathway activation parallels histologic severity of injury and fibrosis in human nonalcoholic fatty liver disease

Authors


  • Potential conflict of interest: Nothing to report.

  • The Nonalcoholic Steatohepatitis Clinical Research Network (NASH CRN) is supported by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) (grants U01DK061718, U01DK061728, U01DK061731, U01DK061732, U01DK061734, U01DK061737, U01DK061738, U01DK061730, and U01DK061713) and the National Institute of Child Health and Human Development. Several clinical centers use support from General Clinical Research Centers or Clinical and Translational Science Awards in the conduct of NASH CRN studies (grants UL1RR024989, M01RR000750, M01RR00188, UL1RR02413101, M01RR000827, UL1RR02501401, M01RR000065, M01RR020359, and UL1RR025741). M.A. is supported by a NIH/NIDDK K23 Career Development Award (K23-DK062116). The analyses described in this study were supported by an NIH/NIDDK grant (PI: to A.M.D.; R01-DK053792 and R01-DK077794) and discretionary funds from the Duke University Division of Gastroenterology.

Abstract

The Hedgehog (HH)-signaling pathway mediates several processes that are deregulated in patients with metabolic syndrome (e.g., fat mass regulation, vascular/endothelial remodeling, liver injury and repair, and carcinogenesis). The severity of nonalcoholic fatty liver disease (NAFLD) and metabolic syndrome generally correlate. Therefore, we hypothesized that the level of HH-pathway activation would increase in parallel with the severity of liver damage in NAFLD. To assess potential correlations between known histologic and clinical predictors of advanced liver disease and HH-pathway activation, immunohistochemistry was performed on liver biopsies from a large, well-characterized cohort of NAFLD patients (n = 90) enrolled in the Nonalcoholic Steatohepatitis Clinical Research Network (NASH CRN) Database 1 study. Increased HH activity (evidenced by accumulation of HH-ligand–producing cells and HH-responsive target cells) strongly correlated with portal inflammation, ballooning, and fibrosis stage (each P < 0.0001), supporting a relationship between HH-pathway activation and liver damage. Pathway activity also correlated significantly with markers of liver repair, including numbers of hepatic progenitors and myofibroblastic cells (both P < 0.03). In addition, various clinical parameters that have been linked to histologically advanced NAFLD, including increased patient age (P < 0.005), body mass index (P < 0.002), waist circumference (P < 0.0007), homeostatic model assessment of insulin resistance (P < 0.0001), and hypertension (P < 0.02), correlated with hepatic HH activity. Conclusion: In NAFLD patients, the level of hepatic HH-pathway activity is highly correlated with the severity of liver damage and with metabolic syndrome parameters that are known to be predictive of advanced liver disease. Hence, deregulation of the HH-signaling network may contribute to the pathogenesis and sequelae of liver damage that develops with metabolic syndrome. (HEPATOLOGY 2012;55:1711–1721)

Nonalcoholic fatty liver disease (NAFLD) is strongly associated with obesity. Because of the current obesity epidemic, NAFLD is currently one of the most prevalent liver diseases in the world and a major cause of cirrhosis and liver-related mortality.1 Fortunately, only some of the many individuals with NAFLD will ever develop progressive liver injury that results in steatohepatitis (SH), cirrhosis, or primary liver cancer. Therefore, efficient, accurate identification of patients who are most likely to develop progressive liver damage is crucial so that such individuals can be targeted for more aggressive surveillance and therapeutic interventions to optimize the outcomes and minimize the costs of the NAFLD epidemic. Success has been stymied by our relatively poor understanding of the processes that regulate the outcomes of fatty liver injury.

Research involving experimental animals is often used to delineate key mechanisms and pilot therapies for human diseases with long and/or seemingly idiosyncratic natural histories. In NAFLD, however, this approach has been hampered by the lack of small animal models of SH and progressive liver fibrosis that also mimic the typical metabolic perturbations of human NAFLD.2 Nonetheless, recent studies in mice demonstrated that the development of SH and fibrosis correlated strongly with the intensity and duration of Hedgehog (HH)-pathway activation that developed during fatty liver injury.3 Other work in cultured cells demonstrated that HH ligands stimulate quiescent hepatic stellate cells to become myofibroblastic, promote proliferation of liver myofibroblasts and progenitors, inhibit apoptosis of these cell types, and up-regulate the production of chemokines for various types of immune cells.4, 5 Therefore, it is conceivable that interindividual differences in HH-pathway activity contribute to the variable outcomes of fatty liver injury in NAFLD patients. This concept was supported by immunohistochemical (IHC) staining of liver-biopsy samples from a small number of NAFLD patients.3 The resultant data showed that the hepatic content of HH-ligand–producing cells, as well as the burden of HH-responsive liver cells, increased in parallel with fibrosis stage. However, the fact that the analysis was performed in only a small number of patients from a single institution raised valid concerns among clinicians who questioned whether the selected cohort was representative of the general NAFLD population. Further investigation of this issue is warranted, and therefore the aim of the present study was to evaluate the relationship between the level of HH-pathway activity and severity of liver inflammation and fibrosis in a large, representative cohort of NAFLD patients.

Abbreviations

BMI, body mass index; CI, confidence interval; COR, cumulative odds ratio; DM, diabetes mellitus; ECs, endothelial cells; ER, endoplasmic reticulum; G, histologic grade; GLI2, glioblastoma 2 transcription factor; H&E, hematoxylin and eosin; HH, hedgehog; HOMA-IR, homeostatic model assessment of insulin resistance; HPF, high-power field; HTN, hypertension; IHC, immunohistochemistry; K7, keratin 7; mRNA, messenger RNA; NAFL, nonalcoholic fatty liver; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; NASH CRN, NASH Clinical Research Network; NKT, natural killer T cells; S, fibrosis stage; SD, standard deviation; SHH, sonic Hedgehog; SH, steatohepatitis; α-SMA, alpha-smooth muscle actin; VIM, vimentin.

Patients and Methods

Study Design and Population.

We performed a cross-sectional analysis using data and liver sections from a representative subset (n = 90) of all subjects in the Nonalcoholic Steatohepatitis Clinical Research Network (NASH CRN) Database 1 Study (n = 1,044).6 Liver histologic data were available for 864 of these 1,044 subjects, but only 232 of those individuals fulfilled the following criteria: (1) age ≥18 years, (2) no significant alcohol consumption (≤14 drinks/week in men or ≤7 drinks/week in women, on average, within the past 2 years) or other coexisting causes of chronic liver disease, (3) liver biopsy of ≥15 mm in length performed within 6 months of enrollment into the CRN Database, (4) unstained tissue sections available for IHC staining, and (5) the corresponding hematoxylin and eosin (H&E)- and Masson-trichrome–stained liver-biopsy slides had already been scored by the NASH CRN Pathology Committee. Our study cohort (n = 90) was comprised of the first 30 consecutive cases from each of the following three histologically defined groups: (1) nonalcoholic fatty liver (NAFL) (i.e., simple steatosis/not definite NASH) with no-to-early fibrosis (stage 0, 1, and 2) (N = 87); (2) definite NASH with early fibrosis (N = 73); and (3) NAFL or NASH with advanced fibrosis (stage 3 or 4) (N=72). Case selection was performed by the NASH CRN Data Coordinating Center. The NASH CRN studies were approved by the institutional review boards at each participating center.

Histologic Evaluation of NAFLD.

Liver biopsies from all of the cases in the present study had been stained with H&E and Masson's trichrome, reviewed, and scored by the NASH CRN Pathology Committee according to the published NASH CRN scoring system.7 Briefly, portal inflammation was graded as 0 (none to minimal), 1 (mild), or 2 (greater than mild). Hepatocyte ballooning was graded as 0 (none), 1 (few), or 2 (many). Fibrosis was staged as 0 (no pathologic fibrosis), 1 (centrilobular or periportal pericellular fibrosis), 2 (centrilobular and periportal pericellular fibrosis), 3 (bridging fibrosis), or 4 (cirrhosis). In addition to data for portal inflammation, hepatocyte ballooning, and fibrosis, we analyzed the complete histologic data set from this cohort to determine whether any other standard histopathological parameter(s) correlated with evidence of HH-pathway activity, as revealed by immunostaining of banked liver sections from the same patients. Because no relationships were demonstrated between HH immunostaining and levels of hepatic steatosis or lobular inflammation, the Results section below details only the findings that were noted with regard to portal inflammation, hepatocyte ballooning, and liver fibrosis.

IHC Evaluations.

Three unstained slides from each patient's formalin-fixed, paraffin-embedded liver biopsy were available for IHC. Representative sections from each histologic subgroup were used to optimize staining conditions; remaining sections were processed to assess the production of sonic Hedgehog (SHH) ligand or accumulation of HH-responsive cells (demonstrated by nuclear staining for the HH-regulated transcription factor, glioblastoma 2; GLI2). Because of the restricted number of sections, technical constraints (e.g., secondary antibody cross-reactivity), and our desire to characterize different types of cells that were HH responsive, representative subgroups of the larger cohorts were stained solely for GLI2, or costained for GLI2 and keratin 7 (K7; a liver-progenitor marker), or GLI2 and vimentin (VIM; a mesenchymal cell marker). To further clarify correlations between HH-pathway activation and fibrosis stage, remaining sections were stained for alpha-smooth muscle actin (α-SMA), a myofibroblast marker that correlates with fibrosis severity.

Details of the IHC methods and antibodies have been previously published.3, 5, 8 Positive staining for SHH, VIM, and α-SMA were semiquantified using 10x objective low-power fields (100x magnification) as a percentage of the total surface area and graded into five categories: grade 1 (less than 20%), grade 2 (20%-39%), grade 3 (40%-59%), grade 4 (60%-79%), and grade 5 (≥80%). For the GLI2/K7 double stain, total number of K7+ cells, and K7+/GLI2+ double-positive cells, were counted in five 40x objective high-power fields (HPFs; 400x magnification) to determine the average number of positively stained cells.

Clinical Information.

All clinical information was collected within 6 months of the liver biopsy. Age, body mass index (BMI; kg/m2), waist circumference (cm), homeostasis model assessment–insulin resistance (HOMA-IR), presence or absence of diabetes mellitus (DM), and hypertension (HTN) were evaluated. Clinical characteristics are reported as the mean ± standard deviation (SD) for continuous variables or as a proportion with a condition for categorical variables.

Statistical Analyses.

To assess the associations between IHC scores and H&E and trichrome scores, we performed Wilcoxon's rank-sum tests or Kruskal-Wallis' tests. For post-hoc comparison, Wilcoxon's rank-sum tests were used. To assess the associations between the level of HH-pathway activity and known clinical risk factors for advanced fibrosis, we performed ordinal logistic regression or linear regression analyses, with and without adjusting for other factors. JMP statistical software (version 7.0; SAS institute, Inc., Cary, NC) was used for analysis, and differences were considered to be statistically significant when the P values were less than 0.05, except for the post-hoc comparison, in which α-levels were adjusted by 0.05/numbers of pairs in a comparison. The plots were created using random scatter in R version 2.13.1 (Free Software Foundation, Inc., Boston, MA).

Results

Clinical and Histologic Characteristics of the Study Population.

Clinical and histologic characteristics of the study population are summarized in Table 1. Mean age and BMI of the study population were 48 ± 13 years and 35 ± 7 kg/m2. Women comprised 59% of the cohort, 30% had DM, 46% had HTN, and 56% had hyperlipidemia. Forty-three percent of our cohort had ballooning grade 2, and 33% had advanced fibrosis (S3-S4).

Table 1. Clinical Characteristics of the Study Population
Clinical Characteristics (N = 90)Summary Statistics
Age (years)48 ± 13
Female gender, %59
Hispanic, %14
Race, % (White:Asian/Pacific islanders:others)81:7:12
BMI (kg/m2)35 ± 7
Type II DM, %30
HTN, %46
Hyperlipidemia, %56
Steatosis grade, % (G0:G1:G2:G3)4:37:33:26
Lobular inflammation grade, % (G0:G1:G2:G3)0:56:38:6
Ballooning grade, % (G0:G1:G2)26:31:43
Portal inflammation grade, % (G0:G1:G2)11:62:27
Fibrosis stage, % (S0:S1:S2:S3:S4)26:25:16:22:11

SHH Expression Correlates With Severity of Ballooning, Portal Inflammation, and Fibrosis Stage in NAFLD.

In animal models of NAFLD, HH-pathway activation has been linked to fibrogenesis.3 Therefore, SHH IHC was performed on liver sections from 84 patients with different stages of fibrosis (S0, n = 21; S1, n = 21; S2, n = 14; S3, n = 19; S4, n = 9). Positive staining was semiquantified (i.e., graded), as described above. SHH positivity was identified in portal tract cells (e.g., bile duct cells, bile ductular cells and endothelial cells; 83.6% of cases), ballooned hepatocytes (58.3% of cases), and periportal hepatocytes (7.0% of cases). In this relatively large cohort of NAFLD patients, the hepatic content of SHH-expressing cells increased with fibrosis stage (Fig. 1A-D). Moreover, the relationship between the level of SHH expression and fibrosis severity was highly significant (Fig. 1E, P < 0.0001; Supporting Figs. 1-6).

Figure 1.

SHH expression correlates with fibrosis stage in NAFLD. Photomicrographs of SHH IHC in patients with S0 (A), S2 (B), S3 (C), and S4 (D) fibrosis show increased numbers of positive cells with increased fibrosis stage (400x magnification). IHC scoring results were semiquantified into five ranks, and the results were plotted according to the fibrosis stage, as scored by the NASH CRN Pathology Committee, using trichrome-stained liver sections (E). Open circles represent individual subjects. The middle line represents the mean value, whereas whiskers with horizontal lines (upper and lower) represent SD, standard deviation. P < 0.0001 (Kruskal-Wallis' test); P < 0.005 (α-level adjusted for 10 post-hoc comparison pairs); *versus stage 1 and 2; **versus stage 0, 1, and 2.

Given that studies in animal models demonstrated that HH ligands stimulate chemokine production by ductular cells and result in the hepatic recruitment of certain types of immune cells,5, 9 we examined the relationship between SHH expression and portal inflammation in these subjects. Portal inflammation was strongly associated with SHH expression (P < 0.0001): Mean rank and SD of SHH expression in patients with grade 0, 1, and 2 portal inflammation were 1.7 ± 0.9, 2.8 ± 1.3, and 4.0 ± 1.0, respectively. SHH expression in subjects with grade 1 portal inflammation was higher than in those with grade 0 portal inflammation (P < 0.015) and lower than in subjects with grade 2 portal inflammation (P < 0.0001) (adjusted α-level = 0.017).

Inflammatory mediators have been implicated in the pathogenesis of NASH and are known to provoke various types of cellular stress, including endoplasmic reticulum (ER) stress. Ballooned hepatocytes exhibit features of ER stress, and treating mouse hepatocytes with tunicamycin to induce ER stress stimulated them to express SHH messenger RNA (mRNA) and protein.10 Therefore, we next evaluated the relationship between hepatocyte ballooning and SHH expression. Ballooned hepatocytes stained strongly for SHH (Fig. 2A-C), and the level of SHH expression strongly correlated with severity of hepatocyte ballooning (Fig. 2D; P < 0.0001).

Figure 2.

SHH expression correlates with hepatocyte ballooning. Photomicrographs of SHH IHC in patients with grade 0 (A), 1 (B), and 2 (C) ballooning show increased numbers of positive cells with increased ballooning grade (400x magnification). SHH expression was plotted relative to the grade of hepatocyte ballooning, as scored by the NASH CRN Pathology Committee, using H&E-stained liver sections (D). Open circles represent individual subjects. The middle line represents the mean value, whereas whiskers with horizontal lines (upper and lower) represent SD. P < 0.0001 (Kruskal-Wallis' test); *P < 0.017 versus grade 1 and 2 (α-level adjusted for three post-hoc comparison pairs).

Numbers of GLI2 (+) Cells Increase With Fibrosis Stage, Portal Inflammation, and Ballooning in NAFLD.

SHH interacts with receptors on the surface of HH-responsive target cells to trigger HH signaling that results in the nuclear localization of the HH-regulated transcription factor, GLI2.11 We did staining for GLI2 on 39 liver sections from a representative subset of our NAFLD patients (S0, n = 5; S1, n = 11; S2, n = 7; S3, n = 9; S4, n = 7) to assess the relationship between SHH production and accumulation of cells with nuclear GLI2 staining. As expected, there was a positive correlation between SHH expression and GLI2 expression in these subjects (P < 0.0001). GLI2 (+) cells were identified in portal tracts (e.g., ductular cells, stromal cells, inflammatory cells, and ECs), small hepatocytic periportal cells, and in lobular fibroinflammatory foci adjacent to ballooned hepatocytes. Hepatic accumulation of HH-responsive cells (i.e., GLI2-positive cells) increased with fibrosis stage (Fig. 3A-D) and was particularly robust in patients with advanced (S3 and S4) fibrosis (Fig. 3E; P < 0.0001).

Figure 3.

Numbers of GLI2+ cells correlate with fibrosis stage and grade of portal inflammation. Photomicrographs of GLI2 IHC in patients with S0 (A), S2 (B), S3 (C), and S4 (D) fibrosis show increased numbers of positive cells (nuclear staining) with increased fibrosis stage (400x magnification). GLI2 staining scores were plotted relative to fibrosis stage (E) as well as portal inflammation grade (F). Open circles represent individual subjects. The middle line represents the mean value, whereas whiskers with horizontal lines (upper and lower) represent SD. Significant stage/grade-related differences are shown: *P < 0.005 (versus stage 0, 1, and 2) (E) and *P < 0.017 versus grade 2 (α-level adjusted for 10 and 3 post-hoc comparison pairs for fibrosis stage and portal inflammation grade, respectively).

Portal inflammation is a potential consequence of HH-pathway activation5 and has also been linked to fibrogenesis in NAFLD.9 The same 39 cases exhibited a range of portal inflammation (G0, n = 2; G1, n = 24; G2 n = 13). Therefore, we assessed the relationship between the accumulation of GLI2-expressing cells and portal inflammation, and found that the severity of portal inflammation was significantly positively associated with numbers of GLI2 (+) liver cells in these subjects (Fig. 3F; P < 0.02).

Ballooned hepatocytes produce HH ligands.10 Hence, we also examined the relationship between hepatic accumulation of GLI2-expressing cells and severity of hepatocyte ballooning. Ballooning scores differed widely in these 39 individuals (G0, n = 7; G1, n = 13; G2, n = 19). Ballooning and GLI2 expression were found to be significantly associated (P < 0.005), although there was considerable overlap in hepatic accumulation of GLI2-expressing cells among subjects with different grades of ballooning.

Hepatic Accumulation of HH-Responsive Liver Progenitors and Myofibroblastic Cells Parallels Fibrosis Stage in NAFLD.

Hepatic progenitors accumulate in parallel with the severity of liver myofibroblast accumulation and liver fibrosis in NAFLD.12 This suggests that injury-related factors might arrest epithelial differentiation of liver progenitors while promoting the outgrowth of myofibroblastic populations. HH ligands generally maintain HH-responsive, epithelial-type progenitors in a relatively undifferentiated state, but promote the growth of HH-responsive myofibroblastic cells.3, 13 Therefore, we did double immunostaining for GLI2 and K7 (which marks a subpopulation of liver epithelial progenitors) in another representative subgroup of liver sections (S0, n = 1; S1, n = 5; S2, n = 2; S3, n = 5; S4, n = 3). The total number of K7+ cells increased in advanced fibrosis (38.9 ± 19.5 for S0-S2 versus 74.4 ± 37.4 for S3-S4; P < 0.05). Regardless of the stage of liver fibrosis, some K7+ cells coexpressed GLI2 (range, 17.8%-72.9%; mean, 46.6 ± 14.9%). Total numbers of K7/GLI2 double-positive cells increased with fibrosis stage (Fig. 4E: P < 0.03). However, the proportion of K7 cells expressing GLI2 remained relatively constant at any given level of fibrosis (44.3% ± 18.3% for S0-S2 versus 49.0% ± 11.4% for S3-S4).

Figure 4.

Hepatic accumulation of liver progenitor cells increases with fibrosis stage, portal inflammation, and hepatocyte ballooning in NAFLD. Photomicrographs of liver sections double stained for K7 (blue) and GLI2 (brown) in patients with S0 (A), S2 (B), S3 (C), and S4 (D) fibrosis show increased numbers of positive cells with increased fibrosis stage (400x magnification). K7-positive cells and GLI2-positive cells were counted in double-stained liver-biopsy sections. Average cell counts (per 400x HPF) were plotted in relationship to fibrosis (E), portal inflammation (F), and hepatocyte ballooning (G). Open circles represent individual subjects. The middle line represents the mean value, whereas whiskers with horizontal lines (upper and lower) represent SD. P < 0.05, 0.03, and 0.004 for fibrosis, portal inflammation, and ballooning, respectively (Kruskal-Wallis' test).

The canals of Hering, the most proximal part of the intrahepatic biliary tree, are thought to provide a niche for liver epithelial progenitors, raising the possibility that the hepatic content of HH-responsive progenitors might be influenced by the level of portal inflammation.5 To address this issue, we correlated numbers of K7/GLI2 double-positive cells with the level of portal inflammation. Portal inflammatory activity correlated significantly with the hepatic content of HH-responsive epithelial progenitors (Fig. 4F; P < 0.03). Because ballooned hepatocytes are a rich source of HH ligands and correlate with fibrosis stage in NAFLD, and progenitor accumulation correlates with fibrosis stage in NAFLD, we also examined the relationship between hepatocyte ballooning and numbers of HH-responsive progenitors. Numbers of K7/GLI2 double-positive cells (Fig. 4G), as well as total K7-positive cells, were strongly positively associated with the severity of hepatocyte ballooning (P < 0.004 and P < 0.03, respectively).

Finally, because expansion of myofibroblastic populations is a hallmark of liver fibrogenesis and tends to parallel the accumulation of immature liver epithelial cells in NAFLD,3 HH-mediated epithelial-to-mesenchymal transition,14 and fibrogenic repair,3 we evaluated the relationship between fibrosis stage and hepatic content of stromal cells that expressed myofibroblast markers and GLI2. The remaining 27 unstained sections were stained for either VIM (n = 12) or α-SMA (n = 15). As was typical of the entire cohort, fibrosis severity ranged from S0 to S4 in this representative subgroup of cases. Sections with advanced fibrosis (S3-S4) had greater numbers of VIM-positive cells than those with less-advanced fibrosis (S0-S2) (Fig. 5A; P < 0.004). Similar results were noted when α-SMA-stained sections were examined (Fig. 5B; P < 0.002). Review of slides that were costained for VIM and GLI2 revealed that the stromal cell populations harbored HH-responsive (i.e., GLI2-positive) cells (Fig. 5C) and demonstrated that numbers of GLI2-positive cells and VIM expression were strongly correlated (P < 0.003, ordinal logistic regression, likelihood ratio test).

Figure 5.

Numbers of myofibroblastic cells increase with fibrosis stage. Liver biopsies were stained with VIM or α-SMA to identify myofibroblastic mesenchymal cells. Liver content of VIM-positive cells (A) or α-SMA-stained cells (B) were quantified as described in Patients and Methods and correlated with fibrosis stage based on NASH CRN Pathology Committee review of trichrome-stained sections. Open circles represent individual subjects. The middle line represents the mean value, whereas whiskers with horizontal lines (upper and lower) represent SD. α-SMA grade and VIM grade were higher in livers with advanced fibrosis (P < 0.004 and P < 0.002, respectively). (C) The photomicrograph illustrates accumulation of HH-responsive mesenchymal cells in fibrotic (i.e., advanced-stage) NAFLD. Arrows demonstrate stromal and vascular cells in fibrotic septa double stained for VIM (blue) and GLI2 (brown).

Clinical Correlates of Liver Fibrosis Significantly Correlate With Hepatic SHH and/or GLI2 Staining.

Tissue samples in the NASH CRN repository are linked to relevant clinical information, providing a unique opportunity to assess relationships between clinical parameters and liver histology. Therefore, we performed uni- and multivariate analysis to identify clinical correlates of liver fibrosis in our study cohort, then assessed the relationship between these parameters and hepatic HH-pathway activity. Univariate analysis demonstrated significant correlations of fibrosis stage with age, BMI, waist circumference, log HOMA-IR, and HTN (Table 2). All of these variables correlated strongly with hepatic expression levels of SHH (Table 2), which (as noted earlier) significantly correlated with hepatic accumulation of GLI2-positive cells (P < 0.0001). Log HOMA-IR correlated with SHH expression, even after adjusting for fibrosis stage (cumulative odds ratio [COR] [95% confidence interval (CI)] = 3.1 [1.5, 6.3]; P < 0.003). Similarly, age, BMI, waist circumference, log HOMA-IR, and HTN were significantly correlated with GLI2 expression (Table 3). After adjusting for SHH expression, only HTN was significantly correlated with GLI2 expression, suggesting that the presence of HTN was independently associated with higher GLI2 expression at a given SHH ligand level (Table 3).

Table 2. Associations of Clinical Variables With Fibrosis Stage and SHH Expression
Clinical VariablesUnadjustedUnadjustedAdjusted
Fibrosis StageSHH ExpressionSHH Expression
COR and 95% CICOR and 95% CICOR and 95% CI
  1. Unadjusted estimates were calculated using ordinal logistic regression, whereas adjusted estimates for SHH were calculated in models including fibrosis stage as a covariate. Estimates for age, BMI, and waist circumference were calculated for 5 unit changes.

Age, 5 units change1.3 [1.1, 1.5] (P < 0.0008)1.3 [1.1, 1.5] (P < 0.005)1.0 [0.8, 1.2] (P = 0.75)
BMI, 5 units change1.6 [1.2, 2.2] (P < 0.001)1.6 [1.2, 2.1] (P < 0.002)1.2 [0.8, 1.6] (P < 0.002)
Waist circumference, 5 units change1.3 [1.1, 1.5] (P < 0.0006)1.3 [1.1, 1.5] (P < 0.0007)1.1 [0.9, 1.3] (P = 0.25)
Log (HOMA-IR)2.5 [1.4, 4.5] (P < 0.002)3.4 [1.8, 6.4] (P < 0.0001)3.1 [1.5, 6.3] (P < 0.003)
DM2.1 [0.9, 4.9] (P = 0.08)2.8 [1.2, 6.5] (P < 0.02)2.2 [0.8, 5.9] (P = 0.12)
HTN3.1 [1.4, 6.9] (P = 0.005)2.6 [1.2, 5.7] (P < 0.02)1.0 [0.4, 2.4] (P = 0.96)
Table 3. Associations of Clinical Variables With GLI2 Expression
Clinical VariablesUnadjustedAdjusted
β ± SEβ ± SE
  1. Unadjusted estimates were calculated using linear regression, whereas adjusted estimates were calculated in models including SHH expression level as a covariate. Estimates for age, BMI, and waist circumference were calculated for 5 unit changes.

  2. Abbreviations: β, beta coefficient computed in linear regression models using GLI2 expression (cell numbers per HPF) as an outcome variable; SE, standard error.

Age, 5 units change20.5 ± 10.5 (P = 0.006)15.0 ± 10.0 (P = 0.145)
BMI, 5 units change55.0 ± 20.5 (P = 0.011)25.0 ± 17.5 (P = 0.169)
Waist circumference, 5 units change4.2 ± 1.9 (P = 0.04)0.7 ± 1.7 (P = 0.656)
Log (HOMA-IR)97.1 ± 41.3 (P = 0.024)−21.9 ± 41.1 (P = 0.597)
DM90.9 ± 61.0 (P = 0.145)8.8 ± 42.3 (P = 0.854)
HTN181.3 ± 48.5 (P = 0.0006)123.8 ± 39.1 (P = 0.0034)

Discussion

This cross-sectional IHC analysis of liver biopsies from a large number of well-characterized patients with NAFLD provides compelling evidence that the severity of liver damage (i.e., hepatocyte ballooning, portal inflammation, and liver fibrosis) parallels the level of HH-pathway activity in this disease. Genetic and pharmacologic approaches that modulate HH signaling in experimental animals and liver cell culture models have proven that the HH pathway regulates several key aspects of liver repair, including the outgrowth of liver progenitor populations,12 hepatic recruitment of inflammatory cells,5 generation and accumulation of liver myofibroblasts,4, 13 and fibrogenesis.3 In animal models of liver injury, transient HH-pathway activation is required for liver regeneration,15 but sustained or excessive HH signaling promotes cirrhosis.3 Thus, although direct proof that deregulated HH signaling mediates NAFLD progression in humans is lacking, the results of the present study demonstrate that this is likely to be true and thus identify novel diagnostic and therapeutic targets to improve NAFLD outcomes.

Although cross-sectional, our data strongly suggest that interindividual differences in the ability to control HH-pathway activity may contribute to the variable outcomes of fatty liver injury. Our univariate analysis supports this concept by identifying strong correlations between hepatic levels of SHH ligand production or nuclear accumulation of the HH-regulated transcription factor, GLI2, and each of the three clinical variables that have been most consistently linked with advanced liver fibrosis in NAFLD (i.e., older age, overweight/obesity, and the diagnosis of insulin resistance/type 2 diabetes). Hepatic production of SHH ligands and/or HH-signaling activity were also demonstrated to correlate with other clinical factors that are associated with liver fibrosis, including waist circumference and HTN, suggesting a relationship between overly exuberant HH-pathway activation in the liver and extrahepatic adverse outcomes of metabolic syndrome. The aggregate data, therefore, suggest that deregulated HH pathway activity might promote, and/or result from, metabolic syndrome and mediate damage to the liver and other tissues that occurs in this condition. Extension of this logic justifies the development of noninvasive tests that quantify HH-pathway activity to identify individuals who are experiencing tissue damage related to metabolic syndrome before irreparable end-organ damage ensues. Such patients could then be enrolled into prospective clinical trials designed to determine whether decreasing HH-pathway activity restores normal tissue repair and prevents (or reverts) progressive tissue damage.

To our knowledge, our study is the first to demonstrate an unequivocal relationship between HH-pathway activity at the tissue level and the severity of damage in that tissue in people with metabolic syndrome. Though novel, evidence that deregulated HH signaling occurs in metabolic syndrome, and is likely to be directly responsible for related tissue pathology, is buttressed by data that were previously reported by our group and others. First, obesity is strongly associated with metabolic syndrome, and it has been proven that the HH pathway is a major, highly conserved regulator of fat mass.16 Pathway activation arrests adipogenesis and promotes the accumulation of adipocyte precursors.17 Mature fat cells themselves are also capable of producing and releasing HH ligands, and ligand generation from adipose depots is increased in obesity.16 Moreover, interaction of a key adipocyte-derived anorexogenic hormone, leptin, with its receptors on target cells, induces HH ligand production and activates HH signaling, which, in turn, directly mediates the effects of leptin in those cells.18 Second, metabolic syndrome is a chronic inflammatory state, and the HH pathway is known to have immunomodulatory functions.19 HH is required for normal thymic development and regulates the viability, tissue localization, and cytokine production of lymphocytes in adults.20 In rodents and humans with NASH, for example, hepatic accumulation of profibrogenic natural killer T (NKT) cells correlates with the level of HH-pathway activity and tissue expression of the NKT cell chemoattractant, CXCL16, an HH-inducible gene. NKT cells, in turn, likely play a key role in fibrosis progression, because in rodents, NASH-related cirrhosis is prevented by NKT cell depletion. Conversely, livers removed from patients undergoing liver transplantation for NAFLD-related cirrhosis are dramatically enriched with NKT cells.9 Third, EC dysfunction and vascular remodeling are characteristics of metabolic syndrome, and the HH pathway is an acknowledged regulator of vasculogenesis/angiogenesis.21 Membranous microparticles released from cells that produce HH ligands (e.g., apoptotic T cells and liver cells) contain biologically active HH ligands that interact with HH receptors on vascular ECs and initiate HH signaling.22 The latter induces EC activation and alters the production of vasoactive substances, such as nitric oxide.23 Such findings have prompted speculation that HH signaling is fundamentally involved in the pathogenesis of EC dysfunction.24 Fourth, obesity and metabolic syndrome are known to increase the risk of cancer in various tissues.25 HH ligands promote the viability and growth of many types of stem and progenitor cells,26 and deregulated HH signaling is well documented in several obesity-associated cancers, including hepatocellular carcinoma,27 which has become one of the main causes of cancer-related death in obese American men.28 Thus, the cumulative evidence strongly supports the concept that deregulated HH signaling is broadly relevant to the pathophysiology of metabolic syndrome. The liver is both a target of, and a contributor to, metabolic syndrome–related pathophysiology, and the present study suggests that both aspects of the relationship are likely to involve the HH pathway. Additional research is needed to examine this issue and to determine whether plasma levels of SHH identify NAFLD subjects with liver injury who have increased fibrogenesis and/or whether treatments that “normalize” HH-pathway activation would improve recovery from NAFLD.

Acknowledgements

The authors thank Miao Yu, M.S., Center for Human Genetics, Duke University Medical Center, for creating the graphs presented in this article and acknowledge that the work provided by Miao Yu was supported by Grant Number UL1RR024128 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH), and NIH Roadmap for Medical Research. The contents are solely the responsibility of the authors and do not necessarily represent the official view of the NCRR or NIH.

Appendix

Members of the Nonalcoholic Steatohepatitis Clinical Research Network are: from clinical centers: Stephanie H. Abrams, M.D., M.S., Leanel Angeli Fairly, R.N., Baylor College of Medicine, Houston, TX; from Case Western Reserve University Clinical Centers: Arthur J. McCullough, M.D., Patricia Brandt, Diane Bringman, R.N. (2004-2008), Srinivasan Dasarathy, M.D., Jaividhya Dasarathy, M.D., Carol Hawkins, R.N., Yao-Chang Liu, M.D. (2004-2009), Nicholette Rogers, Ph.D., PA.-C. (2004-2008), Margaret Stager, M.D. (2004-2009), MetroHealth Medical Center, Cleveland, OH; Arthur J. McCullough, M.D., Srinivasan Dasarathy, M.D., Mangesh Pagadala, M.D., Ruth Sargent, L.P.N., Lisa Yerian, M.D., Claudia Zein, M.D., Cleveland Clinic Foundation, Cleveland, OH; Raphael Merriman, M.D., Anthony Nguyen, California Pacific Medical Center, San Francisco, CA; Parvathi Mohan, M.D., Kavita Nair, Children's National Medical Center, Washington, DC; Stephanie DeVore; Rohit Kohli, M.D., Kathleen Lake, Stavra Xanthakos, M.D., Cincinnati Children's Hospital Medical Center, Cincinnati, OH; Manal F. Abdelmalek, M.D., M.P.H., Stephanie Buie, Anna Mae Diehl, M.D., Marcia Gottfried, M.D. (2004-2008), Cynthia Guy, M.D., Meryt Hanna, Paul Killenberg, M.D. (2004-2008), Samantha Kwan, M.S. (2006-2009), Yi-Ping Pan, Dawn Piercy, F.N.P., Melissa Smith, Duke University Medical Center, Durham, NC; Elizabeth Byam, R.N., Naga Chalasani, M.D., Oscar W. Cummings, M.D., Ann Klipsch, R.N., Jean P. Molleston, M.D., Linda Ragozzino, R.N., Girish Subbarao, M.D., Raj Vuppalanchi, M.D., Indiana University School of Medicine, Indianapolis, IN; Kimberly Pfeifer, R.N., Ann Scheimann, M.D., Michael Torbenson, M.D., Johns Hopkins Hospital, Baltimore, MD; Nanda Kerkar, M.D., Sreevidya Narayanappa, Frederick Suchy, M.D., Mount Sinai Kravis Children's Hospital, New York, NY; Mark H. Fishbein, M.D., Katie Jacques, Ann Quinn, R.D., Cindy Riazi, R.N., Peter F. Whitington, M.D., Northwestern University Feinberg School of Medicine/Children's Memorial Hospital, Chicago, IL; Melissa Coffey, Sarah Galdzicka, Karen Murray, M.D., Melissa Young, Seattle Children's Hospital and Research Institute, WA, Seattle, WA; Sarah Barlow, M.D. (2002-2007), Jose Derdoy, M.D., Joyce Hoffmann, Debra King, R.N., Andrea Morris, Joan Siegner, R.N., Susan Stewart, R.N., Brent A. Neuschwander-Tetri, M.D., Judy Thompson, R.N., Saint Louis University, St. Louis, MO; Cynthia Behling, M.D., Ph.D., Janis Durelle, Tarek Hassanein, M.D. (2004-2009), Joel E. Lavine, M.D., Ph.D., Rohit Loomba, M.D., Anya Morgan, Steven Rose, M.D. (2007-2009), Heather Patton, M.D., Jeffrey B. Schwimmer, M.D., Claude Sirlin, M.D., Tanya Stein, M.D. (2005-2009), University of California San Diego, San Diego, CA; Bradley Aouizerat, Ph.D., Kiran Bambha, M.D. (2006-2010), Nathan M. Bass, M.D., Ph.D., Linda D. Ferrell, M.D., Danuta Filipowski, M.D., Bo Gu (2009-2010), Raphael Merriman, M.D. (2002-2007), Mark Pabst, Monique Rosenthal (2005-2010), Philip Rosenthal, M.D., Tessa Steel (2006-2008), University of California San Francisco, San Francisco, CA; Matthew Yeh, M.D., Ph.D., University of Washington Medical Center, Seattle, WA; Sherry Boyett, R.N., B.S.N., Melissa J. Contos, M.D., Michael Fuchs, M.D., Amy Jones, Velimir A.C. Luketic, M.D., Puneet Puri, M.D., Bimalijit Sandhu, M.D. (2007-2009), Arun J. Sanyal, M.D., Carol Sargeant, R.N., B.S.N., M.P.H., Kimberly Noble, Melanie White, R.N., B.S.N. (2006-2009), Virginia Commonwealth University, Richmond, VA; Kris V. Kowdley, M.D. (original grant with University of Washington), Jody Mooney, M.S., James Nelson, Ph.D., Sarah Ackermann, Cheryl Saunders, M.P.H., Vy Trinh, Chia Wang, M.D., Virginia Mason Medical Center1, Seattle, WA; Elizabeth M. Brunt, M.D., Washington University, St. Louis, MO; Resource centers: David E. Kleiner, M.D., Ph.D., National Cancer Institute, Bethesda, MD; Gilman D. Grave, M.D., National Institute of Child Health and Human Development, Bethesda, MD; Edward C. Doo, M.D., Jay H. Hoofnagle, M.D., Patricia R. Robuck, Ph.D., M.P.H. (project scientist), National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD; Patricia Belt, B.S., Frederick L. Brancati, M.D., M.H.S. (2003-2009), Jeanne M. Clark, M.D., M.P.H., Ryan Colvin, M.P.H., Michele Donithan, M.H.S., Mika Green, M.A., Rosemary Hollick (2003-2005), Milana Isaacson, B.S., Wana Kim, B.S., Alison Lydecker, M.P.H. (2006-2008), Pamela Mann, M.P.H. (2008-2009), Laura Miriel, Alice Sternberg, Sc.M., James Tonascia, Ph.D., Aynur Ünalp-Arida, M.D., Ph.D., Mark Van Natta, M.H.S., Ivana Vaughn, M.P.H., Laura Wilson, Sc.M., Katherine Yates, Sc.M, Johns Hopkins University, Bloomberg School of Public Health (Data Coordinating Center), Baltimore, MD.

Ancillary