Epoxygenase-derived epoxyeicosatrienoic acid (EET) mediators are associated with NAFLD, NASH and fibrosis

Introduction

Nonalcoholic fatty liver disease (NAFLD) is the most prevalent liver disease in the United States, and is projected to become the leading indication for liver transplantation by 2030. Among patients with NAFLD, nearly 25% develop steatohepatitis (NASH), which accelerates progression to fibrosis and cirrhosis. However, reliable biomarkers capable of identifying NASH are lacking. Discovering such biomarkers may facilitate drug development and help identify high-risk patients most likely to benefit from therapy.

Cytochrome P450 (CYP) enzymes metabolize fatty acids to bioactive lipids that exert hepatoprotective effects. The CYP2C and CYP2J epoxygenases convert arachidonic acid into epoxyeicosatrienoic acid (EET) regioisomers (primarily 11,12- and 14,15-EET), which are converted into less-active dihydroxyeicosatrienoic acids (DHET) by soluble epoxide hydrolase (sEH). The sum of total EET (i.e. 11,12- and 14,15-EET) plus total DHET (i.e. 11,12- and 14,15-DHET), and the 14,15-EET to 14,15-DHET ratio, are accepted biomarkers of CYP epoxygenase and sEH function, respectively^1,2. Emerging preclinical data suggest that EET may protect against NASH. In vitro, 14,15-EET treatment protects HepG2 cells from palmitic acid-induced inflammation and oxidative stress, while disrupting the gene that encodes sEH, Ephx2, restores EET levels and attenuates liver injury³. In vivo, NASH is accompanied by suppressed CYP epoxygenase activity and reduced hepatic and circulating EET, while EET administration promotes liver regeneration⁴.

While several clinical studies have evaluated circulating eicosanoids in NAFLD^2,5, none have quantified EET metabolites across the full NAFLD histological spectrum. Thus, we leveraged state-of-the-art, liquid chromatography tandem mass spectrometry (LC/MS/MS) to study circulating 11,12-EET, 14,15-EET and CYP epoxygenase activity in relation to the prevalence and histological severity of NAFLD.

Methods

The study was approved by the Human Research Committee Institutional Review Board. This was a cross-sectional analysis within the prospective Massachusetts General Hospital NAFLD Repository. Since 2010, this cohort has enrolled adults with histologically-confirmed NAFLD and normal (non-NAFLD) controls (see eMethods).

We identified 80 age- and sex-matched adults across four groups: normal liver (non-NAFLD controls), steatosis, NASH with fibrosis stage 0–2 (F0–2), and NASH with fibrosis stage 3–4 (F3–4). The eMethods outlines serum lipid extraction and EET quantification. We focused on eight EET exposures: total EET, total DHET, EET+DHET (i.e. CYP epoxygenase function) and the 14,15-EET:14,15-DHET ratio (EET:DHET; i.e. sEH function).

Statistical analyses are described in the eMethods. Briefly, our primary analysis compared controls to all NAFLD histological categories. We also conducted analyses among NAFLD patients (n=60), comparing steatosis (reference) to NASH F0–2 and NASH F3–4. Using multivariable-adjusted linear regression, we assessed serum EET exposures in relation to NAFLD prevalence and histological severity.

Results

Among 80 subjects, 20 (25%) had normal livers; 60 (75%) had NAFLD (i.e. 20 steatosis; 20 NASH F0–2; 20 NASH F3–4). Table S1 outlines subject characteristics; Figures S1A–B depict EETs by histology group. Compared to controls, total EET, 11,12-EET, 14,15-EET, total DHET, 11,12-DHET and 14,15-DHET were significantly lower in steatosis and NASH (all p<0.05)(Table 1). Among NAFLD subjects, these inverse associations were similar: compared to steatosis, total EET and total DHET were significantly lower in NASH F0–2 (p=0.041 and 0.006, respectively), and NASH F3–4 (p=0.001 and <0.001, respectively). After multivariable adjustment, inverse associations persisted for total EET and total DHET (all p<0.05)(Table 1).

Table 1.

CYP-derived Epoxyeicosatrienoic acid (EET) and Dihydroxyeicosatrienoic acid (DHET) Metabolite Levels According to NAFLD Histological Severity

Metabolite (pg/mL), mean (SD)	Controls N=20	Simple Steatosis N=20	NASH F0–2 N=20	NASH F3–4 N=20	Control vs. SS Pa	Control vs. NASH Pa	SS vs. NASH F0–2 Pa	SS vs. NASH F3–4 Pa
CYP epoxides
14,15-EET	493 (103)	308 (99)	206 (95)	132 (87)	0.008	<0.001	0.017	0.0001
11,12-EET	191 (53)	168 (56)	139 (50)	128 (44)	0.07	0.002	0.032	0.001
Total EET	684 (97)	476 (83)	345 (80)	260 (88)	0.10	0.002	0.041	0.001
Soluble epoxide hydrolase (sEH) diols (pg/mL)
14,15-DHET	1202 (162)	977 (118)	619 (120)	384 (105)	0.18	0.001	0.009	<0.0001
11,12-DHET	541 (61)	470 (49)	355 (53)	242 (32)	0.34	0.003	0.045	0.001
Total DHET	1,743 (295)	1,447 (202)	974 (241)	626 (208)	0.03	<0.001	0.006	<0.001
CYP epoxygenase function (pg/mL)
Sum EET+DHET	2,427 (330)	1,923 (285)	1,319 (306)	886 (343)	0.01	<0.001	0.012	<0.001
sEH function (ratio)
EET: DHET Ratio	0.28	0.32	0.26	0.29	0.56	0.42	0.74	0.65

Data are expressed as mean (SD; standard deviation).

Abbreviations: CYP, cytochrome P450; EET, epoxyeicosatrienoic acid; DHET, dihydroxyeicosatrienoic acid; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; sEH, soluble epoxide hydrolase. For definitions, see Methods and eMethods.

^aStudent’s t-test was performed on log-transformed data followed by FDR correction to account for multiple testing.

^*P-values were derived a multivariable-adjusted model, which accounted for continuous age (years), sex, race/ethnicity, continuous body mass index, type 2 diabetes, hypertension, dyslipidemia, smoking status, use of aspirin or statins, and calendar year. For details, see eMethods and the Table S1 footnotes.

^§NAFLD Severity was coded as an ordinal variable (i.e. 1=no NAFLD; 2= steatosis; 3=NASH with F0–2 fibrosis; 4=NASH with F3–4 fibrosis).

Compared to controls, CYP epoxygenase function (EET+DHET) was significantly lower in NAFLD (p=0.002). Among NAFLD subjects, CYP epoxygenase function declined with progressive NASH (p=0.014)(Table 1), In contrast, EET:DHET did not differ significantly between NAFLD and controls (p=0.17) or with worsening histological severity (p=0.20).

In exploratory analyses, we assessed the performance of EET metabolites for distinguishing NASH from steatosis (Figure S2). EET+DHET accurately identified 93% of NASH cases (AUROC 0.93, 95% CI 0.86–1.00; p<0.001), with adequate calibration (Hosmer-Lemeshow test, χ² P = 1.00).

Discussion

Among adults with histologically-confirmed NAFLD and non-NAFLD controls, circulating EETs were inversely and significantly associated with NAFLD prevalence and histological severity. CYP epoxygenase function progressively declined with worsening NASH fibrosis, and accurately distinguished NASH from steatosis. In contrast, sEH function was not significantly associated with NAFLD prevalence or histological severity. Together, this suggests that EET depletion in NAFLD may result from suppressed EET biosynthesis, and further that EET augmentation could represent a potential therapeutic strategy for NAFLD.

To date, studies quantifying circulating EETs and CYP epoxygenase activity across the NAFLD histological spectrum are scarce. Recently, a small study (n=22) found a positive association between plasma EETs and NASH², in contrast to our findings. However, with only 7 NASH patients, that study could not control for important confounders or assess EETs by NAFLD histological severity. Future prospective studies are needed to clarify this relationship; however, our findings are biologically plausible, given the body of preclinical data indicating that EET depletion from CYP epoxygenase suppression promotes the development of NAFLD, and further that EET augmentation attenuates steatosis, NASH and fibrosis^3,6. Moreover, reduced EETs have also been linked to other metabolic diseases, including diabetes, obesity and heart disease^7,8.

If validated, our findings could carry important implications. Devising non-invasive means to identify high-risk patients with progressive NASH remains an unmet need. Thus, elucidating biomarkers that reflect dysregulated biochemical pathways could identify patients at greatest risk of disease progression, and those most likely to benefit from targeted pathway modulation.

We acknowledge several limitations. As a cross-sectional study, we cannot infer a causal effect of EET on NAFLD pathogenesis, and prospective studies are warranted. Second, we focused upon just 8 hypothesis-driven EET metabolites, and by quantifying free EET, we could not account for the subset of EET esterified to lipoprotein phospholipids^3,4. Accordingly, future studies should validate our findings while broadly assessing free and total EET with other potential mediators. Finally, our cohort was limited to patients with obesity undergoing elective bariatric surgery, and we lacked detailed measures of insulin sensitivity or adiposity, beyond BMI. However, the demographic and clinical profile is representative of the general NAFLD population, supporting the generalizability of our findings.

In conclusion, we demonstrate associations between reduced circulating EET, DHET and CYP epoxygenase function, and the prevalence and histological severity of NAFLD and NASH. Circulating CYP-derived EET may represent potential biomarkers for NASH, and explorations of EET-augmenting therapeutic interventions for NAFLD are warranted.

Abbreviations

CYP

Cytochrome P450

DHET

Dihydroxyeicosatrienoic acid

EET

Epoxyeicosatrienoic acid

LC/MS/MS

Liquid chromatography tandem mass spectrometry

NAFLD

Nonalcoholic fatty liver disease

NASH

Nonalcoholic steatohepatitis

sEH

Soluble epoxide hydrolase