Biomarkers of Hepatic Dysfunction and Cardiovascular Risk

Abstract

Purpose of Review

The objective of this manuscript is to examine the current literature on non-alcoholic fatty liver disease (NAFLD) biomarkers and their correlation with cardiovascular disease (CVD) outcomes and cardiovascular risk scores.

Recent Findings

There has been a growing appreciation for an independent link between NAFLD and CVD, culminating in a scientific statement by the American Heart Association in 2022. More recently, studies have begun to identify biomarkers of the three NAFLD phases as potent predictors of cardiovascular risk.

Summary

Despite the body of evidence supporting a connection between hepatic biomarkers and CVD, more research is certainly needed, as some studies find no significant relationship. If this relationship continues to be robust and readily reproducible, NAFLD and its biomarkers may have an exciting role in the future of cardiovascular risk prediction, possibly as risk-enhancing factors or as components of novel cardiovascular risk prediction models.

Introduction

Non-alcoholic fatty liver disease (NAFLD) is a leading cause of end-stage liver disease worldwide [1]. NAFLD exists on a spectrum with three stages. The first, non-alcoholic fatty liver (NAFL), is characterized by hepatic steatosis. The second stage, non-alcoholic steatohepatitis (NASH), is marked by inflammation (both hepatic lobular and portal) and hepatocyte ballooning degeneration. The final phase begins with the development of bridging fibrosis which progresses to cause cirrhosis and, in some cases, hepatocellular carcinoma [2]. It is important to note that these phases overlap extensively. The diagnosis of NAFLD requires the presence of hepatic steatosis and the absence of alternative etiologies, most notably significant alcohol intake (Table 1). Liver biopsy and imaging are the principal modalities for making this diagnosis, with biopsy considered the gold standard [3]. However, biopsy has significant limitations, including invasiveness and risk of sampling bias [4]. Imaging, typically via ultrasound or magnetic resonance imaging (MRI), also has limitations, including cost and inter-operator variability [5]. In response to these limitations, plasma and serum biomarkers are being explored as an alternative modality for diagnosing NAFLD and tracking disease progression [6].

Table 1.

Proposed nomenclatures for metabolic associated hepatic dysfunction [7••, 42]

	NAFLD	MAFLD	MASLD	MetALD
Terminology	Non-alcoholic fatty liver disease	Metabolic dysfunction associated fatty liver disease	Metabolic dysfunction-associated steatotic liver disease	Metabolic dysfunction-associated steatotic liver disease and increased alcohol intake
Definition	Hepatic steatosis without excessive alcoholic intake, viral hepatitis, and other causes of chronic liver disease	Hepatic steatosis with the presence of either overweight/obesity, type 2 diabetes, or ≥ 2 metabolic dysfunctions	Hepatic steatosis with the presence of ≥ 1 cardiometabolic criteria	Hepatic steatosis with the presence of ≥ 1 cardiometabolic criteria, and significant alcohol intake
Alcohol intake	< 210 g/week for men; < 140 g/week for women	Independent of alcoholic intake	210 g/week for men; 140 g/week for women	210 to 420 g/week for men; 140 to 350 g/week for women
Insulin resistance	Not core criteria	One of the core criteria	Not core criteria	Not core criteria
Diagnostic approach	Liver-centric	Cardiometabolic-centric		Cardiometabolic and alcohol centric

It is important to note that a multi-society consortium including the American Association for the Study of Liver Disease (AASLD) recently released a consensus statement on nomenclature to replace the term NAFLD. They propose metabolic-dysfunction associated steatotic liver disease (MASLD) as a term that more accurately describes the disease phenotype. Diagnosing MASLD also requires hepatic steatosis without significant alcohol intake. A new criterion is the presence of at least one cardiometabolic risk factor, including elevated body mass index (BMI), elevated waist circumference, pre-diabetes, type 2 diabetes (DM2), hypertension, or dyslipidemia. Thus, the MASLD group will include most patients previously defined as NAFLD. If the alcohol threshold is exceeded, a distinct category, MASLD and increased alcohol intake (MetALD), will apply (Table 1) [7••]. Given that most research in the liver biomarker space up until now has utilized the old NAFLD terminology, we will be primarily using this for our review.

NAFLD and Cardiovascular Disease (CVD)

An area of interest in the NAFLD space is the heart-liver axis. CVD is a broad term that involves many pathologies of the heart and the vascular system including atherosclerotic cardiovascular disease (ASCVD), heart failure (HF), valvular heart disease, arrhythmias, and sudden cardiac death (SCD). Given that CVD and NAFLD share risk factors such as DM2, obesity, systemic inflammation, and atherogenic dyslipidemia, it is not surprising that these conditions overlap. However, some investigations highlight NAFLD itself as an independent risk factor for CVD. Dyslipidemia is likely a central mechanism of this relationship. NAFLD is associated with a pro-atherogenic lipid profile and inhibition of certain lipid metabolites can reduce hepatic steatosis and atherosclerosis in rodents [8]. The inflammatory milieu generated in NAFLD is also likely involved, as systemic inflammation leads to endothelial dysfunction and promotes atherosclerosis [8]. Interestingly, inflammatory cytokines are shown to directly induce cardiac arrhythmias [9] and left ventricular (LV) diastolic dysfunction in animal models [10]. Deranged amino acid metabolism is another common thread. Glycine is lowered in both disease states, and glycine-based treatments have been shown to improve hepatic steatosis and atherosclerosis [8]. Alternatively, branched-chain amino acids (BCAA) such as leucine and valine are elevated in both disease states. BCAAs appear to trigger cardiac fat deposition and remodeling [10], although some experiments show beneficial effects of exogenous BCAAs [8]. Epicardial triglyceride (TG) accumulation is another potential mechanism. NAFLD is independently associated with epicardial TG deposition and the presence of epicardial fat has been tied to pro-arrhythmogenic conduction [9], as well as impaired hemodynamics [10].

When evaluating patients from a large Swedish cohort, Simon et al. showed that major adverse cardiac events (MACE) were significantly increased in the biopsy-proven NAFLD population, even after adjustment for confounders such as age and DM2 [11]. MACE and rates of ASCVD in NAFLD patients diagnosed via imaging alone are similarly increased [12]. Interestingly, Mendelian randomization studies have corroborated these findings by showing single nucleotide polymorphisms (SNPs) that predict NAFLD also correlate with CVD risk [13]. In response to this growing body of evidence, the American Heart Association (AHA) published a scientific statement in 2022 naming NAFLD as an independent risk factor for ASCVD [14••]. Moreover, researchers have begun investigating whether biomarkers originally identified for NAFLD may be useful in estimating and refining CVD risk (Fig. 1). Indeed, preliminary investigations suggest that considering NAFLD biomarkers in cardiovascular risk calculations can increase their accuracy, especially for NAFLD patients in whom cardiovascular risk is often underestimated [15]. Given the rapid proliferation of studies, we believe that it is timely and important to review the literature on biomarkers of hepatic dysfunction in NAFLD and their relationship to CVD.

Fig. 1.

Biomarkers of hepatic dysfunction and cardiovascular risk. Increasing NAFLD severity is associated with a progressive enhancement of cardiovascular risk. The magnitude of this risk can be estimated using serum and plasma biomarkers of NAFLD at its various stages. Abbreviations: NAFLD, non-alcoholic fatty liver disease; MI, myocardial infarction; AF, atrial fibrillation; NASH, non-alcoholic steatohepatitis; CK-18, cytokeratin-18; FGF-21, fibroblast growth factor 21; Hs-CRP, high sensitivity C-reactive protein; TNF-alpha, tumor necrosis factor alpha; aPAI1, activated plasminogen activator inhibitor 1; AST, aspartate aminotransferase. Figure created with BioRender.com

NAFL Biomarkers

Hepatic steatosis is the hallmark pathologic feature in NAFLD, and its diagnosis requires steatosis in ≥ 5% of hepatocytes on histology or intrahepatic TG content ≥ 5.5% on MRI. However, ultrasound is frequently used for diagnosis and has an area-under-the-receiver-operator-curve (AUROC) of 0.95 for detecting moderate to severe steatosis when compared to liver biopsy [6]. The following section examines biomarkers of hepatic steatosis and their relationship to CVD (Table 2).

Table 2.

Association between non-invasive biomarkers (NAFL, NASH, and NASH-related fibrosis) and cardiovascular outcomes

Biomarker	Population/dataset	Country	CVD relationship
NAFL
Fatty Liver Index	Adults without CVD [17]	Republic of Korea	Cardiovascular mortality, non-fatal MI, CVA
	KNHIS [18]	Republic of Korea	HF incidence, HF hospitalization
	Chinese Kadoorie Biobank [19]	China	MI, CVA
Hepatic Steatosis Index	Adults without CVD [21]	Republic of Korea	FRS
NAFLD Liver Fat Score	NHANES 1988–1994 [23]	USA	Cardiovascular mortality
	NHANES 1999–2016 [24]	USA	CAD, CHF, angina, cardiovascular mortality
	KNHANES 2008–2011 [25]	Republic of Korea	ACC/AHA PCE
Triglyceride-Glucose Index	PURE [28]	Various	MI, CVA
	NAFLD patients with chest pain [29]	China	Gensini Score, CAD
NASH
CK-18	Adults without chronic liver disease [33]	China	Incidence of cardiometabolic disease
	WELCOME [34]	UK	CIMT progression
	PREVEND [35]	The Netherlands	SCORE
	Post-MI patients [36]	Turkey	LV remodeling
Hs-CRP	Patients with prior MI [39]	Sweden	MACE
	Patients with NAFLD [40]	Republic of Korea	CAC
	Patients without CVD [41]	Taiwan	FRS
	NHANES 1988–1994 [42]	USA	Cardiovascular mortality
TNF-α	Danish Research Center for Prevention and Health [45]	Denmark	Non-fatal MI, CAD mortality
IL-6	MESA [48]	USA	CAD prevalence and severity
aPAI1	Framingham Heart Study [55]	USA	MACE
Arachidonic-acid derived lipid mediators	Patients with prior MI [59]	Taiwan	Recurrent MI
FGF-21	Patients with NAFLD and/or CAD [62]	China	CAD prevalence
	Shanghai Diabetes Study [63]	China	ASCVD prevalence
	Systemic review of 28 studies [64]	Various	Cardiovascular mortality, CAD incidence
	Guangdong Coronary Artery Disease Cohort [65]	China	Cardiovascular mortality
NASH-related fibrosis
Fibrosis-4 Index	Asymptomatic adults with NAFLD [68]	Republic of Korea	CAC progression
	NAFLD patients without CVD [83]	Republic of Korea	CAC progression
	Patients with CVD [69]	Japan, China, USA, Germany	Cardiovascular mortality
	Patients with MAFLD [70]	Republic of Korea	ACC/AHA PCE
	Patients with MAFLD [71]	Iran	CAD
NAFLD Fibrosis Score	Asymptomatic adults with NAFLD [68]	Republic of Korea	CAC progression
	NAFLD patients without CVD [83]	Republic of Korea	CAC progression
	Patients with CVD [69]	Japan, China, USA, Germany	Cardiovascular events, cardiovascular mortality
	Patients with NAFLD [81]	Italy	FRS, Progetto CUORE, SCORE
	Patients with MAFLD [70]	Republic of Korea	ACC/AHA PCE
BARD Score	Patients with NAFLD [81]	Italy	FRS, Progetto CUORE, SCORE
	KNHIS [18]	Republic of Korea	HF incidence, HF hospitalization, cardiovascular mortality
AST to Platelet Ratio Index	NAFLD patients without CVD [83]	Republic of Korea	CAC progression
	Patients with NAFLD [81]	Italy	FRS, Progetto CUORE, SCORE
Forns Index	NAFLD patients without CVD [83]	Republic of Korea	CAC progression
	Patients with NAFLD [81]	Italy	Framingham risk score, Progetto CUORE, SCORE
Hepamet Fibrosis Score	Patients with NAFLD [81]	Italy	Framingham risk score, Progetto CUORE, SCORE

Fatty Liver Index (FLI)

The FLI ranges from 0 to 100 and is generated using BMI, waist circumference, TG level, and serum gamma-glutamyl transferase (GGT) concentration. A score of less than 30 rules out hepatic steatosis with a negative likelihood ratio of 0.2, while a score above 60 rules it in with a positive likelihood ratio of 4.3 [16]. Data from over 3 million patients in the Korean National Health Insurance System (KNHIS) demonstrated a nearly twofold increased risk (hazard ratio (HR) of 1.99, 95% confidence interval (CI) 1.91–2.07) for a composite endpoint of cardiovascular death, non-fatal myocardial infarction (MI), and cerebrovascular accident (CVA) in the highest quartile of FLI. This analysis was adjusted for age, sex, smoking, physical activity, income, body weight, total cholesterol, hypertension, diabetes, and dyslipidemia [17]. In the same database, an analysis of almost 800,000 patients showed increased rates of HF diagnosis (HR = 1.30, 95% CI 1.24–1.36) and HF hospitalization (HR = 1.54, 95% CI 1.44–1.66) in those with FLI over 60. This analysis was adjusted for age, sex, body weight, alcohol intake, smoking, physical activity, income, hypertension, diabetes, dyslipidemia, and glomerular filtration rate (GFR) [18]. In a prospective cohort study using the China Kadoorie Biobank, the highest quintile versus lowest quintile of FLI had a HR of 1.68 (95% CI 1.56–1.80) for MI and a HR of 1.48 (95% CI 1.40–1.57) for CVA after controlling for age, sex, region, educational level, alcohol intake, and smoking [19].

Hepatic Steatosis Index (HSI)

The HSI incorporates aspartate transaminase (AST), alanine transaminase (ALT), BMI, female sex, and presence of DM2. Indices less than 30 rule out NAFLD with a sensitivity of 93.1%, while scores over 36 ruled in NAFLD with a specificity of 92.4% [20]. Kweon et al. investigated over 20,000 Korean adults and stratified them by cardiovascular risk according to the Framingham Risk Score (FRS). Their analyses controlled for age, sex, alcohol intake, smoking, and obesity, as well as dyslipidemia and hypertension treatment. The odds ratio (OR) for patients in the high-risk (> 20% 10-year risk of MI) category having HSI > 36 was 2.43 (95% CI 1.90–3.12) compared to those with HSI < 30. The AUROC for using HSI > 36 as a predictor of elevated cardiovascular risk was 0.612 with a sensitivity of 67.6% and specificity of 51.1%. Interestingly, the FLI had a stronger magnitude of association with high cardiovascular risk, with a HR of 6.36 (95% CI 4.78–8.46) for FLI > 60 relative to FLI < 30 [21].

NAFLD Liver Fat Score (NAFLD-LFS)

The NAFLD-LFS incorporates the presence of metabolic syndrome, presence of DM2, fasting serum insulin, AST, and ALT to create a score that predicts NAFLD with 86% sensitivity and 71% specificity using a cutoff of −0.640 (area-under-the-curve (AUC) 0.87) [22]. Cheung et al. analyzed data from the National Health and Nutrition Examination Survey (NHANES) conducted between 1988 and 1994, along with subsequent follow-up data up to 2006. They observed a HR of 2.24 (95% CI 1.03–4.88) for cardiovascular mortality in those with high NAFLD-LFS (≥ 1.257) despite adjusting for several shared risk factors, including diabetes, hypertension, and smoking [23]. When over 17,000 adults from the NHANES were examined from 1999 to 2016, a high LFS score was again associated with cardiovascular mortality, as well as incident CVD. There was also a higher prevalence of CVD across increasing FLI and HSI in this cohort after controlling for variables such as age, smoking, and statin use [24]. In both lean and obese subjects, LFS greater than −0.640 had an unadjusted correlation with higher CVD risk as estimated by AHA/American College of Cardiology (ACC) Pooled Cohort Equations (PCE). Interestingly, this correlation was stronger in lean patients [25].

Triglyceride-Glucose Index (TyG)

The TyG, calculated from fasting blood glucose and TG levels, was originally developed to screen for insulin resistance (IR). Indices over 4.49 can diagnose IR with a sensitivity 82.6% and specificity 82.1% using the homeostatic model assessment for insulin resistance (HOMA-IR) as a benchmark [26]. The TyG also has utility in diagnosing NAFLD [27]. In a prospective cohort study following patients from 22 countries for 13 years, the TyG was associated with a composite outcome of all-cause mortality, cardiovascular death, non-fatal MI, and CVA with a hazard ratio of 1.21 (95% CI 1.13–1.30). It was also individually associated with MI (HR = 1.24; 95% CI 1.12–1.38) and CVA (HR = 1.16; 95% CI 1.05–1.28). Their analyses were adjusted for cardiovascular risk factors including hypertension, dyslipidemia, diabetes, BMI, and smoking. Interestingly, these associations were higher in magnitude in the subgroup analysis restricted to low-income countries [28]. Additionally, the TyG has been associated with the presence and severity of coronary artery disease (CAD) on coronary angiography in NAFLD patients, independent of age, sex, hypertension, and diabetes [29].

NASH Biomarkers

Following steatosis comes inflammation and ballooning degeneration of hepatic parenchyma. Serum biomarkers of this phase are even more crucial, as steatohepatitis is difficult to detect with imaging [6]. The following section explores NASH biomarkers and their relationship to CVD (Table 2).

Apoptosis

Hepatocyte apoptosis is a key cell death pathway in NASH. Cytokeratin-18 (CK-18), consisting of M30 and M65 antigens, is a cytoskeletal component released during hepatocellular injury and can be used as a serum marker of NASH [30, 31]. Two novel NAFLD susceptibility genetic variants, including patatin-like phospholipase domain-containing-3 (PNPLA3) and glucokinase regulatory protein (GCKR), were found to be associated with higher CK-18 fragment levels [32]. A cross-sectional study in China with 588 participants found that total CK18 (M65) was positively correlated with cardiometabolic disorders such as obesity and dyslipidemia, even after controlling for multiple variables including age, BMI, insulin resistance, hemoglobin A1c (HbA1c), NAFLD diagnosis, and blood pressure [33]. Additionally, reduced carotid intima-media thickness (CIMT) progression has been associated with reductions in CK-18 after adjusting for age, sex, diabetes, smoking, BMI, TG, statin, and antihypertensive use [34]. In a multivariate logistic regression analysis of 312 Dutch patients with NAFLD, the M30 antigen emerged as the only independent predictor of FLI ≥ 60. However, M65 was the only antigen that could predict a very high 10-year CVD risk based on the Systematic Coronary Risk Evaluation (SCORE) algorithm (AUC 0.714). The odds of very high CVD risk increased fivefold at a M65 cutoff of 400 U/L [35]. Interestingly, multivariate regression analyses have also linked elevated CK-18 with LV remodeling after MI and the prognosis of CVA [36, 37].

Inflammation

High sensitivity C-reactive protein level (hs-CRP) is a well-recognized surrogate of chronic inflammation and can help distinguish NASH from NAFL [38]. Hs-CRP is also a well-recognized risk factor for CVD and inflammation plays a key role in its pathophysiology [39]. After controlling variables like age, smoking, and BMI, the development of coronary artery calcification (CAC) is associated with elevated hs-CRP and this association is strengthened in NAFLD patients [40]. The predictive value of NAFLD for FRS ≥ 10% after adjusting for traditional cardiovascular risk factors trends higher in patient with elevated baseline hs-CRP (OR = 2.89 [95% CI 1.40–5.94] versus OR = 1.89 [95% CI 1.23-2.91]) ]) [41]. Of note, elevated hs-CRP is part of the metabolic-associated fatty liver disease (MAFLD) diagnostic criteria and is correlated with cardiovascular mortality in these patients even after controlling for sex, age, race, diabetes, and hypertension [42].

Tumor necrosis factor-alpha (TNF-α) is an inflammatory cytokine that rises with NAFLD progression and is released by liver cells in response to lipid deposition [43]. A meta-analysis of 35 studies reported a higher circulating TNF-α in patients with NASH than in NAFL [44]. Another meta-analysis of six population-based prospective studies adjusted for traditional cardiovascular risk factors observed that increases in TNF-α by one standard deviation corresponds to a 1.17-fold increased relative risk (95% CI 1.09–1.25) of non-fatal MI or CAD mortality [45]. Interestingly, TNF-α inhibition may be associated with improved CVD outcomes [46].

Interleukins (IL) are a diverse group of pro-inflammatory cytokines, some of which are associated with NAFLD and NASH, including IL-1β and IL-6 [47]. Among 668 participants from the Multi-Ethnic Study of Atherosclerosis (MESA), circulating IL-6 was independently associated with the prevalence and severity of subclinical atherosclerosis [48]. This analysis controlled for many variables including age, sex, study site, diabetes, and total cholesterol. The Canakinumab Anti-inflammatory Thrombosis Outcome Study (CANTOS) trial showed that targeted IL-1β inhibition can reduce recurrent CVD events [49]. Preclinical studies investigating the role of IL-1β at the intersection of NASH and CVD have shown that inhibition has benefits in animal models of CVD and steatohepatitis [50, 51].

Other proposed biomarkers of NASH include CXC-chemokine 10 (CXCL10) [52], activated plasminogen activator-inhibitor 1 (aPAI1) [53], and adipokines like adiponectin [54]. Of particular interest is aPAI1, given its overlapping functions in inflammation and coagulation. Indeed, aPAI1 is predictive of cardiovascular risk independent of age, diabetes, hypertension, smoking, total cholesterol, and BMI [55]. CXCL10 and adipokines have been shown to play a role in CVD as well [56, 57].

Oxidative stress

Oxidative stress is a major pathogenic mechanism in NAFLD. Lipidomic studies have found certain lipid oxidation products to be associated with NASH. This discovery has led to the creation of the OxNASH panel, which measures such lipid products with mass spectrometry and demonstrates a commendable level of diagnostic accuracy for NASH with an AUROC of 0.74 [58]. In tandem, products of arachidonic acid oxidation have been associated with MI in patients with known CAD after adjustment for age, sex, and BMI [59].

Lipid and Carbohydrate Metabolism

Fibroblast growth factor 21 (FGF-21) is a hepatically generated hormone that can reduce inflammation, improve insulin sensitivity, and slow hepatic steatosis [60]. Concerning its diagnostic value, a meta-analysis of 25 studies demonstrated a pooled sensitivity and specificity for NASH of 0.62 and 0.78, respectively. It is noteworthy that when combined with CK-18, higher sensitivity (0.92) and specificity (0.85) were demonstrated [61]. In an unadjusted analysis, FGF-21 was associated with NAFLD and CVD individually, but levels were highest in patients with both diseases [62]. Increasing levels of FGF-21 are associated with ASCVD and cardiovascular mortality after controlling for traditional cardiovascular risk factors [63, 64]. Interestingly, one study observed a U-shaped relationship with outcomes in CAD, in which cardiovascular mortality peaked with the lowest and highest concentrations of FGF-21 [65]. This analysis took multiple confounders into account including age, sex, smoking, and diabetes.

NASH-Related Fibrosis Biomarkers

Hepatic fibrosis defines the final phase of NAFLD that ultimately results in hepatic cirrhosis if left unchecked [66]. The following section examines biomarkers of hepatic fibrosis and their connection to CVD (Table 2).

Fibrosis-4 Index (FIB-4)

The FIB-4, initially devised in 2006 for patients with human immunodeficiency virus and hepatitis C virus infections, is used to estimate fibrotic burden in NAFLD. FIB-4 is generated from age, AST, ALT, and platelet count [67]. Among 1173 adults without a history of CVD, the odds ratio for CAC progression was reported to be 1.70 (95% CI 1.12 to 2.58) in patients with NAFLD and intermediate or high FIB-4 scores versus those without NAFLD [68]. This finding was independent of sex, BMI, smoking, hypertension, and other traditional cardiovascular risk factors. A meta-analysis of 12 adjusted cohort studies comprising 25,252 patients with CVD saw a significant increase in CVD events (HR = 1.75, 95% CI 1.53–2.00) at the highest levels of FIB-4. In the same group, CV mortality and all-cause mortality increased approximately twofold [69]. In the presence of MAFLD, elevated FIB-4 (≥ 2.67) was associated with higher ASCVD risk when compared to those with lower FIB-4 (OR = 2.40 versus OR = 1.44) after controlling for variables such as sex, smoking, prior ASCVD history, BMI, and insulin resistance [70]. A prospective study based on 1644 patients showed that by employing a FIB-4 cutoff value of 0.85, CAD can be predicted in MAFLD patients with an AUC of 0.656 (95% CI 0.618–0.693) independent of age, sex, diabetes, smoking, BMI, HbA1c, and dyslipidemia [71].

NAFLD Fibrosis Score (NFS)

By further integrating BMI, diabetes presence, and albumin, the NAFLD fibrosis score achieves improved predictive accuracy for advanced fibrosis in patients with NAFLD [72]. Elevated NFS has been associated with CAC progression, CIMT, and arterial stiffness even with adjustment for traditional cardiovascular risk factors [68, 73]. The Dong-gu study from South Korea showed a HR of 3.60 (95% CI 1.80–6.90) for cardiovascular mortality at high NFS levels despite controlling for sex, smoking, hypertension, and HbA1c, among other variables. They found similar results for the FIB-4 score (HR = 2.60), but not the APRI and BARD scores [74]. In a secondary analysis of the Treatment of Preserved Cardiac Function Heart Failure (TOPCAT) trial, NFS had the strongest predictive value for a composite endpoint of cardiovascular death, aborted cardiac arrest, or HF hospitalization with an AUC of 0.672 (95% CI 0.642–0.702), relative to other liver fibrosis scores. Additionally, across ascending NFS categories, there is a clear and substantial risk gradient for incident atrial fibrillation, exhibiting an unadjusted AUC of 0.678 (95% CI 0.622–0.734) [75]. This superiority could be attributed to the inclusion of serum albumin, which has prognostic significance in incident CVD, mediated through mechanisms such as inflammation and oxidative stress [76]. In parallel with FIB-4, patients with MAFLD and significant liver fibrosis as measured by NFS exhibited a heightened ASCVD risk (OR = 2.77, 95% CI 1.92–4.01) relative to those without MAFLD, surpassing that observed in MAFLD patients without significant liver fibrosis (OR = 1.60, 95% CI 1.21–2.12) [70]. This analysis was adjusted for sex, age smoking, prior ASCVD history, insulin resistance, BMI, and physical activity.

BARD Score

The BARD score is a 4-point scoring system based on three criteria, BMI, AST to ALT ratio, and the presence of T2D. Scores ≥ 2 can predict advanced fibrosis in NAFLD patients with AUROC 0.865 (95% CI 0.793–0.920) [77]. In a nation-wide cohort study in Korea with new-onset T2D patients (n = 139,633), BARD ≥ 2 and FLI ≥ 60 were both associated with a significant increase in MI (HR = 1.26, 95% CI 1.14–1.40), CVA (HR = 1.26, 95% CI 1.15–1.37), and HF (HR = 1.31, 95% CI 1.22–1.41) [78]. This finding was independent of age, sex, smoking, alcohol intake, income level, dyslipidemia, and hypertension. The same secondary analysis of the TOPCAT trial mentioned previously also assessed the BARD score and found a similar predictive value for their primary endpoint (AUC = 0.618, 95% CI 0.588–0.648) [75]. Additionally, BARD ≥ 2 was associated with incident HF, HF hospitalization, and cardiovascular mortality in the KNHIS database after controlling for traditional cardiovascular risk factors [18].

AST to Platelet Ratio Index (APRI)

APRI is a simple fibrotic assessment tool that uses only AST and platelets [79]. An unadjusted cross-sectional study in 1225 patients showed a significant correlation (r = 0.4) between the APRI score and FRS. This correlation is stronger when restricted to patients with APRI > 0.5. In this same study, the metabolic syndrome was associated with a twofold increase in FRS in both men and women. Interestingly, when restricted to patients with APRI > 0.5, the metabolic syndrome conferred an even greater FRS increase [80]. However, the correlation of APRI with CVD has been shown to be weaker relative to other fibrosis biomarkers [81].

Forns Index

The Forns index is calculated based on platelet count, GGT, cholesterol, and age [82]. The AUC for predicting CAC over 100 using the Forns index was 0.659 after adjusting for age, sex, and BMI. This was slightly lower than NFS (0.689) and FIB-4 (0.683), but higher than APRI (0.595) [83]. Forns index also has a strong, unadjusted correlation with different CV risk scores [81]. Interestingly, a study from the UK biobank with more than 250,000 participants showed that the addition of GGT alone to the SCORE model offers a slight but discernible improvement in the 10-year prediction of fatal CV events [84].

Hepamet Fibrosis Score (HFS)

The HFS was created in response to the limitations of other fibrosis scores. It incorporates age, gender, albumin, HOMA-IR, insulin, platelets, and diabetes status. Interestingly, the outcome of HFS is not significantly influenced by age or BMI [85]. A retrospective study of patients with NAFLD and chronic viral hepatitis in Italy revealed a strong, unadjusted association between HFS and cardiovascular risk scores. Notably, HFS ranked third in correlation strength for SCORE, following NFS and Forns Index, and second in both Progetto CUORE and FRS, after NFS [81].

Patented Non-Invasive Tests

This category includes Fibrotest, Hepascore, and the Enhanced Liver Fibrosis (ELF) score. There is less data overall regarding these patented tests in relation to cardiovascular risk and outcomes, and more research is needed to elucidate their role. However, in patients with T2D, high Fibrotest scores were associated with increased rates of MI, unstable angina, coronary revascularization, and CVA. Additionally, high Fibrotest scores were associated with increased rates of these same cardiovascular events in patients with high FRS (> 20%) [86].

Discussion

Given that CVD is the leading cause of death globally [87], the importance of cardiovascular risk prediction is difficult to overstate. This importance is compounded by the increasing burden of metabolic disease in younger adults [88••]. The first widely used cardiovascular risk equation was the FRS based on the Framingham Heart Study (FHS). Other popular cardiovascular risk equations include the AHA/ACC PCE, QRISK, and SCORE. These equations are used worldwide but come with significant limitations [89]. Thus, improving cardiovascular risk estimation is an area of active research [90]. The 2018 Guideline on the Management of Blood Cholesterol introduced risk-enhancing factors, which are meant to augment clinical decision making for patients with intermediate PCE risk (10-year ASCVD risk ≥ 7.5% but < 20%). These factors are family history of premature ASCVD, low-density lipoprotein cholesterol (LDL-C) ≥ 160 mg/dL, metabolic syndrome, history of pre-eclampsia or premature menopause, chronic inflammatory disorders, high-risk ethnic background, chronic kidney disease (CKD), TG ≥ 175 mg/dL, apolipoprotein B (ApoB) ≥ 130 mg/dL, ankle-brachial index (ABI) < 0.9, and lipoprotein (a) [Lp(a)] ≥ 50 mg/dL. Interestingly, hs-CRP ≥ 2 mg/dL is also considered a risk-enhancing factor for ASCVD [91]. However, NAFLD is not currently a part of this list. Including NAFLD as a risk-enhancing factor would further support clinicians in their care for intermediate-risk patients.

Current risk estimators include traditional cardiovascular metrics such as blood pressure and serum cholesterol levels to estimate an individual’s likelihood of developing ASCVD over time. In addition to the consideration of risk-enhancing factors, research has also focused on modifying the equations themselves to yield more accurate results. One of the next steps on this frontier of cardiometabolic science is to incorporate NAFLD biomarkers as variables in CVD risk prediction equations. The hypothesis is that the inclusion of such biomarkers will improve the accuracy of current risk prediction models and could play a key role in the development of new risk prediction algorithms. In the meantime, the connection between CVD and NAFLD requires more scrutiny. Some groups find no correlation after adjusting their models for shared risk factors such as DM2 and obesity [92, 93]. Nonetheless, epidemiological evidence is strong, and it is supported by data from animal models and Mendelian randomization [8–10, 13, 94].

The overarching goal of this research is to discover new risk stratification tools and therapeutics that improve patient outcomes. An exciting possibility of this framework is the management of CVD risk through a NAFLD lens. The first step is evaluating if screening for CVD in patients with NAFLD is beneficial. As of right now, there are guidelines for CVD screening in liver transplant candidates [95], but no guidance on screening patients with NAFLD alone. One potential demographic to benefit from screening would be non-obese NAFLD patients, given that the lean subtype of NAFLD may confer greater cardiovascular risk [25]. The next step is drug discovery. There are several agents, including pioglitazone, sodium-glucose cotransporter-2 (SGLT2) inhibitors, and glucagon-like peptide 1 receptor agonists (GLP1R agonists), that have benefits in both NAFLD and CVD [94, 96]. However, there are no pharmaceuticals for CVD explicitly based on the heart-liver axis. Given that the current understanding of this axis is rudimentary, plausible pharmacologic targets may arise as knowledge expands. Additionally, as new medications are identified for NAFLD and CVD, close attention should be given to secondary cardiovascular and liver outcomes respectively. An FGF-21 analogue, pegozafermin, recently advanced to a phase 3 clinical trial after showing improved fibrosis in NASH patients, although any potential cardiovascular effects are currently unknown [97]. Bempedoic acid, recently evaluated in the CLEAR Outcomes trial for reducing cardiovascular risk in statin-intolerant patients, is another emerging NAFLD treatment that should be investigated further [98]. Amidst all this research and development, we must not forget the fundamental role of lifestyle modification. Countless resources have been spent investigating these two diseases. Despite these efforts, weight loss, dietary changes, and exercise remain some of the most powerful, cost-effective modalities for the treatment and prevention of NAFLD and CVD [99, 100].

Conclusion

There is mounting evidence for an independent link between NAFLD and CVD; however, more research is needed to understand if this is simply an epiphenomenon of shared etiological processes. Interestingly, many biomarkers originally identified for hepatic steatosis, NASH, and NASH-related fibrosis correlate with poor cardiovascular outcomes and cardiovascular risk equations. These biomarkers may play an important role in improving current cardiovascular risk models and creating novel models. Given these developments, NAFLD should be considered for addition to the list of ASCVD risk-enhancing factors described in the 2018 AHA/ACC cholesterol guidelines.