|
The Interplay Between Nonalcoholic Fatty Liver Disease and Atherosclerotic Heart Disease - Editorial
|
|
|
Download the PDF here  
Download the PDF here  
Hepatology April 2019  
Nonalcoholic fatty liver disease (NAFLD), the most common form of chronic liver disease, exists in two predominant histological subtypes: nonalcoholic fatty liver and nonalcoholic steatohepatitis (NASH).1 NASH is the clinically aggressive variant with higher risk of fibrosis progression and mortality. The leading cause of mortality in patients with NAFLD is cardiovascular disease (CVD),2 which is linked to diagnosis of NASH and severity of hepatic fibrosis.3 Although NAFLD is closely associated with cardiometabolic risk factors such as diabetes, obesity, hypertension, and dyslipidemia, the association between CVD and NAFLD may be independent of these risk factors. The intimate relationship between the liver and coronary heart disease likely stems from the central role the liver plays in glucose and lipid metabolism. Development of NAFLD is associated with increased production and secretion of large triglyceride-laden very low-density lipoprotein (VLDL) particles from the liver.4 In circulation, VLDL particles are slowly metabolized and are subject to an exchange process that removes cholesteryl ester from the particle core, replacing it with triacylglycerol, which leads to the formation of highly atherogenic small dense low-density lipoprotein particles (Fig. 1).4 Hepatic production of proinflammatory factors and vasoactive and thrombogenic molecules further contributes to CVD in patients with NAFLD. These putative biological mechanisms result in impaired endothelial function, formation of vulnerable coronary plaque, reduction in coronary artery flow reserve, and poor coronary artery collateral formation with maladaptive response to coronary artery ischemia.5-8 Clinically, these mechanisms culminate into clinically significant cardiovascular outcomes such as myocardial infarction, stroke, and cardiovascular death.9 This intimate relationship between dyslipidemia and NASH is highlighted in post hoc analysis of the PIVENS trial that demonstrated that resolution of NASH improved dyslipidemia.10 Finally, in patients with NAFLD, treatment of dyslipidemiawith statin therapy is associated with substantial risk reduction of cardiovascular events.11  
![TG](../images/041019/041019-3/TG.gif)
The study by Dr. Pais and colleagues in the current issue of Hepatology builds on the published literature and further explores the relationship between early atherosclerosis and steatosis.12 The major limitation of the study is the lack of robust assessment of histological parameters (i.e., steatosis, fibrosis, diagnosis of NASH) and cardiovascular outcomes; however, it does provide insight into systemic atherosclerosis and NAFLD. Using fatty liver index (FLI) as a surrogate for hepatic steatosis, the authors assert that steatosis was associated with carotid and coronary artery atherosclerosis as measured by carotid duplex and multidetector computed tomography. Although the association between carotid plaque, coronary artery calcification, and hepatic steatosis is not new,13, 14 the reported link between multisite atherosclerosis is novel but intuitive. An interesting finding is the quantification of the contribution of hepatic steatosis to patients at intermediate to high risk of having a coronary heart disease event based on Framingham risk score ≥10%. Although the findings of the present study are intriguing, they need to be interpreted in the context of its limitation. First, FLI does not correlate with severity of hepatic steatosis, and therefore, the reported association between the degree of hepatic steatosis and atherosclerosis is nuanced. Due to the cross-sectional nature of the study, it is unclear how atherosclerosis disease at various sites predicts the likelihood of having a coronary heart disease–related event in the future. The clinical implications of the current study are not clear because the findings reported are not linked to clinically significant outcomes. Finally, the effect size of adding hepatic steatosis to identify patients at intermediate to high risk through Framingham risk calculation was relatively modest.  
The study by Pais et al. provides valuable insight into NAFLD and early atherosclerosis and highlights key limitations within the field. It is currently unclear if the relationship between NAFLD and atherosclerosis is due to shared risk factors such as adiposopathy, insulin resistance, and dyslipidemia or if it represents an epiphenomenon. Also, the ideal tool to clinically risk-stratify cardiovascular risk in patients with NAFLD is unknown as all of the present CVD risk assessment tools do not account for NAFLD as a potential risk factor. Although the study notes that the addition of NAFLD to traditional CVD risk parameters improves risk assessment, well-designed prospective studies in patients with histologically proven NAFLD with a priori defined clinical outcomes are necessary to translate the published findings to clinical practice.
In conclusion, multiple studies have reported a close association between NAFLD and atherosclerosis; however, the optimal means of risk-stratifying and reducing CVD morbidity and mortality in patients with NAFLD remains unknown.  
-------------------------------------  
Original Article  
Relationship Among Fatty Liver, Specific and Multiple-Site Atherosclerosis, and 10-Year Framingham Score  
Raluca Pais,1,2,3 Alban Redheuil,5 Philippe Cluzel,5 Vlad Ratziu,1,3,6* and Philippe Giral3,4*  
Abstract  
Despite a well-documented increase in the prevalence of subclinical atherosclerosis in patients with steatosis, the relationship among steatosis and atherosclerosis, specific atherosclerotic sites, multiple-site atherosclerosis, and cardiovascular risk prediction is incompletely understood. We studied the relationship among steatosis, atherosclerosis site, multiple-site atherosclerosis, coronary artery calcification (CAC), and 10-year Framingham Risk Score (FRS) in 2,554 patients with one or more cardiovascular risk factors (CVRF), free of cardiovascular events and other chronic liver diseases, and drinking less than 50 g alcohol/day.  
All patients underwent arterial ultrasound (carotid [CP] and femoral [FP] plaques defined as intima-media thickness (IMT) > 1.5 mm), coronary computed tomography scan (severe CAC if ≥ 100), 10-year FRS calculation, and steatosis detection by the fatty liver index (FLI, present if score ≥ 60). Patients with steatosis (36% of total) had higher prevalence of CP (50% versus 45%, P = 0.004) and higher CAC (181 ± 423 versus 114 ± 284, P < 0.001) but similar prevalence of FP (53% versus 50%, P = 0.099) than patients without steatosis.  
Steatosis was associated with carotid IMT and CAC, but not with FP, independent of age, diabetes, hypertension, and tobacco use (P < 0.001). Fifty-three percent of patients had at least 2-site atherosclerosis and steatosis was associated with at least 2-site atherosclerosis independent of age and CVRF (odds ratio = 1.21, 95% confidence interval 1.01-1.45, P = 0.035).  
Sixty-four percent of patients with steatosis had a FRS score of 10% or more.  
FLI was associated with FRS beyond the CVRF or the number of atherosclerosis sites (P < 0.001). Adding FLI to CVRF predicted an FRS greater than or equal to 10% better than CVRF alone (area under the receiver operating characteristic curve = 0.848 versus 0.768, P < 0.001).  
Conclusion: Steatosis is associated with carotid and coronary, but not femoral atherosclerosis, and with cardiovascular mortality risk. The multiple-site involvement and quantitative tonic relationship could reinforce the prediction of cardiovascular mortality or events over classical CVRF or imaging-based detection of atherosclerosis.  
Discussion  
In this study we investigated the site-specific relationship between steatosis and subclinical atherosclerosis assessed in the carotid, coronary, and femoral arteries territory. Although steatosis predicted both carotid and coronary atherosclerosis beyond the traditional CVRFs, which is in agreement with previous studies,21-23 there was no significant association between steatosis and FP. Steatosis was associated with FP only in men and this relationship disappeared after controlling for CVRFs. Instead, male sex, active tobacco consumption, and clustering of CVRFs were the strongest predictors for the presence of FP. Therefore, our results suggest that steatosis differentially affects the co-existence of early atherosclerosis at different arterial sites. However, because this is a cross-sectional, retrospective study, further insight into this relationship can only be gained by longitudinal studies testing incident early atherosclerotic lesions at different vascular sites in patients with or without baseline steatosis.  
Other studies have shown that CVRFs do not have the same impact at different atherosclerosis sites. In the Pathological Determination of Atherosclerosis in Youth study, smoking selectively increased 3-fold the risk of atherosclerotic lesions of the abdominal aorta, while not influencing the risk of coronary lesions. Glycated hemoglobin was strongly related to coronary atherosclerosis but not with abdominal aorta atherosclerosis.24 In a large cohort from the Offspring Framingham Heart Study,25 Mellinger et al. demonstrated that steatosis was an independent predictor of coronary but not abdominal aortic calcium. It is currently believed that many factors including genetic background, sex, immune status, oxidative stress, chronic low-grade inflammation, and blood flow parameters interact in different and complex ways to generate an atherosclerotic lesion at a particular site.26 Understanding the determinants of atherosclerosis at different anatomical sites is of clinical relevance, as this will translate into different clinical event rates. For example, carotid atherosclerosis is responsible for 25% of strokes, with an annual risk in asymptomatic patients increasing from 1% to 3% according to the severity of stenosis. Peripheral artery atherosclerosis, although often underdiagnosed, is associated with the highest risk of cardiovascular death and cardiovascular events dues to atherothrombosis.27 In asymptomatic subjects from the Aragon Worker’s Health Study, the prediction of severe coronary atherosclerosis by traditional CVRFs was significantly improved when taking into account the presence of FP.11  
Another finding of this study is that patients with steatosis had a higher prevalence of diffuse atherosclerotic disease, involving multiple sites among carotid, coronary, and femoral atherosclerosis. The proportion of patients with multiple-site atherosclerosis in our study gradually increased across FLI tertiles, suggesting a quantitative relationship, with the amount of steatosis playing a major role in the development of diffuse atherosclerosis. Multiple-site atherosclerosis is a major determinant of clinical events: the 1-year clinical event rate (cardiovascular death, stroke, myocardial infarction) significantly increased with the number of symptomatic arterial disease location, ranging from 2.2% in patients with one territory involved to 9.2% in patients with three territories involved.27 More recently, the Registry Reduction in Atherothrombosis for Continued Health Registry (REACH) also highlighted that patients with multiple-site atherosclerosis have a higher rate of cardiovascular fatal and nonfatal events than patients with just one territory affected.28 Interestingly, retrospective long-term follow-up studies with repeat cardiovascular screening for early atherosclerosis (either carotid or coronary)4, 6 suggest that steatosis predates the occurrence of early atherosclerosis. If this is further confirmed by prospective studies and if a quantitative relationship does indeed exist, then it would be important to determine whether steatosis reversal early in the disease process can prevent the occurrence of early atherosclerosis.  
Finally, we have shown that steatosis is associated with cardiovascular risk beyond the traditional CVRFs. In fact, adding steatosis to traditional CVRFs significantly improves cardiovascular risk prediction particularly for patients with high FRS. This was not the case in the Multi-ethnic Study of Atherosclerosis (MESA). In that study, patients with NAFLD had a higher risk of nonfatal cardiovascular events and overall mortality; however, adding NAFLD to traditional CVRFs did not improve the cardiovascular risk prediction when adjusted for age, sex, ethnicity, and traditional CVRFs.29 These discordant results may arise from differences in the study populations. In the current study, patients were at least 6 years younger than in the MESA study, had at least 1 CVRF with a high proportion of patients with dyslipidemia, and had a higher mean FRS. The different methods used for the diagnosis of steatosis (CT scan in the MESA study, FLI in our study) as well as the different design of the two studies may also account for the discordant results.  
Additionally, we have shown that adding steatosis to multiple-site atherosclerosis predicted FRS better than classical CVRFs. This strengthens observations from the global REACH Registry that patients without established atherothrombosis but with risk factors only had a lower risk of cardiovascular events or death when compared with patients with established atherothrombosis with or without prior clinical cardiovascular events.30 The mechanisms whereby steatosis increases cardiovascular risk and cardiovascular mortality by itself (i.e., in addition to traditional CVRFs) is under investigation. Chronic low-grade inflammation and insulin resistance not only play a pivotal role in the occurrence, progression, and complications of atherosclerosis plaques, but also in the progression of liver damage, and is therefore a shared pathophysiological link between NAFLD and atherosclerosis. Inflammatory gene expression in adipose tissue strongly correlates with the progression of liver damage.31 The concomitant presence of NAFLD and systemic inflammation (as assessed by highly sensitive quantification of C-reactive protein) increased the risk of CAC development over 4 years among 1,500 healthy Korean patients.32 Activation of hepatic nuclear factor kappa B and c-Jun kinase pathways was responsible for an increased production of pro-inflammatory cytokines, which further aggravated insulin resistance and favored the progression of liver disease in addition to promoting accelerated atherosclerosis. Interestingly, despite similar C-IMT, patients with steatosis appeared to have a higher inflammation of the arterial wall compared with patients without steatosis.33 This is of interest because the composition and inflammation of atherosclerosis plaques are better correlated with the risk of future cardiovascular events than the degree of stenosis itself.34  
The strength of the current study is the large sample size, with more than 2,500 patients included and multiple-site atherosclerosis evaluation. A limitation of this study is the use of FLI as a surrogate marker for hepatic steatosis instead of the assessment of steatosis by imaging or histology. This is important because at least three components of the FLI were shown to be associated with cardiovascular disease (BMI, waist circumference, and serum triglycerides), and as such, these variables could account for the associations seen in this report. Moreover, despite the numerical range of the FLI index, its ability to quantify is certainly inferior to that of imaging methods that measure directly the hepatic fat signal, such as MRI-based proton density fat fraction.35 However, the FLI has been validated extensively both in the general population17 and in tertiary care referral centers with a good accuracy for discriminating presence from absence of steatosis defined histologically or by ultrasound.36 Moreover, there are data demonstrating that FLI predicts steatosis on liver biopsy better its components.36 Ideally, this study should have incorporated an independent assessment of steatosis; unfortunately, this cohort was not initially designed for the study of liver outcomes. Interestingly, all serum-based biomarker panels proposed so far and used in patients seen in clinics are based on one or several biochemical variables in relation to the metabolic syndrome.17, 36-38 Therefore, these results should be interpreted with caution as they are suggestive but fall short of a definitive demonstration of the association between steatosis and early atherosclerotic lesions.  
It remains to be determined whether there is clinical relevance to the incremental diagnostic gain of adding a surrogate marker of hepatic steatosis to traditional CVRFs for the diagnosis of early atherosclerotic lesions. The improvement in the AUROC reported here is not trivial given the insensitivity of the AUROC statistic to improved prediction by new markers39, 40 and the fact that a 0.85 value is in the range of accepted diagnostic methods with good discriminative value.41 However, our results need to be tested in independent data sets. Ultimately, prospective studies should confirm an increase in predictive value, and cost-efficacy should be demonstrated before the assessment of steatosis can be used for prediction in clinical practice.42  
In conclusion, we have shown that steatosis differently affects early atherosclerosis development depending on the arterial site involved, and found a quantitative relationship between the amount of steatosis and the severity of early atherosclerosis. Steatosis status reinforces the prediction of cardiovascular mortality over traditional CVRFs and over imaging-based measurements of atherosclerotic lesions. The detection of steatosis has important prognostic implications in patients with CVRFs. Conversely, in patients seen primarily for hepatic steatosis, a thorough cardiovascular evaluation is necessary, considering the increased cardiovascular and atherosclerotic risk conferred upon by NAFLD.
| |
|
|
|
|
|