Non-alcoholic fatty liver disease, diastolic dysfunction, and impaired myocardial glucose uptake in patients with type 2 diabetes

Non-alcoholic fatty liver disease (NAFLD) is highly prevalent in patients with type 2 diabetes and is associated with cardiovascular risk. We investigated whether the degree of NAFLD was associated with myocardial dysfunction related to impaired myocardial glucose uptake in patients with type 2 diabetes. Methods In total, 131 patients with type 2 diabetes from a tertiary care hospital were included. Myocardial glucose uptake was assessed using [ 18 F]-uorodeoxyglucose-positron emission tomography. Hepatic steatosis and brosis were determined using transient liver elastography. Echocardiography was performed to evaluate cardiac structure and function.


Introduction
Nonalcoholic fatty liver disease (NAFLD) is currently the most prevailing cause of chronic liver disease worldwide, with a global prevalence of 25.2%. [1] NAFLD is a disease characterized by hepatic steatosis, assessed either by imaging or histology, and a lack of secondary causes of hepatic fat accumulation. [2] Type 2 diabetes is an important risk factor for NAFLD. [1] The overall prevalence of NAFLD in patients with type 2 diabetes is 55.5%, more than two times higher than that in the general population. [1] The association between NAFLD and the risk of cardiovascular mortality is supported by su cien evidence, [3,4] and NAFLD was associated with a two-fold increased incidence of cardiovascular events in patients with type 2 diabetes. [5] In addion to the overt cardiovascular events, NAFLD is associated with subclinical myocardial remodeling and dysfunction. [6,7] Increased left ventricular (LV) volume and diastolic dysfunction have been reported as echocardiographic characteristics related to NAFLD, [6][7][8][9] and altered myocardial insulin resistance has been suggested as a potential mechanism linking NAFLD and cardiac abnormalities. [6,10] The myocardium has a exible metabolic network involving diverse energy substrates such as free fatty acids, glucose, and lactate. [11] In response to increased energy demands, the heart shifts from using fatty acids as the energetic substrate to glucose, and insulin enhances myocardial glucose uptake and metabolism. [11] Myocardial insulin resistance is characterized by the reduced availability of glucose transporters and consequently, decreased glucose uptake. [10,12] This cardiac metabolic switch from glucose metabolism to fatty acid oxidation impairs cardiac e ciency, resulting in heart failure. [13] To evaluate the alterations in myocardial energy substrate uptake, [ 18 F]-uorodeoxyglucose-positron emission tomography ( 18 FDG-PET) is effectively used as a noninvasive method. [14] However, study including subjects con ned with patients with type 2 diabetes that simultaneously evaluated the patients' degree of NAFLD, cardiac structure and function, and myocardial glucose uptake has not yet been conducted. Thus, the mechanistic linkage between NAFLD and myocardial dysfunction related to myocardial insulin resistance in those with type 2 diabetes has not been fully evaluated.
This study investigated whether the degree of hepatic steatosis and brosis was independently associated with myocardial dysfunction related to impaired myocardial glucose uptake in patients with type 2 diabetes using liver FibroScan, echocardiography, and 18 FDG-PET.

Study design and subjects
This study included 171 asymptomatic patients with type 2 diabetes who visited the health promotion center in Severance Hospital, a university-a liated tertiary care hospital from March 2010 to November 2018. Subjects with type 2 diabetes met at least one of the following criteria: (i) fasting glucose level ≥ 126 mg/dL, (ii) postprandial glucose level ≥ 200 mg/dL, (iii) hemoglobin A1c (HbA1c) level ≥ 6.5%, (iv) previous diagnosis by a clinical physician, or (v) use of any antidiabetic medication. Among the 171 patients, we excluded individuals with the following characteristics: (i) a history of heavy alcohol consumption (n = 14), (ii) a history of cardiovascular diseases including coronary artery disease and heart failure (n = 24), or (iii) a history of viral hepatitis B or C (n = 5). Finally, a total of 131 subjects were included. This study was approved by the independent institutional review board of Severance Hospital, Seoul, Korea (4-2017-1082); the requirement of informed consent and was waived. This study adhered to the tenets of the Declaration of Helsinki.
60% and the ratio of IQR to median LSM was ≤ 30%. The CAP value was considered to be valid only when the LSM was reliable for the same signal, at the same volume of the liver parenchyma.
[ 18 F]-uorodeoxyglucose-positron emission tomography and image analysis Imaging protocols of 18 FDG-PET were described in detail previously. [15] Whole-body PET-computed tomography (CT) was performed using either one of the two combination PET-CT scanners: a Biograph TruePoint 40 (Siemens Medical Solutions, Hoffman Estates, IL, USA) or a Discovery 600 (General Electric Medical Systems, Milwaukee, WI, USA). After the subjects had fasted for at least 8 h, blood glucose levels were measured before administering 18 FDG intravenously (approximately 5.5 Bq of 18 FDG per kilogram of body weight). At 60 min after the injection, PET-CT was performed from the skull base to the mid-thigh.
Following CT, PET was performed as follows: 2.5 min per bed position of 21.6 m in a three dimensional acquisition mode (Biograph TruePoint 40) or 2 min per bed position of 15.7 m in a three-dimensional acquisition mode (Discovery 600). The CT images were rebuilt using a 512 × 512 matrix and were converted into 511 eV equivalent attenuation factors for attenuation correction. Reconstructed PET images were obtained using a 128 × 128 matrix with ordered subset expectation maximization and correction for attenuation.
Experts in nuclear medicine calculated the standardized uptake value (SUV) as follows: SUV = (decaycorrected activity [kBq] per ml of tissue volume)/(injected 18 FDG activity [kBq]/body mass [g]) in a clinical data-blinded manner. Measurements of SUV max of the myocardium and the SUV mean of the liver were drawn from multiple regions of interest (ROIs) for a semi-quantitative analysis. Two-dimensional ROIs were drawn through the transaxial images to calculate the SUV max of the LV myocardium within an inner edge. We measured the SUV of the liver from the circular ROI along the periphery of the right lobe, 1 m from the margin. In FDG-PET for detecting malignancies, the liver has been considered as an internal standard to grade the FDG uptake of whole-body lesions, as the SUV of the liver remains stable over time when measuring a mean uptake in the right lobe, even in patients with diffuse fatty liver disease. [23,24] To minimize variability, the heart SUV to the liver FDG uptake ratio (SUV heart/SUV liver) was used to evaluate myocardial glucose uptake. [24,25] The low myocardial glucose uptake group comprised the subjects in the lowest quartile of myocardial glucose uptake.

Statistical analyses
Data are presented as means with standard deviations for normally distributed continuous variables, medians with IQRs for non-normally distributed continuous variables, and as numbers with percentages for categorical variables. The Student's t-test and Mann Whitney U test were used for comparisons of normally and non-normally distributed continuous variables, respectively. Comparisons of categorical variables were conducted using the χ 2 test or Fisher's exact test. Pearson's correlation was performed; Pearson's correlation coe cients (r) were presented to evaluate the correlations between parameters.
Multivariate linear regression models were used to determine the independent determinant factors for a higher E/e' ratio. P-values < 0.05 were considered statistically signi cant. All statistical analyses were performed using the Statistical Package for the Social Sciences software version 23.0 for Windows (International Business Machines Corp., Armonk, NY, USA).

Results
Baseline and echocardiographic characteristics according to the presence of non-alcoholic fatty liver disease (NAFLD) In total, 131 patients with type 2 diabetes (83 patients in the NAFLD group and 48 patients in the no-NAFLD group) were included. Clinical characteristics of the study subjects are listed in Table 1. The mean age of overall study subjects was 60.8 years. Of the total subjects, 77 (55.7%) were male and the sex ratio was not different according to the presence of NAFLD. Patients in the NAFLD group had a signi cantly higher BMI, waist circumference, and hip circumference (all p < 0.001) and were more likely to have hypertension (p = 0.010) than patients in the no-NAFLD group. Patients in the NAFLD group had signi cantly higher fasting glucose and homeostatic model assessment (HOMA) of insulin resistance levels than patients in the no-NAFLD group (all p < 0.001). HbA1c and HOMA of β-cell function levels were similar between the two groups. Serum fasting and postprandial triglyceride levels were higher and serum high-density lipoprotein cholesterol and lipoprotein (a) levels were lower in the NAFLD group than in the no-NAFLD group. Serum liver enzymes and gamma-glutamyl transferase levels were markedly elevated in the NAFLD group. Insulin use was more prevalent in the NAFLD group (p = 0.025) than in the no-NAFLD group, whereas the use of other antidiabetic medications was not different between the two groups. The degree of hepatic steatosis (CAP) and liver stiffness measured by FibroScan were signi cantly associated with BMI, fasting glucose, and serum liver enzyme levels (Supplemental Table S1). Patients with NAFLD exhibited myocardial remodeling and dysfunction (Table 2). LV mass, LV endsystolic and end-diastolic diameters (mm), and LA volume and LA volume index were signi cantly increased in the NAFLD group compared with the no-NAFLD group (all p < 0.05). Furthermore, patients in the NAFLD group revealed worse echocardiographic parameters associated with myocardial diastolic dysfunction. Septal tissue Doppler e' velocity was signi cantly decreased and E/e' ratio was signi cantly increased in the patients with NAFLD (all p < 0.05, Fig. 1A).  Association between the degree of hepatic steatosis and brosis degree and myocardial diastolic dysfunction and myocardial insulin resistance Both hepatic steatosis and brosis were associated with myocardial diastolic dysfunction. The proportion of higher E/e' ratio (above the median) signi cantly increased with an increasing degree of hepatic steatosis based on CAP score categories (p for trend = 0.001, Fig. 1B). Subjects with brosis stage ≥ 2 had a signi cantly increased E/e' ratio (p < 0.01, Fig. 1C); the proportion of higher E/e' ratio also increased signi cantly with an increasing degree of hepatic brosis (p for trend = 0.006, Fig. 1D). Both hepatic steatosis (CAP score, dB/m) and brosis (stiffness, kPa) were positively correlated with E/e' ratio ( Fig. 1E and 1F). Hepatic steatosis was additionally associated with LV mass, LA volume, LA volume index, and septal tissue Doppler e' velocity (all p < 0.05, Supplemental Table S2). Furthermore, even after adjustment for clinical confounding factors, hepatic steatosis and brosis were independent determinant factors for a higher E/e' ratio in multivariate linear regression analyses (r 2 = 0.409, p = 0.041 for steatosis and r 2 = 0.423, p = 0.009 for brosis, Table 3). Association between myocardial glucose uptake and NAFLD and myocardial diastolic function Subjects with NAFLD demonstrated a signi cantly decreased myocardial glucose uptake (p = 0.018, Fig. 2A). The proportion of lower myocardial glucose uptake (above the median) tended to increase with an increasing degree of hepatic steatosis (p for trend = 0.084, Fig. 2B). In addition, myocardial glucose uptake was signi cantly decreased in subjects with brosis stage ≥ 3, and lower myocardial glucose uptake was more likely to be observed in those with advanced brosis stages (p for trend = 0.012, Fig. 2D).
Lower myocardial glucose uptake was also closely associated with myocardial diastolic dysfunction. Subjects with myocardial diastolic dysfunction presenting with a high E/e' ratio showed a marginal signi cant decrease in myocardial glucose uptake (p = 0.058, Fig. 2E and 2F). Myocardial glucose uptake decreased as E/e' increased (r = 0.185, p = 0.037, Fig. 2G). Decreased myocardial glucose uptake was still an independent determinant factor for a higher E/e' ratio in patients with type 2 diabetes after adjustment for potential confounding factors (r 2 = 0.409, p = 0.040, Table 4).

Discussion
The present study demonstrated that the presence of NAFLD was associated with cardiac remodeling and myocardial diastolic dysfunction, and the degrees of hepatic steatosis and brosis were determinant factors for diastolic dysfunction in patients with type 2 diabetes. In addtion, impaired myocardial glucose uptake, examined using 18 FDG-PET, was observed in subjects with a higher degree of hepatic steatosis and brosis and in subjects with diastolic dysfunction. This indicated that myocardial insulin resistance, presenting as impaired myocardial glucose uptake, potentially mediates the pathophysiological association between NAFLD and diastolic dysfunction in those with type 2 diabetes (Fig. 3).
Diabetes mellitus involves cardiovascular complications such as coronary atherosclerosis, but it can also affect cardiac structure and function in the absence of coronary artery disease; this condition is called diabetic cardiomyopathy. [26] Diabetic cardiomyopathy has been de ned as ventricular dysfunction that develops independent of coronary artery disease and hypertension. [26] Diabetic cardiomyopathy is characterized by diastolic dysfunction, which may precede the development of systolic dysfunction. [27] The prevalence of diastolic dysfunction is up to 30% in patients with uncomplicated type 2 diabetes. [28,29] Several mechanistic associations have been suggested, including autonomic dysfunction, abnormalities in ion homeostasis, alteration in structural proteins, and interstitial brosis. [26] These pathogenic mechanisms could be mainly associated with insulin resistance, [30] but sustained hyperglycemia also may result in myocardial stiffness and contractile dysfunction. [26,31] Subclinical myocardial remodeling and diastolic dysfunction are also associated with NAFLD, independent of diabetes. [6,8,9,32] NAFLD was associated with a 29%-increase in diastolic dysfunction risk compared with that in controls. [32] Diastolic dysfunction risk increases signi cantly according to the grades of steatosis and brosis. [6,32] However, limited studies have investigated the independent association between the degree of NAFLD and diastolic dysfunction in subjects con ned to patients with type 2 diabetes. [7,33] In addition, previous evidence supporting that myocardial insulin resistance would be the link in the pathological association between NAFLD and myocardial dysfunction in patients with type 2 diabetes is insu cient. Myocardial insulin resistance is not always accompanied by systemic insulin resistance, and it is considered to develop independent of systemic insulin resistance and hyperglycemia severity. [13,34,35] Thus, to investigate the contributions of myocardial insulin resistance to the association between NAFLD and myocardial dysfunction, cardiac glucose metabolism has to be concomitantly estimated while measuring the degree of NAFLD and cardiac dysfunction. In this study, we examined myocardial glucose uptake using 18 FDG-PET, and we could demonstrate the association between myocardial glucose uptake, hepatic steatosis and brosis, and diastolic function.
The independent association between NAFLD and diastolic dysfunction in patients with type 2 diabetes with underlying systemic insulin resistance and hyperglycemia is noteworthy. Hepatic insulin resistance could be one of the mechanisms that explain this independent association between the liver and heart in type 2 diabetes. Insulin resistance could be either the whole-body or central (hepatic), and hepetic insulin resistance is distinguished from systemic/peripheral insulin resistance which is a major factor in the pathogenesis of type 2 diabetes. [36][37][38] Patients with NAFLD almost universally have hepatic insulin resistance, and there are comparative differences in the hepatic and whole-body insulin resistance along the spectrum of NAFLD. [39,40] Hepatic steatosis with hepatic insulin resistance correlates with epicardial fat volume, which is an independent predictor of diastolic dysfunction. [39,41,42] Hepatic insulin resistance also leads to the overproduction of very-low-density lipoproteins and glucose, and increased lipid accumulation and hyperglycemia-induced oxidation in myocardial cells induce myocardial remodeling, diastolic dysfunction and myocardial insulin resistance. [13,43] Fibroblast growth factor 21 (FGF21), a protein synthetized by the liver, originally improves hepatic insulin sensitivity, but some investigators have suggested that FGF21 resistance observed in NAFLD could be associated with cardiac damage in patients with NAFLD. [44,45] FGF21 resistance may contribute to myocardial insulin resistance due to altered lipid homeostasis in cardiomyocytes. [13,46] In addition, the role of the liver as a generator of circulating mediators that affect cardiac remodeling has been hypothesized. [45] The production of pro-in ammatory and pro-atherogenic cytokines (e.g., interleukin 6, interleukin 12, tumor necrosis factor-alpha, etc.) is increased in patients with NAFLD, and these compounds could be involved in cardiac morphology and dynamics as well as myocardial insulin resistance. [13,45] Several dysregulated hepatokines (e.g., fetuin A, eukocyte cell-derived chemotaxin-2, retinol binding protein, etc.) might also promote in ammatory pathways and cardiac dysfunction. [7] Thus, the superimposed hepatic insulin resistance and in ammation due to the presence of NAFLD in patients with type 2 diabetes might have increased the risk for myocaridal dysfunction.
The clinical relevance of the present study is attributed to several strengths. First, this study was restricted to patients with type 2 diabetes to investigate the independent assocation between NAFLD and myocardial dysfunction, even in the presence of hyperglycemia and systemic insulin resistance. Second, myocardial glucose uptake, re ecting myocardial insulin resistance, was estimated using 18 FDG-PET.
Two previous studies have demonstrated the relation between NAFLD and diastolic dysfunction in patients with type 2 diabetes. [7,33] However, neither of these studies investigated myocardial glucose uptake. To the best of our knowledge, this is the rst study to suggest the potential that myocardial insulin resistance would pathophysiologically mediate the association between NAFLD and myocardial dysfunction in patients with type 2 diabetes by measuring myocardial glucose uptake using 18 FDG-PET.
The present study has some limitations. First, a causal relationship between NAFLD and myocardial dysfunction could not be drawn due to the cross-sectional design of this study. Second, this study was conducted in a single center with a limited number of subjects. A multicenter study should be conducted with a larger number of subjects to generate more confound evidence. Third, subjects were in a fasting condition when myocardial glucose uptake was assessed by 18

Conclusions
In conclusion, hepatic steatosis and brosis were associated with cardiac remodeling and diastolic dysfunction in patients with type 2 diabetes. Myocardial glucose uptake signi cantly decreased according to the degree of hepatic steatosis and brosis along with diastolic dysfunction, suggesting that myocardial insulin resistance could mechanistically mediate the association between NAFLD and diastolic dysfunction in patients with type 2 diabetes. Multicenter longitudinal studies with comprehensive measurements of the hepatic and cardiac parameters, and myocardial insulin resistance are required to con rm the pathophysiological interorgan relationship between the liver and heart in patients with type 2 diabetes. Hepatic steatosis and brosis associated with myocardial diastolic dysfunction (A) E/e' ratio according to the presence of non-alcoholic fatty liver disease (B) Increasing percentage of higher E/e' ratio according to the steatosis stage (C) E/e' ratio according to the presence of moderate to severe brosis ( brosis stage ≥2) (D) Increasing percentage of higher E/e' ratio according to the brosis stage (E) Association between steatosis (controlled attenuation parameter score, dB/m) and E/e' ratio (F) Association between liver stiffness (kPa) and E/e' ratio Associations between NAFLD, diastolic dysfunction, and impaired myocardial glucose uptake in patients with type 2 diabetes

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplementaryInformation.docx