Association of Fatty Liver Index with Incident Diabetes Risk in Patients Initiating Statin–Therapy: A 6-Year Retrospective Study

Background: Statins are associated with new-onset type 2 diabetes (T2D), mainly in patients with metabolic syndrome (MetS). The fatty liver index (FLI) is used as a prognostic score for the diagnosis of non-alcoholic fatty liver disease (NAFLD), which is common in patients with MetS. We aimed to investigate the association of FLI with new-onset T2D in patients initiating statin therapy. Methods: A retrospective observational study including 1241 individuals with dyslipidemia and followed up for ≥3 years. Patients with T2D and those receiving lipid-lowering treatment at the baseline visit were excluded. Models with clinical and laboratory parameters were used to assess the association of FLI with incident T2D. Results: Among the 882 eligible subjects, 11% developed T2D during the follow-up (6 years; IQR: 4–10 years). After adjusting for sex, age and MetS parameters, a multivariate analysis revealed that age (HR:1.05; 95%CI: 1.01–1.09, p < 0.05), fasting plasma glucose (HR: 1.09; 95%CI: 1.06–1.13, p < 0.001) and FLI (HR: 1.02; 95%CI: 1.01–1.04, p < 0.01) were independently associated with T2D risk. The subjects with probable NAFLD (FLI ≥ 60) had a three-fold increased T2D risk compared with the subjects with FLI < 60 (HR: 3.14; 95%CI: 1.50–6.59, p = 0.001). A ROC curve analysis showed that FLI had a significant, although poor, predictive value for assessing T2D risk (C-Statistic: 0.67; 95%CI: 0.58–0.77, p = 0.001). Higher FLI values were associated with reduced T2D-free survival (log-rank = 15.46, p < 0.001). Conclusions: FLI is significantly and independently associated with new-onset T2D risk in patients initiating statin therapy.


Introduction
Non-alcoholic fatty liver disease (NAFLD) is the term used for a range of conditions caused by fat accumulation in the liver in the absence of increased alcohol consumption [1][2][3]. NAFLD is the most common chronic liver disease, with a worldwide prevalence of 32.4% (95% confidence intervals, CI: 29.9-34.9%) [4]. Apart from being the major cause of cirrhosis in the developed countries, the accumulating evidence suggests that NAFLD is the hepatic component of metabolic syndrome (MetS) and is associated with new-onset type 2 diabetes (T2D) and atherosclerotic cardiovascular disease (ASCVD) [4][5][6][7][8]. NAFLD and T2D share common pathophysiological mechanisms, with insulin resistance being the primary one [9]. It is well established that metabolic hepatic steatosis is a phenotype that includes diverse and complex causes, as well as demonstrating a wide spectrum of clinical severity, but also a considerable inter-patient variability [9]. In this context, "MAFLD" (metabolic-dysfunctionassociated fatty liver disease) is preferred over ''NAFLD" as a more appropriate general term [9]. Although liver biopsy remains the gold standard, various prognostic models have been developed for NAFLD diagnosis, such as the fatty liver index (FLI) [10]. On the other hand, statins are the cornerstone therapy in ASCVD prevention and might have a role in the Diagnostics 2023, 13, 503 2 of 9 treatment of NAFLD [11][12][13]. However, statins have been associated with increased T2D risk, mostly in individuals with MetS components, such as prediabetes, obesity, hypertension, and increased triglycerides [14][15][16][17][18]. Thus, there is a need for identifying biomarkers predicting statin-induced T2D.
In this context, the aim of the present study was to investigate the association of FLI with new-onset T2D in patients initiating statin therapy and followed up for ≥3 years.

Materials and Methods
As previously described, this was a retrospective study including adult patients with dyslipidemia who attended the Outpatient Lipid Clinic of the University Hospital of Ioannina in Greece for ≥3 years [18][19][20][21]. The study protocol was approved by the local institutional ethics committee (No. 3231, 5 July 2012) and informed consent was obtained from each patient.
All the study participants were Caucasians. Lipid-lowering drug-naïve patients with dyslipidemia were considered eligible for this study. The patients with T2D at baseline and those not prescribed with statin therapy during the follow-up were excluded. A complete assessment of the clinical and laboratory profile was performed at the baseline visit by a physician, including: (i) sex and age, (ii) body mass index (BMI) and waist circumference, (iii) blood pressure, (iv) fasting plasma glucose (FPG) levels, (v) lipid profile, including total cholesterol (TCHOL), triglycerides (TG), high-(HDL-C) and lowdensity lipoprotein cholesterol (LDL-C), as well as (vi) liver enzymes, including aspartate aminotransferase (AST), alanine aminotransferase (ALT), gamma-glutamyl transferase (gGT), alkaline phosphatase (ALP) and total and direct bilirubin. The insulin levels were available in only few patients. The office blood pressure measurements were performed with a validated upper-arm cuff BP measurement device and an appropriate cuff size. The corresponding instruments measuring the subjects' waist, height and weight were validated and calibrated. The waist circumference was measured by placing a tape measure mid-way between the lower border of the costal margin and the top border of the iliac crest. BMI was calculated as: (weight, kg)/(height, m) 2 . During the study conduct, the glycated hemoglobin (HbA1c) was not routinely measured before the diagnosis of T2D, according to national guidelines [22].

Statistical Analysis
The continuous variables were tested for normality by the Kolmogorov-Smirnov test. The data are presented as the mean ± standard deviation (SD) and median [interquartile range (IQR)] for the parametric and non-parametric data, respectively. For the categorical values, frequency counts and percentages were applied. A chi-square test was performed for the interactions between the categorical values. An independent sample t-test (parametric and non-parametric) was used for the comparison of the continuous numeric values between two groups. A univariate Cox regression analysis was performed to investigate the association of a factor with the investigated outcome of interest during the follow-up. A multivariate logistic regression analysis was conducted using the variables that were statistically significant in the univariate analyses (the backward conditional method was used). The associations with the outcomes of interest are expressed as hazard ratios (HR), with an accompanying 95% confidence interval (CI). A receiver-operating characteristics (ROC) curve analysis was used to analyze the prognostic value of the investigated variables for the corresponding outcomes of interest. The C-statistic (area under the curve) is presented as a unified estimate of sensitivity and specificity. A Kaplan-Meier analysis was performed to study the differences in event-free survival from the investigated outcomes of interest. Two-tailed significance was defined as p < 0.05. The analyses were performed with the Statistical Package for Social Sciences (SPSS) v25.0 software (SPSS Statistics for Windows, Version 28.0. Armonk, New York, NY, USA: IBM Corp).
The data are presented as the mean ± standard deviation (SD) and median [interquartile range (IQR)] for the parametric and non-parametric data, respectively. For the categorical values, frequency counts and percentages were applied. A chi-square test was performed for the interactions between the categorical values. An independent sample t-test (parametric and non-parametric) was used for the comparison of the continuous numeric values between two groups. A univariate Cox regression analysis was performed to investigate the association of a factor with the investigated outcome of interest during the followup. A multivariate logistic regression analysis was conducted using the variables that were statistically significant in the univariate analyses (the backward conditional method was used). The associations with the outcomes of interest are expressed as hazard ratios (HR), with an accompanying 95% confidence interval (CI). A receiver-operating characteristics (ROC) curve analysis was used to analyze the prognostic value of the investigated variables for the corresponding outcomes of interest. The C-statistic (area under the curve) is presented as a unified estimate of sensitivity and specificity. A Kaplan-Meier analysis was performed to study the differences in event-free survival from the investigated outcomes of interest. Two-tailed significance was defined as p <0.05. The analyses were performed with the Statistical Package for Social Sciences (SPSS) v25.0 software (SPSS Statistics for Windows, Version 28.0. Armonk, New York, NY, USA: IBM Corp).

Discussion
To the best of our knowledge, this is the first study to investigate the association between FLI and incident T2D risk in patients initiating statin therapy. The FLI was significantly and independently associated with the risk of new-onset T2D and showed a significant predictive value for the development of T2D in patients initiating statin therapy.
NAFLD is an umbrella term that encompasses a disease continuum ranging from steatosis to steatohepatitis fibrosis and cirrhosis [26]. NAFLD has been associated with AS-CVD, MetS, T2D, obesity, atherogenic dyslipidemia and hypertension [4,[27][28][29]. Genetic factors, including a variant of the PNPLA3 gene, epigenetic mechanisms and gut microbiome, may also play a role in NAFLD pathogenesis [30,31]. A liver biopsy remains the gold standard for NAFLD diagnosis, but it remains an invasive method with complications [1,32]. Thus, liver ultrasonography and biomarkers, such as the FLI, SteatoTest and NAFLD fat score, are commonly used [1].
Our results are in agreement with past evidence indicating an association between the FLI and risk of incident T2D in the general population [33][34][35][36][37]. A population-based cohort study conducted in Korea and including 5,254,786 young adults aged 20-39 years demonstrated that the risk of new T2D was higher in those with FLI ≥ 60 than the control group (adjusted HR: 4.97; 95% CI: 4.90-5.05) [33]. Likewise, a prospective study of 16,648 participants with prediabetes demonstrated that high FLI was independently associated with the risk of T2D onset after 5 years of follow-up (adjusted HR: 6.87; 95% CI: 5.87-8.05 for men, and HR: 5.80 95% CI: 4.86-6.93 for women) [35]. A PREDAPS study, a prospective cohort including 1142 adults with prediabetes, also showed that hepatic steatosis, as defined by FLI ≥ 60, was independently associated with T2D incidence after 3 years of follow-up (HR: 3.21; 95% CI: 1. 45-7.09 in the fully adjusted model) [36]. Similar results were found by a prospective study including 1792 men without MetS showing a higher T2D risk in subjects with FLI ≥ 60 during a mean follow-up of 19 years (HR: 3.19; 95% CI: 2.26-4.52) [37].
The accumulating evidence indicates that NAFLD is associated with T2D onset independently of its components [5]. A meta-analysis included 33 studies with 501,022 individuals (30.8% with NAFLD, diagnosed by imaging techniques or biopsy) of whom 27,953 developed T2D over a median 5 years (IQR: 4.0-19 years) [42]. The patients with NAFLD had a higher risk of incident T2D than those without NAFLD (HR: 2.19, 95% CI: 1.93-2.48) [42]. The patients with more 'severe' NAFLD were also more likely to develop incident T2D (HR 2.69, 95% CI: 2.08-3.49), whereas this risk markedly increased across the severity of liver fibrosis (HR: 3.42, 95% CI: 2.29-5.11) [42]. Of note, all the risks were independent of confounding factors, namely age, sex, adiposity measures and other common metabolic risk factors [42]. This evidence is in accordance with the results of the present study, showing an independent association of FLI with T2D onset in statin-treated patients [42].
This study provides a simplified tool for physicians to identify individuals who are at high risk for new T2D following statin treatment. Therefore, routine FLI calculation in patients with dyslipidemia could be useful in clinical practice, because a high FLI points to further evaluation for NAFLD, but also to preventive strategies to ameliorate the statinassociated T2D risk. In this context, a more frequent monitor of glycemic status, along with a more intensive lifestyle intervention aiming at weight reduction, could probably delay both T2D onset and NAFLD progression [43,44]. At the same time, statin therapy should not be withdrawn but initiated according to the guidelines. Physicians should keep in mind that the sooner they initiate statin therapy, the higher cardiovascular benefit they will accomplish, but also note that NAFLD is not a contradiction to statin therapy [2,13,45]. On the contrary, even patients with steatohepatitis and compensated cirrhosis can be safely treated with statins when recommended [2].
The major limitations of this study are its sample size and retrospective design, along with the lack of data on dietary habits, physical activity and ultrasonography-or biopsydiagnosed NAFLD. Also, the patients were not systematically screened for viral hepatitis at baseline. Moreover, HbA1c was not routinely measured in the non-diabetic individuals during the conduct of this study. Thus, it is possible that a few patients with T2D at the baseline visit could have been included in the present analysis. However, this study had a long follow-up and is the first to evaluate the clinical utility of FLI in the association and prediction of new T2D in statin-treated patients. Although FLI was strongly associated with incident T2D, its predictive value was poor. In this context, larger prospective studies are needed to further investigate this association in statin-treated individuals.

Conclusions
FLI is significantly and independently associated with incident T2D risk in statintreated patients. FLI could be a useful tool, not only for identifying probable NAFLD, but also for predicting new T2D in patients initiating statin therapy.