In conclusion, our study demonstrated that serum vitamin E levels were higher in patients with NAFLD than in those without NAFLD among a Korean diabetic population. Furthermore, both elevated serum vitamin A and E levels were positively associated with an increased risk of NAFLD, while serum folate levels showed no significant association with NAFLD. These findings suggest that higher levels of specific micronutrients, particularly vitamins A and E, may play a role in the development or progression of NAFLD in individuals with diabetes. Given the complex metabolic interactions of these vitamins, especially in relation to lipid metabolism and liver function, our results underscore the need for personalized nutritional strategies in managing NAFLD risk among diabetic patients.

Materials and methods

Data information

We conducted a cross-sectional study using data from the Korean National Health and Nutrition Examination Survey (KNHANES) 2016–2018. KNHANES is a nationwide epidemiological study conducted by the Korea Disease Control and Prevention Agency (KDCA) and provides various health-related information, such as laboratory test results and physical examination results²⁹.

The study protocol was approved by the Institutional Review Board of Sookmyung Women’s University (IRB number: SMWU-2107-HR-063), and the requirement for informed consent was waived since the data were anonymized.

Study population

From a total of 24,269 participants in the KNHANES 7th database, we included Korean adults with diabetes mellitus. Participants were defined as having diabetes if they met one of the following criteria: (a) HbA1c ≥ 6.5% (48 mmol/mol) or fasting blood glucose (FBG) ≥ 126 mg/dL, (b) those with a diagnosis of diabetes by a physician, or (c) those who took antidiabetic drugs on the day of examination. Exclusion criteria included pregnant women, those with excessive daily intake of vitamin A or folate (> 3,000 µg of retinol activity equivalents [RAE] or > 1,000 µg of dietary folate equivalents [DFE], respectively), and those with a previous history of hepatitis B or C, liver cirrhosis, or alcohol use disorder. A history of each liver-related underlying disease was defined as a ‘yes’ response on the questionnaire regarding past medical diagnoses, and a history of alcohol use disorder was defined as a ‘yes’ response on the questionnaire asking whether the individual had ever consulted for alcohol-related issues. We also excluded individuals with missing data.

Diagnostic models for NAFLD

NAFLD was assessed based on the following validated prediction models: the hepatic steatosis index (HSI), Framingham steatosis index (FSI), and comprehensive NAFLD score (CNS). These are simple and non-invasive indices used for screening NAFLD. NAFLD was defined based on the following cut-off values for each score: HSI > 36, FSI ≥ 23, or CNS ≥ 40 ^30,31,32.

Formulae for each diagnostic model are shown below:

1. 1.

   HSI = 8 × alanine aminotransferase (ALT)/aspartate aminotransferase (AST) ratio + BMI (+ 2 if type 2 diabetes; +2 if female).

2. 2.

   FSI = e^x/ (1 + e^x), where x = − 7.981 + 0.011 × age (years) − 0.146 × sex (female = 1, male = 0) + 0.173 × BMI (kg/m²) + 0.007⁷ × triglycerides (TG) (mg/dL) + 0.593 × hypertension (yes = 1, no = 0) + 0.789 × diabetes (yes = 1, no = 0) + 1.1 × ALT/AST ratio ≥ 1.33 (yes = 1, no = 0).

3. 3.

   CNS = 1/ (1 + e^-x) × 100, where x =.

   • for men, x =  0.016 × age (years) + 0.182 × BMI (kg/m²) + 0.089 × waist circumference (WC) (cm) + 0.391 × alcohol (yes = 1, no = 0) + 0.124 × exercise (yes = 0, no = 1) + 0.018 × FBG (mg/dL) + 0.773 × log_e (TG [mg/dL]) − 0.014 × high-density lipoprotein cholesterol (HDL-C) (mg/dL) + 0.145 × uric acid (mg/dL) − 0.674 × log_e (AST [IU/L]) + 1.632 × log_e (ALT [IU/L]) − 21.695.

   • for women, x = 0.320 × BMI (kg/m²) + 0.044 × WC (cm) + 0.533 × diabetes (yes = 1, no = 0) + 0.016 × FBG (mg/dL) + 0.951 × log_e (TG [mg/dL]) − 0.015 × HDL-C (mg/dL) + 0.199 × uric acid (mg/dL) − 0.645 × log_e (AST [IU/L]) + 1.302 × log_e (ALT [IU/L]) + 0.255× menopause (yes = 1, no = 0) − 19.741.

Measurement of the serum vitamin levels

Serum vitamin A (retinol) and vitamin E (α-tocopherol) levels were measured with Agilent1200 (Agilent Technologies Inc., USA) using high-performance liquid chromatography-flame ionization detector Serum folate levels were measured using ARCHITECT i4000Sr (Abbott Laboratories, USA) with chemiluminescent microparticle immunoassay. The cut-off values to define vitamin deficiency were set as follows: serum vitamin A < 0.2 mg/L, serum vitamin E < 5 mg/L, and serum folate < 4 ng/mL^33,34,35.

Definitions of the variables of interest

Obesity was defined using BMI and WC. Those with a BMI ≥ 30 kg/m² or those with abdominal obesity (WC ≥ 90 cm in males, WC ≥ 85 cm in females) with BMI ≥ 25 kg/m² were classified into the obese group³⁶.

Definitions of the variables included in the NAFLD prediction were as follows: patients with hypertension were defined as those who met one of the following criteria: (a) systolic blood pressure (SBP) ≥ 140 mmHg or diastolic blood pressure (DBP) ≥ 90 mmHg, (b) history of diagnosis of hypertension by a physician, or (c) patients who took antihypertensive drugs on the day of the examination³⁷. Alcohol consumption status was categorized into three groups: (a) non-drinkers, those who did not consume alcohol for the past year; (b) heavy drinkers, defined as men who drink ≥ 7 standard unit drinks or women who drink ≥ 5 standard unit drinks ≥ 2 times per week; and others were classified as (c) current drinkers. Regular exercise status was determined as yes if the participant met either one of the following: (a) resistance exercise ≥ 1 time weekly, or (b) ≥ 150 min of moderate-intensity physical activities, or ≥ 75 min of high-intensity physical activities or mixed activities (1 min of high-intensity activity is equivalent to 2 min of moderate-intensity activity).

The covariates included in the analysis were as follows: the presence of dyslipidemia was considered in those with a physician’s diagnosis. Smoking status was divided into three groups: (a) non-smokers, those who had never smoked; (b) past smokers, those who had smoked < 100 cigarettes in their lifetime or those who had smoked ≥ 100 cigarettes in the past but not now; and others were classified into (c) current smokers. Dietary data were surveyed using the 24-hour recall method. Vitamin A intake was measured using retinol equivalents (RE): RE (µg) = retinol (µg) + β-carotenes/6 (µg), according to the Dietary Reference Intakes for Koreans criteria³⁸. Supplement use was determined as yes if the participant had taken dietary supplements for more than two weeks over the past year.

Statistical analysis

To evaluate the differences in the baseline characteristics between the NAFLD and non-NAFLD groups, we used independent Student’s t-tests for continuous variables and chi-square tests for categorical variables. Data are presented as mean ± standard error or number (weighted percentage). Comparison of the serum vitamin levels between the NAFLD and non-NAFLD groups was performed using t-tests. In addition, stratified analysis was performed based on two baseline factors, obesity and elderly, to investigate the consistency of the results across the subgroups.

Participants were classified into quartiles of serum vitamin A, E, or folate levels, with Q1 being the lowest and Q4 being the highest quartile of each serum vitamin level (Q1–Q4). Multivariate logistic regression was performed to investigate the association between the risk of NAFLD and serum vitamin levels. The adjusted odds ratios (AORs) of NAFLD and 95% confidence intervals (CIs) were calculated with the lowest quartile set as a reference. The regression model was adjusted for smoking status, HbA1c, total cholesterol, dyslipidemia, total energy intake, and supplement use. The dietary intake of vitamin A or folate was added as a covariate in the analysis of each vitamin.

All the statistical analyses were performed using SPSS ver. 25.0 (IBM Corp., Armonk, NY, USA), and results were considered statistically significant at p < 0.05.