The National Health and Nutrition Examination Survey (NHANES) is the most in-depth survey administered by the National Center for Health Statistics (NCHS) to assess the health and nutritional status of adults and children in the United States. The NHANES surveys approximately 5,000 individuals annually in 15 different counties across the country in a two-year cycle, and the study cohort is representative of the entire U.S. population through a sample-weighted analysis. This study is based on data from the 2017 to 2018 NHANES, a cycle that included participants’ vibration-controlled transient elastography (VCTE) data used to define MAFLD. The Research Ethics Review Board of the National Center for Health Statistics approved the NHANES study. All participants provided informed consent. According to a large meta-analysis, a controlled attenuation parameter (CAP) of ≥248 dB/m (AUC: 0.823, Sensitivity: 0.688, Specificity: 0.822) was used as the threshold for the diagnosis of hepatic steatosis (13). And a median liver stiffness of ≥8.2 kPa was used to significant fibrosis (14). According to the European Association for the study of the Live (EASL) Clinical Practice Guidelines on non-invasive tests for evaluation of liver disease severity and prognosis, CAP≥275 dB/m might be used to diagnose steatosis (15). Therefore, we also conducted an analysis using the CAP ≥275 dB/m (Supplementary materials). Of the 9,254 participants, those who were not older than 18 years, pregnant, those with missing dietary data used to calculate DII, those with missing CAP data and CAP less than 248 dB/m, those with missing dual-energy X-ray data used to measure skeletal muscle mass, those who did not meet the diagnosis of MAFLD and those with missing data on relevant covariates were excluded, and finally, a total of 917 participants were enrolled in the study (Figure 1).

Based on the presence of imaging evidence of hepatic steatosis in combination with one of the following three conditions: overweight or obesity (defined as BMI ≥25 kg/m² in Caucasians or BMI ≥23 kg/m² in Asians), type 2 diabetes mellitus, and metabolic dysfunction. Metabolic dysfunction was defined as the presence of at least two of the following risk factors for metabolic abnormalities: (1) waist circumference ≥ 102 cm in Caucasian men and ≥ 88 cm in women (or ≥ 90 cm in men and ≥ 80 cm in women in Asians); (2) blood pressure ≥ 130/85 mmHg or specific drug treatment; (3) triglyceride (TG) ≥1.7 mmol/L or specific drug treatment; (4) high-density lipoprotein cholesterol (HDL-C) <1.0 mmol/L in men and HDL-C < 1.3 mmol/L in women or specific drug treatment; (5) prediabetes; (6) homeostasis model assessment of insulin resistance score (HOMA-IR) ≥2.5; and (7) high-sensitivity C-reactive protein (hs-CRP) >2 mg/L (1).

According to the National Institutes of Health recommendations for determining the presence of sarcopenia, appendicular lean mass adjusted for body mass index (ALM_BMI) is used. ALM_BMI = appendicular lean mass (kg)/body mass index (kg/m²), with males <0.789 and females <0.512 considered to have sarcopenia (16). The appendicular lean mass of the extremities was measured by dual-energy X-ray absorptiometry (DXA) whole-body scanning, which was obtained on a Hologic Discovery A optical densitometer (Hologic, Inc., Bedford, Massachusetts) using the Apex 3.2 software version. Trained and certified radiologic technologists perform DXA examinations and more detailed information on the DXA examination program is documented on the NHANES.¹ In the NHANES files “DXDLLLE,” “DXDRLLE,” “DXDLALE” and “DXDRALE,” the specific values of limb lean body mass are recorded.

Daily intakes of dietary components in the NHANES database are obtained from 24-h dietary recall interview, and in this study, the average of two 24-h dietary data was used to calculate the DII. The specific calculation methodology is reported in detail elsewhere (11). In the present study, we used 28 different dietary components to estimate DII, including energy, protein, carbohydrate, dietary fiber, vitamins A, B1, B2, B6, B12, C, D, total fat, total saturated fatty acids, total monounsaturated fatty acids, total polyunsaturated fatty acids, n-3 fatty acids, n-6 fatty acids, cholesterol, vitamin E, β-carotene, niacin, folate, magnesium, iron, zinc, selenium, caffeine and alcohol.

Variables included in this study were gender, age, race, smoking, body mass index (BMI), Alcohol intake, Significant fibrosis, diabetes, cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and glycosylated hemoglobin (HbA1c). Race was categorized as Mexican American, other Hispanic, Non-Hispanic White, Non-Hispanic Black, non-Hispanic Asian, and other races, BMI was weight (kg)/height (m) squared, and smoking was defined as smoking more than 100 cigarettes in one’s lifetime, which was obtained from a questionnaire. The methods of testing TC, TG, HDL-C, and HbA1c are described in detail on the official NHANES website. Diabetes mellitus was defined as “your doctor has told you that you have diabetes mellitus,” or a fasting blood glucose ≥7.0 mmol/L or a random blood glucose ≥11.1 mmol/L or an HbA1c >6.5%, or taking hypoglycemic medication to lower blood glucose or using insulin.

Continuous variables were expressed as mean ± standard deviation and categorical variables were expressed as frequencies and weighted percentages, and weighted linear regression models (for continuous variables), as well as weighted chi-square tests (for categorical variables), were utilized to compare the differences between the two groups. Binary logistic regression was used to analyze the relationship between DII and sarcopenia. Model 1 was unadjusted for variables; model 2 was adjusted for age, race, and BMI; and model 3 was adjusted for age, race, BMI, Alcohol intake, smoking, significant fibrosis, diabetes, TC, TG, HDL-C and HbA1c. In addition, analyses were stratified by age, BMI, and significant fibrosis. Multivariate linear regression was used to estimate the relationship between DII and ALM_BMI. Data were analyzed using the R package, EmpowerStats, and Stata, and p < 0.05 was considered statistically significant.

In this study, 917 patients with MAFLD were enrolled with a weighted mean age of 42.58 years and a prevalence of sarcopenia of 13.07%. Participants were categorized into three groups based on DII tertiles: group T1 (n = 305), group T2 (n = 306), and group T3 (n = 306). There was a statistically significant difference in mean age between the three groups (T1: 44.56 ± 11.10 vs. T2: 41.86 ± 11.57 vs. T3: 41.24 ± 12.24, p < 0.001). In addition, participants with higher DII were more likely to be female (T1: 26.89% vs. T2: 54.12% vs. T3: 63.85%, p < 0.001), have a higher prevalence of diabetes mellitus (T1: 13.07% vs. T2: 12.29% vs. T3: 21.24%, p = 0.004), and a higher BMI (T1: 31.43 ± 5.67 vs. T2: 32.58 ± 6.55 vs. T3: 34.21 ± 6.94, p < 0.001), higher TC (T1: 4.84 ± 0.88 vs. T2: 5.08 ± 0.98 vs. T3: 5.21 ± 1.19, p = 0.009), lower alcohol intake (T1: 15.08 ± 24.75 vs. T2: 8.82 ± 24.59 vs. T3: 6.19 ± 14.46, p < 0.001), lower ALM_BMI (T1: 0.84 ± 0.17 vs. T2: 0.75 ± 0.18 vs. T3: 0.70 ± 0.17, p < 0.001). In these three groups, there were no statistical differences in smoking (p = 0.056), TG (p = 0.904), HDL-C (p = 0.941), HbA1c (p = 0.168), prevalence of significant fibrosis (p = 0.262) and prevalence of sarcopenia (p = 0.111). Detailed information is shown in Table 1. In female participants, we observed that DII was significantly higher in sarcopenia patients than in non-sarcopenia patients (p < 0.001), whereas there was no statistically significant difference in males (p = 0.568) (Table 2).

The association between DII and the risk of prevalence of sarcopenia in NAFLD was analyzed using binary logistic regression models (Table 3). When DII was used as a continuous variable, it was significantly and positively associated with the risk of sarcopenia in women [model 1: odds ratio (OR): 1.42, 95% CI: (1.13, 1.78), p = 0.002]. This relationship remained statistically significant after adjusting for confounders [Model 2: OR: 1.57, 95% CI: (1.24, 1.99), p < 0.001. Model 3: OR: 1.62, 95% CI: (1.27, 2.08), p < 0.001]. In contrast, men had no significant correlation (p > 0.05). When DII was used as a categorical variable, in women, the T3 group had a higher risk of sarcopenia than the T1 group [Model 1: OR: 2.55, 95% CI: (1.09, 5.99), p = 0.031]. After adjusting for age, race, and BMI, the T3 group still exhibited a higher risk of sarcopenia [Model 2: OR: 3.95, 95% CI: (1.65, 9.46), p = 0.002]. After adjusting for age, race, BMI, Alcohol intake, smoking, significant fibrosis, diabetes, TC, TG, HDL-C and HbA1c, the association between DII and the risk of developing sarcopenia did not change [model 3: OR: 4.02, 95% CI: (1.64, 9.82), p = 0.002]. In men, a higher risk of prevalence of sarcopenia in the T2 group than in the T1 group was observed only in Model 2 and Model 3 [Model 2: OR: 2.48, 95% CI: (1.03, 5.96), p = 0.042. Model 3: OR: 2.87, 95% CI: (1.11, 7.41), p = 0.030].

In a stratified analysis according to age (women: P for interaction = 0.938, men: P for interaction = 0.822), BMI (women: P for interaction = 0.357, men: P for interaction = 0.08) were stratified, and the risk of prevalence of DII and sarcopenia among MAFLD patients did not change. The association did not change and still showed a significant positive correlation between DII and the risk of sarcopenia in women, whereas in men, there was no significant correlation. However, after stratifying the participants according to significant fibrosis, higher DII scores related to a higher risk of sarcopenia in men with significant fibrosis [OR: 3.42, 95% CI: (1.08, 10.83), P for interaction = 0.033], and no significant difference among women (P for interaction = 0.580) (Figure 2).

As shown in Table 4, multivariate linear regression analysis showed that DII was negatively associated with ALM_BMI in both men and women [women: Model 1: β: −0.008, 95% CI: (−0.014, −0.003), p = 0.004. Model 2: β: −0.007, 95% CI: (−0.012, −0.002), p = 0.010; men: Model 1: β: −0.007, 95% CI: (−0.013, −0.001), p = 0.042. Model 2: β: −0.007, 95% CI: (−0.013, −0.001), p = 0.049] (Table 4).