Metabolic characteristics and risk profile of non-obese fatty liver disease in an elderly Chinese community cohort

Non-obese non-alcoholic fatty liver disease (NAFLD), or “lean NAFLD”, challenges the assumption that fatty liver predominantly affects obese individuals. In elderly populations, age-related changes in body composition further limit the utility of body mass index (BMI) for metabolic risk stratification. We aimed to characterise the metabolic profile of lean NAFLD and quantify dose–response relationships between key biochemical markers and NAFLD risk in elderly Chinese adults. We analysed health-examination data from community-dwelling adults aged $$\ge$$65 years attending a primary care centre in Xiamen, China (2022–2024). After excluding individuals with excessive alcohol intake, viral hepatitis markers, or missing ultrasound records, 10,586 participants were included. NAFLD was diagnosed ultrasonographically. Participants were classified as lean (BMI <24 kg/m$$^2$$) or overweight/obese (BMI $$\ge$$24 kg/m$$^2$$) per Chinese criteria. We compared metabolic profiles across four BMI–NAFLD phenotypes and performed stratified analyses by triglyceride-to-HDL cholesterol (TG/HDL-C) ratio and serum uric acid categories. NAFLD prevalence was 30.8% (3,258/10,586), with 35.8% (1,167/3,258) occurring in lean individuals. Lean NAFLD participants exhibited higher TG/HDL-C ratios, fasting glucose, glycated haemoglobin, and uric acid levels than Lean Healthy participants, alongside greater prevalence of hypertension (40.3% vs. 25.0%) and diabetes (23.1% vs. 8.9%). NAFLD prevalence increased from 14.2% (TG/HDL-C <1.0) to 59.4% (TG/HDL-C $$\ge$$2.0) among lean participants, and from 15.6% (normal uric acid) to 47.4% (hyperuricaemia). In multivariable models, the highest TG/HDL-C stratum (TG/HDL-C $$\ge$$2.0) conferred 2.26-fold increased odds of NAFLD (95% CI 1.98–2.57) versus the lowest stratum. A composite model combining TG/HDL-C, uric acid, waist circumference, and fasting glucose achieved an AUC of 0.706 (95% CI 0.688–0.724), substantially exceeding TG/HDL-C alone (AUC 0.615). Over one-third of NAFLD cases in this elderly cohort occurred in lean individuals with substantial metabolic disturbances, and 91.4% simultaneously met MASLD cardiometabolic criteria. TG/HDL-C $$\ge$$2.0 functions primarily as a high-specificity rule-in marker (specificity 97.9%, PPV 59.4%) rather than a screening tool; a composite model incorporating TG/HDL-C ratio, uric acid, waist circumference, and fasting glucose provides improved discrimination (AUC 0.706) for identifying lean individuals with hepatic steatosis in primary care settings where BMI-based screening alone is insufficient.