Optimal dose window, formulation efficacy, and molecular mechanisms of curcumin in MASLD animal models: A systematic review and three-level meta-analysis.

Approximately 25% of the global adult population is affected by metabolic dysfunction-associated steatotic liver disease (MASLD), yet no universally approved first-line pharmacotherapy exists. Curcumin is the major bioactive constituent of turmeric (Curcuma longa L.), a plant with a long history of use in traditional medicine systems for treating digestive, inflammatory, and liver disorders. Despite promising multi-targeted activity, the clinical application of curcumin is hindered by poor bioavailability and inconsistent efficacy due to variable dosing and formulations. This systematic review and meta-analysis aimed to: (1) identify the optimal dose range of curcumin for MASLD intervention in preclinical models; (2) quantitatively compare the efficacy of different curcumin-based formulations; and (3) validate its modulation of key mechanistic targets including SREBP-1c, PPARα, TNF-α, and Nrf2. PubMed, Embase, and Web of Science were searched from inception to December 2025 for in vivo studies evaluating curcumin interventions in mouse or rat models of MASLD. A three-level meta-analysis with robust variance estimation (RVE) was used to handle data dependencies from multi-dose studies. The study protocol was preregistered with PROSPERO (CRD420261276896). Non-linear dose-response relationships were modeled using restricted cubic splines (RCS). Efficacy differences across formulation types were quantitatively compared, and effect sizes for key mechanistic targets were calculated. Seventy-two studies were included. Dose-response analysis revealed a non-linear effect on serum ALT reduction, with an optimal window of 50-150 mg/kg; efficacy declined notably at doses >400 mg/kg. Subgroup analyses showed that curcumin derivatives and novel delivery systems were significantly more effective than native curcumin (P < 0.05). Mechanistically, curcumin significantly downregulated SREBP-1c (g = -8.00) and TNF-α (g = -4.36), and upregulated Nrf2 (g = 8.95). Following the FDA body surface area conversion, the 50-150 mg/kg animal dose corresponds to 486-1458 mg/day in humans, which is consistent with dosages used in current clinical trials. This alignment supports the biological plausibility of the animal-derived dose window, though direct translation to human dosing requires clinical validation. This is the first meta-analysis to quantitatively define an optimal dose window for curcumin in MASLD animal models and to demonstrate non-linear dose-response characteristics across different formulations. Our findings provide a pharmacological rationale for dose selection in future preclinical standardization and hypothesis-driven clinical trial design for curcumin-based interventions in MASLD.