Supra-additive neuroprotective effects of berberine-metformin combination in diabetic encephalopathy: Chou-Talalay synergy quantification, AMPK-Nrf2 axis modulation, and pharmacokinetic verification.

Abstract

Type 2 diabetes mellitus (T2DM) increases the risk of hippocampal neurodegeneration and cognitive decline. Berberine and metformin independently activate AMPK and may engage Nrf2-mediated antioxidant defenses, yet their combined neuroprotective interaction has not been formally quantified using validated synergy frameworks, nor has its pharmacokinetic basis been verified. Streptozotocin-nicotinamide diabetic rats were allocated to twelve groups (n = 13/group) receiving berberine (50, 100, 150 mg/kg/day) or metformin (100, 200, 300 mg/kg/day) monotherapy, fixed-ratio 1:2 combinations, or vehicle controls (including a non-diabetic combination group) orally for six weeks. The novel object recognition (NOR) discrimination index served as the predefined primary endpoint for Chou-Talalay combination index (CI) analysis. Hippocampal mechanistic (n = 6/group) and satellite LC-MS/MS pharmacokinetic (n = 6/group) analyses were performed. Diabetes impaired NOR discrimination index (37.2 ± 3.8% vs. 68.4 ± 3.2%; p < 0.001). The reference combination (100 + 200 mg/kg) restored NOR to 67.1 ± 3.6% with CI = 0.65 (95% CI: 0.43-0.91), synergism maintained across the full effect range. All six neuroinflammatory endpoints achieved Benjamini-Hochberg-corrected significance (p_adj = 0.006-0.043; Tier 2). Non-diabetic combination animals showed reduced AMPK activation magnitude (1.53 vs. 2.31-fold; P_adj = 0.067; Tier 3, hypothesis-generating). LC-MS/MS verified bioequivalent drug exposure. Berberine-metformin co-treatment is associated with CI-quantified supra-additive recognition memory recovery in diabetic encephalopathy, with neuroinflammatory suppression as the most statistically robust mechanistic correlate. Pharmacokinetic findings are consistent with a pharmacodynamic rather than pharmacokinetic basis. Causal involvement of the AMPK-Nrf2 axis remains correlative pending direct loss-of-function validation.