Microwave-Assisted Extraction of Garlic Polyphenols: Optimization, Profiling, and In Vitro Digestion

Abstract

Objective: This study aimed to develop and optimize a rapid, eco-friendly microwave-assisted extraction (MAE) process for recovering total phenolic content (TPC) and total flavonoid content (TFC) from garlic (Allium sativum L.), while evaluating the predictive performance of response surface methodology (RSM) versus artificial neural networks (ANNs) and assessing the in vitro gastrointestinal stability of key polyphenols. Methodology: A four-factor, three-level central composite design (CCD) was implemented to evaluate the effects of extraction time, temperature, ethanol concentration, and solvent-to-solid ratio. A second-order polynomial RSM model was benchmarked against a 4-10-2 multilayer perceptron ANN trained by backpropagation. Optimal conditions were derived using the Derringer–Suich desirability function and confirmed experimentally. Individual polyphenols were profiled via LC-MS/MS and monitored across simulated oral, gastric, and intestinal digestion phases. Principal Results: The ANN model demonstrated superior predictive performance (R2 = 0.9999 training, 0.9974 validation, 0.9939 testing) compared to the RSM model (R2 = 0.9721 for TPC and 0.9925 for TFC). Experimental validation under optimal conditions—1.50 min, 55 °C, 75% ethanol, and a 29 mL/g ratio—yielded a TPC of 2.487 mg GAE/g FW and a TFC of 21.356 mg QUE/g FW. During simulated gastrointestinal digestion, significant degradation occurred during the intestinal phase, resulting in low final recoveries for gallic acid (16.9%), caffeic acid (19.9%), and luteolin (22.6%). Conclusions: MAE coupled with ANN modeling provides a highly accurate, rapid, and green extraction strategy for garlic polyphenols. However, the marked degradation of target compounds during intestinal digestion highlights the necessity of encapsulation or protective delivery systems to preserve their biological functionality in food applications.