Abstract
Mild heat treatment of fruit juices in combination with natural aroma compounds has been reported as an alternative to conventional pasteurization to better preserve their nutritional value. However, its antimicrobial efficiency varies from one juice to another. This study aims at developing a secondary predictive model of microbial inactivation scale during such combined process. Carvacrol was used as aroma compound and acid-adapted L. monocytogenes as target microorganism. The inactivation kinetics of this bacteria were followed in simulated fruit juices using a Central Composite Design with pH (2-6), °Brix (0–24), temperature (55–65 °C), and carvacrol concentration (0–60 μL/L) as independent variables. Curves were fitted to the Weibull inactivation model, and data collected used to generate two predictive models of the inactivation scale parameter through multiple regression analysis following an empirical and a mechanistic (based on Gamma concept) approach. The best of the two models was then validated using real fruit (orange, pineapple, and watermelon) juices. The empirical model where only the four variables tested were considered showed a lower statistical performance compared to the mechanistic model where octanol-water partition coefficient (Ko/w) and vapour pressure (Vp) of carvacrol at the treatment temperature were integrated (R2 0.6 and 0.9; Accuracy factor 1.5 and 1.3; Sum of Squared Error 3.6 and 1.1, respectively). No significant difference was observed between inactivation scale values obtained with real juices and the predicted values calculated using this mechanistic model. The Ko/w and Vp of the aroma compound used are key parameters that determine the efficiency of the above-described combined treatment.
[Display omitted]
pH and Brix of fruit juices determine L. monocytogenes mild heat inactivation.Carvacrol supplementation enhances mild heat inactivation of Listeria monocytogenes.This carvacrol effect is determined by its partition coefficient and vapour pressure.Consideration of these two physicochemical parameters increases model prediction.