GROWTH OF PROTEOLYTIC CLOSTRIDIUM-BOTULINUM IN-PROCESS CHEESE PRODUCTS .2. PREDICTIVE MODELING

Improvement of the accuracy of predictive models for proteolytic Clostridium botulinum stability of current and future process cheese compositions was studied. The results of a central composite design were analyzed the effects of pH (5.45 to 5.9), sodium chloride (1.1 to 2.7%, wt/wt), combinations of citrates and phosphates as emulsifying salts (1.5 to 2%, wt/wt) and temperature (15 to 30 degrees C). Supplemental data enabled assessment of the differences in lactate in moisture (1.0 to 2.6%) originating from the cheese raw materials, variations in moisture (50 to 69%) and the percentage of total fat (0.1 to 41%). The time to a 100-fold increase in cell numbers, t100, was modeled using the SAS-LIFEREG procedure, which can compensate for interval-censored and no-growth values. Two quadratic response models were derived to predict the growth of C. botulinum. The Central Composite Model uses significant (P < 0.01) estimates of the combined effects of pH, total salts (NaCl plus emulsifying salts) in moisture (SIM), citrates as a percentage of total salts (CTS), and temperature. The Extended Total Model uses an additional parameter, lactate in moisture (LACM). The role of fat content was insignificant. Moisture content, which is frequently used, appears to be an unreliable predictor of botulinum stability when fat dry basis (FDB) varies. Both models are capable of predicting the observed stability of compositions derived from the literature. They can complement the historic fully quadratic model of Tanaka, they can be used to assess process cheese safety in relation to distribution and/or storage conditions, and they can accelerate product design, minimizing the use of time-consuming product-challenge tests.