Growth/no growth models for heat-treated psychrotrophic Bacillus cereus spores under cold storage

Abstract

The microbiological safety of refrigerated and processed foods of extended durability (REPFED) is linked to spore-forming pathogens, more specifically Clostridium botulinum and Bacillus cereus. In this study two sets of growth/no growth (GNG) models are presented for the spores of two B. cereus strains. The models incorporate both product (water activity (aw) and pH) and process parameters (pasteurization value at 90°C (P90) or heating temperature). The first model evaluates the effect of four different P90-values (P90=0, 4, 7 or 10min, all applied at 90°C) on the germination and subsequent growth of B. cereus spores under different conditions of pH and aw at 10°C. These models show that a heat treatment not only increases the time to growth (TTG), but also significantly increases the minimal aw and pH necessary for germination and subsequent growth: e.g. at aw 0.995 and without heat treatment (P90=0), strain FF355 B. cereus spores were predicted to germinate and grow at pH 5.3. With a P90 of 10min, the minimal pH increased to 5.7. The second set of models for B. cereus spores compares the effect of three heat treatments with the same P90-value (10min) but applied at different temperatures (85, 87 and 90°C), on the germination and subsequent growth at 10°C. The second model shows that lower heating temperatures (85 and 87°C) had less effect on the TTG and minimal aw and pH than a higher temperature (90°C). Finally, the first set of models was validated in broth using spores of seven psychrotrophic B. cereus strains, to evaluate the effect of strain variability on the model predictions. The results of the validation (% growth) were compared to the predicted growth probability. The results showed that the models were prone to fail-dangerous results (i.e. predicting no growth when growth was observed: 17%–34%). Using a very low threshold for growth (0.1% predicted chance of growth was considered to be complete growth), the models were more fail-safe (11%–34%) than fail-dangerous (0.4%–14%).