Physical sample structure as predictive factor in growth modeling of Listeria innocua in a white cheese model system

Abstract

Growth of Listeria innocua at 9 °C was investigated in white cheeses manufactured from ultra-filtrate milk concentrate added varying amounts of skimmed milk powder, NaCl and glucono-delta-lactone. Characterization of the white cheese structures was performed using nuclear magnetic resonance (NMR) T₂ relaxation parameters (relaxation times constants, relative areas and width of peaks) and their applicability as predictive factors for maximum specific growth rate, √μ(max) and log-increase in 6 weeks of L. innocua was evaluated by polynomial modeling. Inclusion of NMR parameters was able to increase the goodness-of-fit of two basic models; one having pH, undissociated gluconic acid (GA(u), mM) and NaCl (% w/v) as predictive factors and another having pH, GA(u) and a(w) as predictive factors. However, the best model fit was observed using √μ(max) as response for the model including pH, GA(u), aw and Width T₂₁ revealing the lowest relative root mean squared errors of 14.0%. As the T₂ relaxation population T₂₁ is assigned to represent immobilized bulk water protons and the width T₂₁ the heterogeneity of this water population, growth of L. innocua in white cheese seemed to be dependent on the heterogeneity of the immobilized bulk water present in cheese.