Quantitative Microbial Risk Assessment Based on Whole Genome Sequencing Data: Case of Listeria monocytogenes

Abstract

The application of high-throughput DNA sequencing technologies (WGS) data remain an increasingly discussed but vastly unexplored resource in the public health domain of quantitative microbial risk assessment (QMRA). This is due to challenges including high dimensionality of WGS data and heterogeneity of microbial growth phenotype data. This study provides an innovative approach for modeling the impact of population heterogeneity in microbial phenotypic stress response and integrates this into predictive models inputting a high-dimensional WGS data for increased precision exposure assessment using an example of Listeria monocytogenes. Finite mixture models were used to distinguish the number of sub-populations for each of the stress phenotypes, acid, cold, salt and desiccation. Machine learning predictive models were selected from six algorithms by inputting WGS data to predict the sub-population membership of new strains with unknown stress response data. An example QMRA was conducted for cultured milk products using the strains of unknown stress phenotype to illustrate the significance of the findings of this study. Increased resistance to stress conditions leads to increased growth, the likelihood of higher exposure and probability of illness. Neglecting within-species genetic and phenotypic heterogeneity in microbial stress response may over or underestimate microbial exposure and eventual risk during QMRA.

A Critical Review of Risk Assessment Models for Listeria monocytogenes in Dairy Products

“…Through this literature search, listeriosis QRA models were retrieved for any foodstuff and then classified into QRA for produce (11 models), seafood (10 models), composite (4 models), meat products (23 models), and dairy (18 models). As defined in the objectives of this study, this review focuses on dairy products only [7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23].…”

“…A total of 18 QRA models, published from 1998, investigating dairy products as sources of listeriosis were recovered (Table 1). Out of them, nine models represented the food production conditions of Europe, covering France (Bemrah et al [13]; Sanaa et al [16]; Tenenhaus-Aziza et al [12]), Italy (Giacometti et al [20]; Condoleo et al [15]), Greece (Koutsoumanis et al [18]), Ireland (Tiwari et al [14]), Denmark (Njage et al [23]), and the EU (Pérez-Rodríguez et al [10]). From the American continent (five QRA models), three of them pertained to the USA and Canada (FDA-FSIS [7]; Latorre et al [19]; FDA-HealthCanada [9]), followed by Mexico (Soto-Beltrán et al [22]) and Brazil (Campagnollo et al [17]).…”

Landscape of Stress Response and Virulence Genes Among Listeria monocytogenes Strains

“…In these studies, the exposure assessment and hazard characterization steps of QMRAs are performed for a pathogenic species as a whole. The current advances in the field of omics technology give opportunities to make use of the greater understanding of intraspecific variability based on various recently published bioinformatics tools ( Brul et al, 2012 ; Den Besten et al, 2018 ; Haddad et al, 2018 ; Rantsiou et al, 2018 ; Fritsch et al, 2019 ; Njage et al, 2020 ). Instead of considering all-hazard strains of a species as equally likely to cause disease or equally likely to survive the food chain, WGS data could give support to rank subtypes with respect to their virulence potential ( Chen Y. et al, 2011 ; Collineau et al, 2019 ) or to groups subtypes with respect to their differences in robustness or fitness to reach the consumer stage ( Den Besten et al, 2018 ).…”

Microbial Stress: From Model Organisms to Applications in Food, Microbiotechnology and Medicine

“…In these studies, the exposure assessment and hazard characterization steps of QMRAs are performed for a pathogenic species as a whole. The current advances in the field of omics technology give opportunities to make use of the greater understanding of intraspecific variability based on various recently published bioinformatics tools (Brul et al, 2012;Den Besten et al, 2018;Haddad et al, 2018;Rantsiou et al, 2018;Fritsch et al, 2019;Njage et al, 2020). Instead of considering all-hazard strains of a species as equally likely to cause disease or equally likely to survive the food chain, WGS data could give support to rank subtypes with respect to their virulence potential Collineau et al, 2019) or to groups subtypes with respect to their differences in robustness or fitness to reach the consumer stage (Den Besten et al, 2018).…”