Toward an Integrated Genome-Based Surveillance of Salmonella enterica in Germany

Abstract

Despite extensive monitoring programs and preventative measures, Salmonella spp. continue to cause tens of thousands human infections per year, as well as many regional and international food-borne outbreaks, that are of great importance for public health and cause significant socio-economic costs. In Germany, salmonellosis is the second most common cause of bacterial diarrhea in humans and is associated with high hospitalization rates. Whole-genome sequencing (WGS) combined with data analysis is a high throughput technology with an unprecedented discriminatory power, which is particularly well suited for targeted pathogen monitoring, rapid cluster detection and assignment of possible infection sources. However, an effective implementation of WGS methods for large-scale microbial pathogen detection and surveillance has been hampered by the lack of standardized methods, uniform quality criteria and strategies for data sharing, all of which are essential for a successful interpretation of sequencing data from different sources. To overcome these challenges, the national GenoSalmSurv project aims to establish a working model for an integrated genome-based surveillance system of Salmonella spp. in Germany, based on a decentralized data analysis. Backbone of the model is the harmonization of laboratory procedures and sequencing protocols, the implementation of open-source bioinformatics tools for data analysis at each institution and the establishment of routine practices for cross-sectoral data sharing for a uniform result interpretation. With this model, we present a working solution for cross-sector interpretation of sequencing data from different sources (such as human, veterinarian, food, feed and environmental) and outline how a decentralized data analysis can contribute to a uniform cluster detection and facilitate outbreak investigations.

What WGS Reveals about Salmonella enterica subsp. enterica in Wildlife in Germany

“…For this analysis, we included 1864 isolates from farm and food from our in-house collection (NRL- Salmonella database) and 1046 human isolates collected in 2020 in the frame of the ongoing GenoSalmSurv real-time surveillance project [ 4 ], plus an additional 71 human isolates included in recent outbreak investigations. An overview of the included isolates grouped by serovar and database is given in Table 2 .…”

“…We combined three different strategies for the compilation of the data set: Firstly, we included 30 wildlife isolates, which were sequenced previously, as part of routine Salmonella surveillance (published under BioProject PRJEB31846). Secondly, we included 64 wildlife S. Choleraesuis isolates sequenced in 2020 in the frame of the GenoSalmSurv project [ 4 ]. Thirdly, we applied the R function sample_n to randomly select wildlife Salmonella enterica isolates from the large NRL- Salmonella collection of previously not-sequenced isolates.…”

“…In addition, we included human and non-human isolates collected in 2020, in the frame of the ongoing real-time surveillance project GenoSalmSurv [ 4 ] ( n = 1896 (5 March 2021)).…”

“…However, Salmonella can persist under various conditions and are also able to survive outside the animal or human gut for longer time periods [ 2 , 3 ]. Due to their human health impact, most countries have established national monitoring and control programs on the occurrence of Salmonella in livestock [ 4 ]. Equally, most investigations focus on Salmonella in livestock populations and food products.…”

Decentralized Investigation of Bacterial Outbreaks Based on Hashed cgMLST

“…For each method combination, the pairwise distances of all sample pairs were obtained and a distance threshold was chosen, such that samples could be grouped into possible related clusters. Though no single fixed threshold is sufficient and advisable for the differentiation of outbreak to non-outbreak strains ( Simon et al, 2018 ; Radomski et al, 2019 ), an allele threshold of 10 is considered suitable for an initial clustering ( Besser et al, 2019 ; Uelze et al, 2021 ). For comparability with cgMLST clustering results, an SNP threshold of 17 was chosen (see section “Comparison With National Center for Biotechnology Information Pathogen Single-Nucleotide Polymorphism Results” for details), which is in agreement with Pightling et al (2018) .…”