NGS-Based S. aureus Typing and Outbreak Analysis in Clinical Microbiology Laboratories: Lessons Learned From a Swiss-Wide Proficiency Test.

Valérie Barbié,Simone Oberhaensli,Frank Imkamp,Claire Bertelli,Aitana Lebrand,Peter M Keller,Daniel Wüthrich,Gilbert Greub,Onya Opota,Dominique Blanc ,Christophe Dessimoz,Stefan Neuenschwander,Laurent Falquet,David Dylus,Alban Ramette,Stefano Leo,Helena M B Seth-Smith,André Kahles ,Adrian Egli,Vladimir Lazarević ,Trestan Pillonel,Jacques Schrenzel,Sacha Laurent

doi:10.3389/fmicb.2020.591093

Abstract

Whole genome sequencing (WGS) enables high resolution typing of bacteria up to the single nucleotide polymorphism (SNP) level. WGS is used in clinical microbiology laboratories for infection control, molecular surveillance and outbreak analyses. Given the large palette of WGS reagents and bioinformatics tools, the Swiss clinical bacteriology community decided to conduct a ring trial (RT) to foster harmonization of NGS-based bacterial typing. The RT aimed at assessing methicillin-susceptible Staphylococcus aureus strain relatedness from WGS and epidemiological data. The RT was designed to disentangle the variability arising from differences in sample preparation, SNP calling and phylogenetic methods. Nine laboratories participated. The resulting phylogenetic tree and cluster identification were highly reproducible across the laboratories. Cluster interpretation was, however, more laboratory dependent, suggesting that an increased sharing of expertise across laboratories would contribute to further harmonization of practices. More detailed bioinformatic analyses unveiled that while similar clusters were found across laboratories, these were actually based on different sets of SNPs, differentially retained after sample preparation and SNP calling procedures. Despite this, the observed number of SNP differences between pairs of strains, an important criterion to determine strain relatedness given epidemiological information, was similar across pipelines for closely related strains when restricting SNP calls to a common core genome defined by S. aureus cgMLST schema. The lessons learned from this pilot study will serve the implementation of larger-scale RT, as a mean to have regular external quality assessments for laboratories performing WGS analyses in a clinical setting.

Highlights

Hospitals and clinical laboratories are increasingly using generation sequencing (NGS) technology to address a multitude of questions
The design aimed at disentangling the variability in the final outcome that might arise from differences in sample preparation, raw data processing for SNP calling and choice of phylogenetic methods
In increment 1, we obtained reads from five sequencing centers, all based on Illumina sequencing technology, except one that included a mix of MinION and Illumina reads

Summary

Introduction

In clinical microbiology, whole genome sequencing (WGS) has been used for typing (cgMLST, SNP calling) [e.g., (Mellmann et al, 2017; Abdelbary et al, 2019; Zakham et al, 2019; Magalhães et al, 2020)], and enables addressing strain relatedness using high resolution data, e.g., for outbreaks within hospitals or in the community [e.g., (Deurenberg et al, 2017)], or at a larger geographic scale, e.g., for food-borne pathogens (Hendriksen et al, 2018), or other environmental pathogens [e.g., (Wüthrich et al, 2019)]. In the Netherlands, a multicenter PT of bacterial outbreak analyses was implemented with 10 isolates each of methicillin-susceptible S. aureus strains, Enterococcus faecium, and Klebsiella pneumoniae, for which participants were asked to identify outbreak clusters from FASTQ datasets (personal communication from Jordy Coolen)

Objectives

Methods

Results

Conclusion