Whole Genome Sequencing for Surveillance of Diphtheria in Low Incidence Settings

Abstract

Corynebacterium diphtheriae (C. diphtheriae) is a relatively rare pathogen in most Western countries. While toxin producing strains can cause pharyngeal diphtheria with potentially fatal outcomes, the more common presentation is wound infections. The diphtheria toxin is encoded on a prophage and can also be carried by Corynebacterium ulcerans and Corynebacterium pseudotuberculosis. Currently, across Europe, infections are mainly diagnosed in travelers and refugees from regions where diphtheria is more endemic, patients from urban areas with poor hygiene, and intravenous drug users. About half of the cases are non-toxin producing isolates. Rapid identification of the bacterial pathogen and toxin production is a critical element of patient and outbreak management. Beside the immediate clinical management of the patient, public health agencies should be informed of toxigenic C. diphtheriae diagnoses as soon as possible. The collection of case-related epidemiological data from the patient is often challenging due to language barriers and social circumstances. However, information on patient contacts, vaccine status and travel/refugee route, where appropriate, is critical, and should be documented. In addition, isolates should be characterized using high resolution typing, in order to identify transmissions and outbreaks. In recent years, whole genome sequencing (WGS) has become the gold standard of high-resolution typing methods, allowing detailed investigations of pathogen transmissions. De-centralized sequencing strategies with redundancy in sequencing capacities, followed by data exchange may be a valuable future option, especially since WGS becomes more available and portable. In this context, the sharing of sequence data, using public available platforms, is essential. A close interaction between microbiology laboratories, treating physicians, refugee centers, social workers, and public health officials is a key element in successful management of suspected outbreaks. Analyzing bacterial isolates at reference centers may further help to provide more specialized microbiological techniques and to standardize information, but this is also more time consuming during an outbreak. Centralized communication strategies between public health agencies and laboratories helps considerably in establishing and coordinating effective surveillance and infection control. We review the current literature on high-resolution typing of C. diphtheriae and share our own experience with the coordination of a Swiss-German outbreak.

Keywords: Corynebacterium diphtheriae; diphtheria; molecular epidemiology; public health; surveillance; toxin; whole genome sequencing.

Figure 1

Minimum spanning tree showing relationship of all available Corynebacterium diphtheriae genomes by cgMLST. All C. diphtheriae assemblies available from NCBI on 27.02.2019 were retrieved, and compared to isolates from our laboratory, as well as all reads available from NCBI on 08.10.2018, and those published, which were assembled using unicycler (168), and duplicate samples removed, giving n = 419 genomes. The cgMLST scheme of (35) was used within Ridom SeqSphere+ v4.1.6, with clusters given between nodes with five or fewer differences. Nodes are colored according to country of isolation.

Figure 2

Phylogenetic overview of all available C. diphtheriae genomes. All C. diphtheriae reads available from NCBI on 08.10.2018 were retrieved, and those published, and compared to isolates from our laboratory, as well as all assemblies available from NCBI on 27.02.2019, which were shredded to reads using wgsim in samtools (https://github.com/lh3/wgsim), and duplicate samples removed, giving n = 419 genomes. All reads were mapped against the reference genome CP003210 (155) within CLC Genomics Workbench 10.1.1, also used to generate a single nucleotide polymorphism (SNP) phylogeny with parameters that differed from the default as: variant calling with 10x minimum coverage, 10 minimum count and 70% minimum frequency, and SNP tree creation with 10x minimum coverage, 10% minimum coverage, 0 prune distance and including multi-nucleotide variants (MNVs). Metadata was retrieved from the NCBI database and was associated with the phylogeny using phandango (169). Colors use the same key as Figure 1; years are shown on a continuous scale. The bottom clade shows the clearly separate cluster proposed as C. diphtheriae subsp. lausannense.