WGS for surveillance of foodborne infections in Denmark

Benefits and potential drawbacks

Eva Møller Nielsen

Head of unit, PhD

Foodborne Infections

Statens Serum Institut

Copenhagen, Denmark

WGS - Benefits and drawbacks

Performance

- Confidence in clusters and links

- Variety of outputs based on one laboratory procedure

- Flexibility – not pathogen specific, avoid batches

- No phenotype

- Education and experience needed for analysis and interpretation

Costs

- Cost-effective alternative to classical typing

- Flexibility – same person can analyse different pathogens

- Expensive to establish (equipment, education) – but can be shared

across departments

WGS for food safety – views from the Danish public health

Implementation of WGS for surveillance of infections in Denmark

Small country’s perspective on implementation of WGS

Building on established and functioning surveillance system

Implementation for minimal extra resources (no extra funding)

• Infrastructure, equipment, personnel

• Limited parallel use of old and new methods

WGS implemented for routine surveillance in 2013

- Starting with WGS of all Listeria, replacing other methods

- Outbreaks of all foodborne pathogens

- STEC/VTEC surveillance from 2015

Laboratory-based surveillance of human infections

Real-time typing/characterisation of isolates from patients:

- Detect clusters

- Outbreak investigations/ case definition

- Linking to sources/reservoirs

- Determine virulence potential

- Antimicrobial resistance

Salmonella Typhimurium infections

MLVA types

In-house resources for WGS

2011-2012:

- Batches of project isolates were sequenced by external facilities

- Limited bioinformatics competences in our department

2013:

- Purchase of MiSeq – shared by all microbiology groups at SSI

- Bioinformatician hired

2016:

- Two MiSeqs – and need for more capacity

- Three bioinformaticians + more microbiologists have improved skills

Same laboratory staff performing WGS and classical methods

Same microbiologists/epidemiologists doing surveillance and taking action

Large outbreak: confidence in link to food source (41 cases)

August 2013

Real-time WGS of human isolates

July 7: 5 cases from 2014 in outbreak

July 16: matching food isolate

Clear case definition

10

0

90

80

70

Cluster

Cluster

Cluster

Cluster

Date

2014 Jan

2013 Marts

2013 Juli

2014 Marts

2013 Jan

2013 Jan

2013 April

2013 April

2013 Aug

2013 Sept

2013 Juni

2014 Jan

ST-1 isolates (n=12):

SNP analysis of WGS data of Listeria isolates belonging to ST-1

More clusters detected

Listeriosis:

75 patients part of cluster

68 patients sporadic

Linking “sporadic” cases + linking to food facility

10 cases 2013-15 (specific clone of ST-6):

Sep 2014:

Isolates from cold smoked fish from Company A identical to isolates from patients.

Food control intervention

Spring 2015:

New cases – have eaten smoked fish from supermarkets that sell products from Company A

Product and environmental samples at Company A were again positive for the ST-6 clone

SNP tree: Outbreak related and non-related ST-6 isolates

Squares: isolates related to the outbreak

Circles: isolates not part of the outbreak

Maximum parsimony tree:

All ST6 isolates from the years 2013-15

From a variety of laboratory methods to WGS

Mix of lab-techniques

serotyping, antimicrobial resistance, PCR, PFGE, MLVA, sequencing

Whole-genome-sequencing

Analysis of sequence data for different purposes (typing, virulence,…)

”Backward comparability” for some characteristics

Backward comparability

Pathogenic E. coli (e.g. STEC/VTEC)

Expensive and time consuming characterisation:

- PCR or hybridisation: virulence profile → pathogroup, virulence potential

- Classical O:H-serotyping with antisera → expected epidemiology, sources/reservoirs

- PFGE or MLVA: high-discriminatory typing → outbreak detection and investigations

Cost-effective to replace these methods by WGS

- Virulence genes can be extracted

- O:H-serotype can be predicted

- SNP-analysis gives a very high discrimination for cluster detection

12

Workflow – routine surveillance

13

Sequence

data

Serotype

SNP analysis

Outbreak

investigations

MLST

nomenclature

MLSTAntimicrobial

resistanceVirulence

genes

Risk

assessment

Treatment,

interventions

Performance

Defining clusters/outbreaks

- More confident definition of clusters/outbreaks

- Better case definition

- Interpretation of data (- as for all typing methods)

- Re-define “rules” for a cluster (time span, similarity)

Improved source tracing

- More certain microbiological evidence for linking to sources

- Potential for correlation to time/evolution

Analysis still under development

- Validation

- Interpretation of data in relation to epidemiology

- International comparability

14

Costs, flexibility

Costs

- Investments in equipment

- Expensive reagents

- Education of staff

Workflow, flexibility

- One lab method for all bacteria and all typing needs

- Same overall approach for all bacterial pathogens

• organism-specific data analysis if relevant, e.g. for backward comparability

- Cost-effective if replacing several classical methods

Changes

- Need for other skills among laboratory staff

- Need for bioinformatics

- Organisation of lab work and data analysis (less organism specific)

- More money spend on reagents/kits, less on personnel?

Perspective for using WGS (Denmark/Europe)

Partly implemented in Denmark and other European countries

- Public health, routine surveillance of foodborne infections

- Food/veterinary labs

- Will be implemented with different pace in different countries

- Parallel use of different methods (same situation as with other methods)

European-level surveillance of foodborne infections

- Surveillance system for rapid detection of dispersed international outbreaks

• Presently based on isolate typing by PFGE and MLVA

• Plans for using WGS

- Joint human-food databases about to be established

International perspectives

- Comparability

- Nomenclature