The European Food Safety Authority (EFSA) asked the Panel on Biological Hazards (BIOHAZ) to deliver a scientific Opinion on the evaluation of molecular typing methods for major food-borne microbiological hazards (i.e. Salmonella, thermophilic Campylobacter, Shiga toxin-producing Escherichia coli (STEC) and Listeria monocytogenes)and their use for attribution modelling, outbreak investigation and surveillance, including data-related issues.
Following that request, the BIOHAZ Panel adopted, on 5 December 2013, a Scientific Opinion addressing the evaluation of molecular typing methods and their suitability for the different applications that were considered (EFSA, 2013a). Important conclusions of that Opinion were that data from strain characterisation should be linked with epidemiological data, that the selection of isolates must be unbiased and statistically representative of the population to be assessed and that international harmonisation of molecular characterisation outputs by means of standardisation or appropriate quality control procedures is essential. Important recommendations were that cross-sector (humans, food, food animals and related environments) and international coordination of method validation is required as a priority, and that development and improvement of international initiatives with regard to harmonised platforms for sharing of data should be urgently prioritized, including the integration of Whole Genome Sequencing (WGS) into such international platforms.
In the current scientific Opinion, the BIOHAZ Panel has addressed data-related issues, in particular: (i) the evaluation of the requirements for the design of surveillance activities for food-borne pathogens, especially regarding the selection of statistically representative group of isolates to be included in molecular typing investigations and attribution modelling; and (ii) the requirements for harmonised data collection, management and analysis, with the final aim of achieving full integration of efficient and effectively managed molecular typing databases for food-borne pathogens. In order to provide a comprehensive overview of the applicability of molecular typing methods for the aforementioned food-borne pathogens in the given applications, both the Opinions should be referred to.
In the scope of this Opinion, the term ‘monitoring’ has been applied to describe a system of collecting, analyzing and disseminating data on the occurrence of zoonoses, zoonotic agents and antimicrobial resistance of public health relevance in the food chain. ‘Surveillance’ is understood as the systematic ongoing collection, collation and analysis of information related to food safety and the timely dissemination of information to appropriate persons so that action can be taken. Public health surveillance has been defined as the ongoing, systematic collection, analysis and interpretation of health data, essential to the planning, implementation and evaluation of public health practice, closely integrated with the dissemination of these data to appropriate persons and linked to prevention and control.
Surveillance programmes based on active and harmonised sampling are most suitable for statistical analysis which may be used for testing hypotheses. They provide the most complete, accurate and representative data and are more likely to be suitable for source attribution and detailed/advanced epidemiological investigations and risk assessments, as long as the datasets are sufficiently large to support robust statistical analyses. Typing results of isolates collected from routine laboratory submissions where the isolates are linked to limited information can still be valuable and may help support food-borne outbreak investigations, generation of hypotheses, early detection of emerging pathogen subtypes and genetic studies of bacterial populations, but sampling bias should be taken into account when formulating conclusions.
The introduction of molecular typing-based surveillance should include the establishment of a continuous information cycle to provide accurate and representative data over time and space, to include the relevant typing characteristics of specified food-borne pathogens (i.e. Salmonella, STEC, L. monocytogenes and thermophilic Campylobacter spp.) in food animal species and key points in the food production chain. Currently, various non-comparable methods are applied for the molecular typing of these pathogens worldwide. Pulsed-field gel electrophoresis (PFGE), is still the most widely used method for subtyping of Salmonella, STEC and L. monocytogenes. For S. Typhimurium and S. Enteritidis, PFGE may be used together with Multi-Locus Variable number tandem repeat Analysis (MLVA); although MLVA is increasingly being used as the sole method. Multi-locus sequence typing (MLST) has been the method of choice for thermophilic Campylobacter but is being superseded by WGS. Routine molecular typing of Campylobacter jejuni/coli has not been shown to add value for outbreak detection but may contribute to source attribution studies for campylobacteriosis.
Integrated analyses will be optimised if surveillance activities incorporate complete datasets containing all relevant information on the isolate. Examples of such datasets are those related to the genotype and other characteristics such as serovar or antimicrobial resistance profile, coupled with accurate data on the effect on the host and related epidemiological data. At present, prototype databases cannot be used for surveillance purposes since they are not widely linked to epidemiological data. Thus, the development of linkage mechanisms to access complex genetic and epidemiological data within different databases may be required.
A key priority in relation to integrated public health surveillance is to determine a threshold value for the level of genetic variation amongst isolates that can still be regarded as epidemiologically related. This threshold will vary according to the organism under investigation, time frame, population size and geographical scope of the investigation of the chain of transmission. The discriminatory power of a method describes its capacity to assign different subtypes to epidemiologically unrelated strains in the population studied, and is thereby a tool for describing the threshold for separation of epidemiologically related and unrelated isolates. A high discriminatory power will often lead to the division of panels of isolates into many subtypes, where the probability of categorizing unrelated isolates to the same subtype is small. With increasing discriminatory power, the probability of assigning related strains to different subtypes may also increase. In contrast, a relatively low discriminatory power will result in fewer subtypes and the probability of categorizing related isolates to different subtypes is small, but the probability of including unrelated isolates in the defined subtype is likely to increase. In the integrated analysis of typing data and epidemiological data it is important to optimise the discriminatory power/threshold for separation in a way which gives the most meaningful grouping of isolates from an epidemiological perspective to obtain the highest level of epidemiological concordance.
The collection of data for molecular typing-based surveillance of food-borne pathogens in animals, feed and food should be based on active sampling and an agreed sampling design should be prioritized for the purposes of molecular surveillance of pathogens in the food chain and from human cases. The use of alternative sources of data and strains should be carefully evaluated according to the required outcome and to a set of established criteria. The applied molecular typing methods should be based on both the pathogen to be characterised and the level of discriminatory power required depending on the required application of the surveillance results. Furthermore, molecular typing data should be coupled with a minimum required set of epidemiological data including, for example, information on the sampling context and population/sample set under study. Datasets generated should be comparable and suitable for joint analysis with other data from parallel surveillance in humans and/or relevant samples. Surveillance activities should be primarily aimed at investigating the priority source/pathogen combinations and be robust and statistically based. Rules for assembling strain collections and associated provenance data from general surveillance of pathogens should be agreed and introduced as EU standards.
When assessing requirements for integrated and harmonised data collection and management activities, the data collection process and the characteristics of the data repository should ensure the highest level of both the reproducibility of data and analyses, over time and space, and maximise the compatibility and interoperability among different datasets. This would be best accomplished by providing the overall architecture of a surveillance programme that includes the highest level of harmonisation with either international standards, if available, or a uniform approach to collection, management and analysis of data. Opportunities for harmonisation are facilitated by European Union Reference Laboratories (EU-RLs) which have an important role to support harmonisation in the laboratory characterization of food-borne hazards and active involvement in coordination of development and implementation of new molecular typing methods will be an important priority within the remit of EU-RLs in future years. Development of molecular methods for characterisation of food-borne pathogenic bacteria in animals, feed and food should be harmonised with those adopted for the surveillance of similar food-borne pathogens in the human population. Reference methods and materials, including sequence data, should be adopted for typing characterization of food-borne pathogens, and upload of data should be allowed only for approved laboratories.
Since the rapid development of sequence-based methodology is likely to outstrip the capabilities of individual centres of expertise, ongoing international expert consultation and oversight is required to optimise the opportunities offered by WGS. This should involve specialist centres, specialist scientists, bioinformaticians, risk assessors and risk managers from public health, veterinary, food production and retail sectors to identify issues and design a consensual ‘one health’ approach. Finally, the BIOHAZ Panel strongly recommends the establishment of a joint EFSA-ECDC-EU-RLs committee for the support of cross-sectoral molecular surveillance, to represent a balance of expertise from the public health and veterinary/food sectors as well as epidemiologists and microbiologists.