Introduction
Salmonellosis is one of the most frequent zoonotic diseases worldwide. In Europe,
Salmonella enterica ssp.
enterica continues to be the second most reported zoonosis and a common cause of foodborne outbreaks responsible for about 20% of all outbreaks in the EU in 2014 [
1]. Symptoms range from self-limiting diarrhoea to life-threatening sepsis. Molecular typing of
Salmonella strains aids in improving the identification, investigation, and control of an outbreak source so that prevention of further spread is effective. In the EU, the European Centre for Disease Prevention and Control’s (ECDC’s) disease programme for Food- and Waterborne Diseases and Zoonoses (FWD DP) is responsible for the surveillance of salmonellosis. In order to improve the surveillance of foodborne infections, molecular typing data have been reported to the European Surveillance System (TESSy) by the Member States since 2012 [
2]. For
Salmonella, this included pulsed field gel electrophoresis (PFGE) data for all serovars and multiple locus variable number of tandem repeat analysis (MLVA) data for
Salmonella Typhimurium. The molecular typing for surveillance is based on the capacity of the laboratories in the European FWD network (FWD-Net) to produce typing data of good quality to allow for reliable cross-border surveillance and outbreak detection. The current molecular typing methods used for EU-wide surveillance and outbreak investigation of
Salmonella, are PFGE suitable for typing all serovars [
3,
4], and MLVA for
S. Typhimurium [
5,
6] and
S. Enteritidis [
7], two of the most prevalent serovars in the EU/EEA [
1]. MLVA provides a higher discriminatory power compared to PFGE within
S. Typhimurium, and has been used for surveillance and outbreak investigations in the past decade [
8‐
10]. Likewise, there was a need for a more discriminatory method than PFGE for detecting outbreaks of serovar Enteritidis. Therefore, a multi-laboratory validation study of a 5-locus MLVA scheme for
S. Enteritidis was recently carried out [
7] and since June 2016 MLVA data for
S. Enteritidis have also been included in TESSy. Thus, MLVA for
S. Enteritidis will be included in future EQAs. Protocols for PFGE and both MLVA methods have been standardized to produce comparable results across laboratories [
7,
11‐
15].
Since 2012, the National Public Health Reference Laboratories (NPHRLs) in the FWD-Net have been given the opportunity to participate in External Quality Assessments (EQAs) for the molecular typing methods used in the European surveillance of foodborne pathogens. The evaluation of the PFGE results of the first round of EQAs (2012-EQA) for the three pathogens
Salmonella,
Listeria, and verotoxin-producing
E. coli is reported by Schjørring et al. 2016 [
16]. In the present study, the results of the following three EQA rounds related to molecular typing of
Salmonella strains by PFGE for all serovars and MLVA for
S. Typhimurium will be presented. The aims of each EQA round were to assess the quality of PFGE typing and the comparability of the test results among the NPHRLs in the FWD-Net, as well as to determine and ensure the quality and integrity of the
S. Typhimurium MLVA results among the participating laboratories. Here, we combined the results of the three rounds which took place in the years 2013–15 and assessed the development in the capability of the NPHRLs in producing molecular typing data that can be used for supporting the surveillance of
Salmonella infections in Europe. We will focus on development in the capacity to perform the standard methods, the quality and comparability of the typing data, and the identification of common quality issues.
Discussion
Definitive characterization of foodborne pathogens by molecular typing is essential for surveillance, outbreak detection, and source identification. The globalization of food product trade implies that multi-country food-associated outbreaks may occur more frequently [
22‐
25]. Cross-border comparison of molecular typing data is dependent on robust and comparable standard typing methods. For
Salmonella, standard PFGE and MLVA protocols [
11,
13‐
15] and nomenclature have been developed [
12,
21]. Laboratory procedures for the methods accepted for EU-wide surveillance are available at the ECDC website (
http://ecdc.europa.eu/en/healthtopics/food_and_waterborne_disease/surveillance/Pages/index.aspx). However, the full potential of cross-border surveillance relies on ability of the laboratories to generate high-quality typing data routinely and share data as close to real-time as possible. Therefore, this study addressed the quality of the typing data reported from the European NPHRLs expected to share data in TESSy in order to identify critical parameters for improving surveillance, laboratories in need of technical assistance, and performance development.
PFGE has been the international gold standard for molecular subtyping of Salmonella, and almost all (27/29) participating NPHRLs registered to this part of the assessment scheme. The quality of a PFGE gel depends on strict adherence to the standard protocol. Even slight deviations from the procedure may result in reduced quality and comparability. This was reflected in the low number of gels (7%) obtaining a total 100% score. Only three laboratories (10%) managed to produce perfect gels, which highlights the challenges related to this method. However, an excellent gel quality is not a necessity for inter-laboratory comparison of profiles, and only two laboratories were unable to produce profiles of high enough quality to be useful for comparison. To improve the performance of these two laboratories, direct technical assistance or training might be useful, as the detailed feedback on their specific problems through individual evalution reports did not circumvent their low performance.
The comparability of profiles relies primarily on the use of correct running conditions, good quality image acquisition, and distinct bands. Indeed, these gel parameters, “image acquisition and running conditions” and “bands”, generated non-acceptable profiles in all EQA rounds. The additional gel parameters received an acceptable score in all cases and seem to be more robust to deviations.
It is of utmost importance to use the PFGE protocol for the relevant organism; however, some (n = 4) laboratories were suspected of applying incorrect running conditions, e.g., the electrophoresis conditions for E. coli O157 were used for Salmonella. Normalization of bands according to an international standard directly depends on correct standard running conditions and adequate numbers of reference lanes. Thus, deviation from the standard renders the profiles incapable of being compared. Laboratories performing PFGE routinely should be expected to use correct running conditions as detailed in the provided protocols. In addition to the use of correct running conditions, proper image acquisition and distinct bands are essential. Production of clear and distinct bands is highly susceptible to deviations from the protocol, and indeed the parameter “bands” obtained the lowest average score through the three EQA rounds.
Fuzzy and/or thick bands were common causes for non-acceptable and low “band’ scores. In a few cases (4/14 poor band scores), entire lanes were distorted as well. Several factors cause bands to appear fuzzy; poor image capture by improper focusing, use of too narrow wells, and thick gel slices >2 mm. In the specific feedback, laboratories were encouraged to evaluate their in-house procedure systematically to identify even slight deviations from the standard protocol. Low-quality image acquisition highly affects the band quality. The presence of weak bands most likely led several laboratories to increase the exposure time and/or enhance the contrast of the gel image, causing the bands to appear too thick to separate double bands. Thus, laboratories presumably producing a gel of acceptable quality failed to document this due to improper image capture. A thorough protocol on correct image acquisition was distributed to the participants in the 2014- and 2015-EQA rounds, and the quality increased from 2014 to 2015. The combined gel parameter “image acquisition and running conditions” could be separated into two different parameters in future EQA rounds to assess the parameters individually.
It can be speculated that the overall quality of PFGE gels produced in NPHRLs not performing this method routinely would diminish. Thus, the substitution of PFGE with whole genome sequence (WGS) typing methods will reduce gradually the need for PFGE EQA schemes. Future assessment schemes should represent the different methodologies applied in parallel in the EU, including PFGE, MLVA and WGS-based analyses.
Participants could select which parts of the EQA to perform, but although the majority (27/29) of the participating NPHRLs performed PFGE typing, only 19 performed the subsequent profile analysis. Laboratories potentially lack access to the software or lack experience in analysing PFGE profiles. However, PFGE-based analysis requires the capacity to analyse PFGE profiles. The performance level displayed by those laboratories participating in the gel analysis was high and improving during the 3-year period. In the EQA-2015, all laboratories demonstrated the ability to produce an acceptable gel analysis in accordance with the guidelines.
The overall performance level for MLVA typing was high and improving through the EQA rounds; all participants reported 100% correct allelic profiles in at least one EQA round. A low rate of 4% (25/420) of misreported allelic profiles was seen among the 420 profiles reported in the 3 years. These errors could have been avoided by proofreading of the results and by taking into consideration the known characteristics of the individual locus, e.g., allele out of expected range, allele not seen before, or locus always present. Absence of amplification of STTR3 was reported a total of five times by two laboratories, although absence of this locus should be considered highly unlikely. In the MLVA scheme, only absence of the loci STTR6 and STTR10 is likely [
5]. Furthermore, two of the four incorrect allele numbers reported for STTR3 (allele 409 and 410) are uncommon alleles. STTR3 is a combined locus of two repeat sizes, consisting of 0–5 27-basepair repeat units and 8–14 33-basepair repeat units. However, not all combinations of these seem to occur [
12]. Since the incorrect allele numbers of STTR3 were only reported in two of the ten test strains in one of the three EQA rounds, it was not considered as a general error in the participant’s procedure. Critical evaluation of results and the use of checkpoints to control the quality and to assess if values are within the expected range and is of high importance.
In summary, lack of proofreading contributed considerably to the errors identified in the MLVA typing. Although these errors were few and should be easy to avoid, uploading of incorrect allelic profiles to TESSy can have significant impact on outbreak detection, since MLVA data are not curated. Thus, proofreading by experienced personnel should be done before sharing of data. TESSy could improve by curation of MLVA data, i.e., confirmation when rare alleles occurs.
Recently, a standard operating procedure for MLVA typing of
S. Enteritidis was published [
26], and MLVA data for this serovar are now collected in TESSy. Future EQA rounds will include this method to foster harmonization and ensure data quality in the same way as for
S. Typhimurium. In Europe,
S. Enteritidis and
S.Typhimurium are the two most common serovars reported in humans, and MLVA typing provides high resolution with good epidemiological concordance for both serovars [
27]. In comparison to whole genome sequence (WGS) typing methods, MLVA is low-cost, and easier to perform and interpret.
Conclusion
The EQA schemes strengthened the laboratory network of FWDs at the EU/EEA level as the number of laboratories participated increased, and results improved during the 3-year period. More harmonized and reliable typing methods, particularly molecular typing techniques, allow: (a) effective analysis to detect unusual events, detect uncommon increase of cases and possible national and cross-border outbreaks, (b) increased capacity for further characterization of human isolates, more accurate delineation of outbreaks, and (c) source identification and verification, comparison with data from the food/animals sector.
The participation in the PFGE laboratory part of the assessment was high, whereas the S. Typhimurium specific MLVA method received a lower participation. This emphasizes the need for continued assessment of the different methodologies applied in the NPHRLs in Europe, including both classical methods and new WGS-based typing methods, as it should be anticipated that several methods will be used in parallel by the different NPHRLs in a transition period. This poses a challenge for European-wide surveillance, but the historical data on PFGE and MLVA, combined with the possibility of producing WGS-based typing data on different resolution levels, improves the possibility, of to some degree, correlating results obtained by different methods. Currently, a harmonized procedure for WGS data analysis in routine surveillance remains to be developed. Future EQA shemes could foster the development and harmonization of WGS-based surveillance at the EU level; however, in the transition period, PFGE and MLVA will be a valuable tools for comparison.
External quality assessment is a valuable tool for identifying areas for improvement, and strong and improved performance levels were demonstrated through the course of the three EQA rounds. For PFGE, the main gel quality problems were related to image acquisition, running conditions, and indistinct bands. The current EQA scheme have especially supported the European NPHRLs to use the standard running conditions for PFGE, uniform PFGE band assignment procedures, use of MLVA reference strains, as well as uniform MLVA allelic nomenclature. The European NPHRLs’ interest in improving the molecular typing data quality and the participation in EQA schemes is important for the quality of the European-wide surveillance of foodborne infections.