Clostridioides difficile in national food surveillance, Slovenia, 2015 to 2017

Background Clostridioides difficile is an important human and animal intestinal pathogen. Because of increasing indications of an association between C. difficile and food, in 2015, the Administration of the Republic of Slovenia for Food Safety, Veterinary Sector and Plant Protection (UVHVVR) included C. difficile in its national food surveillance. Aim We aim to report the results and experience with a nationwide and long-term testing of food for C. difficile as a part of a regular national food surveillance programme. Methods Retail minced meat and meat preparations (beef, pork and poultry) were sampled within a three-year period, 2015 to 2017. Selected raw retail vegetables, leaf salads and root vegetables, and ready-to-eat salads were only sampled during 2016 and 2017. Seafood was only sampled in 2017. Results Altogether, 434 samples were tested, with 12 of 336 (3.6%) meat samples and 6 of 98 (6.1%) raw vegetables contaminated with C. difficile. Twelve of 18 recovered food isolates were toxigenic (toxinotypes 0, III, V, XII). The isolates belonged to 13 different PCR ribotypes, 001 being most common (5 isolates). Several food types with an increased potential of being contaminated with C. difficile were detected by surveillance. Conclusion The three-year C. difficile testing within the national food surveillance revealed a low proportion of C. difficile-contaminated food and high genotype variability. Because the risk of C. difficile infection associated with C. difficile-contaminated food is unknown, no measures were recommended in the case of positive results.


Introduction
Clostridioides difficile is an important cause of intestinal infections in humans. The global rates of C. difficile infections (CDI) have increased considerably over the past two decades [1][2][3][4]. In Slovenia, the reporting of CDI is obligatory. According to the report on infectious disease surveillance in Slovenia for the year 2017, provided by the National Institute for Public Health, the number of cases increased from 316 in the year 2013 to 665 in the year 2017 [5]. No community outbreaks were reported in Slovenia during this time and only a few outbreaks were reported in hospitals [6]. The reason for this increase is not completely clear, but could include improved reporting and changes in laboratory practice.
For more than a decade, the zoonotic potential and food-borne transmission of C. difficile have been hypothesised to contribute to its spread [7][8][9][10][11]. The zoonotic transmission is supported by the occurrence of C. difficile in animals, and the shared PCR ribotypes between humans and animals. A more precise wholegenome analysis has confirmed an overlap of genetically related strains between humans and animals [12][13][14][15].
Food studies have mainly focused on meat and meat products; however, raw and ready-to-eat vegetables, seafood and milk can also contain C. difficile [10,11,16]. Most of the reported prevalence rates vary from not detected to 12.5% in meat and 7.5% in vegetables [10,11,17], but can rise above 50.0% in seafood and root vegetables [18][19][20][21]. As with animal strains, PCR ribotypes of food isolates can also overlap with human PCR ribotypes and can be genetically undistinguishable from human isolates by MLVA, a method with higher discriminatory power than PCR ribotyping [16].
Because of increasing indications of moderate to high rates of C. difficile contamination of certain food, in 2015, the Administration of the Republic of Slovenia for Food Safety, Veterinary Sector and Plant Protection (Uprava za varno hrano, veterinarstvo in varstvo rastlin (UVHVVR)) included C. difficile in the national surveillance programme of raw meats. In 2016, the C. difficile surveillance was expanded to selected vegetables. The objective was to determine the burden of C. difficile in retail food in Slovenia based on systematic data collection and analysis. Here we present the results of C. difficile detection in Slovenian retail food based on 3 years of national food surveillance.

Participating laboratories
Two different institutions provide the sampling and microbiological analysis of food samples in national food surveillance program. Sampling and analysis of meat and seafood is performed by the National Veterinary Institute, which operates under the auspices of the Veterinary Faculty, University of Ljubljana. Food of non-animal origin is sampled and analysed by the National Laboratory for Health, Environment and Food (NLZOH). Both institutions followed the annual sampling instructions released by the UVHVVR. Results are reported to UVHVVR at the end of yearly surveillance and included in its annual report.

Sampling
Different types of raw meat, including bovine or/and pork minced meat and meat preparations, poultry meat and poultry meat preparations were collected at food markets and grocery stores in all Slovenian regions from April to December 2015 and from March to December in both 2016 and 2017. Meat preparations are defined in European Union regulation (EC) 853/2004 as fresh meat, including meat that has been reduced to fragments, which has had foodstuffs, seasonings or additives added to it or which has undergone processes insufficient to modify the internal muscle fibre structure of the meat and thus to eliminate   Meat preparations refer to raw meat with additional ingredients. c PCR ribotype 001 was detected in two samples.

Table 3
Number The toxinotyping results were congruent with toxin testing results with rapid membrane enzyme immunoassay C. DIFF QUIK CHEK COMPLETE (Alere, Galway, Ireland) for all the isolates.
the characteristics of fresh meat [22]. Seafood was only sampled from March to December 2017 (excluding October).
Like retail meat samples, vegetables were sampled systematically in all Slovenian regions, but only from March to December in 2016 and 2017 (excluding April 2016 and November 2017). Samples were obtained from various retailers, including restaurants, food markets, grocery stores and supermarkets.
Information on sampling location and date was recorded for each collected sample.

Origins of samples
C. difficile was found in our previous study on imported retail potatoes, indicating a potential way for C. difficile to spread between countries [21]. For this reason, information of food origin was also included in the analysis of the samples from the national food surveillance.
Of 319 meat and meat preparation samples, 266 were labelled 'made in Slovenia' (70.8%), 21 'made in Austria' (6.6%), 11 'made in Croatia' (3.4%), nine 'made in Germany' (2.8%), one 'made in Hungary" (0.3%) and one 'made in Denmark' (0.3%). A further five samples included ingredients from more than one country while another five had no information on origin available. Samples of shrimps (n = 9) originated from Lithuania, Denmark, Ecuador and the Czech Republic, while all samples of bivalve molluscs (n = 8) originated from Slovenia.

Isolation of Clostridioides difficile
Retail meat samples and seafood were analysed by the microbiological laboratory of the Institute of Food Safety, Feed and Environment, which is a part of the National Veterinary Institute. The isolation protocol for meat samples was previously established and evaluated by Biasizzo et al., 2018 [23], where samples spiked with 1,000, 100 and 10 CFU per sample were tested and 100% sensitivity was determined for all spiked samples.
Cycloserine cefoxitin fructose broth (90 ml) supplemented with taurocholic acid and lysozyme (TCCFB + L; prepared according to Biasizzo et al.) was added to 10 g of each meat sample in a sterile plastic blender bag and homogenised in a peristaltic blender (BagMixer, Interscience, St Nom la Bretèche, France) for 1 min. The samples were then incubated under anaerobic conditions at 37 °C for 7 days. The enriched samples were tested for C. difficile by quantitative real-time PCR and plated after spore selection by ethanol treatment onto blood agar plates (blood agar supplemented with 5% defibrinated bovine blood), C. difficile agar base with added C. difficile selective supplement and 7% (v/v) defibrinated horse blood ( cycloserine cefoxitin fructose agar (CCFA); all from Oxoid, Hampshire, United Kingdom (UK)) as previously described [23] and onto chromID C. difficile agar (bioMerieux, Lyon, France).
Retail vegetables were tested by NLZOH. Root and leafy vegetables were prepared differently, but for both, 25 g of the sample was analysed and the cultivation method was the same. Isolation was based on previously described cultivation method of Tkalec et al., 2019 [21] with detection threshold for root and leafy vegetables between 1 and 10 CFU per tested sample.
The recovery of C. difficile from root vegetables was assessed using sterile sponges (Polywipe, Medical Wire and Equipment (MWE), Corsham, UK), enrichment in BHIST broth (BHI supplemented with sodium taurocholate, yeast extract, L-cysteine, cycloserine and cefoxitin), spore selection by ethanol treatment and plating onto C. difficile selective agar (chromID C. difficile agar (bioMerieux)) [21]. Sampling with sponges is considerably more sensitive than using swabs [24] and can, together with enrichment step in cultivation, enable detection of 1 to 10 spores for swabbed potatoes [21].
The leafy vegetables were homogenised and incubated in blender bags (BagMixer, Interscience) in 200 ml of BHIST. C. difficile was isolated after the spore selection and plating onto chromogenic agar plates (bioMerieux) as described for root vegetables.

Molecular characterisation of Clostridioides difficile isolates
All isolates were characterised by toxinotyping and PCR ribotyping. Toxinotyping was done by PCR amplification and restriction analysis of A3 and B1 fragments of tcdA and tcdB genes as previously described by Rupnik and Janezic [25]. The binary toxin gene was detected by partial amplification of cdtB [25].
PCR ribotyping was performed with Janezic primers as described by Janezic [26]. Profiles obtained by agarosebased PCR ribotyping were analysed by BioNumerics software version 7.6 (Applied Maths NV, Sint-Martens-Latem, Belgium) and compared with PCR ribotypes in our large PCR ribotype library that includes ca. 300 different PCR ribotypes, comprising isolates from Slovenia and abroad. Some of the reference library strains are representatives of internationally recognised ECDC/ Leeds PCR ribotypes. If there was no match with these reference strains, the ribotype has a local designation (SLO+number).

Statistical analysis
The data was analysed by R version 3.4.4 (R Foundation, Vienna, Austria). The analysis was performed for each categorical variable where the proportions were given. Proportions were compared using Fisher's exact test. Statistical significance was held at p < 0.05.

Results
In years 2015, 2016 and 2017, a total of 345, 363 and 462 food samples of animal origin were tested, respectively (Table 1) (Table 1).This means that a total of 434 retail food samples collected as part of the national surveillance programme were tested for C. difficile. The overall prevalence of C. difficile in retail meat was 3.6% (12/336) ( Table 2) and in retail vegetables it was 6.1% (6/98) ( Table 3).

Clostridioides difficile in retail meat and seafood
In  (Table 2). Overall, C. difficile was isolated from 12 of 336 meat samples; five poultry meat preparations, three beef and/or pork meat preparations, three raw poultry meat and one bivalve molluscs. C. difficile was not detected in any of the 56 tested pork and/or beef minced meat samples. Of the 12 C. difficile-positive samples, 10 were made in Slovenia (3.8% of all tested samples made in Slovenia) and two were made in Austria (9.5% of all tested samples made in Austria).
There was no significant statistical difference in C. difficile prevalence between different meat types/ categories or sampling locations (food markets vs grocery stores) (p > 0.05).

Clostridium difficile in retail fresh vegetables
In 2016, two of 48 tested samples were C. difficile-positive and four of 50 samples tested positive in the year 2017 (Table 3). Collectively, C. difficile was isolated from six of 98 (6.1%) tested samples, in particular from three of 21 raw leaf salads, one of 20 root vegetables and two of 57ready-to-eat salads ( Table 3). Comparison of C. difficile prevalence according to sample type (root, unprocessed salad, ready-to-eat salad) and location of sampling (supermarket, food market, farm, catering/restaurants) showed no significant statistical difference between the samples (p > 0.05).
Vegetables from three of six C. difficile-positive samples were grown in Italy, two in Slovenia and one in Poland. None of the positive vegetables shared sampling time point or location of sampling.

PCR ribotypes and toxinotypes in retail food samples
The 18 C. difficile isolates obtained from food belonged to 13 different PCR ribotypes (Tables 2 and 3). The most prevalent PCR ribotype was PCR ribotype 001 (n = 5), followed by 010 (n = 2). Both were detected in food of animal and non-animal origin. The other 11 PCR ribotypes were detected only once. One nontoxigenic PCR ribotype, SLO 283, was new to our PCR ribotype library. It was isolated from a beef/pork meat preparation of Slovenian origin.
The majority of PCR ribotypes were already found in Slovenia, either in patients, animals, food or the environment (Table 4).

Discussion
C. difficile can be found in different types of food and the reported contamination rates vary between food types, countries and studies. The main novelty of this report is that it describes the first documented longterm, national C. difficile food surveillance performed thus far.
We found that the proportion of C. difficile positive meat and vegetable samples numerically increased every year. However, the number of tested samples was too low to speculate on any possible trends. The changes in samples sizes and produce types could have contributed to this observation. Except for vegetables in 2017, the average annual positivity rate for food of animal or non-animal source was 4.6% or lower.
The proportion of meat samples positive for C. difficile detected during Slovenian food surveillance varied from 2.5% to 4.6% annually, and was 3.6% for entire 3-year interval. There are no published data available for Slovenia on meat contamination with C. difficile. In previous small sporadic pilot studies performed by NLZOH where 124 meat samples of pork, beef, sheep and poultry were tested, all were negative for C. difficile (Tkalec and Rupnik; data not shown). Therefore, it seems that the systematic sampling and optimised methodology [23] used in national surveillance contributed to better detection. Early studies about possible presence of C. difficile in food of animal origin from North America reported high contamination rates of meat and meat products (20% in beef and veal ground meat in Canada [27]; 14.3% to 62.5% depending on the meat sample type, in the US [28]). These high contamination rates have not been reported in subsequent European or other non-European studies [8,10,11,17]. In general, the meat testing results were comparable to other European countries reporting from none to 6.3% [10,11,17] and to most of the recent reports from non-European countries reporting from none to 12.5% contamination rate in different types of meat [10,17,29]. In our sample set, beef/pork minced meat were always negative for C. difficile; this was also the case in some reports from Switzerland and Austria [30,31]. However, other studies from Austria, Belgium, France and Sweden reported positivity rates from 1.9% to 4.7% [11,17]. Poultry meat or meat preparations are not often included in the testing. In contrast to our finding of overall 4.4% (8/180) of C. difficile positive poultry samples, a 2.7% positivity rate in poultry meat was reported from the Netherlands while C. difficile was not detected in any poultry meat sample in Austria or Sweden [30,32].
The contamination rate of bivalve molluscs (5.9%; n = 1) was determined on a small number of samples (n = 17). But similar percentages (3.9%, 4.5% and 4.8%) were reported for Italy, Texas and Canada, respectively [17,29]. It seems that presence of C. difficile in mussels varies widely, as two other Italian studies found 49% (26 of 53 samples, and 75% (four of six samples) of mussels collected at farms or retailers in Naples, Italy to be positive for C. difficile [18,19].
The proportion of C. difficile-positive vegetable samples (6.1%) is comparable to the low to moderate percentages of contaminated vegetables and ready-to-eat salads reported by others [10,[33][34][35]. Again, such comparisons are difficult not only because of the difference in methodology, but also because of heterogeneity of sampled vegetables. For this reason, the proportion of C. difficile-positive vegetable samples found during national C. difficile food surveillance cannot be directly compared with our previous study (9.4%) [21]. Leaf salads were relatively often contaminated with C. difficile in Slovenian food surveillance (14.3% of 21 samples). There is not much data on the testing of raw leaf salads, but a previous Slovenian survey found 8.8% of retail leaf salads positive for C. difficile [21], while a report from Iran found 5.7% positive and two studies from the US found 4.6% and 13.8% positive, respectively [29]. Ready-to-eat salads have been tested in France and Scotland, with 2.9 and 7.3% of samples positive, respectively [33,36]. However, the type of the ready-to-eat salads is vaguely described in most studies.
Of the root vegetables we tested (carrot, parsley and beetroot), only a single parsley sample was contaminated with C. difficile. An Australian study that tested organic root vegetables reported much higher positivity rates; 22.2% for beetroot, 5.6% for onions and 5.3% for carrots [20]. The vegetable reported by us and others to have a very high contamination rate (from 28.0% to up to 55.6%) is the potato [20,21], but potatoes were not included in the Slovenian food surveillance programme as these data were not available when the programme was started.
A large variety of PCR ribotypes and toxinotypes was found and it was slightly higher in vegetables than in meat. The six strains isolated from vegetables belonged to six different PCR ribotypes while the 12 strains isolated from meat belonged to nine different PCR ribotypes. Only ribotypes 001 and 010 were shared between meat/seafood and vegetables. All the strains of PCR ribotype 001 originated from Slovenian samples, while samples yielding strains of PCR ribotype 010 were from Italy (salad) and Slovenia (meat). Both PCR ribotypes 001 and 010 are rather diverse, and whole genome sequencing should be used to confirm the genetic (non)clonality. Of the 13 detected PCR ribotypes, only one was new to our PCR library. Some of the detected PCR ribotypes, such as 001, 010 and 078, are among the most prevalent in different human, animal and environmental studies from several countries, suggesting that certain PCR ribotypes are well adapted to survival in several different niches.
Some C. difficile-positive food samples in the food surveillance were of non-Slovenian origin. Another group of positive samples included foodstuffs that were composed of several ingredients, some of which had non-Slovenian origin. Positive samples or only their ingredients originated from Italy, Poland, Austria, Belgium and Lithuania. Therefore, as suggested earlier by the authors [21], food in general could possibly contribute to cross-border transmissions.
Theoretically, the C. difficile contamination sources of food could be very heterogeneous. For meat, the most direct way is faecal contamination during slaughtering, but post-production processing could also play a role. Despite a very low, close to zero, reported prevalence of C. difficile in poultry before slaughter by some groups [8,37], higher contamination rates, up to 12.8%, of raw poultry meat have been documented in North America [37,38] and in our surveillance (5.0%). This suggests a post-slaughter contamination source in the production line. A possible source for the contamination of both meat and vegetables could also be via contaminated hands during handling. The main sources of vegetable contamination are soil or indirect faecal contamination via irrigation or manuring [39]. Non-toxigenic strains are often found in rural soils [40] and were also commonly isolated from vegetables during our surveillance. This constant environmental exposure of vegetables to C. difficile spores is probably one of the reasons for higher contamination rates of vegetables compared to meat and meat products.
C. difficile food surveillance had several limitations. One of the limitations is that no baseline contamination rates were known for Slovenia. It is for this reason that testing for C. difficile was initially added for all samples of raw meat, meat preparations and raw vegetable planned to be sampled for other microbiological parameters in the surveillance programme. Another limitation is the difference in the methodology used by both institutions providing the testing. Institutions used their own optimised methods that differed in the quantity of the sample (higher for vegetables), enrichment medium and use of molecular methods for screening of positive enrichment cultures, used for meat only. This difference in the methodology could have contributed to the higher proportion of vegetable samples positive for C. difficile (6.1%) compared with meat samples (3.6%). However, both methods were previously shown to have detection limits of at least 10 spores per sample. Furthermore, because of different sampling procedures, different types of tested food, bias because of the low sample number and, more notably, the differences in detection methods [11], comparisons among published studies should be made cautiously.
One of the aims of the national food surveillance is consumer protection. However, the detection of C. difficile in food during the surveillance has not resulted in any recommended measures because the risk associated with contaminated C. difficile food is unknown. Given the low detected prevalence of C. difficile contaminated food samples, the food safety risks for the tested food types are likely very low. On the other hand, although the levels of food contamination with C. difficile spores are usually low [41], the constant exposure to the low spore levels in combination with a disrupted gut microbiota or immune incompetence could represent increased risk for C. difficile infection (CDI).
In summary, C. difficile testing within the national food surveillance programme over the course of 3 years revealed a low percentage of C. difficile-contaminated food and a high genotype variability. Surveillance detected that some food types included in regular food surveillance, such as meat preparations and leafy salads, are more likely to be contaminated. Because the risk associated with C. difficile contaminated food is unknown, no measures have been recommended in terms of any of the 18 foods positive for C. difficile.