Introduction
Enteric yersiniosis, primarily caused by Yersinia enterocolitica (YE), was the fourth most common zoonotic disease in European Union countries in 2023 with 8 738 confirmed cases and a notification rate of 2.4 per 100 000 population, with an increase of 13.5% compared to the rate in 2022 [1]. Germany accounted for the highest number of reported cases (n = 1 931), followed by France (n = 1 384) and Spain (n = 1 317).
In Europe, the most common type of YE reported in human infections is bioserotype 4/O:3 (87%), followed by 2/O:9 (11%) [1]. Pigs are an important reservoir of YE, especially bioserotype 4/O:3 [Reference Fosse, Seegers and Magras2, Reference Martinez3], whereas cattle seem to be associated with bioserotype 2/0:9 [Reference Le Guern4]. Most reported YE infections in Europe appear to be sporadic; yet, several foodborne outbreaks have been documented and linked to the consumption of pasteurized or unpasteurized milk [Reference Ackers5–Reference Hamad8], raw vegetables, and ready-to-eat products [Reference Karlsson9], as well as contaminated water [Reference Sabina10]. The route of human contamination is mainly through the ingestion of contaminated food and more rarely through a direct or indirect contact with animals [Reference Sabina10].
In France, microbiological surveillance of human yersiniosis is conducted by the National Reference Centre (NRC) for Plague and Other Yersiniosis. Since 2018, routine microbiological surveillance is based on the characterization of each isolate – species and lineage – and inclusion in a genomic cluster, using core genome multilocus sequence typing (cgMLST) schemes [Reference Savin11, Reference Le Guern12]. Santé publique France (the French National Public Health Agency) is contacted for the assessment of clusters which meet the outbreak alert criteria: at least 5 isolates belonging to a same genomic cluster isolated within 8 weeks [Reference Le Guern12]. This method is useful for public health surveillance and has proven effective for outbreak detection [Reference Savin13].
Outbreak detection
On 22 October 2024, the NRC notified Santé publique France of a cluster consisting of 14 cases of isolates of YE belonging to a same genomic cluster (number 4452) isolated between 17 September and 07 October 2024. An outbreak investigation team with experts from Santé publique France, the NRC, the French Agency for Food, Environmental and Occupational Health and Safety (ANSES), and the French General Directorate of Food (DGAl) launched extensive epidemiological, microbiological, and food investigations to identify the vehicle of transmission and propose appropriate control measures.
Methods
Epidemiological investigation
Outbreak case definition and case finding
A French outbreak case was defined as a resident in mainland France with laboratory-confirmed YE infection whose isolate was isolated between 17 September 2024 and 07 March 2025, characterized as lineage 2/3-9b (corresponding to bioserotype 2/O:9) [Reference Savin11] and belonging to the genomic cluster 4452 using the 1 727-gene Y. enterocolitica cgMLST scheme followed by single-linkage clustering with a three-allele mismatch threshold [Reference Le Guern12]. French yersiniosis cases were identified by the routine microbiological surveillance conducted by the NRC on Yersinia isolates sent on a voluntary basis by the French clinical microbiology laboratories.
A cross-border case was defined as a laboratory-confirmed case residing in a European country other than France, with an isolate’s genomic sequence belonging to cluster 4452 as identified for French isolates, with the isolation date between 17 September 2024 and 07 March 2025. Cross-border cases were identified from European countries through EpiPulse [14], the European surveillance portal for infectious diseases, run by the European Centre for Disease Prevention and Control (event 2024-FWD-00114), and genomic sequencing of raw data together with demographic and epidemiological data (when available) shared with France.
Data collection and analysis
Demographics (age, sex, and postal code of residence) of cases, as well as YE isolation dates, were available from medical laboratories’ declaration forms. All French cases (or their legal representatives) with isolation dates between 17 September and 8 November were contacted by phone for interviews. Interviews took place between 22 October and 28 November and were temporarily suspended between 28 November and 16 December, following the implementation of control measures. Interviews were resumed between 16 and 24 December and targeted cases with isolation dates at least seven days after the start of the product recalls. Interviews were conducted by epidemiologists using a standardized trawling questionnaire, covering clinical aspects (symptoms and onset date), medical history, detailed food consumption, contact with other persons experiencing diarrhoea, travel history, and contact with animals during the seven days prior to symptom onset. Place of purchase and loyalty card information when available were also collected to trace back supermarket purchases.
Food product and environmental investigations
Food trace back investigations were conducted by the departmental direction in charge of the protection of populations (DDecPP) and by the DGAl. Suspected food products (30 cheeses with different ripening periods) and environmental samples (14 milk filters from 4 milk producers) were taken by DDecPP and sent to the ANSES Ploufragan Laboratory, France, for YE detection and characterization.
Microbiological investigation
Methods used, respectively, by the NRC and the ANSES, to isolate, characterize, and compare isolates for clinical, food, and environmental samples were described previously [Reference Savin13]. Relationships between genome isolates were visualized using a minimum spanning tree (Bionumerics v7.6.3), which was reconstructed by single linkage clustering based on allelic profiles obtained with the 1 727-gene Y. enterocolitica cgMLST. Phylogenetic inference was performed on the 1 727 loci that were first extracted with the ‘Export’ plugin available at https://bigsdb.pasteur.fr/yersinia, concatenated for each isolate, and subsequently aligned using MAFFT v7.467 [Reference Katoh and Standley15]. The resulting alignment served as input for IQ-TREE2 [Reference Minh16], where the optimal nucleotide substitution model was identified with ModelFinder (integrated in IQ-TREE2), and a maximum-likelihood tree was constructed with 1 000 ultrafast bootstrap replicates.
Results
Epidemiological investigation
Description of the outbreak in France
As of 02 May 2025, 167 French cases had been identified with isolation dates between 17 September 2024 and 07 March 2025 (Figure 1a).
Distribution of French outbreak cases of Yersinia enterocolitica lineage 2/3–9b (bioserotype 2/O:9) (n = 167), France, September 2024–March 2025, by (a) sampling week, interview status, and reported consumption of the incriminated cheese type and (b) postal code of residence. (a) The solid vertical line marks the date of 12 November 2024, when an inspection was carried out at the cheese factory. The dotted vertical line marks the date of 21 November 2024, the initiation of the recall of products. (b) The red triangle marks the location of the cheese manufacturer from whom isolates were collected.
Figure 1. Long description
Panel A is a stacked bar chart. The y-axis is labeled Number of cases ranging from 0 to 25. The x-axis is labeled Isolation week with dates from 16/09/24 to 17/03/25. The bars are composed of individual squares representing cases. The legend indicates three categories: Not interviewed in grey, Interviewed - no consumption in light blue, and Interviewed - consumption in dark blue. Cases begin in mid-September 2024, peaking at 21 cases in early December 2024. Dark blue squares (consumption) are concentrated between late September and mid-November. A solid red vertical line is placed at 12/11/2024 and a dashed red vertical line at 21/11/2024 (which marks initiation of incriminated product recalls). Following the dashed line, the majority of cases are grey (not interviewed).
Panel B is a map of France showing the geographical distribution of cases. Blue dots representing residence locations are scattered across the country. There is a high concentration of cases in the East, specifically near the border with Switzerland. A single red triangle, marking the cheese manufacturer, is located within this eastern cluster. Other smaller clusters are visible in the North near Paris and in the West along the Atlantic coast. A single case is shown on the island of Corsica in the Southeast.
Median age was 50 years (range: 2–91), 4 cases were children under 5 years, 45 cases were aged over 65 years, and 84 (51%) were female. Cases originated from 11 of the 13 regions in mainland France, with 75 (45%) cases coming from the Bourgogne-Franche-Comté region (Figure 1b). Most isolates were collected from foecal samples (n = 160, 96%), four were isolated from blood samples, and two from other samples (dialysis fluid and pus).
All isolates were assigned to the cluster 4452 (Figure 2, Supplementary Figure S1) consisting of 229 clinical isolates identified between 01 August 2020 and 07 March 2025 (3 in 2020, 9 in 2021, 23 in 2022, 12 in 2023, 169 in 2024, and 13 in 2025).
Minimum spanning tree of Yersinia enterocolitica lineage 2/3–9b cluster 4452 isolates (n = 232) from human cases (n = 229), including French outbreak cases (n = 167) and cross-border cases (n = 5), and from incriminated cheese (n = 2) and milk filter (n = 1), May 2020–March 2025, by (a) year of isolation and (b) outbreak case definition. cgMLST, core genome multilocus sequence typing. Numbers close to the branches correspond to the allelic distance between two nodes according to 1 727-gene Y. enterocolitica cgMLST [Reference Le Guern12].
Figure 2. Long description
A two-panel figure labeled A and B displaying cgMLST minimum spanning trees.
Panel A is categorized by year of isolation. The central core consists of large, multi-segmented nodes dominated by light blue representing 2024 Clinical cases. Radiating from this core are numerous smaller light blue nodes. The upper-left branches contain a mix of red 2021 Clinical, yellow 2023 Clinical, and dark purple 2024 Cross border nodes. The top-most branch features light purple 2022 Clinical nodes. A single black node representing 2024 Cheese is located near the upper-right of the central cluster. Small numbers on the connecting lines indicate allelic distances, mostly ranging from 1 to 3.
Panel B uses the same tree structure but is categorized by outbreak case definition. The central core and the majority of the lower and right-hand branches are colored red, representing Clinical Outbreak cases. The upper-left and far-peripheral nodes are primarily green, representing Clinical Non outbreak cases. A few blue segments within the central nodes represent Cross border Outbreak cases. A single yellow node for Milk filter and a cyan node for Cheese are positioned on the periphery. The legend for Panel B includes Clinical Non outbreak in green, Clinical Outbreak in red, Cross border Outbreak in blue, Milk filter in yellow, and Cheese in cyan.
Forty-five cases were interviewed: 39 cases between 22 October and 28 November 2024, and 6 cases between 16 and 24 December 2024 (Figure 1a). The cases were aged between 2 and 82 years old (mean age: 45), 25 (56%) were females, and 25 (56%) resided in the Bourgogne-Franche-Comté region. The onset dates of the symptoms ranged from 05 September to 26 November 2024. Most cases experienced abdominal pain (n = 39) and diarrhoea (n = 33) (28 cases presented both abdominal pain and diarrhoea). Fewer cases had fever (n = 17), nausea (n = 17), vomiting (n = 6), or bloody diarrhoea (n = 2). Three cases had been hospitalized (short stay: 3 to 5 days) and two patients had an emergency consultation. No death was reported.
Identification of the source of the outbreak
Food exposures of confirmed cases and trace back investigation
In total, 28 (72%) of the 39 cases interviewed between 22 October and 28 November reported consumption of a same cheese type: a soft cheese made in the Bourgogne-France-Comté region from raw cow’s milk and wrapped in a wooden box. Cases reporting consumption of the suspected cheese type resided in 8 of the 11 affected regions and the symptom-onset dates ranged from 09 September to 28 October 2024. Among them, 20 had purchased the suspected cheese in cheese shops and 6 in supermarkets (2 could not specify the place of purchase). A cheese manufacturer in the Bourgogne-Franche-Comté region was identified as the suspected source. Three cases had bought the cheese directly from the suspected manufacturer’s cheese shop and five additional cases reported purchasing the suspected cheese type produced by the suspected manufacturer, although the cheeses were bought in different places. No other food item was consumed by more than 50% of the cases. Among the six cases who mentioned purchase of the suspected cheese type in supermarkets, four cases had loyalty cards. Purchase of the cheese type could be confirmed for one case and trace back identified the suspected cheese manufacturer.
Among the six cases interviewed in December, five reported consuming the suspected cheese type and trace back investigations identified the suspected cheese manufacturer for three cases. The three cheeses were purchased before the recalls.
Site inspections and food and environmental samples analysis
The DDecPP conducted a site inspection at the cheese manufacturer on 12 November 2024. The inspection did not reveal any hygiene deficiencies or non-conformities with regulations. Thirty samples were taken from 30 cheeses at the manufacturer and 14 milking machine filters at the 5 milk producers supplying the cheese manufacturer. On 20 November 2024, the ANSES isolated YE biotype 2 from two cheeses from different batches and one milking machine filter, sequenced their genomes, and shared the sequencing data with the NRC. On 28 November 2024, the NRC confirmed that the three isolates were YE lineage 2/3-9b (corresponding to bioserotype 2/O:9) and belonged to cluster 4452 (Figure 2). The two cheese isolates had 2 allelic distance (AD) between them and each was closely related to at least one outbreak case’s isolates (0 and 1 AD). The milk filter isolate was 6 AD from the closest outbreak case’s isolate (the closest isolate in the cluster was dated January 2025, 3 AD). On the phylogenetic tree (Supplementary Figure S1), the three food/environmental isolates were close to the outbreak cases’ isolates.
Outbreak control measures
Following the inspection of the manufacturer, all stocks were blocked at the manufacturer from 14 November 2024, and the production was stopped on 15 November 2024. A withdrawal and a recall of all batches of different cheeses produced since September 2024 from the manufacturer were initiated on 21 November 2024. The milk producer, whose milking filter tested positive for YE, had sold all its production to the cheese manufacturer between October and November 2024. At the cheese manufacturer, raw milk cheese production was allowed to resume on 21 November, without the milk from the suspected farm. Sales were authorized once the ANSES conducted microbiological analyses on 29 November 2024 on all batches of cheese produced since 21 November, and the results came back negative.
Given the post-transfusion sepsis risk associated with YE [Reference Guinet, Carniel and Leclercq17], the French Blood Establishment included a two-month deferral criterion, from 20 November 2024 to 02 January 2025, for donors who reported having consumed the incriminated type of cheese during the seven days prior to donation.
Cross-border investigation
European health authorities were informed of the outbreak through EpiPulse (2024-FWD-00114), with genomic data for one outbreak isolate (BIGSdb id = 23297) on 15 November 2024, and a Rapid Alert System for Food and Feed (RASFF notification 2024.8664) on 22 November 2024. The cheese was exported to 16 European countries and 3 Asian countries. Five cross-border cases were identified between October and December 2024 in Luxemburg (n = 1) and Belgium (n = 4 cases, not interviewed). The case from Luxemburg (strain isolation date 30 October 2025), a 68-year-old male, had mild diarrhoea and mentioned no travel abroad. He reported a regular consumption of the incriminated type of cheese (no information on the manufacturer could be retrieved).
Discussion
The microbiological routine surveillance in place in France allowed the identification of an outbreak of YE 2/O:9 infections with isolates of the same genomic profile. Following the alert, prompt epidemiological and trace back investigations provided strong evidence pointing to a soft, cow raw milk cheese produced by one manufacturer as the source of the outbreak, which was reinforced by the identification of the outbreak strain from cheese samples and one milk filter. This same bioserotype (2/O:9) was identified in another large outbreak of YE, which occurred earlier in 2024 in France [Reference Savin13]. The peculiar feature of these outbreaks, beyond the serotype involved, was their association with the consumption of raw milk cheese. The implicated cheeses varied in manufacturing processes and milk source (cow versus goat); however, in both cases, the outbreak strains were isolated directly from cheese samples obtained at the manufacturer. A large number of cases occurred, especially in the region of production, and the outbreaks were ultimately contained through the withdrawal and recall of the contaminated cheese products. These two outbreaks highlighted the risk associated with raw milk products’ consumption for French and international consumers.
Raw milk cheeses are part of the French dairy culture with 190 964 tons produced in 2023, with the leading region of production being Bourgogne-Franche-Comté with over 102 040 tons [18]. Their consumption has often been implicated in foodborne outbreaks worldwide [Reference Costard19, Reference Choi20] caused by various pathogens such as Salmonella, enterohaemorrhagic Escherichia coli, Listeria monocytogenes, and less frequently YE [Reference Ackers5–Reference Gruber7]. In the French food industry, routine testing for the presence of YE is not compulsory. No precise data are available on the frequency of this microorganism in raw milk. Contamination of milk may be associated with poor hygienic practices and can occur at different stages (during milking, storage, transport, or cheese production). In this regard, control and prevention of contamination by pathogens are of primary importance to ensure public health. For several years, French health authorities [21] have recommended that vulnerable populations (young children <5 years, pregnant women, elderly over the age of 65, or immunosuppressed persons) avoid consuming raw milk or raw milk products.
In total, 167 outbreak cases were identified in France. This likely underestimates the true number of individuals who developed enteric yersiniosis after the consumption of the contaminated products as (1) only a portion of cases are identified from surveillance due to limitations in healthcare seeking and laboratory testing and (2) surveillance of enteric yersiniosis in France relies on a network of medical laboratories sending isolates on a voluntary basis. A second limitation is the small proportion of cases interviewed after the recall. Among those interviewed, the three who reported consumption of the incriminated cheese had purchased it prior to the recall. The observed delay in the decline in cases after implementation of control measures can likely be explained by several factors, including the delays between consumption and symptoms’ onset, delays to healthcare seeking, and time required for cases to be integrated in surveillance data. It is also possible that some consumers who purchased the cheese before the recalls consumed it long time after without being aware of the recall notice. Nevertheless, as the number of new cases substantially decreased from January onwards, control measures were considered to be effective.
The genomic cluster 4452 had been identified as early as 2020 in the French microbiological surveillance, with 3 to 23 cases per year (2020–2023), without reaching the alert criteria (at least 5 isolates belonging to a genomic cluster isolated within 8 weeks). Furthermore, no seasonality, demographic, or geographic grouping could be observed. This outbreak was identified in October 2024 when the sudden increase across the country in the number of isolates belonging to this genomic cluster met the outbreak alert criteria. After control measures were implemented, the number of cases decreased but strains belonging to the genomic cluster are still being identified infrequently (after 07 March 2025, 17 isolates were identified in the cluster with isolation dates between 11 April and 29 August 2025). This suggests a continuous circulation of the strain at low levels. This continuous circulation is also supported by the fact that the cheese type which caused this outbreak is produced only between mid-August and mid-March and is sold over a limited period of time from mid-September to mid-May (even though consumption of frozen cheese cannot be excluded). Moreover, the close phylogenetic relatedness of all isolates in genomic cluster 4452, independently of the year of isolation, shows a slow diversification of this strain (as previously observed for the lineage [Reference Le Guern12]) and cannot reveal a clear genomic distinction between the cases/strains related to the outbreak and background noise.
In the case of genomic clusters that span over several months or even years, as for the present cluster, genomic analyses and epidemiological data may not be sufficient to understand the circulation of a strain. Although cgMLST is a powerful tool to identify clusters and improve investigation outcomes [Reference Pires, Brandal and Naseer22], the investigation of this cluster is an illustration of the necessity to interpret genomic data in the light of epidemiological and ecological information to comprehend the circulation of a strain and risk factors for human contamination [Reference Besser23].
A more comprehensive understanding of the epidemiology of bioserotype 2/O:9, including its prevalence in raw milk products and its distribution among different reservoirs, including wildlife and environment, is required. Elucidating the contamination pathways within herds through an integrated ‘One Health’ approach that links animal health, food safety, and public health could contribute to mitigating the public health risk posed by YE.
Conclusion and recommendations
Microbiological, epidemiological, and environmental evidence all pointed towards raw milk cheese as the vehicle of this YE 2/O:9 outbreak. Quick identification and withdrawal of incriminated cheese controlled this outbreak. Our study is a reminder of the risks associated with consumption of raw milk cheeses and highlights the importance of applying good hygiene practices on farms to minimize the risk of milk contamination.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0950268826102209.
Data availability statement
Interviews’ data that support the findings of this study cannot be shared with third parties because consent for this was not obtained from the participants. Genomic sequences and metadata of all isolates have been deposited and are freely accessible in the BIGSdb database at https://bigsdb.pasteur.fr/yersinia.
Acknowledgements
The authors would like to acknowledge all the people involved in the investigation including: Catherine Ragimbeau from Laboratoire National de Santé (Luxembourg), Wesley Mattheus and Elena Portell Buj from Sciensano, National Reference Centre for Yersinia (Belgium) for sharing genomic data, E. Fay from Direction départementale en charge de la protection des populations (DDecPP) du Doubs for technical participation in this investigation; Arnaud Felten from Anses for molecular analysis, Lauriane Szpakowski-Perrot, Laure-Marine Houel, and Marie Barba-Vasseur from Regional Health Authority Bourgogne-Franche-Comté for participation in this investigation; Syria Laperche from Etablissement Français du sang for participation in control measures implementation.
We would like to thank all cases who agreed to be interviewed.
Author contribution
Conceptualization: C.S., A-S.L.G., F.C.; Data curation: F.C.; Investigation: S.T., C.S., E.T., J.R., P.P., N.F., L.G., C.M., E.H., C.G., F.C., L.E., J.M., O.R., M.D., A-S.L.G., F.C.; Methodology: C.S., M.D., A-S.L.G.; Resources: J.R., L.G., C.M., E.H.; Software: C.S., A-S.L.G.; Supervision: A-S.L.G., F.C.; Validation: S.T., C.S., E.T., N.F., F.C., M.D., A-S.L.G., F.C.; Visualization: S.T., C.S.; Writing - original draft: S.T., C.S., A-S.L.G., F.C.; Writing - review & editing: S.T., C.S., E.T., J.R., P.P., N.F., L.G., C.M., E.H., C.G., F.C., L.E., J.M., O.R., M.D., A-S.L.G., F.C.
Open access
