Listeriosis is a foodborne disease caused by Listeria monocytogenes, which causes invasive syndromes in people with altered immunity (i.e. immuno-compromised, the elderly, pregnant women, foetuses and the newborns). Symptoms can include severe sepsis or infection of the central nervous system that can lead to lifelong sequelae or death. Pregnancy related listeriosis can result in preterm birth, miscarriage or stillbirth [1, 2]. In healthy people, L. monocytogenes may cause a self-limiting febrile gastroenteritis [3].

Within the European Union, notification of listeriosis cases is mandatory. However, in Italy, different notification rates are reported among the various Regions, and to date national coverage is not guaranteed [4]. Moreover, systematic subtyping and comparison with food isolates is fragmented, hindering the investigation and control of possible outbreaks [5]. Lombardy is the most populous region in Italy with 17% of the Italian population [6], and since 2005 it holds an enhanced listeriosis surveillance system that involves a laboratory-based network with voluntary referral of clinical isolates to the Regional Reference Laboratory for foodborne diseases (LabSS). The LabSS collects clinical isolates of L. monocytogenes and carries out serotyping and molecular subtyping with Pulsed-Field Gel Electrophoresis (PFGE) and Multilocus Sequence Typing (MLST). This allowed for the period 2006–2014 to report in Lombardy a mean incidence of 0.56 cases/100000 inhabitants in view of the Italian one of 0.20 cases/100000 inhabitants) (data not shown). Between 2006 and 2010, 23% of all collected isolates were identified as Sequence Type 38 (ST38) by MLST, and were grouped in the same PFGE cluster [7]. Given the elevated proportion and the recurrent presence of this ST, there was a possibility these cases were related and belonged to a single outbreak that went undetected by the traditional surveillance system in the Lombardy Region.

Conventional epidemiology typically is inefficient in detecting listeriosis outbreaks as foods with long shelf lives, long incubation periods, and infrequent infections in spite of presumably frequent exposures, may allow a listeriosis outbreak to occur as a succession of apparently unrelated cases [1, 7–9]. Therefore, it is essential that conventional epidemiology be supported by adequate molecular subtyping methods. To date, PFGE is the gold standard for L. monocytogenes subtyping, however sequence-based techniques, like MLST, Multi-Virulence-Locus Sequence Typing (MVLST) and core genome MLST [10], have also been successfully applied as they yield unambiguous and highly informative data, which can be easily accessed and exchanged through public databases [11]. MVLST, based on the sequencing of 6 virulence gene fragments, shows excellent epidemiologic concordance [12] (i.e. the ability of a subtyping system to classify epidemiologically related isolates derived from a presumably single-clone outbreak into the same clone [13]); and allows the identification of Epidemic Clones (ECs), strains originated from a single ancestral cell and involved in geographically or temporally unrelated outbreaks [12] .

Our aim was to clarify the epidemiology of listeriosis cases in Lombardy by gaining a deeper insight into ST38 isolates. First, the isolates collection period was expanded by 4 years to include all clinical L. monocytogenes isolates up to 2014. Subsequently, all isolates were subtyped using serotyping, PFGE and MLST and those belonging to ST38 were further subtyped using MVLST. These data were analyzed to evaluate the geospatial and temporal distribution of the strains. This study provides an integrated model to better detect and control future listeriosis outbreaks.

Based on temporal and epidemiologic data, we hypothesize that strains belonging to the previously detected Cluster 11 [7] were part of an outbreak that went undetected by the traditional surveillance system. Our findings reveal that between 2006 and 2014, isolates belonging to ST38 represented the most prevalent L. monocytogenes subtype detected in Lombardy. However, the overall mean age and case fatality rate of ST38 cases were comparable to those previously reported [7]. Notably, the presence of ST38 strains may have played a role in finding in a previous study serotype 1/2a more prevalent than serotype 4b in Lombardy [7]. This trend is consistent with the worldwide scenario, where serotype 1/2a is now competing with 4b as the leading serotype responsible for clinical cases [20, 21].

The observed peak of cases (2009 and 2011), identified by all subtyping methods used, is consistent with the occurrence of an outbreak rather than a surveillance artifact. In fact, the plotting of cases over time yields a typical epidemic curve and the presence of an epidemic event was confirmed by spatiotemporal analysis of the distribution of cases belonging to ST38 using a mathematical model [22]. Moreover, in Bergamo – the province with the highest number of cases – such epidemic event may have caused the increase of listeriosis incidence for the years 2010 and 2011 as registered by the notification system.

MVLST divided the ST38 isolates in two VTs (VT80 and VT104), characterized by a different spatiotemporal distribution, where the peak in the number of cases between 2009 and 2011 is mainly due to VT104 (Figs. 2 and 3). The two VTs differ only in the 90th nucleotide of the dal gene fragment analyzed, which is a synonymous mutation. The first VT80 clinical cases originated in Bergamo province in 2006, while the first cases of VT104 were reported in 2009 especially in the Eastern provinces of Lombardy (i.e. Cremona, Lodi, Milan and Bergamo) (Fig. 3).

We can reasonably hypothesize that Taleggio cheese was the implicated food, based on the following considerations: i) all patients reported cheese as suspected food and four specifically recalled consumption of Taleggio cheese, ii) a direct correlation between a clinical case and the Taleggio cheese sample retrieved from the patient’s refrigerator was confirmed by PFGE and MVLST, iii) all routinely collected food isolates with the same pulsotypes and VTs of those in the ST38 cluster are from Taleggio cheese samples, and iv) the molecular type found (ST38, Cluster 11 pulsotypes [7], VT104) is infrequent and never reported before in other food sources [17, 19]. Based on these assumptions and to the best of our knowledge, we can designate VT104 as the first outbreak clone detected in Italy.

Typically, many L. monocytogenes outbreaks originate from food processing plants that are colonized by the outbreak clone, which cause the contamination of the final product [23]. In this light, PDO Taleggio producing plants were identified as the hypothetical source: VT104 was found in a production plant (Plant A) in Bergamo province following the sampling relative to the epidemiologic investigation linked to a patient; while VT80 was found in a different PDO Taleggio production plant (Plant B), also located in the Bergamo province, during routine monitoring carried out by IZSLER.

We can hypothesize, as supported by the timeline (Fig. 2), that VT80 may have first appeared in one plant in the Bergamo province and caused an undetected outbreak (Figs. 2 and 3, orange). Then either in the same plant or after being transferred to a second one, in the same province (Bergamo), it evolved into VT104 and caused a second outbreak detected herein (Fig. 3, blue). The two VTs have different but overlapping distribution in Western and Eastern Lombardy, likely due to the fact that the processing plant colonized with VT80 may have also shipped contaminated product to the East part of Lombardy and the processing plant colonized with VT104 also shipped product to the West. It is interesting to note that no VT80 or VT104 cases have been reported in Lecco and Sondrio provinces (Fig. 3, Table 1), even though these provinces showed the highest listeriosis incidences together with Bergamo (Table 1) and have a similar dairy production. This may be explained by the fact that in Italy dairy products have a strong artisanal and local connotation that may limit the geographic diffusion of L. monocytogenes strains. Nevertheless, considering that many Italian local products are also exported, it may be possible that a contaminated dairy product will be responsible for a listeriosis case abroad while not cause any case in a neighboring producing area.

Interestingly, the PFGE profiles of isolates in Cluster11 were closely related to the pulsotype (AscI.0087) of one clinical case involved in the 2012 US outbreak linked to ricotta salata produced in Southern Italy and of two clinical cases detected in Lombardy in 2011 [24]. The isolates linked to the ricotta salata outbreak were typed as VT80 [19, 25], while the two Italian cases were subtyped as ST101, which is grouped in the same Clonal Complex (CC) as ST38 (i.e. CC101). On the other hand, VT80 has also been found in other cheeses as PDO Taleggio and Gorgonzola cheese, two products that have an overlapping producing area (limited exclusively to Lombardy and Piedmont) [19, 26]. These findings suggest that this molecular type (VT80, corresponding to ST38 and ST101, both belonging to CC101) might be common in dairy production facilities in Italy, particularly those manufacturing cheese. Haase et al. [27] reported that CC101 isolates were frequently isolated from clinical cases in the mid-1950s, but only rarely in recent times, and this may have hampered the designation of a correlated EC. For instance, it is still unknown if the ricotta salata and the taleggio outbreak have some kind of connection, but we can hypothesize the involvement of a common mean of transmission between dairy processing facilities, such as milk and/or transportation vehicles. Therefore, considering that the definition of an EC is a strain or group of strains originated from a single ancestral cell and involved in geographically or temporally unrelated outbreaks [12], it can be argued that the Taleggio outbreak clone (CC101/ST38/VT104) and the ricotta salata outbreak clone (CC101/ST101/VT80) can be considered as a novel EC, namely ECXI.

Unfortunately, the epidemiologic interviews were available only in 13 out of the 43 cases (with 11 cases belonging to VT104 and two to VT80). This shortage of information limited the possibility of identifying with more certainty the food vehicle associated to VT80, which in turn could have disclosed possible connections with the ricotta salata outbreak.

However, this study suggests how the application of sequence based typing methods can support the rapid detection and control of outbreak events, especially when cases do not immediately appear correlated (e.g. they are spread in time and space). Also, this study highlights the importance of public typing databases, pivotal in supporting epidemiological investigations. Moreover, the specific type found (ST38 or VT80/104) had only been rarely reported before and it appears now to have adapted to the dairy production environmental niche. Therefore, the results of this study could be useful in the investigation of other outbreaks linked to this molecular type and in the monitoring of strain persistence in processing plants.

Our findings revealed a major outbreak in Northern Italy linked to PDO Taleggio cheese in 2009–2011, which went undetected by local health authorities. The closely related outbreak clones causing this outbreak represent a novel epidemic clone (ECXI) that was also linked to the U.S. ricotta salata outbreak. Our results suggest that traditional molecular subtyping methods (e.g. PFGE) should be integrated with DNA sequence-based methods (e.g. MLST and MVLST) and space-time analysis models, in order to rapidly detect and control outbreaks that otherwise would have been neglected [28].