Molecular diversity and antimicrobial susceptibility of Listeria monocytogenes isolates from invasive infections in Poland (1997–2013)

The epidemiology of invasive listeriosis in humans appears to be weakly characterized in Poland, the sixth most populous member state of the European Union. We obtained antimicrobial susceptibility data, PCR-serogroups and genotypic profiles for 344 invasive isolates of Listeria monocytogenes, collected between 1997 and 2013 in Poland. All isolates were susceptible to the 10 tested antimicrobials, except one that was resistant to tetracycline and minocycline and harbored the tet(M), tet(A) and tet(C) genes. Overall, no increasing MIC values were observed during the study period. Four PCR-serogroups were observed: IVb (55.8%), IIa (34.3%), IIb (8.1%) and IIc (1.8%). We identified clonal complexes (CCs) and epidemic clones (ECs) previously involved in outbreaks worldwide, with the most prevalent CCs/ECs being: CC6/ECII (32.6%), CC1/ECI (17.2%), CC8/ECV (6.1%) and CC2/ECIV (5.5%). The present study is the first extensive analysis of Polish L. monocytogenes isolates from invasive infections.


Results
During the study period, 344 L. monocytogenes invasive isolates were collected. The isolate submission rate to the NRCBM differed throughout the years (Fig. 1). The overall incidence rate across the study period was 0.05 per 100 000 population, with the highest incidence peak in 2012 (0.17/100 000). All of patients were hospitalized (100%). Information about gender was available for 98.6% of cases, with an observed male-to-female ratio of 1.8 to 1. Age was known in 98.3% of cases, with patients ranging from 1 day to 91 years (median age 67). Specific diagnosis was available for 290 patients (84.3%). Cases with unspecific diagnosis were all categorized as bloodstream infections (BSI) since all such isolates were cultured from blood. During 2009-2013, BSIs were more common among patient ≥65 years of age in comparison to younger patients (60.8% vs. 38.9%, respectively; p = 0.005). On the other hand, meningitis was more prevalent among patients 45-64 years of age than in all other patients (60% vs. 46.9%, respectively, p = 0.04). The distribution of cases according to clinical manifestation and age is shown on Fig. 2. Among all cases there were 40 pregnancy-associated (including four from a mother-child pair). The clinical outcome was known only for 175 cases (50.9%). Overall, the case fatality rate (CFR) was 40% when only cases with known outcome were considered and 20.3% when considering all cases with unknown outcome as cured cases (n = 344) (Table 1).

Discussion
The results described here, based on laboratory voluntary surveillance in Poland, revealed a remarkable increase in listeriosis incidence rates between 1997 and 2013 ranging from 0.01 in 1997 to 0.12 in 2013 per 100,000 inhabitants. The epidemiological trend observed in Poland seems to be consistent with that observed in other European countries 6,17 . However, the NRCBM received limited numbers of isolates each year before 2008 and thus the rising trend observed in Poland may be attributed only partially to an actual increase of listeriosis incidence and mostly to the enhancement of the laboratory-based surveillance system 6,16 . Interestingly, when compared with 2011 and 2013, an increase of reported listeriosis cases was observed in 2012 alongside with significant rise of CC1 contribution (data not shown). No food-borne listeriosis outbreaks were identified in Poland that year which could explain this observation 18 . Over the last years, an increase of listeriosis cases in people 65-years-old and above has been observed in the EU and US 6,18,19 . Comparably, in our study, the median age was 67. Surprisingly, the highest number of cases was reported for patients in two age groups: i) 65-year-old and above and ii) the age group 45-64. In this respect, our findings differ from the age distribution trends observed elsewhere 3,6,[18][19][20] . As in other studies, we noticed significantly more BSIs cases among patients 65-year-old and older compared to younger population. However, this was observed only during 2009-2013 while the Polish surveillance system was being enhanced. During the same period of observation, meningitis was significantly more common among patients 45-64 years of age, especially in 2012.
Fatality rate from invasive listeriosis is known to be high but variable, ranging from 20% to 30%, and thus the disease has been described as the most frequent cause of foodborne infection-related death in Europe 1,6,17 . In our study, the overall CFRs in cases with known outcome (40%) and in all cases (20.3%) were higher than the mean CFR in EU (15.7%) 6,17 . However, these results must be interpreted with caution due to significant differences in surveillance systems and data completeness in EU countries 17 . During the study period in Poland for example, the enhancement of a laboratory surveillance system in 2008 improved patient data completeness and available outcome data of patients increased from 31.8% to 59.8% (data not shown).
As expected, a fatal outcome was observed more frequently in patient ≥65 years of age but also in the age group 45-64. However, we also observed a high CFR (46.7%) in newborns, although outcome was known only for 38.5% cases and this was not statistically significant (p = 0.79). High fatality rates of listeriosis in newborns were also reported in Lombardy, Italy (30%), and in England (24%) 21,22 . Recently, a French study suggested that CFR might be generally higher in pregnancy-related cases than reported to date. They observed more than 80% of infected mothers experiencing major fetus or neonatal complications including death 4 . Therefore, the implementation of prevention strategies during pregnancy as well as continuous improvement of pregnant women's awareness on listeriosis and dietary recommendations is still needed.
During the last few years, an increase in listeriosis caused by serogroup IIa was noticed in most of the European countries where this serogroup is more frequently isolated 3,15,23,24 . In our study however we did not observe an increased prevalence of PCR-serogroup IIa. The most common PCR-serogroup was IVb which accounted for more than 55% of invasive listeriosis cases. The highest frequency of PCR-serogroup IVb during the study period was fairly constant with the exception of 2011, when PCR-serogroup IIa was the most common 25 . Interestingly, most studies on non-clinical isolates from Poland (obtained from animals, meat, fresh and smoked fish) showed PCR-serogroup IIa as the most frequent, with PCR-serogroup IVb observed only sporadically [26][27][28] . There is only one report from Poland showing a higher percentage of PCR-serogroup IVb (31.4%) from RTE products (ready-to-eat cakes and delicatessen items, excluding meat products) 29 . It hypothesized previously that PCR-serogroup IVb may have increased virulence and may be less able to survive in foods compared to the other serogroups 6,27,30 . Our findings further confirm the strong association of PCR-serogroup IVb with invasive infections. Most of the isolates (55.3%) analyzed in this study belonged to three of the four previously described hyper-virulent clones (i.e. CC1, CC2 and CC6) 15 . Interestingly, CC6 was the most common CC (32.6%) observed herein, in contrast to other studies worldwide reporting lower rates of this clone (approximately 3-13%) 3,12,24 . Among our CC6 isolates, 65.2% were responsible for meningitis. This is in line with a study from the Netherlands showing a significant contribution of CC6 to meningitis 31 . They observed CC6 to be associated with unfavorable outcomes in listerial meningitis but not with death 31 . Consistent with their findings, we did not observe a correlation between CC6 and meningitis fatality. Overall in our study, CC6 was the most represented clone in all types of clinical diagnosis, followed by CC1. Similarly, CC1 and CC6 were the two most common CCs related to meningitis and bacteremia in France 15 . In the same study, CC6 was also the fourth most common CC associated with pregnancy-related cases 15 .
Another VT, which has been recently associated with a new EC (ECVIII) 33 is VT14, found in 14 isolates of our study (10 recovered from BSI and four from meningitis). VT14 was observed in a 1994 US multi-state outbreak due to the consumption of contaminated chocolate milk which caused at least 45 cases of febrile gastroenteritis 34 , but was also responsible for sporadic cases of listeriosis in Lombardy (2008-2013) 33 .
The correspondence in phylogenetic clustering and discriminatory power between CCs as determined by MLST and ECs as determined by MVLST has been demonstrated 11 . However, the definition of EC "a group of isolates that are genetically related and have been implicated in temporally-and geographically-unrelated outbreaks" 34 , assumes that L. monocytogenes strains differ in their potential to cause outbreaks. In contrast, the definition of CC is framed in terms of evolutionary biology, as "group of isolates that descended from a common ancestor and accumulated differences mainly through mutations" 35 , and makes no reference to the involvement in epidemics. In this respect, we believe that the characterization based on ECs may be more informative when evaluating the epidemiology of clinical cases.
Because of its severity and associated high fatality, invasive listeriosis requires an effective antimicrobial therapy, which usually involves either ampicillin or penicillin, or meropenem and co-trimoxazole 1,4 . All Polish L. monocytogenes isolates were susceptible to the antibiotics tested, with the exception of one isolate resistant to tetracycline and minocycline. Our results are comparable with observations from other countries and confirm that the frequency of acquired resistance in isolates from human listeriosis cases is low 7,8,36 . Conversely, resistance is more commonly observed in animal and food isolates 26,37,38 . As suggested by Hansen et al., a possible explanation as to why resistance to antibiotics is not a problem in human clinical isolates of L. monocytogenes is that resistant bacteria appear in environments where large amounts of antibiotics are used (e.g. animal farms or hospitals), and since L. monocytogenes is not a common inhabitant of patients in hospitals, it is consequently seldom exposed to antibiotics 36 . During the 17-year span of our study, MIC values for antimicrobials did not change significantly, as also seen in other countries 4,7,23,36 . However, some isolates showed borderline MIC breakpoint values, particularly for ampicillin, which is the drug of choice in the management of neuro-listeriosis 4,7 . The borderline MIC breakpoints were observed after 2006, however this observation was significant only for ampicillin and co-trimoxazole. Our finding underlines the necessity for active and continuous antimicrobial resistance surveillance for early detection of any shift in the antimicrobial susceptibility, which could limit therapeutic options of listeriosis.
As mentioned above, we found only one strain resistant to tetracycline and minocycline, suggesting ribosome protection mechanism, which was confirmed by the detection of tet(M) and integrase associated with the Tn916-Tn1545 transposon genes. These transposons are widespread in enterococci and streptococci, and therefore their presence in L. monocytogenes is likely due to the acquisition of genetic information originating from these bacteria 9,10 . Additionally, the tetracycline-resistant strain carries tet(A) and tet(C) determinants, which are responsible for activation of efflux pumps 10 . Coexistence of these two mechanisms in the same strain could contribute to the high MIC value observed for   tetracycline (64 µg/ml). tet(M) is the most often detected determinant among tetracycline-resistant isolates of various origins [8][9][10] . Conversely, tet(A) and tet(C) have been found sporadically in L. monocytogenes and mostly in isolates from animals, food and food-processing plants 10,37,38 . To the best of our knowledge, this is the first study reporting coexistence of tet(M), tet(A) and tet(C) genes in a human isolate from meningitis. Additionally, WGS analysis of the strain revealed the presence of sul (sulfonamides), norB (quinolone) and fosX (fosfomycin) resistance genes, but their presence did not correlate with phenotypic resistance (data not shown), except for fosfomycin, to which L. monocytogenes is considered to be naturally resistant 39 40 . The present study is the first extensive analysis of L. monocytogenes isolates from Poland from human invasive infections using molecular techniques, including MLST and MVLST, showing predominance of PCR-serogroup IVb isolates in invasive listeriosis. We identified CCs and ECs previously involved in the outbreaks worldwide, allowing the inclusion of Polish data into the international database to track the spread of invasive strains and clones of L. monocytogenes, and contributing to the understanding of the global epidemiology of listeriosis.

Materials and Methods
Country background. Estimates of the national census for 31st December of every year were used as the denominator for the calculation of annual incidence rates. On average, the total Polish population was 38,276,946 during the duration of the study [Polish Central Statistical Office; http://stat.gov.pl/obszary-tematyczne/roczniki-statystyczne/]. A listeriosis case for laboratory reporting was defined as isolation of L. monocytogenes from a normally sterile body site of a patient with clinical symptoms of infection. Finally, all cases were grouped into meningitis or bloodstream infections (BSI). Listerial meningitis was defined by the isolation of L. monocytogenes from cerebrospinal-fluid or the clinical diagnosis of meningitis with L. monocytogenes isolated from blood. A pregnancy-associated listeriosis case was defined by isolation of L. monocytogenes from a pregnant woman or newborn (age <28 days) with meningitis or BSI. For all cases, only one isolate per patient was included in the study.
Bacterial identification and PCR-serogrouping. The study encompassed L. monocytogenes isolates obtained during laboratory surveillance between January 1997 and December 2013 conducted by the NRCBM.
Species identification was confirmed biochemically (API ® Listeria or Vitek-2 System; BioMerieux, France) and by PCR amplification of the hly, iap, and lmo2234 genes 41,42 . Identification of the L. monocytogenes PCR-serogroups was made by multiplex PCR 42-44 . Molecular characterization of isolates. The isolates were categorized into Lineages using a multiplex PCR method 45 . Sequence types (STs) were identified for all isolates either by MLST (n = 293) 46 or were extracted from the WGS data (n = 51) obtained as a part of the European Listeria Typing Exercise (ELiTE WGS) organized by the European Centre for Disease prevention and Control-ECDC 47 . Raw sequence data of these 51 isolates have been deposited at the European Nucleotide Archive (Project ID PRJEB26063) ( Table 4).
The tetracycline resistant strain was not the part of ELiTE WGS, therefore the WGS was performed with Ion Torrent PGM (Thermo Fisher Scientific, Waltham, MA, USA) and obtained reads were assembled using CLC Genomic Workbench v.8 (CLC bio, Aarhus, Denmark). This Whole Genome Shotgun project has been deposited at DDBJ/ENA/GenBank under the accession QPLL00000000. The version described in this paper is version QPLL01000000 (Table 4).
Markers for four ECs (ECI-ECIII and ECV) were screened by PCR 32,42 . Subsequently, MVLST was performed on all isolates negative for those four EC markers. Additionally, all ECV-putative isolates identified by PCR were confirmed by MVLST 42 . Overall, MVLST was carried out on 170 isolates. Allele numbers and VTs were determined according to the MVLST database [https://sites.google.com/site/MVLSTdatabase/home, last accessed 23 July 2018].
Bacterial susceptibility testing. Antimicrobial susceptibility to the following 10 antibiotics was tested for all isolates: ampicillin, penicillin, meropenem, erythromycin, trimethoprim-sulfamethoxazole (co-trimoxazole), levofloxacin, gentamicin, vancomycin, tetracycline, and rifampicin, using the broth microdilution method as recommended by the European Committee on Antimicrobial Susceptibility Testing (EUCAST) 48 . Different clinical breakpoints were used for interpretations according to the EUCAST: those specific for L. monocytogenes were used for ampicillin, penicillin, meropenem, erythromycin, and co-trimoxazole 49 ; those specific for Streptococcus pneumoniae were used for levofloxacin, vancomycin, tetracycline and rifampicin 36,49 . For gentamicin, interpretation was carried out according to the clinical breakpoint value for Staphylococcus spp. 36,49 . Finally, minocycline was tested only for the tetracycline-resistant strain, which was additionally screened by PCR for the presence of resistance and integrase genes 50,51 . Statistical methods. The diversity index (DI) was calculated as described by Grundmann et al. 52 . Chi square test or Fisher's exact test was used to analyze the differences in frequencies; p-values ≤ 0.05 were considered significant.