(See the editorial commentary by Hedberg, on pages 665–6.)
National Salmonella surveillance systems from France, England and Wales, Denmark, and the United States identified the recent emergence of multidrug-resistant isolates of Salmonella enterica serotype Kentucky displaying high-level resistance to ciprofloxacin. A total of 489 human cases were identified during the period from 2002 (3 cases) to 2008 (174 cases). These isolates belonged to a single clone defined by the multilocus sequence type ST198, the XbaI-pulsed-field gel electrophoresis cluster X1, and the presence of the Salmonella genomic island 1 variant SGI1-K. This clone was probably selected in 3 steps in Egypt during the 1990s and the early 2000s and has now spread to several countries in Africa and, more recently, in the Middle East. Poultry has been identified as a potential major vehicle for infection by this clone. Continued surveillance and appropriate control measures should be implemented by national and international authorities to limit the spread of this strain.
Salmonella infection is a major public health problem worldwide. Various animals (especially poultry, pigs, cattle, and reptiles) are reservoirs for Salmonella species, and humans generally become infected by eating undercooked or contaminated food . More than 1.6 million cases of human laboratory-confirmed Salmonella infections were reported during 1999–2008 in 27 European countries . In high-income regions of North America, there are an estimated 1.7 million Salmonella infections per year, and ∼2800 are fatal . Although most infections produce mild gastroenteritis, life-threatening disseminated infections are common among elderly and immunocompromised patients. Fluoroquinolones, such as ciprofloxacin and extended-spectrum cephalosporins, are the drugs of choice for these severe infections. Dissemination of antimicrobial resistance among nontyphoidal Salmonella isolates in humans is thought to be predominantly due to the use of antimicrobial agents in food animals [3, 4]. Furthermore, infections with such drug-resistant Salmonella species are associated with increased morbidity and mortality [5, 6].
High-level resistance to ciprofloxacin, defined as a minimal inhibitory concentration (MIC) >1 μg/mL or ≥4 μg/mL according to the guidelines of The European Committee on Antimicrobial Susceptibility Testing (EUCAST) and the Clinical Laboratory Standards Institute (CLSI), respectively, has been reported rarely in nontyphoidal S. enterica isolates in humans [7, 8]. These isolates of serotypes Typhimurium, Choleraesuis, and Schwarzengrund were mainly reported in Asia [9–12]. During the period 2002–2005, we reported 17 cases of salmonellosis in French travelers returning from northeast and eastern Africa, from whom S. enterica serotype Kentucky isolates resistant to ciprofloxacin with an MIC ≥4 μg/mL (hereafter referred to as CIPR Kentucky) were recovered . At that time, no investigation of risk factors for infection was conducted .
Further spread of CIPR Kentucky was observed in 2003 and 2004 during an outbreak of nosocomial infection in 2 hospitals from the Slovakia . The index case had acquired the infection in Egypt. In 2005, a Belgian traveler infected in Libya with CIPR Kentucky required treatment with a carbapenem due to secondary extended-spectrum cephalosporin resistance after multiple treatment failures . To estimate the circulation magnitude of and the molecular subtypes responsible for the CIPR Kentucky emergence, we gathered surveillance data from 3 European countries (France, England and Wales, and Denmark) and from the US and performed a comprehensive molecular epidemiologic analyses on human and nonhuman isolates from these 4 countries and on isolates from putative reservoir countries.
National Surveillance System of Human Salmonella Infections
The national Salmonella surveillance systems of France, England and Wales, Denmark, and the United States are laboratory based. Identification of Salmonella isolates is performed by using conventional methods, and serotyping is performed on the basis of the White-Kauffmann-Le Minor scheme . Epidemiological data (date and site of isolation, sex, age, and international travel data) were recorded for each isolate. The national surveillance systems differ in coverage and representativeness, as follows:
In France, ∼65% of all human Salmonella isolates recovered in clinical practice are reported to the French National Reference Center. During 2000–2008, 98360 serotyped Salmonella isolates (∼11000 per year) from humans were registered at the Institut Pasteur. Antimicrobial susceptibility testing (AST) is performed every year on a selection of ∼500 isolates belonging to the major serotypes and on all isolates of selected serotypes, such as Kentucky. Real-time subtype surveillance by pulsed-field gel electrophoresis (PFGE) is not routinely performed.
England and Wales
In England and Wales, ∼90% of all human Salmonella isolates isolated in clinical practice are sent to the Health Protection Agency (HPA), Laboratory of Gastrointestinal Pathogens. During the period 2000–2008, 118 922 nontyphoidal Salmonella isolates (∼13000 per year) from humans were reported on by HPA. AST is performed every year on ∼99% of the isolates. Real-time subtype surveillance is performed by phage typing on 10 serotypes, including Typhimurium and Enteritidis.
In Denmark, ∼99% of all laboratory-confirmed cases of salmonellosis in Denmark are reported to the Statens Serum Institut by hospitals and clinical laboratories. During the period 2000–2008, 19326 serotyped human Salmonella isolates (∼2000 per year) were registered in Denmark. AST is performed every year on ∼75% of the isolates. Real-time subtype surveillance is performed only on serotype Typhimurium isolates by PFGE and/or multi-locus variable number of tandem repeats analysis since 2003.
In the US, 80%–95% of all laboratory-confirmed cases of human salmonellosis are reported to the US National Salmonella Surveillance System at the Centers for Disease Control and Prevention (CDC), Atlanta, Georgia, through the Public Health Laboratory Information System (PHLIS) . During the period 2000–2008, 328470 Salmonella isolates were reported (∼36000 per year), of which ∼5% were tested for antimicrobial susceptibility by the National Antimicrobial Resistance Monitoring System (NARMS) at the CDC , and ∼65% were subtyped by PFGE.
Statistical analyses were performed using the Student’s t test to compare mean ages, the χ2 test, odds ratios with 95% confidence intervals (CIs), and the Fisher exact test (for small numbers) to compare the distribution of categorical French data. Analyses were performed using Stata software, version 8 (STATA).
A total of 197 S. enterica serotype Kentucky isolates were extensively characterized as follows: 80 human isolates from the 3 European countries isolated during 2000–2008 and selected on the basis of their diversity (geographic area and year of isolation, susceptibility to ciprofloxacin); all 67 nonhuman CIPR Kentucky isolates during 2005–2008; 49 isolates (40 human and 9 nonhuman) from the French National Reference Center collection isolated from 1959 to 1999; and the reference strain (98K) . Of the 67 recent nonhuman CIPR Kentucky isolates, 44 were collected from chickens in Nigeria during 2007–2008, 20 (13 from seafood and 7 from turkey meat) were isolated in Morocco during 2007–2008, 2 (1 from river water and 1 from dried herbs imported from North Africa) were isolated in France in 2008, and 1 was isolated from chicken in Ethiopia in 2005.
Antimicrobial Susceptibility Testing
Antimicrobial susceptibility testing was performed on all the European S. enterica serotype Kentucky isolates (n = 1309) and on 5% of isolates from the US (n = 23) using either the disk diffusion breakpoint or broth microdilution [13, 18]. Results were interpreted using CLSI (US and Denmark), French Society for Microbiology (France), or British Society for Antimicrobial Chemotherapy (BSAC) guidelines [7, 8, 13]. To standardize the results in the present study, we defined resistance to nalidixic acid and ciprofloxacin as an MIC of >64 μg/mL and 1 μg/mL, respectively. MICs were determined by a 2-fold dilution method or Etest (Solna, Sweden), as described previously [9, 13].
PulseNet standard PFGE of XbaI-digested chromosomal DNA was performed on all of 197 human and nonhuman S. enterica serotype Kentucky isolates . PFGE profiles were compared by using Bionumerics software, version 5.10 (Applied Maths). A search of the US PulseNet database (containing profiles of 309215 Salmonella isolates typed from 1991 to present) was performed.
Multilocus sequence typing (MLST) was performed on a subset of 66 isolates selected on the basis of their diversity (geographic area and year of isolation, susceptibility to ciprofloxacin), as described previously [21, 22]. Data were submitted to the MLST database website (http://mlst.ucc.ie/mlst/dbs/Senterica).
Determination of Resistance Mechanisms
For 55 representative isolates (including 39 CIPR), the gyrA, gyrB, parC, and parE (encoding subunits of the DNA gyrase and the topoisomerase IV, 2 targets of fluoroquinones) were sequenced, as described elsewhere . For the 39 CIPR isolates, the search of the plasmid-mediated quinolone resistances genes, qnrA, qnrB, qnrS, qnrD, aac(6’)-Ib-cr, and qepA was performed as described elsewhere [23–26]. Presence and variant of the Salmonella genomic island 1 (SGI1) was determined for 102 isolates, as described elsewhere [27, 28].
Occurrence of S. enterica Serotype Kentucky in Humans
A total of 497 nonrepeated clinical isolates of S. enterica serotype Kentucky were reported in France from 2000 to 2008 (0.5% of all human Salmonella isolates). In 2007 and 2008, 113 and 139 S. enterica serotype Kentucky isolates were reported, respectively, compared with 24–54 annual isolates between 2000 and 2006 (Figure 1). This increase was primarily due to the emergence of CIPR Kentucky isolates (MIC, 4–16 μg/mL), which represented 200 (40.2%) of 497 S. enterica serotype Kentucky isolates. In 2008, 99 (71.2%) of all 139 S. enterica serotype Kentucky isolates were CIPR. In 2009, 131 human infections due to CIPR Kentucky were reported in France. The first CIPR Kentucky was isolated in 2002 from a French tourist who had gastroenteritis during a cruise on the Nile River in Egypt. This isolate was also resistant to amoxicillin, streptomycin, spectinomycin, gentamicin, sulfamethoxazole, and tetracyclines. Before 2002, this multidrug resistance profile with decreased susceptibility to ciprofloxacin (CIPDS; MIC, 0.125–1 μg/mL) was mainly observed in isolates recovered from patients also returning from Egypt.
The origin of the isolates was stool (426 cases [86%]), urine (31 cases [6%]), blood (6 cases [1%]), bile (6 cases [1%]), or unknown (28 cases [6%]). Patients ranged in age from 1 month to 92 years (mean age, 36 years), and 58% were female. No significant differences were found in age, sex-ratio, or source of clinical specimen when we compared patients infected with CIPR Kentucky to those infected with S. enterica serotype Kentucky isolates susceptible to ciprofloxacin (CIPS Kentucky). Patients infected with CIPR Kentucky were more frequently hospitalized than those infected with CIPS Kentucky (39% vs 28%; OR, 1.7; 95% CI, 1.1–2.5; P = .009).
England and Wales
During 2000–2008, 698 S. enterica serotype Kentucky isolates were recorded at the HPA (0.6% of all Salmonella isolates), and 244 were CIPR (35%) (Figure 1). The first CIPR Kentucky was isolated in 2004 and represented 50% of all serotype Kentucky isolates in 2008 (62 of 124).
During 2000–2008, 114 S. enterica serotype Kentucky isolates were reported to the Statens Serum Institut (0.6% of all Salmonella isolates), and 45 were CIPR (40%) (Figure 1). The first CIPR Kentucky was isolated in 2002 and represented 56% of all serotype Kentucky isolates in 2008 (13 of 23).
During 2000–2008, 679 S. enterica serotype Kentucky isolates were reported to the CDC (0.2% of all Salmonella isolates; data not shown). No CIPR resistance was identified among 23 serotype Kentucky isolates submitted to NARMS during that period of time.
Infections With S. enterica Serotype Kentucky Resistant to Ciprofloxacin Are Mainly Travel Related
Travel information was available for 307 patients (63%) infected with CIPR Kentucky. Of these 307 patients, 272 (89%) traveled internationally in the 2 weeks before illness onset and 35 patients (6 French, 26 English, and 3 Danish patients) did not. The countries visited are indicated in the Table 1. During 2002–2005, the majority of the patients traveled to northeastern and eastern Africa. Since 2006, patients reported travel to northeastern and eastern Africa, North Africa, West Africa, and the Middle East. During 2007–2008, 2 patients reported travel to another European country, such as Spain. In 2009, 3 French patients reported travel to additional countries, such as Mauritania, Togo, and South Africa (data not shown).
NOTE. FR, French patients; E&W, English and Welsh patients; and DK, Danish patients
Identification of a Clonal Population of CIPR Kentucky
Genetic relatedness of S. enterica serotype Kentucky isolates was studied by MLST and PFGE (Figures 2 and 3). The 66 isolates studied by MLST were grouped into 9 sequence types (STs) that defined 7 distinct clonal complexes (having sequence types differing by at least 2 alleles). The ST198 clonal complex was most prevalent (49 [74%] of 66) (Table; online only). All CIPR Kentucky isolates belonged to this clonal complex. The oldest ST198 strain was ciprofloxacin susceptible and was isolated from poultry in the US in 1937 . Human and nonhuman ST198 CIPR Kentucky isolates from Africa (irrespective of the location and date of isolation) clustered into the main PFGE group X1 (based on ≥80% similarity). ST198 CIPDS isolates recovered in patients returning from Africa also clustered into the X1 group, whereas those acquired in China, India, or Thailand clustered into the X2 group. Forty-five XbaI matches were found for profile X1 (PulseNet profile, JGPX01.0007) among human and food isolate data submitted to the US PulseNet national database, including 5 isolated from spices imported from North Africa during 2002–2009. Susceptibility testing of the spice isolates showed that 3 were CIPR Kentucky. Recent CIPS Kentucky isolates (of various sequence types) displayed different PFGE profiles belonging to neither the X1 nor the X2 group.
Beyond quinolone resistance, additional resistance was seen in some CIPR Kentucky isolates (Table; available online). The most prevalent patterns were AST1 (246 [50%] of 489), which included resistance to amoxicillin, streptomycin, spectinomycin, gentamicin, sulfamethoxazole, and tetracycline; AST2 (46 [9%]), which included resistance to amoxicillin; and AST3 (73 [15%]), which included resistance only to nalidixic acid and ciprofloxacin.
Ciprofloxacin resistance in all the 39 CIPR Kentucky isolates tested was due to gyrA and parC mutations (Table; available online). All contained a double mutation in gyrA (Ser83 and Asp87) and a single parC substitution (Ser80 encoding an isoleucine residue). No isolates had gyrB or parE mutations. In gyrA, all isolates contained a change at codon Ser83 to phenylalanine, whereas mutations in the Asp87 codon resulted in substitutions to asparagine, tyrosine, or glycine residues depending on the geographic origin of the isolates. All 3 mutations in Asp87 codon were found in isolates from Egypt, whereas isolates from north Africa and the Middle East had the mutation that resulted in an asparagine residue; those from east Africa had the mutation that resulted in a tyrosine residue; and those from west Africa had the mutation that resulted in a glycine residue. No plasmid-mediated quinolone resistance genes, such as qnr, aac(6’)-Ib-cr, and qepA, have been found in CIPR Kentucky isolates.
A genomic island that harbors a multidrug-resistant gene cluster was integrated between the 2 chromosomal genes thdF and yidY in all 48 tested ST198-X1 S. enterica serotype Kentucky isolated since 1996, regardless of their susceptibility to ciprofloxacin (Table; available online). The extensive genetic characterization of 28 isolates identified several variants of the SGI1 that belonged to the SGI1-Ks, -Ps, and -Qs subgroups. Insertion sequences (ISs) IS26 and ISVch4 were always inserted in the upstream S044 and S010 regions of the SGI1, respectively. Highly diverse genetic rearrangements mediated by various ISs and transposons were observed in the multidrug-resistance region of these SGI1-Ks, -Ps, and -Qs variants, accounting for resistance profile diversity among CIPR Kentucky (Figure 4). Finally, the SGI1-K or derived variants were not found among non-ST198-X1 isolates.
Origin of the ST198-X1 Kentucky Clone
Among historical isolates tested, the oldest ST198-X1 was isolated in West Africa in 1961 (Table; available online). The sequential temporospatial acquisition of new resistance determinants by the ST198-X1 S. enterica serotype Kentucky clone is summarized in Figure 5.
We report an increase in nontyphoidal salmonellosis caused by S. enterica serotype Kentucky isolated in European countries during the period 2005–2008. This increase is due to the emergence of the ST198-X1 CIPR Kentucky clone, which infected almost 500 patients in France, England and Wales and Denmark during 2000–2008. The number of cases is likely underestimated due to limitations in the catchment area of these national surveillance systems. Although the first infections were reportedly acquired in Egypt during 2002–2005, the geographic spread has increased to include countries throughout Africa and the Middle East. However, the importance of this clone as a cause of human infections in the countries of expected origin is unknown. The absence of reported international travel or the notion of travel limited to another European country among some patients infected with CIPR Kentucky suggests that infection may have also occurred in Europe, probably through consumption of contaminated imported foods or through secondary contaminations. The report of CIPR Kentucky isolates from humans in Canada in 2006  and from imported foods in the US indicates that the strain has reached North America. Although the X1 PFGE profile has been found rarely among human and food strains documented in the US PulseNet database, a more extensive sampling and antimicrobial susceptibility testing data for isolates submitted to PulseNet would be necessary to estimate the current burden of ST198-X1CIPR Kentucky in North America.
S. enterica serotype Kentucky has been closely associated with poultry since 1937, where it was isolated for the first time from a chick in the US (ST198 isolate 98K) . This serotype was rarely associated with human illness (81 human cases in 2005) in the US, although widespread in the food supply chain, particularly in poultry, in which it has been the most frequent serotype encountered [17, 30]. However, 157 (93%) of the 168 nonhuman S. enterica serotype Kentucky isolates from the US found in the MLST Salmonella database belonged to the ST152 clonal complex, which is unrelated to ST198 (sharing 0/7 alleles). In the present study, the ST198-X1 CIPR Kentucky clone was isolated from chickens and turkeys from 3 noncontiguous African countries (Ethiopia, Morocco, and Nigeria), suggesting that poultry is an important vehicle for infection with this strain. The common use of fluoroquinolones, including ciprofloxacin, enrofloxacin, and ofloxacin, in chicken and turkey production in Nigeria and Morocco (unpublished data) may have contributed to its rapid spread . Shellfish, possibly contaminated through the aquatic environment, itself contaminated by humans or poultry, may constitute a secondary food reservoir in areas of endemicity, as observed in Morocco. Another potential food vehicle in Europe and the US may be spices or raw vegetables imported from areas of endemicity. Two parameters may account for the rapid dissemination of the strain: consumption of poultry, which constitutes an important source of protein in Africa (particularly in Muslim countries), and the development of local semi-intensive poultry industries [32, 33]. How the ST198-X1 CIPR Kentucky clone entered in the poultry sector in various parts of Africa remains to be determined. This clone was found in at least 2 species of poultry (chicken and turkey). Furthermore, a preliminary investigation revealed that poultry industries of Nigeria, Morocco, and Ethiopia used indigenous domestic fowl, arguing against the dissemination of a common contaminated poultry lineage throughout Africa. The ancestor of the ST198-X1 CIPR clone probably appeared in Egypt during the 1990s, where it caused infection in humans and successively acquired resistance determinants. Unfortunately, the reservoir of this strain remains unknown due to the lack of data on nonhuman isolates in this country.
In the absence of surveillance data on farms and on feed and fertilizer components in countries affected over time, and although ST198 appears to be associated with poultry, it may be instructive to speculate about the involvement of other possible sources in the genesis of the multidrug-resistant ST198-X1 clone. Therefore, intensive aquaculture reliant on large amounts of antimicrobial agents may have played an initial role through the acquisition of the genomic island SGI1-K. Although there is ongoing debate on the potential aquacultural origin of the SGI1 in S. enterica serotype Typhimurium DT 104, the origin of S. enterica serotype Paratyphi B variant Java isolates carrying SGI1 has been clearly associated with tropical fish farmed in Asia [34–36]. Intensive pond aquaculture was introduced in Egypt in the mid-1990s, and today, Egypt is responsible for 80% of the farmed fish production on the African continent [37, 38]. The presence of an ISVch4 element from the aquatic environmental bacteria Vibrio cholerae in all the SGI1-Ks, -Ps, and -Qs variants harbored by the ST198-X1 CIPR clone points to the role of the aquatic ecosystem in the acquisition of the SGI1. Furthermore, SGI1 variants were reported for at least 2 other serotypes of Salmonella (ie, Newport and Haifa, both from Egypt; unpublished data and [39, 40]). The independent acquisition of SGI1 by these 3 distinct serotypes suggests that its transfer occurred repeatedly in a single geographic area. The acquisition of the resistance determinants by the poultry-associated ST198 S. enterica serotype Kentucky clone in the aquatic environment may have occurred in farms practicing integrated aquaculture. In these facilities, poultry manure and poultry-derived products are commonly used to fertilize ponds, and aquacultural waste and rice bran, produced by rice-fish farming, may be used as poultry feed supplements [37, 41]. Manufactured poultry feeds, which may include ingredients of animal and plant origin, could also be a source . Although it will be very difficult to confirm the speculation on the role of aquaculture in the acquisition of SGI1-K, because it probably occurred once in the mid-1990s, the potential role of contaminated feed supplements in the current dissemination of ST198-X1 CIPR Kentucky in African poultry can be quite easily assessed. Whatever the route of contamination, the current emergence of ST198-X1 CIPR Kentucky highlights the need to set up a global integrated national surveillance system for humans and all sectors of the food production chain, focused on the main foodborne pathogens in Africa.
In conclusion, recent experience with multidrug-resistant S. enterica serotype Typhimurium DT 104 demonstrates the potential for global spread of resistant Salmonella infection [43, 44]. In our study, multinational surveillance allowed prompt identification of the epidemic ST198-X1 CIPR Kentucky clone at an international level. Heightened awareness by national and international health, food, and agricultural authorities is necessary to implement measures to monitor and limit spread of this strain.
This work was supported by the Institut Pasteur; the Danish Research Agency (274-05-0117); the World Health Organization Global Foodborne Infections Network; the Institut National de la Recherche Agronomique; the Health Protection Agency; the Statens Serum Institute; the Centers for Disease Control and Prevention; the Institut Pasteur du Maroc; the University of Ibadan; the Agence Française de Sécurité Sanitaire des Aliments; the Réseau international des Instituts Pasteur et instituts associés; and the Institut de Veille Sanitaire.
Permanent financial support came from the institutes, universities, or national agencies of the authors. The work conducted at the National Food Institute, Technical University of Denmark was supported by the World Health Organization Global Foodborne Infections Network (www.who.int/gfn) and the Danish Research Agency (274-05-0117).
We are grateful to Chritina Svendsen, Charlotte Hoe, Elizabeth De Pinna, Lisa Theobald, and Véronique Guibert for outstanding technical assistance in performing MIC and PFGE on isolates from Nigeria and Denmark, England and Wales, the United States, and France, respectively. We thank Christine Keys and David Melka for analysis of nonhuman US isolates.
The MLST data are publicly available at http://mlst.ucc.ie, which is currently funded by a grant from the Science Foundation of Ireland (05/FE1/B882).