Reptiles in Guadeloupe (French West Indies) are a reservoir of major human Salmonella enterica serovars

The epidemiology of human Salmonella enterica infections in Guadeloupe (French West Indies) appears to be specific, with a higher prevalence of the subspecies enterica serovars Panama and Arechavaleta (Panama and Arechavaleta) than in other regions. A study was performed in Guadeloupe to identify the reservoir of Salmonella serovars by comparing their distribution in warm- and cold-blooded animals and in humans living in Guadeloupe and mainland France. Furthermore, a case–control study was conducted in 2012–2013 to identify the main epidemiologic risk factors for S. enterica infection among children under 15 years of age. Between June 2011 and December 2014, feces from 426 reptiles (322 anoles, 69 iguanas and 35 geckos) and 50 frogs distributed throughout Guadeloupe and nearby islands were investigated. The frequency of S. enterica carriage was 15.0% (n = 64) in reptiles but varied by species. The only significant risk factor for S. enterica infection was a more frequent presence of frogs in the houses of cases than in those of controls (P = 0.042); however, isolates were not collected. Panama and Arechavaleta were the two serovars most often recovered between 2005 and 2014 from humans living in Guadeloupe (24.5% (n = 174) and 11.5% (n = 82), respectively), which is in contrast to the low prevalence in mainland France (0.4%). Their presence at low frequencies in wild reptiles (4.6% (n = 3) and 3.1% (n = 2), respectively) and pigs (7.5% (n = 5) and 1.5% (n = 1), respectively) suggests a broad host range, and humans may be infected by indirect or direct contact with animals. These serovars are probably poorly adapted to humans and therefore cause more severe infections. The unusual subspecies houtenae serovar 43:z4,z32:- was a major subspecies in wild reptiles (24.6%, n = 16) and humans (9.4%, n = 67) but was not recovered from warm-blooded animals, suggesting that reptiles plays a key role in human infection.


Introduction
All serovars of Salmonella belong to two species, S. enterica and S. bongori, although more than 99.5% of isolates are assigned to S. enterica. S. enterica comprises six subspecies: enterica, salamae, arizonae, diarizonae, houtanae, and indica. Most cases of human illness arise from enterica subspecies. Serovars Typhi, Paratyphi A, Paratyphi B and Paratyphi C are grouped as typhoidal Salmonella, and other serovars are described as non-typhoidal Salmonella (NTS). Typhoidal Salmonella are human host-restricted bacteria that cause typhoid and paratyphoid fever, which are systemic diseases, whereas a large number of NTS serotypes are generally responsible for acute gastroenteritis. The incidence of NTS infections is lower in industrialized than in developing countries, at about 45 per 100 000 inhabitants [1]. Invasive infections caused by NTS, such as meningitis and septicemia, can occur, particularly among young children, the elderly, malaria-infected and malnourished children, and immunocompromised people [2].
NTS can be host-generalists, capable of infecting or colonizing a broad range of vertebrate species, or host-specialists, adapted or restricted to particular non-human species [2]. In industrialized countries, NTS are transmitted to humans predominantly through the consumption of commercially produced food contaminated with livestock feces (e.g. meat, dairy products, poultry, and eggs) [2]. Outbreaks and sporadic cases of NTS have also been reported after direct or indirect contact with reptiles [3,4], as NTS are commonly found in the digestive tracts of reptiles (crocodiles, lizards, snakes, turtles) and amphibians (frogs, newts) [5][6][7]. Approximately 1.4 million human cases of Salmonella infection occur each year in the USA, of which about 74,000 are a result of exposure to reptiles and amphibians [8]. Within the European Union, it is estimated that less than 1% of cases of human salmonellosis are associated with exposure to reptiles. Reptile-related salmonellosis has been associated with young age, a high rate of hospitalization, and invasive disease [9,10]. Of the Salmonella serovars, 40% have been cultured predominantly from reptiles and are rarely found in other cold-and warm-blooded animals, suggesting that human infections with these serovars are of reptile origin.
Guadeloupe, a French overseas territory in the Caribbean, is a very high-resource country according to the Human Development Index in 2013. Although few data are available on the epidemiology of Salmonella in humans in the Caribbean, infections appear to be specific. In Guadeloupe, Panama and Arechavaleta were the most prevalent serovars recovered from 171 infants and children infected with S. enterica who were seen at the university hospital in Pointe-à-Pitre (Guadeloupe) between 2010 and 2014. The two serovars represented 50% of all Salmonella isolates in that study [11]. Surprisingly, they have been rarely encountered in mainland France or in other regions of the world [12]. In addition, these serovars are significantly associated with bacteremia (P < 0.001) [11]. Four cases of Panama meningitis were recently described in exclusively breastfed infants in French Guiana, suggesting a specific reservoir [13]. In Guadeloupe, wild reptiles and amphibians (e.g. anoles, geckos, iguanas and frogs) are commonly found in and around houses. We therefore conducted a study in Guadeloupe to identify the reservoir of Panama and Arechavaleta by comparing the distribution of Salmonella serovars in warm-and cold-blooded animals and in humans. A matched case-control study was also conducted to determine the main epidemiologic risk factors among children with S. enterica infection.

S. enterica isolates from human samples
S. enterica clinical isolates were received for serotyping between January 2004 and December 2014 by the French national reference center for Escherichia coli, Shigella and Salmonella (FNRC-ESS) (Institut Pasteur, Paris, France) from public and private clinical laboratories in Guadeloupe and mainland France. If more than one isolate with the same serovar was recovered from the same patient, only the first was included. Epidemiologic data (date and site of isolation, age and gender of the patient, history of travel abroad) were recorded when available.

S. enterica isolates from warm-blooded animals
Salmonella spp. were isolated from samples collected at poultry farms (droppings, dust, eggs, and poultry meat) and pig and beef farms (feces, carcasses, and meat) in Guadeloupe during 2010-2014 for sanitary inspections. Serotyping was performed at the Institut Pasteur of Guadeloupe, at the FNRC-ESS and at the French Agency for Food, Environmental and Occupational Health and Safety (ANSES). Data were compiled from the different databases.
S. enterica isolates from wild cold-blooded animals. Between June 2011 and December 2014, a single cloacal swab was taken from 322 endemic anoles of 3 species and 11 sub-species at 85 sampling sites distributed throughout Guadeloupe and nearby islands (Les Saintes, Marie-Galante, La Désirade, and Petite-Terre) (S1 Table). Feces from iguanas living in colonies were collected at 10 sampling sites: 45 from the endemic Iguana delicatissima and 24 from the invasive I. iguana (S1 Table). Fifty frogs and 35 geckos at 8 and 12 sampling sites, respectively, were trapped and placed in sterile vials (S1 Table). Fecal droppings were collected within 24 h of capture. After sampling, all frogs and lizards were released at the capture sites. All samples were placed at +4˚C immediately after sampling and were processed within 4 h.
All procedures were approved by the regional environment, planning and housing agencies and by the Guadeloupe National Park. The project was also approved by the Committee for Samples were incubated in 9 ml of buffered peptone water for 16-20 h at 37˚C. Three drops (75μl) of pre-enrichment broth were inoculated onto modified semi-solid Rappaport-Vassiliadis (MSRV) agar and incubated for a maximum of 48 h. Positive MSRV spots were streaked onto a specific medium, xylose-lysine deoxycholate (XLD) agar, and incubated for up to 48 h. Presumptive Salmonella colonies (H2S positive) on XLD agar were isolated and identified on API 20E test strips (bioMérieux, Marcy L'Etoile, France). Serotyping was performed on five colonies from each sample.

Serotyping of S. enterica isolates
Isolates were serotyped on the basis of somatic O and both phase 1 and phase 2 flagellar antigens in agglutination tests with antisera (Bio-Rad, Marnes-La-Coquette, France) as specified in the White-Kauffmann-Le Minor scheme [14].

Epidemiological study
A case-control study was carried out in 2012-2013. Cases were patients admitted to the pediatric department of the university hospital of Pointe-à-Pitre with an S. enterica infection (acute gastroenteritis or bacteremia). At the inclusion of each case, a trained scientist administered a standardized questionnaire by telephone with the parents to collect demographic, environmental and lifestyle information on the S. enterica infection; the same person selected agematched (± 5 years) children without S. enterica infection or carriage and administered the same questionnaire to the parents.
The study protocol was approved by the French Advisory Committee on Information Processing in Material Research in the Field of Health (CCTIRS . Written informed consent to participate in the study was obtained from the parents of all children included in the case-control study.

Data analysis
Statistical analyses were performed with R software [15]. Conditional tests were used because assumptions of traditional parametric tests were not met, with small samples and non-normal distributions. Thus, resampling procedures were implemented with the "coin" package, which does not assume random sampling from well-defined populations. Resampling provides especially clear advantages when assumptions of traditional parametric tests are not met. In a twoway contingency table, inference was based on 9999 Monte-Carlo resampling. Statistical differences were considered significant for two-sided P values < 0.05.
For the 180 Panama isolates collected between 2010 and 2014, information on travel abroad was available for 61 (33.9%) patients: 54 (88.5%) had traveled in the French West Indies or in South or Central America, and 7 (5.6%) had no history of travel.
Serotyping was performed on five colonies from each reptile, but only one anole contained two different serovars: one belonging to Newport and one to 43:z 4 ,z 32 :-(houtenae subspecies).
Significant associations between prevalent serovars and location were found only for Pomona and Schwarzengrund. Pomona was found exclusively in anoles sampled on two islands (Marie-Galante and La Desirade), and Schwarzengrund was isolated from endemic iguanas on two islands (Petite-Terre and La Désirade) (83.3%, 20/24) but not in invasive iguanas sampled on the main island.

Epidemiological study
A total of 75 children were enrolled in the study (50 cases and 25 controls); 45 were boys. The mean age at inclusion was 33.2 months, and the age distribution was: 19 (25.3%) aged 0-11    Table 3. No significant difference was found between cases and controls in environment or lifestyle factors, except for a more frequent presence of amphibians in the houses of cases than in those of controls (P = 0.042) ( Table 3).

Discussion
The two most frequent Salmonella serovars in humans living in Guadeloupe were Panama and Arechavaleta. Panama was also the most prevalent serovar recovered from humans in two French overseas territories, Martinique in the Caribbean (35% of all isolates investigated between 1990 and 1994) and French Guiana in South America (11.7% in 2011) [12,16]. Although few data are available on the epidemiology of Salmonella infections in humans in this region, the Panama serovar appears to be highly prevalent. It was also the most prevalent serovar in humans in Colombia and Chile [17,18]. The prevalence is higher than those in other regions of the world, including mainland France, where it was found only rarely (0.4% of all isolates investigated, most collected from patients with a history of travel to either the French West Indies or South or Central America). To the best of our knowledge, the only cases of human infection with Arechavaleta have been reported in New Zealand [19], but at a lower frequency than in Guadeloupe, with nine cases reported between 1997 and 2016. As in mainland France and more generally in Europe and other parts of the world, the monophasic variant of the serovar Typhimurium (4, [5],12:i:-) has predominated in Guadeloupe since the mid-2000s [20,21]. Nevertheless, although Typhimurium remains the most frequent serovar in mainland France, 4, [5],12:i:-replaced its biphasic variant in Guadeloupe after 2008. Unsurprisingly, 4, [5],12:i:-was the second most frequent serovar isolated from livestock in Guadeloupe during the study period, supporting the role of pigs and poultry in its transmission to Salmonella in reptiles in Guadeloupe humans [22]. The other major serovars found in humans are commonly associated with human infections, except for 43:z 4 ,z 32 :-(houtenae subspecies). To our knowledge, this serovar was recovered in one case of osteomyelitis in a Taylor's cantil pit viper but has not been isolated in humans [23]. As these unusual serovars were not (43:z 4 ,z 32 :-) or rarely (Panama and Arechavaleta) found in the warm-blooded animals sampled, we investigated a reptilian source of contamination in order to identify their reservoir. The overall frequency of S. enterica carriage in the reptiles studied in Guadeloupe was 15%, which is in the lower range of the reported values (13-57%). Salmonella carriage rates differed by species in our study: no isolates were recovered from geckos, in agreement with the low prevalence reported in most studies [24,25]. The rate in anoles was 11.4%, between the two reported values (0 and 33%) [5,26], and Salmonella was found in 40.5% of iguanas tested, also in the middle of the range of reported values (26-98%) [12,[27][28][29]. Several factors might explain the variation among studies in the recovery rates in reptiles. The frequency of carriage might differ by species, and each group is composed of several species. Carriage frequency also differed by habitat. Crowding of reptiles favors the transmission of NTS, as seen in the high frequency of Schwarzengrund in iguanas from the small islands Petite Terre (1.68 km 2 ) and La Desirade (21.42 km 2 ). In addition, reptiles are intermittent Salmonella shedders [30,31]. In most studies, including ours, specimens were taken only once, which results in underestimates of the rate of Salmonella carriage. Lack of a standard method for Salmonella isolation and differences in the sampling technique (cloacal swabbing of protected animals versus fresh fecal samples from sacrificed animals in other studies) are possible explanations [32].
The case-control study of the main epidemiologic risk factors of Salmonella infection showed, despite the small sample, that the presence of frogs in homes was significantly associated with [24,25] Salmonella infection. However, we were unable to isolate Salmonella from 50 frogs, perhaps because of selection bias, as we could not sample frogs from the houses of cases. Furthermore, an observer bias is possible since frogs are more easily observed in houses than anoles or geckos. Previous studies reported very low rates of carriage in frogs [24,33]. In contrast to the rates observed in other studies (< 50%), in our study most of the isolates recovered from reptiles were assigned to subspecies enterica (69%), which is commonly isolated from warm-blooded animals. Although the subspecies salamae, arizonae, diarizonae and houtanae are known to be harbored by reptiles, we recovered only houtenae.
The subspecies houtenae serovar 43:z 4 ,z 32 :-was prevalent in reptiles and humans in Guadeloupe but was not found in livestock, suggesting a strict reptilian origin of human infections caused by this uncommon serovar. The presence of Arechavaleta and Panama in anoles and iguanas in Guadeloupe also indicates a reptilian reservoir for both serovars. This is not surprising, as Panama has previously been isolated from frogs, toads, turtles, lizards, and snakes [12,34,35] and Arechavaleta from cane toads [36]. However, their host range is certainly much larger, as shown by its presence in pigs in Guadeloupe. Both serovars were also found in previous studies in warm-blooded animals; Panama was found in wild birds, pigs, poultry, and Indian mongooses [37][38][39][40][41] and Arechavaleta in dogs and Indian mongooses [36,39,42]. In Guadeloupe, wild reptiles and amphibians (e.g. anoles, geckos, iguanas and frogs) are commonly found in and around houses. Guadeloupe is also a highly anthropized island, suggesting that close promiscuity between warm-and cold-blooded animals is at the origin of inter-species transmission. Therefore, we hypothesize that Panama and Arechavaleta are transmitted to humans either by direct contact with animals, in particular reptiles, or indirectly, through the consumption of food contaminated with livestock or reptile feces.
The low prevalence of serovars Panama and Arechavaleta in cold-and warm-blooded animals contrasts with that observed in human Salmonella infections. Evolutionary models suggest that host-adapted Salmonella serovars, such as serovars Typhi and Paratyphi A in humans, tend to be of high virulence, causing higher mortality rates than those with a broad host range, such as serovars Typhimurium and Enteritidis [43]. These host-adapted Salmonella serovars can cause illness in all age groups, whereas those with a broad host range tend to be more frequently associated with disease in young animals than in adults, suggesting that they are not optimally adapted to cope with a fully mature immune system [43]. Finally, chronic carriage, which develops more frequently following systemic infections by that host-adapted Salmonella serovars increases transmissibility [43]. Serovars Panama and Arechavalata are lesser virulent than serovars Typhi and Paratyphi A, as illustrated by the fact that they cause lesser mortality rates (0 to 13% versus 12 to 32%) and that they cause disease primarily opportunistically [11,[43][44][45][46]. The number of bacteremia cases associated with serovars Panama and Arechavalata were higher in infants and children than in adults during a 5 year-survey (January 2010 to December 2014) among all patients with Salmonella infection admitted to the emergency room at the University Hospital in Pointe-à-Pitre (31 cases versus one in a 87-year-old woman). In addition, no asymptomatic carriers or secondary cases were identified (unpublished data). Reptile-related salmonellosis is also known to lead to invasive disease in young age [9,10]. All these elements suggest that serovars Panama and Arechavaleta are probably poorly adapted to humans.
In conclusion, the data reported here add to understanding of the epidemiology of Salmonella in Guadeloupe and, by extension, in the Caribbean. Panama and Arechavaleta were the two serovars most often recovered in humans. Their presence in wild reptiles and pigs suggests a broad host range and that human infections may result from indirect or direct contact with animals.
Supporting information S1 Table. Details of GPS coordinates of sampling locations of cold-blooded animals, number of positive animals and isolated serovars. (XLSX)