INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Infections with nontyphoidal Salmonella enterica serovars are most often the result of ingestion of contaminated foods and are only rarely attributed to other causes. Salmonella is the most frequently detected cause of bacterial illness in parts of Europe and Korea and is the second most frequently detected cause of bacterial illness in the United States, England, and Australia (20, 28, 30). The estimated annual cost of food-borne salmonellosis in the United States is $0.9 to 3.5 billion (3). Around the world, Salmonella is the most important agent causing food-borne illness, with Salmonella enterica serovar Enteritidis and Salmonella serovar Typhimurium predominating (28). In Canada, Salmonella serotype Typhimurium and Salmonella serotype Enteritidis have also caused a majority of cases of sporadic diseases and are responsible for most outbreaks due to Salmonella since 1983 (5, 6, 8, 14-18). While other serotypes, including Salmonella serotypes Heidelberg, Hadar, Agona, Infantis, Thompson, and Newport, are important causes of human disease in Canada, for the most part these organisms are thought to be acquired through ingestion of contaminated food (11, 18). The characterization of a Salmonella serotype Infantis outbreak in Canada in 1999 (P. N. Sockett et al., unpublished data) led to the identification of pet treats produced from pig ears as a common source of human illness for many of the outbreak cases, suggesting that there are alternate ways for these bacteria to infect the human population.

Salmonella serotype Infantis is a pathogen of both humans and animals. This organism has caused outbreaks of human disease in England (2), Finland (12), the United States (21), and Canada (23), and isolates of Salmonella serotype Infantis have been recovered from pigs in Africa (22). A Danish surveillance program for Salmonella in fresh meat, instituted after an epidemic of Salmonella serotype Infantis in 1993 in that country, found 3.1% of pork cuts were contaminated with Salmonella serotype Infantis (27). In Finland, Salmonella serotype Infantis has been isolated frequently from chickens and cattle and is the third most common cause of human salmonellosis (24, 29). Isolates from a cattle outbreak of Salmonella serotype Infantis associated with contaminated feed in Finland were analyzed by pulsed-field gel electrophoresis (PFGE), plasmid analysis, ribotyping, and IS200 typing (19).

Salmonella serotype Infantis has been among the top 10 Salmonella serovars from humans in Canada characterized by the National Laboratory for Enteric Pathogens (NLEP) in recent years (5, 6,14-17), ranking second as a cause of human illness in Canada in 1983 and then decreasing to consistently rank sixth or seventh by the early 1990s (18). This was accompanied by a corresponding decrease in prevalence in nonhuman isolates during this time. Illness due to this organism has been seen in all provinces, although more than half of the human cases identified from 1993 through 1996 were from Ontario, Canada. In the absence of identified outbreaks, it has been difficult to trace the source of human infection by this organism.

In this study, we describe the typing and fingerprinting methods used to characterize isolates from pet treats associated with the Salmonella serotype Infantis outbreak in Canada in 1999. The results indicate that contact with pet treats made from pig ears or with dogs who had eaten these pet treats was the predominant cause of recent human Salmonella serotype Infantis illness. A variety of other pet treats were also contaminated with additional Salmonella serotypes. Although none of these isolates appeared to be associated with specific outbreaks, the disease burden in terms of sporadic cases due to serotypes other than Salmonella serotype Infantis to human disease is not known. Possible reasons for the predominance of Salmonella serotype Infantis in the present outbreak remain to be determined.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial isolates. Bacterial isolates were obtained from a number of provincial government and private laboratories across Canada. Isolates were forwarded either to the NLEP in Winnipeg, Manitoba, Canada, or to the Laboratory for Foodborne Zoonoses in Guelph, Ontario, Canada. After subculture, strains were prepared for long-term storage by inoculation onto slants of Institut Pasteur maintenance medium or were frozen at 80°C in brain heart infusion broth containing 15% (vol/vol) glycerol.

Isolation and characterization of Salmonella. Salmonella was isolated from human stool and urine samples in hospital laboratories by standard methods. During the outbreak investigation, screening of pet treats and animal products for Salmonella was conducted at Provincial Laboratories of Public Health, the Canadian Food Inspection Agency laboratories, and private laboratories by methods established by the Canadian Health Protection Branch (7, 26).

Phage typing. All Salmonella strains isolated were phage typed when appropriate using schemes developed at the Central Public Health Laboratory, Colindale, United Kingdom, and at the NLEP, Winnipeg, Manitoba, Canada (1; R. Khakhria, D. Duck, and H. Lior, Conjoint Meet. Inf. Dis., abstr. BC = 26, 1982).

PFGE. PFGE was done according to previously described protocols (4, 10), summarized as follows. Bacteria grown overnight in brain heart infusion broth were collected by centrifugation, washed in TE wash buffer (100 mM Tris [pH 8.0], 100 mM EDTA), and suspended in TE wash buffer to give a density reading of 0.50 with a Dade turbidity meter. Proteinase K was added to washed cells, which were then directly embedded in 1.2% SeaKem Gold agarose (Mandel Scientific Co., Guelph, Ontario, Canada) prepared with TE buffer (10 mM Tris [pH 8.0], 1 mM EDTA) containing 1% sodium dodecyl sulfate. Solidified plugs were transferred to 1.5 ml of lysis buffer (50 mM Tris [pH 8.0], 50 mM EDTA, 1% sarcosine, 0.5 mg of proteinase K/ml) and incubated at 54°C in a water bath with shaking at 150 rpm for 2 h. Plugs were washed twice at 50°C with distilled water (18 M quality) and then washed three times for 15 min with TE buffer with shaking. Plug slices were equilibrated in 100 µl of buffer H (Roche Diagnostics, Laval, Quebec, Canada) at 37°C for 15 min and digested at 37°C for 2 h with 80 U of XbaI in 100 µl of fresh buffer H. Selected isolates were also restricted with SpeI. After digestion, plugs were equilibrated in 0.5× TBE containing 0.89 M Tris, 0.89 M boric acid, and 0.02 M disodium EDTA (pH 8.4) (Roche Diagnostics) for 5 min. Electrophoresis was performed with the CHEF DR III unit (Bio-Rad Laboratories Canada, Ltd., Mississauga, Ontario, Canada) in 1% PFGE agarose at 14°C. Initial and final switch times were 2.2 and 63.8 s, respectively, and the total run time was 22 h. Following electrophoresis the gels were stained with ethidium bromide (0.5 µg/ml) and imaged with the Alpha Imager 2000 (Canberra Packard Canada, Mississauga, Ontario, Canada). The interpretation of the PFGE patterns was aided by use of Molecular Analyst Software Fingerprinting PLUS, version 1.6 (Applied Maths, Kortrijk, Belgium). All associations obtained using these software packages were checked visually by at least two people. For the isolates analyzed in this study, each unique pattern was given its own number. Dendrograms were created using the unweighted pair group method with arithmetric means within the Molecular Analyst software.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Isolation of Salmonella spp. from humans and dog treats. Of 94 pig ear samples obtained from retail outlets at the time of the investigation, 48 (51%) were positive for Salmonella. Other retail products tested at this time included treats containing lamb, turkey, or beef products and those labeled as beef chew, rawhide, chew stick, beef hoof, braided chew, lamb chunks, hoof delight, and roulé rôti, or just "pet treats." Several (15 of 39, or 38%) of these treats also contained Salmonella. In addition to retail products, pig ears and other pet treats or animal parts were obtained directly from production plants during the investigation. Salmonella was found in products from 5 of the 12 plants investigated in Alberta, Manitoba, Ontario, Québec, and New Brunswick and was found in plants from three of the five provinces. Overall, 49 of 171 (29%) of pig ear pet treats from these plants contained Salmonella; 7 other pet treat products (as listed above) were also contaminated with Salmonella. Some products contained as many as four different serotypes of Salmonella, although the relative levels of each serotype in these products were not determined.

Molecular typing of Salmonella serotype Infantis: evidence for transmission of bacteria from pig ear dog treats to humans. Phage typing and PFGE using XbaI and SpeI were used to discriminate among isolates confirmed as Salmonella serotype Infantis by biochemical testing and serological identification (Table 1). Nine phage types (PTs) were found among the strains tested, as well as one isolate that gave an atypical phage typing pattern and another that was untypeable (Table 2). Of the 156 isolates tested by phage typing, 96 (62%) were PT 4, and 20 (13%) were PT 26. Together, these two PTs accounted for 116 (74%) of all isolates tested. PT 4 isolates comprised 53% of isolates from human patients, 85% of isolates from pig ears, and approximately one-third of the isolates from chickens. Other PTs were found in isolates from a more restricted number of hosts: PT 3 and PT 7 in isolates from humans and pigs, PT 8 in isolates from humans and an iguana, PT 9 in isolates from humans and chickens, and PTs 10, 13, and 29 in humans only.

View larger version (72K): [in this window] [in a new window]   FIG. 1.   PFGE profiles of Salmonella serotype Infantis strains restriction enzyme digested with XbaI showing representative PFGE patterns. Lanes 1, 5, 10, and 15 contained the Salmonella serotype Newport standard strain AM01144.

View larger version (102K): [in this window] [in a new window]   FIG. 2.   Dendrogram of Salmonella serotype Infantis PFGE types.

Molecular typing of other Salmonella serotypes. Phage and PFGE typing information are available only for some of the other serotypes of Salmonella isolated from pig ear pet treats (Table 5). Relatively low numbers of some serotypes were isolated from humans in Canada in 1998 and 1999. These isolates included the Salmonella serotypes Agoueve, Banana, Bovis-morbificans, Brandenberg, California, Derby, Havana, Livingstone, Mbandaka, Meleagridis, Montevideo, Muenster, Ohio, Panama, Schwarzengrund, and Uganda. PFGE was not done with human isolates of these serotypes unless they were involved in outbreaks of human disease during the course of the year. Pig ear pet treat isolates of a few serotypes were analyzed by phage typing and by PFGE to give baseline data for future comparisons with human isolates (see Table 5).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The evidence presented here indicates that pig ears and other pig products are a potential source of human infection by many serotypes of Salmonella. Though pet treats made from pig ears and other porcine products were contaminated with many different kinds of bacteria, only Salmonella serotype Infantis was identified as causing an increased number of human infections. While the reasons for this are not clear, it may be that this organism was more virulent or was present in higher numbers in the pet products; however, quantitative assessments of Salmonella serotype Infantis in these products were not made. A second possibility is that Salmonella serotype Infantis grows better in dogs than other serotypes, thus serving to amplify the organism, and that dogs are an intermediate source of infection for many humans. During the outbreak, infection occurred in patients whose sole exposure was to pig ears. Few isolations from dogs were attempted, though case reports suggested that dogs become sick as a result of exposure to pet treats. It is not clear that methods currently available would be adequate for this kind of analysis. Identifying the absolute numbers of bacteria contaminating these treats may be of limited value until the role of dogs in amplifying pathogen numbers and causing infection is understood.

Salmonella serotype Infantis characterized in this study constituted a diverse group of organisms. Isolates from humans showed considerable variation in their PFGE patterns and PTs, and in some cases particular types may be associated with specific sources or routes of infection. The presence of identical PFGE and PTs in strains of Salmonella serotype Infantis associated with both humans and pig ear pet treats supports the idea that pig ear pet treats were the original source of these strains. Pigs and pig ear pet treats appeared to be contaminated with a limited variety of Salmonella serotype Infantis PFGE and PTs, suggesting that limited type diversity is not a likely reason for the isolation of identical PTs and PFGE types from both pig ear dog treats and humans. The prolonged outbreak of Salmonella serotype Infantis appeared to result from multiple introductions of the organism into pigs or pig ears and from there into the human population, rather than introduction of bacteria from a point source. Support for this premise comes from the fact that the PFGE patterns 1 and 3 responsible for most of the human cases are widely separated in the dendrogram of Salmonella serotype Infantis patterns and have less than 80% pattern similarity.

A single pig ear isolate with a PFGE type 19 pattern was found in Québec, where there were also a number of isolates from human patients with identical PFGE patterns. The finding of these and closely related types only in Québec suggests a possible route of transmission from pet treats to humans specific for this area and indicates that there may be regional differences in the Salmonella serotype Infantis types contaminating these treats. It is possible that the prevalence of the type 1 and 3 patterns among Salmonella serotype Infantis isolates from humans largely reflects the distribution of pet treats from suppliers in western Canada throughout the rest of the country (Sockett et al., unpublished).

Salmonella serotype Infantis is present in pigs, poultry, chicken eggs, cattle, and animal feeds. In 1996, Salmonella serotype Infantis of animal origin was predominantly recovered from pork or pork products, especially in Québec and Prince Edward Island (14). Very few isolates of this organism were found in pork in 1995 (15). Poultry, especially in Ontario, was the predominant source of nonhuman Salmonella serotype Infantis in 1993 and 1994 (16, 17). Interestingly, a single isolate of Salmonella serotype Infantis was obtained from dog snacks in 1994, although not in 1995 or 1996. It is not clear how this pathogen emerged in pork, though its present prevalence seems to be relatively recent.

Both PFGE and phage typing were required for an adequate description of the strain characteristics of all Salmonella serotypes analyzed in this study. Phage typing discriminated among most of the Salmonella serotype Infantis PFGE types and less frequently among PFGE types of other Salmonella serotypes. PFGE similarly subdivided PTs of most Salmonella serotypes tested. It is important to note that pentaresistant Salmonella serotype Typhimurium DT 104 was isolated from pig ears, suggesting that this may constitute a source of human infection with this organism as well.

The sharp increase in the numbers of Salmonella serotype Infantis infections is a recent trend in human infection, and dog treats may have contributed to this increase. Clearly, the causal relationship between human cases and pig ear dog treats was clarified by molecular typing, which also served as the trigger for an epidemiological investigation of this outbreak. Comprehensive and detailed descriptions of the phenotypic, genetic, and molecular fingerprint characteristics of the Salmonella isolates associated with pig ear pet treats are necessary to evaluate the effects of interventions designed to eliminate these treats as a source of human infection. For instance, the fact that phage typing and PFGE appear to measure biological traits that vary independently of each other suggests that the use of both methods is necessary for adequate typing of related isolates. The laboratory and epidemiological components of the National Enteric Surveillance Program for enteric bacteria currently track trends in Salmonella infections throughout Canada to evaluate the effect of such interventions and to determine the contribution of pig ear pet treats to the disease burden.

Pig ears and other dog treats from similar sources, produced under similar conditions, may have been a continuing source of human disease for quite some time and may have contributed to the background of sporadic cases seen in Canada and elsewhere. Continued epidemiological and laboratory monitoring of changes in the background of sporadic human cases, especially of uncommon serotypes, will further define the scope of the problem.