Development and Validation of a Stochastic Model for Predicting the Growth of Salmonella Typhimurium DT104 from a Low Initial Density on Chicken Frankfurters with Native Microflorat

The presence of native microflora is associated with increased variation of Salmonella growth among batches and portions of chicken meat and as a function of temperature. However, variation of Salmonella growth can be modeled using a 95% prediction interval (PT). Because there are no reports of predictive models for growth of Salmonella on ready-to-eat poultry meat products with native microflora and because Salmonella is usually present at low levels on poultry meat, the current study was conducted to develop and validate a stochastic model for predicting the growth of Salmonella from a low initial density on chicken frankfurters with native microfiora. One-gram portions of chicken frankfurters were inoculated with 0.5 log CFU of a single strain (ATCC 700408) of Salmonella Typhimurium DT104. Changes in pathogen numbers over time, N(t), were fit to a two-phase linear primary model to determine lag time (X), growth rate (p.), and the 95% PT, which characterized the variation of pathogen growth. Secondary quadratic polynomial models for natural log transformations of X, p., and P1 as a function of temperature (10 to 40°C) were obtained by nonlinear regression. The primary and secondary models were combined in a computer spreadsheet to create a tertiary model that predicted the growth curve and PT. The pathogen did not grow on chicken frankfurters incubated at 10 to 12°C, but p. ranged from 0.003 log CFU/glh at 14°C to 0.176 log CFU/ glh at 30°C to 0.1 log CFU/glh at 40°C. Variation of N(t) increased as a function of time (i.e., PT was lower during lag phase than during growth phase) and temperature (i.e., PT was higher at 18 to 40°C than at 10 to 14°C). For dependent data (n = 338), 90.5% of observed N(t) values were in the PT predicted by the tertiary model, whereas for independent data (n = 86), 89.5% of observed N(t) values were in the PT predicted by the tertiary model. Based on this performance evaluation, the tertiary model was considered acceptable and valid for stochastic predictions of Salmonella Typhimurium DT104 growth from a low initial density on chicken frankfurters with native microflora. Salmonella is frequently isolated from raw poultry and red meat but is infrequently isolated from ready-to-eat products such as frankfurters, which are heat processed. Palumbo et al. (11) reported that the normal thermal process used to manufacture frankfurters completely inactivates gram-negative bacteria such as Salmonella, but more heatresistant gram-positive bacteria can survive and eventually cause spoilage. Heat processing reduces the total microfiora of frankfurters on average from 105 to 102 CFU/g (11), indicating that although frankfurters are thermally processed and ready to eat, they are not sterile products. Although Salmonella are not expected to survive thermal processing, they can still be found on the finished product, usually as a result of cross-contamination during peeling and packaging. In September 2000 during routine microbiological testing, the U.S. Department of Agriculture Food Safety and Inspection Service detected Salmonella in frankfurter samples obtained at a commercial plant, prompting a product recall (1). * Author for correspondence. Tel: 410-651-6062; Fax: 410-651-8498; E-mail: [email protected]. t Mention of trade names or commercial products in this publication is solely for providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture. Although Palumbo et al. (11) did not observe growth of Salmonella Senftenberg 775W or Salmonella Dublin on frankfurters held at 5 or 37°C, growth of Salmonella on frankfurters held at other temperatures has been reported. Bayne and Michener (3) found Salmonella Ententidis growing on frankfurters at 20°C but not at 7°C. Rice and Pierson (12) observed growth of Salmonella Infantis and Salmonella Enteritidis on frankfurters stored at 15 and 27°C, and Whichard et al. (15) reported growth of Salmonella Typhimurium DT104 on chicken frankfurters held at 22°C for 24 h. These studies indicate that Salmonella is capable of growing at temperatures encountered during storage and handling of the product after manufacture. However, the data available on growth of Salmonella on frankfurters are not sufficient to allow development and validation of a model for predicting food safety. A technical hurdle for modeling pathogen growth in food with native microfiora has been the lack of a naturally occurring strain with a phenotype that can be followed and enumerated in the presence of other microorganisms. Recent studies have revealed that a multiple-antibiotic—resistant strain of Salmonella Typhimurium DT104 (ATCC 700408) that occurs in nature could be used to investigate and model growth on ground chicken breast meat with na1136 OSCAR J. Food Prot., Vol. 71. No. 6 live microfiora; this strain has a phenotype that can be followed and enumerated in the presence of other microorganisms (9). The level of Salmonella on chicken frankfurters, although not reported, is likely to be low (<2 log CFU/g), as suggested by the results of Palumbo et al. (11). Previous studies with sterile (7) and nonsterile (JO) chicken indicate that growth of Salmonella Typhimurium from low initial densities (3 log CFU/g). These results suggest that it might be best to develop a model for chicken frankfurters using a low initial density of Salmonella. Recently, a most-probable-number (MPN) drop-plate method was used to successfully model growth of Salmonella Typhimurium DT104 from a low initial density (0.6 log CFU/g) on ground chicken breast meat contaminated with native microflora (9). Another important factor to consider when developing a model for predicting growth of Salmonella on food with native microflora is the heterogeneity of the food matrix. In a previous study (9), growth of Salmonella Typhimurium DTI04 differed among batches and portions of ground chicken breast meat and as a function of temperature. This growth variation was attributed to variation of the number and types of native microfiora among batches and portions of chicken meat, the nonuniform distribution of the native microflora in the meat, and changes in the types and numbers of native microflora as storage temperature increased (9, 10). In a more processed product, such as frankfurter, that contains additional ingredients such as antimicrobial chemicals, the heterogeneity of the food matrix in terms of both abiotic (i.e., chemical) and biotic (i.e., microbiological) factors at the microniche or microscopic level is likely greater than that in fresh meat. To model the variation of Salmonella growth in a heterogeneous food matrix, a 95% prediction interval (P1) that captures experimental error, the uncertainty of the curve fit, and the scatter of the growth data around the curve has been used successfully (9). The objective of the present study was to develop and validate a stochastic model for predicting the growth of Salmonella Typhimurium DT104 from a low initial density on chicken frankfurters with native microflora. The ability of the model to predict growth of Salmonella on other frankfurter formulations, to predict growth of other strains of Salmonella on frankfurters, and to predict growth from other initial densities of Salmonella were not addressed in this study but will be addressed in future studies. MATERIALS AND METHODS Organism. A multidrug-resistant strain of Salmonella Typhimurium definitive phage type 104 (DTI04; ATCC 700408, American Type Culture Collection, Manassas, Va.) was used for model development and validation because this strain occurs in nature, has a phenotype that can be followed, can be enumerated in the presence of other microorganisms, and has growth patterns similar to those of other strains of Salmonella (9). The stock culture was maintained at —70°C in brain heart infusion (BHI) broth (Becton Dickinson, Sparks, Md.) that contained 15% (vol/vol) glycerol (Sigma, St. Louis, Mo.). Preparation of chicken frankfurter portions. A single brand of chicken frankfurters (Gwaltney, Smithfield, Va.) was purchased weekly from local retail outlets. Listed ingredients were mechanically separated chicken, water, corn syrup, modified food starch, salt, potassium lactate, sodium phosphate, sodium diacetate, flavorings, sodium erythorbate, and sodium nitrite. Composition of frankfurters was reported on the label as 9 g of total fat, 2.5 g of saturated fat, 50 mg of cholesterol, 760 mg of sodium, 5 g of total carbohydrates, 0 g of dietary fiber, 1 g of sugars, and 5 g of protein per frankfurter or per 56-g serving. The listed formulation and listed composition did not change during this study. On a weekly basis, frankfurters from a newly purchased package were sliced and trimmed to yield circular 1-g portions that were transferred to individual wells of a 12-well tissue culture dish (Falcon Multiwell 12-well polystyrene, Becton Dickinson) for subsequent inoculation. In this study, the term "batch" refers to the weekly preparation of frankfurter portions from a single package. Challenge trials. On the day before a challenge trial, 2 ILl of the thawed stock culture was added to 5 ml of BHI broth in a 25-ml Erlenmeyer flask. The flask was then sealed with a foam plug and incubated at 30°C and 150 rpm for 23 h to obtain stationary phase cells for inoculation. Just before inoculation of the chicken frankfurter portions, the 23-h culture (10.2 log CFU/ml) was serially diluted in buffered peptone water (BPW; Becton Dickinson), and 2 il of the 107 dilution was spot inoculated onto the surface of each chicken frankfurter portion for an average initial density of 0.5 log CFU/g. Inoculated chicken frankfurters were incubated at 10, 11, 12, 14, 18, 22, 26, 30, 34, or 40°C. Two (10, 11, and 12°C) or four (14, 18, 22, 26, 30, 34, and 40°C) trials were conducted per temperature, with a different batch of frankfurters and a different inoculation culture in each trial. Pathogen enumeration. At selected incubation times, a l-g frankfurter portion was homogenized (model 80 Stomacher blender, Seward, London, UK) for 1 min in 9 ml of BPW. For samples with a low density (0 to 3.28 log CFU/g) of Salmonella Typhimurium DTI04, the sample homogenate was used in a 3 X 4 MPN assay (9) for pathogen enumeration. The MPN assay samples were prepared and incubated in BPW for 24 h at 38°C before pathogen detection was performed by drop plating onto XLHCATS, which is xylose lysine agar medium (Becton Dickinson) supplemented with 25 mM HEPES (N-[2-hydroxyethyl]piperazine-N'-[2-ethanesulfonic acid]) and 25 .tg/ml chloramphenicol, ampicillin, tetracycline, and streptomycin (Sigma). The MPN tubes that were positive for Salmonella Typhimurium DT104 formed a black drop on XLH-CATS after 24 It of incubation at 38°C. The log MPN per gram was calculated using the method of Thomas (14) as described by Oscar (9). For samples with a higher density (>3 log CFU/g) of Salmonella Typhimurium DT104, the sample homogenate was serially diluted in BPW, and 50 p.1 was spiral plated (Whitley Automatic Spiral Plater, Microbiology International, Frederick, Md.) onto XLH-CATS plates, which were incubated at 38°C for 24 h. Colonies were counted with an automated counter (Protocol, Microbiology International). Primary modeling. Growth of Salmonella Typhimurjum DT104 on chicken frankfurters was not observed at 10 to 12°C, but growth was observed at 14 to 40°C. When growth was observed, pathogen enumeration data for all trials within a temperature were combined and graphed as a function of time and were fit by least squares regression (version 5.0, Prism, Graph pad SoftJ. Food Pro.. Vol. 71 No. 6 MODEL FOR GROWTH OF SALMONELLA ON CHICKEN FRANKFURTERS 1137 ware, Inc., San Diego, Calif.) to a two-phase linear primary model