Food as a Vehicle for Transmission of Shiga Toxin Producing Escherichia coli

Transcription

1 2426 Journal of Food Protection, Vol. 70, No. 10, 2007, Pages Copyright, International Association for Food Protection Review Food as a Vehicle for Transmission of Shiga Toxin Producing Escherichia coli MARILYN C. ERICKSON AND MICHAEL P. DOYLE* Center for Food Safety and Department of Food Science and Technology, University of Georgia, 1109 Experiment Street, Griffin, Georgia , USA MS : Received 2 March 2007/Accepted 8 May 2007 ABSTRACT Contaminated food continues to be the principal vehicle for transmission of Escherichia coli O157:H7 and other Shiga toxin producing E. coli (STEC) to humans. A large number of foods, including those associated with outbreaks (alfalfa sprouts, fresh produce, beef, and unpasteurized juices), have been the focus of intensive research studies in the past few years (2003 to 2006) to assess the prevalence and identify effective intervention and inactivation treatments for these pathogens. Recent analyses of retail foods in the United States revealed E. coli O157:H7 was present in 1.5% of alfalfa sprouts and 0.17% of ground beef but not in some other foods examined. Differences in virulence patterns (presence of both stx 1 and stx 2 genes versus one stx gene) have been observed among isolates from beef samples obtained at the processing plant compared with retail outlets. Research has continued to examine survival and growth of STEC in foods, with several models being developed to predict the behavior of the pathogen under a wide range of environmental conditions. In an effort to develop effective strategies to minimize contamination, several influential factors are being addressed, including elucidating the underlying mechanism for attachment and penetration of STEC into foods and determining the role of handling practices and processing operations on cross-contamination between foods. Reports of some alternative nonthermal processing treatments (high pressure, pulsed-electric field, ionizing radiation, UV radiation, and ultrasound) indicate potential for inactivating STEC with minimal alteration to sensory and nutrient characteristics. Antimicrobials (e.g., organic acids, oxidizing agents, cetylpyridinium chloride, bacteriocins, acidified sodium chlorite, natural extracts) have varying degrees of efficacy as preservatives or sanitizing agents on produce, meat, and unpasteurized juices. Multiple-hurdle or sequential intervention treatments have the greatest potential to minimize transmission of STEC in foods. PREVALENCE OF STEC IN FOODS Since 1982, a large number of reported outbreaks and sporadic cases of human infections occurred with Shiga toxin producing Escherichia coli (STEC). Although serotype O157:H7 is predominantly associated with human infections in the United States, United Kingdom, Canada, and Japan, 100 STEC serovars associated in part with the O26, O91, O103, O111, O118, O145, and O166 antigen serogroups are known to cause human illness (204). Although the major hosts for STEC are ruminants (i.e., cattle, sheep, and goats), there are many routes by which STEC is transmitted, including food and water (vegetables, fruits, and drinking and recreational waters) (49, 212, 234, 245, 280), person-to-person contact (212, 215), and animal-toperson contact (53, 212). To determine whether food or environmental pathways present the greatest risk, a risk analysis study conducted on E. coli O157 outbreaks in Scotland from 1994 to 2003 showed that approximately 40% of the outbreaks were foodborne, 54% were environmental, and 6% involved both transmission routes (256). However, when the total number of cases was considered, * Author for correspondence. Tel: ; Fax: ; mdoyle@uga.edu. foodborne transmission was the predominant form of transmission (83%). In the United States, foods are the most frequently identified vehicles of STEC outbreaks, with 52% of outbreaks of E. coli O157:H7 infection between 1982 and 2002 associated with consumption of contaminated food (212). A 2005 review of the global prevalence of E. coli O157 in beef reported rates ranging from 0.1 to 54.2% in ground beef, 0.1 to 4.4% in sausage, 1.1 to 36.0% in unspecified retail cuts, and 0.01 to 43.4% in whole carcasses (114). In many situations, increased prevalence rates were reported in recent years and were attributed to improved sampling techniques and analytical techniques. An exception has been the change reported by the U.S. Department of Agriculture, Food Safety and Inspection Service in prevalence of E. coli O157:H7 in ground beef in the United States between CY 2000 and CY 2002 (0.8 to 0.9%) and the prevalence between CY 2003 and CY 2006 (0.17 to 0.30%) with no change in testing procedures (266). Isolation rates of non-o157 STEC from foods have been higher than E. coli O157, with prevalence rates ranging from 2.4 to 30.0% in ground beef, 17.0 to 49.2% in sausage, 11.4 to 49.6% in unspecified retail cuts, and 1.7 to 58.0% in whole carcasses (114). Prevalence of STEC in raw milk

2 J. Food Prot., Vol. 70, No. 10 TRANSMISSION OF STEC BY FOOD 2427 TABLE 1. Selected studies (2003 to 2006) on the prevalence and virulence profile of STEC isolates from meat Product type Source Prevalence (%) O157 Non- O157 No. of samples examined Cell no. in positive samples (log CFU/g) Virulence profile (% of positive) stx 1 and stx 2 stx 1 stx 2 eaea ehxa Reference Preevisceration carcass U.S., large Midwestern beef processing plant Postintervention carcass U.S., large Midwestern beef processing plant Minced beef Ireland, butcher shops and supermarkets ND a ND 467 ND ND ND 17 ND ND 21 ND ND ND ND 1, (50% of positives) 0.5 (50% of positives) Beef India, Mangalore ND ND 6 ND Bison carcasses U.S., Midwestern bison 1.1 ND 355 ND processing plant Carcasses, preevisceration Southern U.S., beef 23.3 ND 511 ND ND ND ND ND ND 220 processing plant Carcasses, preevisceration Northern U.S., beef 26.8 ND 522 ND ND ND ND ND ND 220 processing plant Carcasses, postintervention Southern U.S., beef 0 ND 499 ND ND ND ND ND ND 220 processing plant Carcasses, postintervention Northern U.S., beef 0.8 ND 520 ND ND ND ND ND ND 220 processing plant Meat cube Gaborone, Botswana, 5.2 ND 134 ND ND ND ND ND ND 153 retail outlets Minced meat Gaborone, Botswana, 3.8 ND 133 ND ND ND ND ND ND 153 retail outlets Fresh sausage Gaborone, Botswana, 2.3 ND 133 ND ND ND ND ND ND 153 retail outlets Ground beef U.S. federally inspected plants and retail outlets ( ) 0.7 ND 26,521 ND ND ND ND ND ND 186 Beef trimmings Ireland, abattoir 2.4 ND 1, Beef carcasses Ireland, abattoir 3.0 ND Bovine head meat Ireland, abattoir 3.0 ND Ground beef Queensland, Australia, ND retail butchers Lamb cutlets Queensland, Australia, ND retail butchers Ground meat São Paulo, Brazil ND ND 4 ND ND ND 42 Ground beef, fresh U.S., Seattle, retail outlets ,750 ND ND ND ND ND ND 223 Beef cuts Midwest, U.S., commercial meat-processing facilities 0.3 ND 1,022 ND ND ND ND ND ND 254 Cattle carcasses a ND, not determined. Istanbul, Turkey, abattoirs ND ND 12 ND ND 282 was also reviewed in 2005, with rates ranging from 0.75 to 16.2% (115). Reports of recent studies on the prevalence of STEC in foods are summarized in Tables 1 and 2. Prevalence on preevisceration beef carcasses is attributed to fecal contamination of the hide (10, 67), with the brisket and rump areas being the most and least contaminated areas, respectively (214). Prevalence on postevisceration carcasses is less than prevalence on preevisceration carcasses and reflects the effectiveness of intervention steps applied after evisceration (to be discussed later in this article) in the processing of the carcass (220). Geographical locale had the greatest influence on prevalence on preevisceration carcasses, whereas differences in harvest and processing practices and treatments at different processing plants had a greater difference on prevalence on postintervention carcasses. Contamination of processed samples, however, was affected by the season in which animals were slaughtered. Prevalence of positive samples in minced meat collected in Ireland from January to April (4.8%) was much higher than other periods (1.0 to 3.2%) (36). Similarly, a seasonal trend was identified in the rate of E. coli O157:H7 positive raw ground beef samples

3 2428 ERICKSON AND DOYLE J. Food Prot., Vol. 70, No. 10 TABLE 2. Selected studies (2003 to 2006) on the prevalence and virulence profile of STEC isolates from dairy, produce, and seafood products Product type Source Prevalence (%) O157 Non- O157 No. of samples examined Virulence profile (% of positive) stx 1 and stx 2 stx 1 stx 2 eaea ehxa Reference Raw milk Hungary, farms 0.4 ND a 250 ND ND ND ND ND 112 Raw ovine and caprine Western Spain, dairy milk plants Ovine and caprine Western Spain, dairy ND ND ND ND ND 217 fresh cheese curds plants Ovine and caprine Western Spain, dairy ND ND ND ND ND 217 cheeses plants Produce (23 different Minnesota, southern 0 ND 605 ND ND ND ND ND 173 types) and central farms Fresh produce (leafy greens, herbs, and cantaloupe) Southern United States, farms and packing sheds 0 ND 398 ND ND ND ND ND 123 Fresh produce (leafy greens, herbs, cantaloupe, and vegetables) Mushrooms Sprouts Shellfish (mussels, oysters, or cockles) a ND, not determined Southern United States, packing sheds (produce imported from Mexico) 0 ND 466 ND ND ND ND ND 123 Seattle, Washington, retail outlets ND ND ND ND ND 223 Seattle, Washington, retail ND ND ND ND ND 223 outlets French coastal areas collected from federally inspected establishments in the United States; however, this seasonal trend was not significant in samples from retail outlets (186). In Ireland, packaging influenced occurrence of positive samples in minced meat, with greater prevalence of E. coli O157:H7 in fresh packaged burgers (4.5%) than in fresh unpackaged minced meat (2.0%) purchased from supermarkets (36). In most studies of the prevalence of STEC in foods, cell numbers of contamination are not reported. In those few studies in which cell numbers are reported, they are small (Table 1). Of particular concern is the presence of a few cattle in a herd that are described as super-shedders and shed E. coli O157 at levels exceeding 10 4 /g (161, 195, 197). Such high-shedding animals pose an elevated risk of contaminating hides or carcasses of other cattle either before or during slaughter. Moreover, if the carcass is destined to be ground, the entire batch can become contaminated from inclusion of the contaminated meat. When meat remains intact, however, variation in contamination between cuts occurs. Based on 1,022 samples analyzed, E. coli O157:H7 was exclusively isolated from cuts derived from the sirloin area of the carcass (top sirloin butt and butt, ball tip), suggesting that contamination of beef cuts is influenced by the region of the carcass from which they are derived (254). Furthermore, mechanical tenderization can translocate pathogens from the surface to interior sites of beef, requiring thorough cooking of whole meat cuts (203). Surveys of foods other than beef for STEC have also been undertaken (Table 2). Produce has the potential to be contaminated with STEC via manure applied to fields, contaminated irrigation or process water, poor worker hygiene, and poor equipment sanitation (24). In two separate studies surveying a wide range of produce items, however, E. coli O157:H7 was not detected (124, 173). In contrast, STEC was recovered in 6.0 and 4.0% of retail sprouts and mushrooms, respectively (223), and was also found in retail lettuce and spinach samples that were associated with outbreaks in the fall of STEC has also been reported in shellfish (mussels, oysters, or cockles) whose filter-feeding activities concentrate and retain the pathogenic microbes present in contaminated waters (Table 3). Interestingly, STEC prevalence was greater in cockles that reside in muddy sediments than in oysters that reside in the water column (94). VIRULENCE PROFILES OF STEC ISOLATED FROM FOODS Pathogenic STEC strains are typically characterized by the expression of one or two toxins, Shiga toxin 1 (Stx 1) and Shiga toxin 2 (Stx 2) (187). Other virulence factors may or may not be present but include the factor responsible for intimate attachment to the intestinal surface (initimin) and for enterocyte damage (enterohemolysin) encoded by the genes eaea and ehxa, respectively. In toxin-profiling studies of E. coli O157:H7 clinical isolates, Ostroff et al. (199) demonstrated that patients infected with isolates

4 J. Food Prot., Vol. 70, No. 10 TRANSMISSION OF STEC BY FOOD 2429 TABLE 3. Selected studies (2003 to 2006) evaluating the survival and growth of E. coli O157:H7 in different foods Food product Initial concn (log CFU/ml) Temp ( C) ph Survival (days) Growth Reference Whole milk (28) a b (0 days) (68) (24 h) a 5 (days 4 10) Whole milk w/cinnamon and lemon (14) b (0 days) (68) (24 h) 5 (days 4 10) Skim milk (18) (0 days) (68) (24 h) 5 (days 4 10) Unpasteurized whey (days) (21) (days) (day 0) (14) 4.1 (day 28) Pasteurized whey (28) (day 0) (28) (7 days) 4.2 (day 28) Raw goat milk lactic cheeses (42) 268 Yogurt (16 22) 63 Fresh-sliced cactus pear fruit packaged (day 0) (14) (14 days) 52 in air 5.6 (day 14) Fresh-sliced cactus pear fruit in modified atmosphere packaging (day 0) 3.7 (day 14) 5.9 (day 0) 5.5 (day 14) 5.9 (day 0) (14) (14) (10) (14 days) (14 days) 4.0 (day 14) Minimally processed and packaged artichokes (16) (16 days) 226 Fresh whole tomato (10) (1 day) (10) (7) (4 days) (10) Processed tomato juice (23) (23) (23) (23 days) (23) Processed tomato sauce (16) ( 8 days) (16) (16) ( 8 days) (16) Ketchup (6) (21) (2) (8) Orange juice, fresh (35) 63 Strawberry juice (3) 98 Fruit concentrates (apple, orange, (84) 201 pineapple, and white grape) Banana puree (84) Fruit concentrates (cranberry, lemon, (0.25) 190 and lime) Lemon juice, single strength (3) 72 Lime juice, single strength (3) Chocolate ND ( ) 20 Biscuit cream ND (58) Mallow ND (29) a, indicates food was positive for E. coli O157:H7 for or at designated number of days or hours. b, indicates food was either negative for E. coli O157:H7 at designated number of days or no growth occurred.

5 2430 ERICKSON AND DOYLE J. Food Prot., Vol. 70, No. 10 carrying only stx 2 were 6.8 times more likely to develop severe disease than those infected with strains carrying stx 1 or both stx 1 and stx 2, whereas another study showed that the carriage of stx 2 by an isolate increased its association with hemorrhagic colitis and hemolytic-uremic syndrome fivefold (28). Hence, in an effort to understand the clinical significance of STEC strains associated with foods, several studies have surveyed the virulence determinants of STEC isolates (Tables 1 and 2). Rey et al. (217) determined a close relationship exists between the STEC virulence profile and the animal species from which STEC isolates were isolated. For example, in western Spain, all sheep strains harbored stx 2, whereas goat strains carried both the stx 1 and ehxa genes. Results of a study conducted in India showed stx 1 was detected only in STEC isolates from beef samples, whereas seafood isolates carried either the stx 2 gene or both stx 1 and stx 2 (142). In contrast, stx 2 predominated in isolates from ground meat samples in Brazil, whereas stx 1 prevailed among strains of human origin (42). Within the United States, Barkocy-Gallagher et al. (17) reported a higher prevalence of stx 2 and both stx 1 and stx 2 genotypes among cattle serotype O157:H7 isolates than among non-o157 STEC and suggested that virulence profiles contribute to the greater occurrence of O157:H7 strains associated with human disease. MODE FOR CONTAMINATION OF FOODS BY STEC Ruminants, and in particular cattle, have been identified as primary reservoirs of STEC (166). Contamination of carcasses with STEC usually occurs during slaughter and subsequent processing through fecal contamination originating directly or indirectly from the rectal-cecal area. A study in the United States showed a positive correlation of E. coli O157 contamination with total animal prevalence (prevalence in feces and on the hide) and E. coli O157 prevalence on derived carcasses (67). Studies by McEvoy et al. (163) showed that in a typical Irish beef abattoir, carcass contamination with E. coli O157:H7 occurred during both hide removal and bung tying (tying off the rectum with a plastic bag). With pulsed-field gel electrophoresis (PFGE), a subsequent study showed that cross-contamination of STEC from the hide onto the meat most likely occurred during skinning (13). A preevisceration washing step that occurs prior to bung bagging was suggested by Stopforth et al. (253) as the most likely source of STEC carcass contamination. They determined that surfaces of carcasses following postwash bagging were 5% E. coli O157:H7 positive compared with 1.7% for carcass surfaces sampled following prewash bagging. ATTACHMENT AND INTERNALIZATION OF STEC IN FOODS Whereas carriage of STEC occurs in ruminants, foods are contaminated via environmental exposure following shedding of the pathogen into ruminant feces. STEC is well adapted to survive in animal feces (271), water containing animal feces or rumen content (286), and soil (86). In addition, periods of heavy rainfall can transport bacteria into rivers and lakes by runoff (85). Survival of STEC in feces and its redistribution has led to contamination of fresh produce when manure contaminates irrigation water or is used for fertilization (120, 121, ). The degree of exposure to such samples of contamination can affect the amount of produce contaminated. For example, the number of lettuce plants testing positive following a single exposure to E. coli O157:H7 through spray irrigation was greater than the number testing positive following surface irrigation (249, 250). Multiple exposures increase the amount of E. coli O157:H7 contamination on the plant (249). Although these direct sources of contamination are most prevalent, indirect contamination of produce and other food items with STEC also occurs, with birds, rodents, mammals, and insects playing potential roles. For example, Mediterranean fruit flies exposed to fecal material enriched with green fluorescent protein labeled E. coli became contaminated and were capable of transmitting E. coli to intact apples in a cage model system (231). Once a food is exposed to STEC, adhesion to food surfaces may involve several different mechanisms that could include extracellular polymeric substances (128), the presence or absence of fimbriae (83), cell surface hydrophobicity (58), or bacterial surface charge (264). Contradictory results from examining these factors have been reported and may be related to the heterogeneous nature of different food surfaces investigated as well as differences in cell surface composition between the studies. For example, attachment duration on lean fascia-covered tissues was much greater (54%) than on adipose fascia-covered tissue (18%) following simple water rinses (244). The low surface hydrophobicity of E. coli O157:H7 (240) likely accounts for these differences in carcass attachment as well as for less attachment to hydrophobic lettuce cuticle compared with hydrophilic cut lettuce tissue (260). Consequently, treatment with a hydrophobic surfactant (Span 85) detached 80% of the pathogen from the intact lettuce surface; however, it was ineffective at detaching the pathogen from cut edges (102). Studies have shown increased attachment of E. coli O157:H7 to lettuce in the presence of ionic calcium; hence, divalent cations may serve as bridges between surface polymer molecules of the pathogen (102). E. coli O157:H7 cells were observed microscopically to clump on the leaf surface when in the presence of calcium ions (102). E. coli O157:H7 that expresses curli, a thin, wiry, and aggregative surface appendage, has also been observed to clump (47). However, when curli-expressing cultures were compared with noncurli-expressing cultures, only minor differences in attachment were observed, suggesting that curli is not a factor for attachment to food surfaces. Studies of the attachment of E. coli O157:H7 to apple and lettuce surfaces showed greater adhesion was associated with the production of capsules, and neither electrostatic and hydrophobic interactions nor surface proteins appeared to play an important role (103). In contrast to these studies, Solomon and Matthews (248) concluded that not only were cell surface proteins not required but also that active bacterial processes (e.g., increased expression or production of cellular molecules) were not essential for the

6 J. Food Prot., Vol. 70, No. 10 TRANSMISSION OF STEC BY FOOD 2431 interaction of adherence of E. coli O157:H7 to lettuce. In their study, no differences in levels of attachment to lettuce tissue were found between Fluospheres and live or killed cells of E. coli O157:H7. Hence, these authors suggested that interaction or attachment of E. coli O157:H7 to foods is likely a simple physical entrapment. Such a scenario is supported by the observation of E. coli O157:H7 cells penetrating into the stomata and junction zones of cut lettuce leaves and becoming entrapped 20 to 100 m below the cut-edge surface (233). The extent of internalization, however, varies with the food type. In apples, no E. coli O157: H7 cells were found below 6 m on the surface of washed apples (131), whereas in carrots, bacteria were found mainly at cell junctions and in intracellular spaces up to 50 m deep (12). In beef, E. coli O157:H7 was found mainly between muscle fibers and within connective tissue (at a depth of 25 m), whereas in cheese, the bacteria occurred within the protein matrix of the cheese either singly or in small clumps of up to 10 cells (12). Internalization in response to disruption of the physical state of the food was variable. Although mechanical disruption to the seminal root and root hairs of spinach plants did not result in the internalization of E. coli O157:H7 in the aerial leaf tissue (107), the pathogen was internalized at a frequency of 2.5 to 3.0% in oranges with simulated peel punctures and at an uptake level of 0.1 to 0.01% when submerged in contaminated solutions (65). In the latter study, the size of the peel puncture affected E. coli O157:H7 uptake, with pathogen uptake occurring in 31% of oranges containing 0.91-mm surface holes and in only 2% of oranges containing 0.68-mm holes. In addition, the age of the puncture or cut influenced uptake, with penetration of E. coli O157:H7 up to 5.5 mm within and 2.6 mm horizontally away from fresh punctures in apples and no penetration beyond their boundaries in apples with 48-h-old punctures (75). Once internalized, E. coli O157:H7 can grow and survive in apple tissue, causing degradation of plasma membranes and release of cytoplasm contents of apple cortical cells into the central vacuole (122). CROSS-CONTAMINATION OF STEC IN FOODS E. coli O157 can survive for substantial periods of time on stainless steel (74) and plastic (4). Hence, these surfaces can serve as intermediate sources of contamination during food processing operations. Avery et al. (13) confirmed the occurrence of cross-contamination in preslaughter and skinning operations of abattoirs by isolating identical E. coli O157:H7 PFGE subtypes from hides, lairage environments, and beef carcasses. Different scenarios were evaluated to model the transfer of E. coli O157:H7 between beef tissue and high-density polyethylene board surfaces by Flores et al. (81). Results showed that all the treatment variables, i.e., high-density polyethylene board surface roughness (rough and smooth), type of beef tissue (fat and fascia), and beef tissue condition (wet and dry surface), interacted significantly in determining the transfer coefficient. However, in general, fat tissue had a lower transfer coefficient than fascia tissue, whereas transfer coefficients from dry beef tissues were larger than from wet beef tissues. Processing equipment and utensils used in the preparation of fruit juices have also been linked to cross-contamination events. Mean pathogen counts for cutting boards, knives, and extractors used in orange juice squeezing operations with fresh oranges inoculated with E. coli O157: H7 (3.6 log CFU/cm 2 ) ranged from 0.4 to 1.9 log CFU/ cm 2 (160). Subsequent squeezing of uninoculated oranges with these contaminated utensils resulted in E. coli O157: H7 contamination of orange juice. Meat grinding is a unit operation that may be a significant contributor to cross-contamination resulting from multiple contacts with surfaces in the mixing, blending, cutting, and forming actions. With a small-scale grinder (5 to 10 g/s), Flores and Tamplin (80) determined that E. coli O157: H7 accumulated on the grinder collar, which attaches the die to the blade, at cell numbers proportional to the inoculum level. Similarly, E. coli O157:H7 contamination was detected on the collar of a mid-size commercial grinder (34 g/s) (79). Likewise, a bowl cutter, used to grind products and mix ingredients and spices into the ground product, became contaminated with E. coli O157:H7 when inoculated beef trim was fed by hand into an uncontaminated beef-trim batch (78). Areas of the bowl cutter that were most likely to be contaminated were the top of the comb or knife guard on which material remained and the knife. Equipment surroundings were also contaminated when material overflowed the cutter bowl. Cross-contamination events are not limited to foodprocessing environments but also occur in food service operations and the home. Simulating the handling procedures used in a restaurant associated with an outbreak of E. coli O157:H7 infection, 100% of lettuce leaves were contaminated when one inoculated dry lettuce piece was mixed with a large volume of dry lettuce and stored in water in the refrigerator (269). Because many of the operations in food service operations and home kitchens involve human hand contact, several studies have investigated the transfer rates between foods and hands and vice versa. During patty formation, gloved fingers became contaminated with 2.76 and 2.35 log of E. coli O157:H7 following handling of ground beef containing the pathogen at levels of and CFU/g, respectively, and subsequently transferred approximately 10 and 1% of the pathogen to lettuce, respectively (270). While these investigators found similar levels of E. coli O157:H7 transfer with gloved hands and bare hands, Montville et al. (170) determined a 1,000-fold decreased transfer by gloves as opposed to bare hands. Food preparation surfaces are also subject to contamination upon exposure to contaminated food products. Contact of E. coli O157:H7 contaminated hamburger patties ( CFU/g) with plastic cutting board surfaces for 5 to 10 s resulted in an average transfer of approximately CFU E. coli O157:H7 per 50 cm 2 (270). Although overnight storage significantly decreased the numbers of recoverable bacteria on these cutting boards, a 15-s water rinse failed to remove substantial numbers ( 0.4 log CFU/ cm 2 )ofe. coli O157:H7 whether the rinse was applied immediately after contamination or following overnight room temperature storage. Lettuce leaves placed in contact

7 2432 ERICKSON AND DOYLE J. Food Prot., Vol. 70, No. 10 with these contaminated cutting boards also became contaminated; however, fewer bacteria were transferred from the contaminated cutting boards to the leaves applied in succession. Unfortunately, washing procedures used in kitchens for the cleaning of utensils and dishes are not always sufficient to eliminate E. coli O157:H7 from contaminated surfaces (162). When dishes were inoculated at 10 3 CFU per dish and washed in water (38 to 48 C) containing 0.12% (vol/vol) detergent, dishes (100%) were still contaminated with E. coli O157:H7. Moreover, the washing-up sponge, the dish towel used to dry the dishes, and any sterile dishes washed subsequently all became contaminated. Hence, the potential for survival and cross-contamination of E. coli O157:H7 in the kitchen environment is significant. SURVIVAL AND GROWTH OF STEC IN FOODS Transmission of E. coli O157 by foods is dependent on the survival of the pathogen in the product as well as the composition of the product (129). Studies continue to address survival in a variety of products and focus on commodities associated with recent outbreaks, including commodities in which postprocessing contamination could occur, as well as commodities whose composition is typically less favorable (e.g., low ph) for survival of pathogens. Conditions characterizing survival and growth of enterohemorrhagic E. coli have been reviewed by Bell (22). In general, nonrefrigerated temperatures ( 7 C) are sufficient for growth of E. coli O157:H7; however, survival time is markedly extended under refrigeration temperature conditions (2 to 5 C). In many survival studies, both high and low levels of inoculum have been evaluated. In these studies, higher contamination levels ( 10 2 CFU/g) have been used to determine the rate of decline in a given medium, whereas lower contamination levels ( 10 2 CFU/g) have been used to represent levels that would most likely be encountered in actual circumstances. Selected survival studies published from 2003 to 2006 are highlighted in Table 3. The first group of entries addresses survival in dairy products. Milk is an excellent medium for the survival of E. coli O157:H7; however, the survival period at 4 C is cut in half when flavored and skim milks are contaminated compared with whole milk (157). Although pasteurization is typically used for inactivation of pathogens in milk, in many countries, including the United States, unpasteurized milk has been used for making cheese, especially in small farmstead operations. E. coli O157:H7 survived all stages of raw milk cheese production for up to 70 days postmanufacturing in a smear-ripened cheese (154) and also in a raw goat milk lactic cheese (268). Moreover, survival occurs in contaminated whey produced after separation of cheese (159). Survival in this latter product was affected to some degree by the presence of lactic acid bacteria, as the duration of survival in the unpasteurized whey was significantly lower than that in the pasteurized samples. Nevertheless, E. coli O157:H7 is generally considered tolerant to acid (51), and this would explain the pathogen s ability to survive in other foods of high acidity, including apple cider (ph 3.7 to 4.0), buttermilk (ph 4.3), yogurt (ph 4.17 to 4.39), and sour cream (ph 4.3) (60, 167). In these types of products, acid adaptation is also of concern. For example, in processed tomato sauce, a sharp decline in E. coli O157:H7 populations occurred for up to 8 days; however, after this period, the counts increased, possibly as a result of acid adaptation (73). Survival of STEC in food products that have been associated with contamination has been studied to determine the hazard potential under a range of storage conditions. While E. coli O157:H7 survived for only limited periods (1 to 21 days) in shelf-stable, dairy-based, pourable salad dressings (25) and margarine or yellow fat spreads (106), there was a greater concern with contamination in several products (chocolate, biscuit cream, or mallow) of reduced water activity (0.4 to 0.8) due to the extended survival (42 to 366 days) of the pathogen at 10 and 22 C (20). To prevent contamination during manufacture of these latter products, it was recommended that controls, similar to those used to prevent contamination by Salmonella, be implemented. In light of the recent isolation of multiantibiotic-resistant strains of STEC from foods (227, 228), the survival of multiantibiotic-resistant cultures has been compared with antibiotic-sensitive cultures. While McGee (164) suggested a possible link between the acquisition of antibiotic resistance and the survival of E. coli O157:H7 in a low ph medium, Duffy et al. (63) determined that multiantibioticresistant E. coli O157:H7 died off in yogurt and orange juice significantly more rapidly than in the antibiotic-sensitive STEC strains. Because only one multiantibiotic-resistant strain was investigated in the latter study, the possibility that strain differences played a role was not discounted. Consideration of STEC serotypes other than O157:H7 should also be addressed in survival studies. Studies of STEC O91:H21 by Molina et al. (169) showed a stress resistance that was greater than that of two O157: H7 strains tested under several acidic conditions. Similarly, Baylis et al. (20) determined that E. coli of serotypes O111 and O26 survived in chocolate at 38 C longer than E. coli O157:H7. The source of an isolate may also influence its survival. In a study comparing the survival of isolates derived directly from animals versus laboratory cultures, the isolate from orally inoculated cattle survived for up to 10 weeks longer in water than a strain that was subcultured previously in the laboratory (229). Hence, results from isolates repeatedly subcultured in the laboratory may not adequately represent the survival that may occur under lownutrient conditions compared with an isolate that recently survived passage through the gastrointestinal tract and is more likely to serve as a source of food contamination. Modeling the survival and growth of E. coli O157:H7 in foods has been a major focus of several research groups (89, 90, 137, 261, 284). With these models, researchers attempted to predict the survival or growth rate of the pathogen under specific conditions of temperature, ph, gaseous atmosphere, chemical preservatives, or aqueous activity, either alone or in combination. Although the three most common types of models used to describe relationships between combinations of factors and growth parameters are the Ar-

8 J. Food Prot., Vol. 70, No. 10 TRANSMISSION OF STEC BY FOOD 2433 rhenius equation, the square root model, and the response surface model (275), artificial neural networks are considered more powerful, as they are capable of predicting not only growth but also survival of E. coli O157:H7, such as in a mayonnaise-type system (284). The application of these and future versions of predictive models to risk assessments will serve as useful tools in developing strategies to reduce the occurrence of E. coli O157:H7 contamination. PHYSICAL AND CHEMICAL INTERVENTIONS TO REDUCE TRANSMISSION OF STEC IN FOODS Thermal interventions. Thermal processing is one of the most common methods applied to foods to inactivate pathogens such as E. coli O157:H7. Hence, the heat sensitivity of this pathogen has been studied extensively, with D-values (decimal reduction time or time required to destroy 90% of the organisms) and z-values (degrees required for the thermal destruction curve to traverse 1 log cycle) of STEC inactivation in meat and poultry recently reviewed by O Bryan et al. (193). In general, E. coli O157:H7 is not considered heat resistant (130), being more heat sensitive than indigenous bacteria in beef (126) and more heat sensitive than Salmonella or Listeria monocytogenes in ground pork (174) and chicken-fried beef patties (198). The thermal tolerance of the pathogen is affected by the composition of food. For example, in white grape juice concentrate, the heat resistance of E. coli O157:H7 was greater than in single-strength juice and was greater than L. monocytogenes or Salmonella (71). Smith et al. (247) determined that E. coli O157:H7 was more heat resistant in beef containing 19% fat than in beef containing 4.8% fat, which was attributed largely to the low thermal conductivity and reduced water activity of fat. Emulsification of fat during turkey grinding may reduce the protective thermal effect of fat, as Kotrola and Conner (141) did not observe a protective thermal effect with emulsified turkey products containing different fat contents. Other variables may also influence the thermotolerance of E. coli O157:H7, such as the strain, the age of the culture, and the environmental conditions used for culturing the inoculum (130, 208). For example, latestationary-phase cells have greater heat resistance than logarithmic-phase cells (130), and heat resistance of emulsiongrown cells is greater than that of broth-grown cells (208). A principal purpose for heating of foods is to convert a raw food into a form preferred by consumers. Different methods of heating have emerged that vary in their form of delivery and in their effectiveness for pathogen inactivation. For example, microwave heating involves the use of electromagnetic waves with frequencies of 2,450 and 915 MHz (55). Compared with conventional heating, a product is heated to a desired temperature in a shorter time period by microwaves. However, the widespread use of microwave heating in commercial operations is hampered by nonuniformity of heating and inability to ensure thermal inactivation of pathogens throughout the product (59). For example, studies by Apostolou et al. (8) on microwave heating fresh chicken as small portions of breasts and carcasses showed E. coli O157:H7 survived in chicken breast portions that appeared to be well cooked (endpoint temperature of 69.8 C). Survival also occurred in whole chicken carcasses that appeared to be thoroughly cooked (subsurface wing and thigh temperatures between 60.2 and 92 C). Some other forms of cooking also do not always ensure E. coli O157:H7 inactivation because of nonuniform heating. Grilling with a double-sided grill proved more effective than grilling with a single-sided grill, especially when the product was only turned once (218). Most recent research on thermal inactivation of E. coli O157:H7 has focused on three main areas: (i) validating thermal processes; (ii) decontaminating raw products; and (iii) determining the combined effects of heat and additives on thermal sensitivity. Thermal process validation studies are needed to comply with hazard analysis critical control point regulations that require processors to document their processes to meet performance standards (175, 177). Such studies evaluate the reliability of models to predict pathogen inactivation under a defined set of conditions. With these models, D- and z-values must be known for the product in question; however, key parameters between the referenced study and commercial process should be similar (193). Moreover, because most situations involve a transient heating process (product temperature changes with time), thermal pathogen lethality is estimated by calculating process lethality, which is described as the time needed to cause the required log reduction in pathogen numbers at a specific reference temperature (176). Mild heat treatments or high-temperature, short-time heat treatments have also been evaluated in recent years as a process for pathogen inactivation on raw produce and meat. When chemical treatments proved minimally effective in decontaminating seeds for sprouts, wet and dry heat treatments, either by themselves or in combination with various commercial sanitizing solutions, were investigated (108, 230, 274). In general, low temperatures (23 C) fail to kill pathogens, whereas hot water treatments (50 to 70 C) reduce the populations of pathogenic bacteria but may also reduce the germination rate of the seeds (230, 274). Timetemperature conditions required for sufficient inactivation (a 5- to 7-log reduction) of pathogens while maximizing germination rate vary with the seed type (274). Moreover, seeds tolerate longer exposures to dry heat than moist heat; a 4-day treatment at 55 C killed all pathogenic E. coli without affecting germination rate (108). Thermal treatments for decontaminating hides and carcasses have been studied extensively, resulting in three major methods for application of heat being adopted by the industry: (i) hot water rinses; (ii) steam pasteurization; and (iii) steam vacuuming. Several recent reports provide overviews of these methods to reduce pathogens on meat (59, 113, 136). Because none of the interventions is 100% effective, a multiple-hurdle intervention system is usually employed by processors to ensure the safety of their products, such as including a steam vacuuming step after knife trimming and then a preevisceration wash of hot water or organic acid (136). Recent research has focused on areas in which thermal treatments were not previously applied. For example, Retzlaff et al. (216) determined the operating conditions for a steam pasteurization unit that could be incor-

9 2434 ERICKSON AND DOYLE J. Food Prot., Vol. 70, No. 10 porated into small-scale meat processing operations, and Logue et al. (150) determined the effect of condensing steam on the decontamination of meat primals postfabrication. When the pathogen survived high-temperature, short-time treatments, Huang (109) suggested that either the surface temperature never reached the temperature needed to kill the bacteria or that the bacteria were entrapped beneath the surface and not exposed to the destructive effects of the steam. Carcass discoloration (113) and increased Shiga toxin secretion by surviving cells (285) are hypothetical concerns regarding the use of thermal treatments. Carcass surface discoloration is only a temporary response (40, 206), and concerns about Shiga toxin secretion are speculative with little evidence of practical significance. Exposure of STEC to chemical additives prior to or in conjunction with thermal treatments has been the subject of recent studies to elucidate increased thermal tolerance due to expression of cross-protective stress proteins (127, 235). Novak and Yuan (192) did not observe any changes in the D 60 -values for E. coli O157:H7 on beef exposed to ozone (3 ppm of O 3 for 5 min); however, Blackman et al. (27) suggested that enhanced thermotolerance could arise in the presence of an iron and nucleotide oxidative complex. Acid-adapted cells of E. coli O157:H7 have also been observed to be more resistant to heat (235), implying that pasteurization conditions necessary to achieve inactivation in fruit juices or other acidic milieu need to be more severe. The type of acidulant can influence microbial inactivation by heat with acetic acid, rendering E. coli O157:H7 more sensitive to the lethal effect of heat than lactic acid (127). Alternatively, addition of sodium lactate did not affect the thermal resistance of E. coli O157:H7 in ground beef (110), nor were there significant differences in D 65 -values for acid-adapted and nonadapted E. coli O157:H7 in stored (4 or 20 C) ground beef (243). In contrast, thermal inactivation of E. coli O157:H7 in apple cider was significantly increased when the cider contained 0.02% glycerol monolaurate (7). With this additive, a 5-log CFU/ml reduction could be achieved by heating at 50 C for 5 min. Freezing interventions. In contrast to thermal interventions, freezing does not substantially reduce pathogen cell numbers. For example, Conner and Hall (50) reported that 50% of E. coli O157:H7 cells survived during 18 months of storage at 20 C in ground chicken breast meat. Injury and loss of viability in apple juice, however, can be increased with multiple freeze-thaw cycles (278) or through frozen storage combined with other treatments, such as the addition of chemical additives (119, 265). The growth conditions of the pathogen can influence the effect of freezing on E. coli survival; for example, cold stress (4 C for 4 weeks) or starvation (cells suspended in water at 37 C for 6 h) led to increased freeze-thaw survival of E. coli O157: H7 (68, 91). Drying interventions. Drying is not an effective intervention to kill STEC. Although many nonpathogenic E. coli strains are sensitive to desiccation, studies with 35 strains of STEC (15 each of serotypes O157 and O26 and 5 of serotype O111) showed all survived 24 h of drying at 35 C (105). These results suggest that desiccation tolerance is a discriminative property of STEC that is expressed regardless of serotype or toxin type. Furthermore, STEC survival is enhanced in the presence of sugar and fat. Studies have shown that STEC can survive for up to 2 years in a desiccated state when stored at refrigeration conditions (105). Pretreatment of food with acidulants prior to drying can influence the antimicrobial activity of drying treatments. For example, E. coli O157:H7 inactivation on dried apple slices was increased by pretreatment with common household acidulants compared with slices dried without treatment (56). Beef jerky slices pretreated with an acidic marinade (5 to 9% acetic acid) also exhibited increased inactivation of E. coli O157:H7 during drying (37, 283). High-pressure interventions. High hydrostatic pressure (HHP) is an emerging technology proposed as an alternative to thermal processing to inactivate pathogens in foods. HHP, also known as high-pressure processing or ultrahigh-pressure processing, subjects liquid and solid foods, with or without packaging, to pressures between 100 and 800 MPa. Investigations into the mechanism for inactivation by HHP have concluded that it is multitargeted, with the cytoplasmic membrane being the key target and extensive solute loss, protein coagulation, inactivation of key enzymes, and ribosome conformational changes also factoring into the process (158). When exposed to HHP in the range of 450 to 690 MPa, inactivation of E. coli O157:H7 in whole milk had a logarithmic relationship to treatment pressure (46), although differences in resistance to HHP were strain-dependent (23, 155). With Fourier-transform infrared spectroscopy, spectral differences between these strains presumably correspond to variations in the cell envelope s lipid and carbohydrate signals (156). The pressure required for effective inactivation of STEC in foods makes HHP, when used alone, a very costly technology. A combination of treatments with lower pressures have been used successfully to reduce operating costs. Early studies with milk and poultry by means of this technology showed that the combined use of pressure and elevated temperatures was more effective in inactivating E. coli O157:H7 than either treatment alone (205). The antimicrobial effectiveness of high-pressure treatments has also been increased by incorporating a rapid decompression step. For example, in orange juice subjected to a 250-MPa treatment, rapid decompression ( 2 ms) provided a 4.8-log CFU/ml reduction in viable E. coli O157:H7 compared with a 3.2-log CFU/ml reduction with slow decompression ( 30 s) (191). Alternatively, dynamic high pressure employing multiple exposures to reduced pressures has been effective in inactivating E. coli O157:H7 in milk and orange juice (259, 267). Because sublethal injury can occur with HHP treatments (125), it is important to include a recovery step to enable the repair of cell injury and prevent overestimation of the microbiological safety provided by the process (31). Pulsed-electric field interventions. High-intensity pulsed-electric field processing is another emerging intervention technology for inactivating E. coli O157:H7 in

10 J. Food Prot., Vol. 70, No. 10 TRANSMISSION OF STEC BY FOOD 2435 foods. Such treatments involve the application of pulses of high voltage (5 to 80 kv/cm) to foods placed between two electrodes, with cell death attributed to membrane structural or functional alteration (158). Exposing dialyzed egg white, egg yolk, and whole egg products at 0 C to 500 pulses with an electric field strength of 15 kv/cm reduced E. coli O157: H7 by 1, 3, and 3.5 log CFU/g, respectively (5). Further studies showed that increasing treatment temperature increased the E. coli O157:H7 inactivation rate in both liquid egg white (6) and liquid whole egg (21). Inactivation may also be facilitated by subjecting pulsed-electric field sublethally injured cells to an acidic milieu following treatment (87). For example, the combination of a pulsed-electric field treatment of apple juice at 25 kv/cm for 400 s and subsequent storage under refrigeration for 48 h provided a 5-log CFU/g reduction of E. coli O157:H7 compared with a 1-log CFU/g reduction immediately after treatment (88). Ultrasound interventions. Ultrasound treatments involve the application of sound waves of 20 khz to foods in a liquid medium that generate locally high pressures and temperatures and, as a result, lyse microbial cells through intracellular cavitation. Ultrasonography has been used experimentally to dislodge bacteria from the surface of poultry carcasses during chilling in chlorinated water (145). Application to larger carcasses such as beef or swine is not practical, because immersion in a liquid would be required. With more suitable applications involving liquid foods such as apple juices, continuous flow ultrasound treatment (sonifier probe at 20 khz, 100% power level, 150 W acoustic power, and 118 W/cm 2 acoustic intensity) combined with mild heat (57 C) reduced E. coli O157:H7 cell numbers by 6 log CFU/ml (54). Ionizing radiation interventions. Low-dose ionizing radiation is an effective treatment for killing pathogenic bacteria in foods through the formation of thymine dimers and toxic-free radicals. However, commercial use of irradiation in foods presently is limited, largely because of poor consumer acceptance, increased product cost for the treatment, and negative effects on odor and flavor in high dose treated products. Research has focused on evaluating the effectiveness of low- and medium-dose irradiation treatments to minimize the undesirable quality characteristics. Low and medium doses of irradiation (1 to 2 kgy) of E. coli O157:H7 contaminated beef steaks and ground beef were effective in reducing E. coli O157:H7 by 4 to 5 log CFU/g without creating negative effects on flavor and color (11, 84). Apple cider has also successfully been treated with medium-dose irradiation, with an irradiation treatment of 2.47 kgy, reducing acid-resistant E. coli O157:H7 by 5 log CFU/ml (273). The effectiveness of low- and medium-dose irradiation treatments is dependent on the type of food being treated. For example, the D-values (dose required for a 1-log reduction) for inactivation of E. coli O157:H7 were higher in chicken than in beef or trout (26). D-values were also substantially greater in broccoli seeds (1.11 to 1.43 kgy) than in alfalfa seeds (0.55 to 0.60 kgy) (210, 262). However, dosages higher than 2 kgy adversely affect sprouting and the subsequent yield of sprouts (211). Combining irradiation (2.0 kgy) with dry heat at 50 C for 60 min completely inactivated E. coli O157:H7 on alfalfa seeds and did not decrease the germination percentage or sprout length (15). Similarly, a treatment combination of 200 ppm of chlorine and low-dose irradiation (1.05 kgy) significantly reduced levels of E. coli O157:H7 ( 7 log CFU/ml) on fresh cilantro while retaining product quality (82). A combination treatment of irradiation and dry heat on mung bean or radish seeds, however, affected sprout length (15). Such results illustrate the importance of a critical evaluation of combination treatments prior to their commercial application, especially in light of the finding that ph-dependent stationary-phase acid resistance provided an enterohemorrhagic E. coli strain cross-protection to subsequent irradiation (34). UV light interventions. UV light has been the subject of several recent studies on the inactivation of E. coli O157: H7 in alfalfa seeds and fruit juices and on the surfaces of fresh produce and fish. UV light possesses germicidal properties in wavelength ranges of 100 to 280 nm and is attributed to photochemical transformation of pyrimidine bases in bacterial DNA to form dimers, thus destroying the ability of bacteria to multiply and cause disease. UV light (253.7 nm) applied at a dosage of 24 mj/cm 2 reduced E. coli O157:H7 on apples and leaf lettuce by ca. 3.3 and 2.79 log CFU/g, respectively (281). UV doses of 14 mj/cm 2 inactivated 5 log CFU of E. coli O157:H7/ml in apple cider, with levels of inactivation dependent on the strain and on the apple cultivar used to make the cider (19). There was no statistically significant relationship between the Brix, ph, or malic acid content and inactivation of E. coli O157: H7. Absorptivities of fruit juices did influence the effectiveness of UV light treatments, with D-values for the inactivation of E. coli O157:H7 increasing as the absorbance of the media increased (200). Stirring fruit juice containing suspended particles or reducing the liquid film thickness minimized the influence of suspended particles on the antibacterial effect of UV light (200). Alternatively, multiple exposures of apple cider to UV light increased the effectiveness of this treatment (61). The efficacy of UV light to inactivate microorganisms may also be improved through pulsation of the light source (64). Pulsed UV light (5.6 J/cm 2 per pulse, 3 pulses per s, 60 to 90 s) applied to inoculated alfalfa seeds reduced E. coli O157:H7 cell numbers by greater than 4 log CFU/g, with greater reductions occurring on seeds closer to the light source (236). Subjecting raw salmon fillets to pulsed UV light (5.6 J/cm 2 per pulse, 3 pulses per s, 60 s) reduced E. coli O157:H7 cell numbers from an initial population of 6 to 0.86 log CFU/g on skin side and 1.09 log CFU/g on muscle side; however, the surface temperature increased to 100 C during the treatment and led to color and quality changes (202). Biological interventions. Fermented foods are produced by the metabolic activity of microorganisms that breaks down macromolecular compounds to bring about desirable changes, such as acid production. The acidity produced can help preserve food and extend its shelf life. An