A total of 482 veal cutlet, 555 ground veal, and 540 ground beef samples were purchased from retail establishments in the mid-Atlantic region of the United States over a noncontiguous 2-year period between 2014 and 2017. Samples (325 g each) were individually enriched and screened via real-time PCR for all seven regulated serogroups of Shiga toxin–producing Escherichia coli (STEC). Presumptive STEC-positive samples were subjected to serogroup-specific immunomagnetic separation and plated onto selective media. Up to five isolates typical for STEC from each sample were analyzed via multiplex PCR for both the virulence genes (i.e., eae, stx1 and/or stx2, and ehxA) and serogroup-specific gene(s) for the seven regulated STEC serogroups. The recovery rates of non-O157 STEC from veal cutlets (3.94%, 19 of 482 samples) and ground veal (7.03%, 39 of 555 samples) were significantly higher (P < 0.05) than that from ground beef (0.93%, 5 of 540 samples). In contrast, only a single isolate of STEC O157:H7 was recovered; this isolate originated from 1 (0.18%) of 555 samples of ground veal. Recovery rates for STEC were not associated with state, season, packaging type, or store type (P > 0.05) but were associated with brand and fat content (P < 0.05). Pulsed-field subtyping of the 270 viable and confirmed STEC isolates from the 64 total samples testing positive revealed 78 pulsotypes (50 to 80% similarity) belonging to 39 pulsogroups, with ≥90% similarity among pulsotypes within pulsogroups. Multiple isolates from 43 (67.7%) of 64 samples testing positive had an indistinguishable pulsotype. STEC serotypes O26 and O103 were the most prevalent serogroups in beef and veal, respectively. These findings support related findings from regulatory sampling studies over the past decade and confirm that recovery rates for the regulated STEC serogroups are higher for raw veal than for raw beef samples, as was observed in the present study of meat purchased at food retailers in the mid-Atlantic region of the United States.
Higher recovery rates of regulated STEC serogroups were found in raw veal than in beef.
Recovery of STEC was not associated with state, season, packaging, or store type.
Recovery of STEC was associated with brand and fat content.
The most prevalent STEC serogroups in beef and veal were O26 and O103, respectively.
STEC with the same pulsotype had the same virulence profile.
Shiga toxin–producing Escherichia coli (STEC) poses a significant threat to public health as evidenced by its occasional recovery from raw meat and by the number and magnitude of recalls and associated illnesses attributed to STEC and beef that continue to occur (6, 13, 36, 53). Although serotype O157:H7 strains of E. coli have been a concern for beef processors, researchers, and regulators since the early 1980s, more recently six additional serogroups of STEC (“the big six”), namely O26, O45, O103, O111, O121, and O145, were also declared adulterants when present in raw beef and veal (45). The incidence of non-O157 STEC infections in the United States surpassed that of O157 in 2013 (14). Non-O157 STEC strains cause about twice as many foodborne illnesses per year in the United States as do strains of serotype O157:H7 STEC, with an estimated ca. 97,000 STEC O157 infections versus ca. 170,000 non-O157 STEC infections (11, 42, 55). According to Heiman et al. (23), ca. 65% of U.S. STEC O157 outbreaks from 2003 to 2012 were of foodborne origin, with the remainder due to person-to-person contact (10%), indirect or direct contact with animals (10%), water (4%), or a different or unknown mode of transmission (10%). In contrast, ca. 82% of non-O157 STEC illnesses are estimated to be foodborne in origin (42). The incidence of U.S. infections attributed to O157 STEC declined in 2017 compared with 2006 to 2008, probably at least in part because of targeted regulations from the U.S. Department of Agriculture (USDA) Food Safety and Inspection Service (FSIS) and subsequent control measures implemented by industry over the past decade that in general resulted in a lower prevalence and decreased levels of STEC associated with ground beef (31). The overall incidence of U.S. STEC infections in 2016 increased from 1.09 to 1.69 cases per 100,000 population; O157 STEC infections decreased from 0.96 to 0.62 cases per 100,000, and non-O157 STEC infections increased from 0.13 to 0.89 cases per 100,000 (14). Despite sporadic outbreaks and/or recalls of raw veal products linked to STEC (47, 49, 50, 53), the global veal market will likely continue to grow over the next few years because consumers are increasingly demanding more tender and juicy cuts of meat such as veal (2).
Since the early 1980s much has been published on the recovery of E. coli O157:H7 from ground beef, whereas less information is available on the recovery of non-O157 STEC from ground beef and even less has been published on recovery of all seven regulated STEC serogroups from veal. In a handful of studies, the prevalence of STEC in veal was noticeably higher than that in beef, thus fueling concerns over a disproportionately greater public health risk from veal than from beef, even though in the United States only ca. 80 million lb (36.3 million kg) of veal but ca. 26 billion lb (11.8 billion kg) of beef are produced annually (1, 9, 32, 43). The USDA FSIS (46, 48) reported a higher public health risk associated with STEC from raw ground beef components (RGBC) derived from veal (8.0% of 312 samples in 2012 and 4.8% of 250 samples in 2013) than from those derived from beef (0.8% of 5,342 in 2012 and 0.4% of 7,011 in 2013); however, far fewer veal samples (562) than beef samples (12,353) were analyzed. In another regulatory sampling exercise by the USDA FSIS (51) from 2014 to 2015, the Nationwide Microbiological Baseline Data Collection Program: Beef-Veal Carcass Baseline Survey (BVCBS), the estimated national prevalences of E. coli O157:H7 and non-O157 STEC were 0.06 and 0.21%, respectively, on beef carcasses and 0.50 and 8.54%, respectively, on veal carcasses. This trend toward higher recovery of non-O157 and E. coli O157:H7 STEC from veal than from beef was also observed in the STEC sampling program for RGBC in 2018 (50). Regarding raw beef testing, evaluation of RGBC in 2018 revealed that 0.19% (6 of 3,203) and 0.16% (5 of 3,139) of beef samples were positive for O157:H7 and non-O157 STEC, respectively (50). All 35 RGBC veal trim samples analyzed for O157:H7 were negative in the STEC sampling program for RGBC in 2018, whereas 8.82% of 34 veal samples were positive for non-O157 STEC (50).
Although considerable information has been published on recovery rates and to a lesser extent levels and subtypes of serotype O157:H7 strains in raw beef, much less information is available on the comparative prevalence (plus levels and types) of the big six STEC serogroups associated with raw veal and/or beef, particularly in retail products. In the studies that have been published, estimated recovery rates were based on analysis of a very limited number of veal samples. The objective of the present study was to collect raw ground veal and ground beef samples from food retailers across the U.S. mid-Atlantic region to establish and compare the true prevalence of the seven regulated serogroups of non-O157 STEC and of E. coli O157:H7 in raw veal versus raw beef and subsequently to characterize the virulence potential and genetic relatedness of these strains.
MATERIALS AND METHODS
Collection of samples
Raw veal and beef were purchased on a weekly basis from May 2014 to June 2015 and again from July 2016 to April 2017 in the mid-Atlantic region of the United States: Delaware, Maryland, Pennsylvania, Virginia, and North Carolina. During this 2-year noncontiguous sampling period, 482 veal cutlet, 555 ground veal, and 540 ground beef samples were randomly collected from national and local supermarkets (122 stores total) and specialty stores or meat markets (14 stores total). The list of stores was generated manually by searching for appropriate food retailers online before each shopping trip. The 136 stores visited were confined to a limited number of zip codes near or within 3 to 14 metropolitan areas in each state as follows: Delaware, 30 stores (15 zip codes); Maryland, 28 stores (20 zip codes); Pennsylvania, 36 stores (22 zip codes); Virginia, 31 stores (22 zip codes); and North Carolina, 11 stores (6 zip codes). Shoppers were instructed to (i) obtain 20 samples (13 to 15 veal samples and 5 to 7 beef samples) per week, (ii) purchase boneless veal cutlets, ground veal, and ground beef (ca. 1.0 kg total of each by compositing multiple packages if necessary) of one brand with the same use-by or sell-by date, (iii) place each sample in a separate Ziploc plastic bag (W.W. Grainger, Lake Forest, IL) to avoid possible cross-contamination during transport to the testing laboratory, (iv) measure and record product temperature at the time of purchase with a handheld infrared thermometer (Fisherbrand Traceable, Fisher Scientific, Pittsburgh, PA), and (v) store samples in a cooler with ice packs during shopping and during transportation from the food retailer to the research laboratory (ca. 1 to 6 h). The samples collected in Maryland, Virginia, and North Carolina were also separately transferred into Ziploc bags and placed in coolers containing ice packs for shipment via overnight carrier to the Food Safety Laboratory (University of Maryland Eastern Shore, Princess Anne) and/or the USDA Agricultural Research Service laboratory (Wyndmoor, PA). In addition to purchasing raw veal and beef samples, shoppers also collected associated metadata for products and stores.
Metadata of meat samples
Upon arrival at the laboratory, samples were stored at 4 ± 2°C and processed well within the sell-by or frozen-by date, which was typically within 72 h. The following metadata were recorded for each sample: (i) date purchased at retailer and date received at testing laboratory, (ii) state and store names and addresses where the samples were purchased, (iii) brand and/or supplier information, (iv) temperature of the sample at purchase and upon arrival at the laboratory, (v) type and species of the sample (veal cutlet, ground veal, or ground beef), (vi) percent fat as stated on product label, (vii) presence of safe handling instructions and/or cooking instructions or temperatures on product label, (viii) use-by or sell-by dates, and/or (ix) whether the sample was ground and/or packed at the store or by the manufacturer. After taking photos of each sample and its attendant label, the product was removed from the original package and repacked in a sterile bag (30.5 by 38.1 cm; Prime Source Packaging, Houston, TX). A unique letter-number code was assigned to each sample to blind the brand and store information. Samples were then stored at 4 ± 2°C for 18 h until being screened for STEC.
Screening meat samples for STEC and recovery of isolates from presumptive-positive samples
The protocols from the USDA FSIS Microbiological Laboratory Guidebook sections 5B.05 and 5.09 (52) were used with only minor modifications (26) to isolate cells of the seven regulated serogroups of STEC from raw meat (Fig. 1). The repacked ground meat was emptied onto a Styrofoam tray (1012S, Genpak, Glens Falls, NY) and mixed well by hand. Veal cutlets were separately emptied onto Styrofoam trays and cut into small pieces (ca. 1.25 by 1.25 by 4 cm) with a sterilized knife and fork. A sample (325 ± 32.5 g) of ground meat or chopped veal cutlet was aseptically transferred into a stomacher bag (BA6042, Seward, Worthing, UK) containing 975 ± 19.5 mL of modified tryptic soy broth (mTSB; Oxoid, Basingstoke, UK) with casamino acids (10 g/L of mTSB; VWR, Solon, OH). Residual meat samples were vacuum sealed (Multivac A300/16, Sepp Haggemuller KG, Wolfertschwenden, Germany) and stored at −20°C indefinitely. Meat samples in mTSB were macerated for 2 min at 180 rpm (Stomacher 3500 series, Seward) and then incubated at 42°C for 18 to 24 h. Following incubation, ca. 500 mL of enriched sample was transferred into a filter bag (XX-C003, Microbiology International, Frederick, MD). Two 20-mL aliquots of each filtered enrichment broth were separately combined with 20 mL of brain heart infusion (BHI; Difco, BD, Sparks, MD) broth containing 20% glycerol (Fisher Bioreagents, Fair Lawn, NJ), mixed gently, and then stored at −20°C for further screening if needed. A 25-μL portion of the enrichment broth from each sample was then screened for STEC with the BAX real-time PCR assay (BAX system Q7, Hygiena Diagnostics, Wilmington, DE) according to the manufacturer's instructions. STEC strains O26:H11, O45:H2, O103:H2, O111:H, O121:H19, O145:NM, and O157:H7, representing each of the seven regulated O serogroups, were used as positive controls for enrichment and BAX assays (26). All control strains harbored genes stx1, stx2, and eae. Enriched samples testing positive by BAX for both stx and eae and for at least one of the six O serogroups of non-O157 STEC or E. coli O157:H7 were considered as presumptive positive for STEC via enrichment. Two aliquots (750 μL) of each presumptive-positive enrichment broth were separately combined with 750 μL of BHI broth containing 20% glycerol, mixed gently, and then stored at −80°C for further screening if needed.
Presumptive-positive enrichment samples were subjected to immunomagnetic separation. A 980-μL aliquot of the enriched sample was processed with 20 μL of serogroup-specific immunomagnetic beads (Abraxis, Warminster, PA) in a magnetic particle processor (Kingfisher Flex, Thermo Scientific, Waltham, MA). A 50-μL aliquot of the undiluted or diluted (1:10) eluate of the six O serogroups of non-O157 STEC or E. coli O157:H7 from the particle processor was separately plated onto modified Possé (mPossé) (39) or modified Rainbow (mRainbow) (45) agar, respectively, and the plates were incubated at 37°C for 18 to 24 h. Compared with the original Possé medium (39), mPossé contains reduced concentrations of potassium tellurite (decreased from 2.5 to 0.5 mg/L) and novobiocin (decreased from 8 to 5 mg/L) for enhanced detection of non-O157 STEC (28, 44). Up to five morphologically distinct colonies of a specific color as elaborated by Possé et al. (39) were subjected to serogroup-specific latex agglutination testing (E. coli non-O157 identification kit, Pro-Lab, Woodbridge, Ontario, Canada). If a positive or desired agglutination reaction was observed, the remaining portions of each of the same five colonies were separately transferred to BHI broth and incubated overnight at 37°C for subsequent molecular characterization.
Confirmation of STEC isolated from raw meat
Cultures were grown in BHI broth from each of the up to five colonies from a single meat sample testing presumptive positive by latex agglutination for the regulated STEC serogroups. A 100-μL aliquot of this culture then was subjected to a unique 11-plex PCR assay (3). This assay was designed to detect the O antigens of the seven regulated serogroups of STEC and the associated four virulence genes (stx1, stx2, eae [for intimin], and ehxA [for hemolysin]). The amplified PCR products were visualized with an imaging system (Gel Doc XR, Bio-Rad, Hercules, CA). For samples with the genotype targeted by the 11-plex PCR assay (i.e., stx1 and/or stx2, eae, and genes encoding for surface antigens for a regulated STEC serogroup), a loopful of cells from the corresponding tube of freshly grown cells in BHI broth was streaked for purity on an mPossé agar plate. A portion of a single mPossé agar colony with the expected color and morphology was again subjected to a serogroup-specific latex agglutination test, and when positive, the remaining portion of that specific single colony was transferred to 3 mL of BHI broth and incubated overnight at 37°C. Three 750-μL portions of each overnight culture were separately combined with 750 μL of BHI broth containing 20% glycerol, mixed gently, and then stored at −80°C for further characterization. On rare occasions, mPossé agar colonies had more than one color or morphology during the confirmation process. In these instances, each distinct colony type was again subjected to a latex agglutination test and the 11-plex PCR assay to identify and retain a pure culture of the desired genotype. As described elsewhere (26, 30), for quality control purposes food contact, nonfood contact, and equipment surfaces in our laboratory were selected at random for sponge sampling (typically up to five samples weekly when veal and beef samples were being analyzed) to screen for STEC. During the study, only 3 of 101 total control samples tested positive for the stx and eae genes by BAX; however, none of these three samples tested positive for any of the regulated serogroups of STEC (data not shown).
In total, 270 STEC (one to five isolates from each of the 64 positive meat samples) were subjected to pulsed-field gel electrophoresis (PFGE) using the restriction enzyme XbaI (50 U per sample; New England Biolabs, Boston, MA) essentially as described (12). Genotypic relatedness was delineated at 90% similarity, with the Dice correlation set at 1% position tolerance (BioNumerics software version 7.6, Applied Maths, Austin, TX).
All metadata were recorded in an Excel spreadsheet (Microsoft, Redmond, WA), and all categorical variables were coded in a systematic manner. Season was established based on the purchase date of the samples: spring (March through May), summer (June through August), fall (September through November), and winter (December through February). After merging of the microbiology data and metadata by means of a unique letter-number code for each sample, the distribution of sample by categorical variables was tabulated using frequency tables in STATA/MP 15.0 (Stata Corp, College Station, TX). The prevalence of samples positive for the regulated seven STEC within each categorical variable was compared utilizing the PROC FREQ with CHISQ, Fisher's exact test, and/or Cochran-Mantel-Haenszel options in SAS (SAS Institute, Cary, NC). When significant differences were found, P values were adjusted with the Bonferroni method in PROC MULTTEST to examine significant pairwise comparisons within a variable (P < 0.05).
RESULTS AND DISCUSSION
Distribution of meat samples collected
A total of 1,577 samples (57 unique brands) were purchased during 81 shopping trips made on 78 noncontiguous weeks from May 2014 to April 2017. Of the 57 brands, 21 were veal cutlets, 19 were ground veal, and 46 were ground beef. Some brands produced all three types of products tested. A total of 136 stores in 85 zip codes across five states were visited. Of these 136 stores, 122 were large chain stores and 14 were specialty stores or meat markets. Eighteen of the stores were visited only once (four stores in North Carolina, two in Maryland, five in Pennsylvania, two in Delaware, and five in Virginia), and the remaining 118 stores were visited more than once (mean, 19 visits per store; range, 2 to 49 visits). For each sample, the state, season, brand, location where meat was ground or packed, fat content, intact versus nonintact status, and inclusion or not of safe handling and/or cooking instructions on the product label (Table 1) were recorded for possible correlation with pathogen source, persistence, virulence potential, and diversity. During phase I (May 2014 to June 2015), 258 veal cutlet, 363 ground veal, and 196 ground beef samples were collected, and in phase II (July 2016 to April 2017) 224 veal cutlet, 192 ground veal, and 344 ground beef samples were collected. With the possible exception of North Carolina where we shopped intermittently, meat samples were usually purchased from multiple stores within one of the four other states on a weekly basis, and stores in each of the states were visited about once per month. The numbers and types of samples purchased each week were subject to product availability. For veal cutlet and ground veal samples, 33.8% of the 482 samples were collected in spring and 35.3% of the 555 samples were collected in summer, whereas 59.3% were collected in fall and 71.5% were collected in winter. For ground beef samples, distribution based on season was as follows: 38.1% were collected in spring, 19.3% in summer, 23.3% in fall, and 19.3% in winter.
In addition to season, samples were also categorized by the location where the meat was packed or ground; 23.7% of veal cutlet samples (114 of 482), 18.7% of ground veal samples (104 of 555), and 33.5% of ground beef samples (181 of 540) were packed or ground at retail stores, indicating that most meat samples were packed directly by the processor. Veal cutlet samples also were categorized as intact (97.9%, 472 of 482 samples) or as nonintact (2.1%, 10 of 482 samples). Viable STEC were not recovered from the 10 nonintact cutlets tested (data not shown). The percent fat in each sample as stated on the product label was also recorded; 60.5% of ground veal samples contained 20 to 31% fat, and 54.4% of ground beef samples contained 8 to 20% fat. Regardless of meat species or sample type, 97.1% (1,074 of 1,106) of meat products packed or ground by a processor and 79% (283 of 399) of those ground or packed by a retailer included a safe handling instruction on the label. Overall, for meat packed by the processor, no significant differences (P > 0.05) were observed for the presence of safe handling instructions on the labels of ground veal (99.3%, 414 of 417 packages) and veal cutlet (92.8%, 333 of 359 packages) compared with the presence of these instructions on the labels of ground beef (99.1%, 327 of 330 packages). In contrast, for samples packed or ground at retail stores, the inclusion of safe handling instructions on the labels of ground veal (46.5%, 48 of 104 packages) and veal cutlets (53.5%, 61 of 114 packages) was significantly lower (P < 0.05) than that on packages of ground beef (96.1%, 174 of 181 packages). The observation that store-packaged veal samples were less likely to have safe handling instructions affixed to the label than were veal packages directly from the manufacturer suggests that additional resources should be directed to provide interventions for food retailers and to better educate store employees about potential food safety issues related to the absence of label instructions. Of the 1,505 (95.4%) of 1,577 meat samples tested for which label information was available, 68.5% (1,031 samples) had both safe handling instructions and cooking instructions on the package label. As observed with the safe handling instructions, the number of samples that contained cooking instructions on the package label was significantly lower (P <0.05) among samples packed or ground by retailers (146 of 399, 36.6%) than among samples packed by the manufacturer (851 of 1,106, 76.9%). Cooking instructions were included on the product label for 24 (21.5%) of 114 veal cutlet samples, 23 (22.1%) of 104 ground veal samples, and 99 (54.7%) of 181 ground beef samples packed or ground at retail. However, for samples packed by the manufacturer, 196 (54.6%) of 359 veal cutlet, 387 (92.8%) of 417 ground veal, and 268 (81.2%) of 330 ground beef samples included cooking instructions on the product label. Only 7 (1.5%) of 482 veal cutlet, 263 (47.4%) of 555 ground veal, and 285 (52.8%) of 540 ground beef samples displayed detailed cooking information on the package label, such as “For food safety, cook thoroughly to an internal temperature of 160°F as measured by a food thermometer,” “Cooking instructions: bake, grill, or pan fry until brown or internal temperature of 160°F,” or “Cooking this beef to 160°F helps reduce harmful bacteria which could cause serious or fatal illness. Put cooked beef on a clean plate. Use a meat thermometer.” In general terms, label instructions recommended that consumers cook the meat to an end-point temperature of 145, 160, and 165°F (63, 71, and 74°C) for veal cutlets, ground beef, and ground veal, respectively.
Recovery of STEC from raw meat
Overall, the non-O157 STEC serogroups were recovered more frequently (P < 0.05) from ground veal samples (39 of 555, 7.0%) and veal cutlet samples (19 of 482, 3.9%) than from ground beef samples (5 of 540, 0.9%) (Table 2). No significant differences (P > 0.05) in the proportion of positive samples were observed between phase I and phase II samples of ground beef, ground veal, or veal cutlets. For all meat types and species, the prevalence of E. coli O157:H7 was lower than that of non-O157 STEC; a single E. coli O157:H7 isolate was recovered from ground veal (1 of 555, 0.18%), whereas this pathogen was not recovered from any veal cutlet or ground beef samples. Thus, 5 (0.9%) of 540 ground beef samples, 19 (3.9%) of 482 veal cutlet samples, and 39 (7.0%) of 555 ground veal samples harbored viable STEC with the requisite surface antigens and virulence genes indicative of one of the regulated non-O157 serogroups. The single ground veal sample that tested positive for O157:H7 STEC also tested negative for the six non-O157 STEC serogroups.
Among the 57 brands that were analyzed, 56.7% (894 samples) of the 1,577 total samples tested were from the following 5 brands: brand A (14.8%, 234 samples), brand AA (7.9%, 124 samples), brand C (17.6%, 277 samples), brand D (8.2%, 129 samples), and brand H (8.2%, 130 samples). Brands AA, D, and H produced ground veal, ground beef, and veal cutlets, whereas brands A and C produced only ground veal and veal cutlets. All 5 brands were among the 14 brands of the 64 samples from which STEC were recovered: brand A accounted for 57.8% (37 samples) of these positive samples, brand D for 7.8% (5 samples), brand H for 7.8% (5 samples), brand C for 6.25% (4 samples), brand GG for 6.25% (4 samples), and brands AA, B, BBB, E, NN, PP, RR, S, and XX for 1.6% each (1 sample each). The 52 (81.3% of 64 positive samples) samples of brands A, AA, B, BBB, C, E, H, NN, and XX were prepacked at national or local supermarket stores, and all 52 samples had safe handling instructions and/or cooking instructions or temperatures on the product label, whereas the 12 (18.8% 64 positive samples) samples of brands D, GG, PP, RR, and S were ground or packed by the retailer. Except for brand S (1 sample) that had safe handling instructions but not cooking instructions or temperatures on the label, brands D, GG, PP, and RR (11 samples) did not have safe handling instructions and/or cooking instructions or temperatures on the label.
Of the five STEC-positive samples from brand D, three samples were collected at the same store on three separate visits between August and October 2016, and two samples were collected at the same store or chain but from different locations in Maryland (December 2016) and Virginia (February 2017). All four samples from brand GG were collected at the same small specialty retailer on four visits between August 2014 and February 2015. Multiple isolates (one to five) from each of 64 samples testing positive for at least one of the seven regulated serogroups of STEC were retained for further characterization via PFGE.
The prevalence of the stx, eae, and stx plus eae genes in ground veal samples was 40.0, 35.7, and 28.3%, respectively, which was significantly higher than the prevalence of these genes in ground beef or veal cutlet samples (P < 0.05) (Table 2). However, 7.0% of ground veal samples yielded a confirmed isolate of non-O157 STEC that harbored both the stx and eae genes and one of the targeted six somatic O antigens. As reported elsewhere (5, 7, 15, 32), samples positive for stx and/or eae by PCR assay do not necessarily harbor surface antigens for the regulated non-O157 STEC. Bacteria other than STEC, such as Shigella sonnei and Shigella boydii, also can test positive via PCR for stx or eae (24, 38). Recovery of the seven regulated STEC serogroups from food samples may be difficult due to the presence of substantial levels and types of background organisms, the physiological status (e.g., injured cells) of cells of the target serogroup of STEC, and/or the limitations of the recovery media and methodology. Although the methodology utilized in this study was approved by the USDA FSIS (52) and validated by other investigators (22), both the media and methodology used are subject to many of the same inherent limitations as other PCR-based detection methods, notably amplification of target DNA from inactive cells. Recovery of STEC is further complicated by the lack of any unique phenotypic differences among non-O157 STEC, such as the absence of β-glucuronidase activity and the inability to ferment sorbitol as is characteristic for cells of serotype O157:H7 STEC (25, 37). Despite these limitations, 5.4% of ground beef samples tested presumptive positive by the BAX PCR assay for non-O157 STEC, and 0.9% harbored a viable isolate of the big six STEC (Table 2). Wasilenko et al. (56) reported that 20 (6.5%) of 308 retail ground beef samples tested positive for stx and eae via the BAX PCR assay; however, only 2 of these 20 samples yielded a viable isolate that carried both stx and eae and had the somatic O antigen of a regulated serogroup, in this case both were STEC O103. Liao et al. (29) reported that 9 (0.8%) of 1,192 retail ground beef samples were presumptive STEC positive by the BAX assay and harbored stx, eae, and one or more genes associated with one of the six regulated serogroups of non-O157 STEC; however, no viable isolates of a target genotype were recovered. Differences in the prevalence of non-O157 STEC among these studies may result from (i) use of different methodologies and sample sizes (325 versus 25 g), (ii) type of enrichment broths (e.g., mTSB vs TSB), (iii) concentration of the selective or differential antibiotics (e.g., novobiocin), (iv) use of different suppliers of immunomagnetic beads, and/or (v) use of different selective agars (e.g., mRainbow versus mPossé agar) (28). Recovery rates for STEC from food can also be affected by the presence of a selective agent (e.g., novobiocin) in the enrichment broth, the type and concentration of a selective agent in the plating medium (e.g., potassium tellurite), and the PCR primer sequences and DNA extraction methods (27, 54, 55).
Several investigators reported recovery of STEC from retail fresh ground meats and poultry (4, 19, 40, 41), thus confirming that foods of animal origin are a significant source of this pathogen and its associated illnesses. Doyle and Schoeni (19) recovered STEC O157:H7 from retail ground beef (6 of 164 samples, 3.7%), pork (4 of 264 samples, 1.5%), poultry (4 of 264 samples, 1.5%), and lamb (4 of 205 samples, 2.0%). Barlow et al. (4) reported recovery of viable STEC from 16 and 40% of retail ground beef (285 samples) and lamb cutlets (275 samples), respectively. Samadpour et al. (41) reported the presence of E. coli isolates positive for Shiga-like toxin genes 1 and 2 from among 23% (14 of 60) of ground beef samples, 18% (9 of 51) of pork samples, 63% (5 of 8) of veal samples, and 48% (10 of 21) of lamb samples procured at retail. In addition to recovery of STEC from raw ground veal and beef in the present study, we also occasionally tested ground lamb collected at food retailers. Of 117 lamb samples tested, 2 samples yielded viable STEC with surface antigens for O103 and harboring the stx1, eae, and ehxA genes (data not shown). Additional studies are planned to assess the prevalence, levels, and types of regulated STEC in raw lamb and other less common species of meat and poultry. Regarding the prevalence of STEC in beef products at retail, although Samadpour et al. (40, 41) reported that ca. 16% of retail ground beef samples collected from food retailers in Washington state were positive for stx, no attempt was made to recover viable STEC isolates from the stx-positive samples. The prevalence of stx in ground beef samples reported by Samadpour et al. (40, 41) was similar (16.8%) to the results in the present study (18.7%) (Table 2). A somewhat higher prevalence (1,006 of 4,133 samples, 24.3%) of stx-positive samples was reported for commercial finished ground beef samples from 18 ground beef producers across the United States from 2005 to 2007 (8). At least one STEC isolate was recovered from 300 (7.3%) of the 4,133 samples, and 10 of these 300 samples harbored pathogenic STEC, including three of the seven regulated serogroups of STEC (O26, O103, and O145). In another study, the recovery rate of non-O157 STEC from 249 retail ground beef samples collected between 2009 and 2010 in the vicinity of Washington, DC, was 5.2%; none of the 13 samples harbored cells of any of the seven regulated serogroups of STEC (25).
Unlike for ground beef, to the best of our knowledge, little if any data on the recovery of STEC in retail ground veal have been published. As part of the BVCBS by the USDA FSIS from 2014 to 2015, the prevalence of E. coli O157:H7 and non-O157 STEC on beef carcasses (n =1,368) was estimated at 1.83 and 6.14%, respectively, after hide removal and 0.66 and 0.73%, respectively, before chilling (51). In this same report, the prevalence of E. coli O157:H7 and non-O157 STEC on veal carcasses (n = 274) was estimated as 0.73 and 23.72%, respectively, after hide removal and 0.73 and 9.85%, respectively, before chilling. USDA FSIS reported the prevalence of non-O157 STEC from raw ground beef and its components during 2012 to 2013 and 2016 to 2017: veal samples harbored STEC more frequently than did beef samples (46, 48). More specifically, the prevalence of non-O157 STEC in beef versus veal in 2016 and 2017 was 0.6% (20 of 3,192 samples) versus 4.5% (5 of 110 samples) and 0.4% (25 of 5,582) versus 1.8% (7 of 398), respectively. Bosilevac et al. (9) tested sponge samples of matched hide and preevisceration carcass samples from 803 animals and reported that 90.3% (725 of 803) of veal hide samples and 68.2% (548 of 803) of matched veal carcass samples tested positive for non-O157 STEC by PCR assay for stx, eae, and one or more targeted O-group genes. In that study, only carcass samples were further screened for viable STEC, and 56.8% (311 of 548) of the carcass samples harbored viable non-O157 STEC (9). With culture-based methods, the prevalence of E. coli O157:H7 on these same 803 matched hide and carcass veal samples was 20.3 and 6.7%, respectively (9). These findings complement the findings of the present study, that the prevalence of non-O157 STEC is higher in veal than in beef, and the prevalence of non-O157 STEC is higher than that of E. coli O157:H7 in both raw veal and raw beef. The reasons for the higher prevalence of non-O157 STEC in ground veal than in ground beef have not yet been fully elucidated. However, the observed differences in the recovery of the seven regulated serogroups of STEC in veal versus beef is more likely due to the physiological nature of the animal than any differences in processing of the harvested meat (9, 10, 34, 35). Mir et al. (34) reported on an association between STEC colonization and the diversity of gut microflora related to animal age. A more diverse microflora was found in older calves, which had less STEC colonization, and STEC shedding in cattle decreased with cattle age. Because the present study was conducted over a 2-year period in the mid-Atlantic region of the United States, seasonal and geographical conditions could have affected the recovery of STEC. Thus, the prevalence of non-O157 STEC by state, season, brand, fat level, or location where meat was ground or packed was compared statistically. The recovery of non-O157 STEC from raw veal and beef was significantly associated with the brand of meat and with fat content (P < 0.05) but was not significantly associated with state, season, or location where the meat was ground or packed (P > 0.05).
Distribution of serogroups
The number of STEC isolates recovered as delineated by their O-serogroup designation is shown in Table 3. Except for a single veal cutlet sample that yielded strains of two STEC serogroups (O111 and O103, one isolate each), each positive sample yielded cells displaying only one STEC serogroup. As detailed below, although multiple isolates from a positive sample were of one serogroup, such isolates on occasion had different restriction endonuclease digestion profiles (REDP). The most prevalent serogroup of isolates recovered from veal cutlets and ground veal was O103; 15 veal cutlet and 33 ground veal samples harbored this serogroup. Other investigators also reported that serogroup O103 isolates were predominant among the six regulated non-O157 STEC associated with raw meat. Bosilevac and Koohmaraie (8) conducted a survey to establish the prevalence and serogroup of non-O157 STEC in commercial ground beef in the United States and found a prevalence of 0.24% (10 of 4,133 samples) for non-O157 STEC. One isolate from each sample was subsequently delineated as serogroup O103 (four samples), O26 (two samples), O145 (one sample), O117 (one sample), and an untypeable O-group STEC (two samples) (8). According to the USDA FSIS results of the 2013 sampling of RGBC for the six regulated non-O157 STEC, of the 14 trim samples (4,511 total samples) positive for non-O157 STEC, 10 samples were positive for serogroup O103 (46). Of the eight trim samples (3,173 total samples) testing positive for non-O157 STEC as part of the 2018 RGBC testing program, five samples were positive for serogroup O103 STEC (46, 50). Serogroup O103 also was the most prevalent serogroup of the six regulated non-O157 STEC among isolates recovered from fecal samples of feedlot cattle in the United States (17, 18, 20). In the present study, strains with serogroup O45, O121, and O157 surface antigens of STEC were not recovered from veal cutlet samples, and serogroup O111 and O121 strains were not recovered from ground veal samples. From the 5 of 540 ground beef samples that tested positive, isolates with O26 surface antigens were recovered from four samples and isolates with O103 surface antigens were recovered from one sample.
Genetic diversity and virulence potential
The 270 STEC isolates recovered from the 64 confirmed positive samples of raw veal or beef were subtyped by PFGE (Fig. 2). For the multiple isolates (up to five per positive sample) retained from the majority (43 of 64 samples, 67.2%) of samples testing positive for STEC, all such isolates had indistinguishable REDP (aka pulsotypes). Isolates from the remaining 21 samples testing positive for STEC had two to four REDP per sample; however, isolates with different REDP from the same sample were 50 to 80% related or similar. When two isolates from the same sample shared the same fingerprint, such isolates were considered as one unique strain. When strain A from food sample A and strain B from food sample B were indistinguishable by PFGE because they shared the same fingerprint, such strains were grouped within a single or unique REDP; the metadata for these strains were separately compared with subsequently assess their potential epidemiological association. Using these criteria, the 270 total isolates retained from the 64 STEC positive samples were categorized by PFGE into 39 pulsogroups (≥90% similarity) and further delineated as 97 isolates displaying 78 distinct REDP.
The most prevalent REDP based on number of isolates with a specific pulsotype (not inclusive of the summation of multiple isolates from a given sample displaying an indistinguishable fingerprint) were REDP 19 (three isolates total from three samples), REDP 46 (three isolates total from three samples), and REDP 48 (three isolates total from three samples). Twenty of the 39 pulsogroups were composed of STEC isolates with 2 to 14 REDP per pulsogroup, whereas the remaining 19 pulsogroups were composed of STEC isolates with a single REDP. Among the 20 pulsogroups with multiple REDP, the largest pulsogroup by number of isolates was pulsogroup AA, This pulsogroup contained 18 isolates (ca. 18.6% of the 97 total isolates) that collectively had 14 REDP. These 18 isolates were recovered from among 14 positive samples (one or two isolates per sample) representing five brands that were purchased on single or multiple visits to 13 stores. The remaining 38 pulsogroups (79 of 97 isolates) contained one to eight isolates each that collectively displayed 64 REDP; these isolates were recovered from among 55 positive samples (one to four isolates per sample) representing 14 brands (inclusive of the abovementioned five brands) that were purchased on single or multiple visits to 31 stores. Of the 78 REDP, 62 were unique to a single sample (62 isolates), whereas 16 were represented by isolates recovered from two (26 isolates) or three (9 isolates) positive samples (35 isolates total). These 16 REDP were identified in ca. 48.4% (31) of the 64 samples testing positive. These 31samples were purchased in 22 (64.7%) of the 34 stores and belonged to 9 (64.3%) of 14 brands. Of these 14 brands that tested positive for STEC, 57.8% (37) of the 64 positive samples were from brand A. The 62 (63.9%) of 97 isolates recovered from brand A had 26 pulsogroups (66.7%) comprising 50 (64.1%) of 78 REDP. These samples were purchased at 18 stores either over multiple shopping visits (10 stores) or during a single shopping visit (8 stores). For the remaining 13 brands, 2 had four to five positive samples that were purchased in three to four stores, 1 had three positive samples purchased at the same store over three shopping visits and two samples purchased at the same store chain but at different locations, 1 had four positive samples purchased at the same store over four shopping visits, and 9 had only a single positive sample that was purchased at nine stores on a single visit. Further scrutiny of positive samples revealed that samples from brands D and GG that tested positive for STEC were collected at multiple locations and/or over multiple visits, and such samples were ground or packed by the retailer rather than the producer. For the five isolates recovered from brand D (from five samples), three isolates were recovered from three samples purchased at the same store on three visits between August and October 2016. Of these three isolates, two (one from a veal cutlet sample and one from a ground beef sample) had the same pulsogroup and REDP (pulsogroup K and REDP 19). The four isolates recovered from four brand GG samples (two ground veal, one veal cutlet, and one ground beef sample) purchased at a small specialty retailer between August 2014 and April 2015 had four pulsogroups (J, O, P, and DD) and six REDP (18, 21, 22, 24, 45, and 46). It would be of interest to compare the REDP found in this study with PFGE profiles of STEC from animal, environmental, abattoir, and food samples in other data repositories and/or peer-reviewed publications.
As expected, the nature and number of REDP varied among strains belonging to different serogroups. A total of 20 isolates of STEC O103 from 15 veal cutlet samples clustered within 10 pulsogroups (O, R, V, X, AA, BB, CC, DD, EE, and FF), and 57 isolates of STEC O103 from 33 ground veal samples were clustered within 23 pulsogroups (A, C, D, F, G, O, P, Q, T, U, V, W, X, Y, Z, AA, CC, DD, EE, GG, HH, II, and MM). Seven pulsogroups (O, V, X, AA, CC, DD, and EE) were found in isolates recovered from both veal cutlet and ground veal samples but not ground beef samples. Regardless of species or type of meat, with the exception of STEC O145 strains from two veal cutlet samples that harbored stx2 (pulsogroups JJ, KK, and LL), all non-O157 STEC isolates carried the stx1, eae, and ehxA genes. The single O157:H7 isolate (pulsogroup L, REDP 19) recovered in this study, which was isolated from ground veal, harbored the stx1 and stx2 genes and the eae and ehxA genes. All isolates confirmed as STEC in this study possessed the ehxA gene except for a single STEC O145 strain (pulsogroup LL) from ground veal. The ehxA gene, which encodes an enterohemolysin, is an accessory virulence gene that is usually carried by STEC that also carry eae (for intimin), saa (for STEC auto-agglutinating adhesion), and subAB (for production of subtilase cytotoxin) (33). The stx gene alone is insufficient for STEC to cause human illness, and clinical isolates usually carry all four of these accessory virulence genes. Thus, in addition to genes encoding the seven serogroup-specific somatic O antigens, it is critical to identify and characterize genes carried by STEC that are potentially responsible for pathogenicity and that are linked to human illness to definitively state that a sample is STEC positive and that such cells are pathogenic for humans.
There are different approaches for screening meat samples for the presence of cells of non-O157 STEC. Bosilevac and Koohmaraie (8) initially screened meat samples for the presence of the stx1 and stx2 genes by PCR amplification, and samples were subsequently screened via a phenotypic assay to recover stx-containing non-O157 STEC isolates. Isolates positive for a regulated O-antigen serogroup were further characterized for the presence of other virulence factors, such as the eae gene. The USDA FSIS (45) considers a sample positive for non-O157 STEC when a viable isolate (i) tests positive for one of the six regulated serogroups (O26, O45, O103, O111, O121, and O145), (ii) tests positive for the stx and eae genes, and (iii) is confirmed biochemically as E. coli. Thus, in this study we used the USDA FSIS approved protocol and initially tested samples for the presence of the stx and eae genes via the BAX real-time PCR assay; only samples that harbored both genes were then further screened for the six non-O157 regulated serogroups or serotype O157:H7 (52). This may at least partially explain why all the non-O157 STEC isolates recovered in this study harbored both the eae and stx genes.
Although the present study was focused on the recovery of the seven regulated serogroups of STEC in retail veal and beef samples, E. coli isolates of the target serogroups without the stx and/or eae genes or without surface antigens for the regulated proven serogroups but harboring the stx and/or eae genes were also recovered (data not shown). The most common virulence profile of such isolates was the presence of the eae and ehxA genes but no stx genes; strains of E. coli serogroups O26 and O145 had this virulence profile. Although such isolates did not possess an stx gene, enteropathogenic E. coli strains carrying eae but not stx1 or stx2 are also considered pathogenic to humans (16). The prevalence of E. coli without an stx gene in retail veal and beef products should not be dismissed as innocuous. Ferdous et al. (21) emphasized the public health importance of stx-negative STEC due to its virulence potential. Shiga toxin genes are encoded on lysogenic bacteriophage, and sometimes E. coli lose the bacteriophage encoding for stx during host infection and/or as a consequence of repeated laboratory manipulation and passage (21). Thus, all E. coli isolates recovered in this study will be subsequently characterized by whole genome sequencing to gain insight into their virulence potential, epidemiological relevance, and potential impact on human health.
In summary, 7.03% of ground veal (39 of 555 samples), 3.94% of veal cutlets (19 of 482 samples), and 0.93% of ground beef (5 of 540 samples) harbored cells of at least one of the regulated non-O157 STEC. A serotype O157:H7 STEC strain was recovered from only 1 (0.18%) of 555 raw ground veal samples. The prevalence of STEC in ground veal (39 of 555 samples, 7%) was significantly higher than that in ground beef (5 of 540 samples, 0.9%). Regarding serotyping of the isolates recovered from samples testing positive for STEC, similar to previous reports most isolates from veal were serotype O103, whereas most isolates from beef were serotype O26. Based on the 78 REDP among the 97 unique isolates of the 270 total isolates recovered and analyzed, molecular subtyping via PFGE revealed this collection of STEC strains to be diverse, thus suggesting the existence of multiple opportunities for STEC contamination of meat in the veal supply chain. Although the sample size might be somewhat limited for rigorous statistical analysis, these data may be used to estimate the comparative prevalence of the seven regulated STEC serogroups among retail veal and beef samples. These data also will be useful for supporting risk assessments, informing policy decisions, and developing interventions to reduce the risk of STEC associated with veal and beef products that may be undercooked and/or improperly handled or stored.
The authors extend their appreciation to David Renter, Jianfa Bai, Tiruvoor G. Nagaraja, and Natalia Cernicchiaro (Kansas State University, Manhattan) and Mykeshia McNorton (Washington, DC) for contributing their time, talents, resources, and/or expertise to this study. Our gratitude is also extended to Joy Mudoh, Breann Hrechka, Sylvia Ossai, and Arquette Grant for technical assistance and to Meshack Mudoh for collection of some samples (University of Maryland Eastern Shore, Princess Anne). This project was supported in part by Agriculture and Food Research Initiative Competitive Grant 2012-68003-30155 from the USDA National Institute of Food and Agriculture. Mention of trade names or commercial products is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the USDA.