E. coli O157: Pre-harvest opportunities & Challenges

E. coli O157: Pre-harvest opportunities & Challenges

Emerging Options to Control E. coli O157:H7 Pre-Harvest Guy H. Loneragan Epidemiologist West Texas A&M University Canyon, Texas, USA SteakExpert June 24-25, 2009, Angers, France Background Born and raised in Mudgee, New South Wales, Australia Parents have a 500 hectare property Cattle, sheep, crops, ~15 hectares shiraz/cabernet Veterinary training at the University of Sydney, Australia Degree awarded 1994 Epidemiology training at Colorado State University, Fort Collins, Colorado, US

MS and PhD awarded 1998 and 2001, respectively West Texas A&M University, Canyon, Texas, USA Focus on food safety, particularly prior to harvest E. coli O157, Salmonella, antimicrobial drug resistance Comments Prior to Presentation E. coli O157:H7 and non-O157 VTEC are not geographically constrained Found worldwide E. coli O157:H7 is the primary cause of HUS in North America, many EU member states, South American countries, Australia, etc. Focus of my research has been on the North American production system Mostly confined feeding As such, most of the information I present is related to confined cattle feeding operations (i.e., feedlots, dairies)

Some of the options may provide areas for further discussion and opportunities for France and rest of EU Discussion E. coli O157:H7 and non-O157 VTEC should be thought of as commensal organisms of cattle In other words, they are (likely) beneficial bacteria of cattle and potentially other ruminants In many instances, we encourage commensals Historically, we have not had to ponder too much about how to eliminate/reduce commensal bacteria Over time, we have developed tools to identify and control pathogens For example, Salmonella Typhimurium, M. bovis, Brucella Initial pre-harvest efforts to control E. coli O157:H7 was to approach the problem as if it were a pathogen Isolation, containment, prevent exposure

Not very effective Chronology From a North American perspective, our initial flawed approach resulted in effective pre-harvest interventions arising relatively late Most initial control at abattoirs Highly effective but still failures occurred Increase in illnesses and recalls because of contamination in 2007 Finally, we began to understand the ecology of VTEC O157 and understood that it was not a pathogen for cattle Consequently, interventions were developed that exploit some tool or characteristic of the bacterium Very specific (vaccine) or broad sweeping approaches (chlorate) Chronology There now exists a large collection of scientific literature

that provides evidence that it is possible to control E. coli O157 in cattle Use of appropriate interventions given the production system can reduce the burden of E. coli O157 leaving farms and entering abattoirs As an epidemiologist, I am most interested in field studies Interventions A variety of interventions have been developed/evaluated: 1 Biological DFM Bacteriophage

Chemical Neomycin Sodium chlorate Immunomodulation Vaccine 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0

0.001 0.101 0.201 0.301 0.401 0.501 0.601 0.701 0.801 to 0.1 to 0.2 to 0.3 to 0.4 to 0.5 to 0.6 to 0.7 to 0.8 and more Discussion predicated that I am most interested in field studies of effect Challenges models do not take into account important determinants of shedding (namely cohort level) L. acidophilus-based probiotics Continuously fed to cattle Only suitable to confinement operations In the USA, achieved FDA GRAS status BUT no FDA-approved label claim for control of E. coli O157 Most widely evaluated product is that of Nutrition

Physiology Corporation BOVAMINE, includes LAB NP51 (NPC 747) 9 peer-reviewed manuscripts demonstrating significant reduction using L. acidophilus NP51 Articles from 4 groups of researchers Systematic review, Sargeant et al, ZPH 2007;54:260 Standardized approach Evidence for efficacy of LAB NP51 Meta-analysis Meta-analysis is a method to combined data from different studies Uncertainty associated with individual studies Meta-analysis combined data, reduces uncertainty, provides best-estimate of effect Powerful analytical approach It is not, however, perfect and can be affected by statistical biases

Performed meta-analysis of NP51 Both published and unpublished data Summary of LAB NP51 U1 U2 U3 A1 A2 A3 A4 N1 J1 J2 C1 T1 R1 Combined .1 .5 .75 1 Fecal: Relative Risk

2 4 Summary of LAB NP51 A1 A2 A3 A4 N1 J1 C1 T1 R1 Combined .1 .5 Hide: Relative Risk

.75 1 2 4 Feedlot studies of NP51 Meta-analysis; average reduction: 40% reduction of prevalence in feces (26.5 versus 12.7%) 47% reduction of prevalence on hides (20.4 versus 11.3%) Some evidence for an effect on concentration Fecal load in positive animals reduced by greater 99.5% NP51 was associated with reduced number of positive animals and reduced number of bacteria in animals that remained positive LABm and Control of E. coli O157:H7

in Ground Beef Another Probiotic: PROBIOS FS Enterococcus, Chr. Hansen Fed for last 14 days on feed Average reduction: 56% feces Hides not evaluated Has FDA GRAS status in USA Limited use Limited availability??? Direct-Fed Microbials Easy adoption within confined feeding operations Efficacy in a variety of feeding systems

Dairies, feedlots, etc. While not dramatic in all studies, evidence of improvements in animal performance No easy implementation for pastured cattle No USA FDA label claim for E. coli O157 control Regulatory agencies in the USA will never endorse Better positioned if had an approved label claim for control/reduction of E. coli O157 Immunomodulation Is it possible to trick an animal into developing immunity to a commensal? In the past, I believed this to be impossible Three products have been evaluated in commercial settings Vaccine: 8 peer-reviewed manuscripts 7 showed significant reduction in shedding/colonization Data generated from 3 groups of researchers

Bioniche Product Ontario, Canada Vaccination against Type III secreted proteins Tool the bacteria use to colonize the gut Field work really began with study published by Potter et al., 2004 Bioniche Product Ontario, Canada Bioniche Product Ontario, Canada Available in Canada as Econiche Soon to be Conditionally licensed in the USA Cannot use trade name In studies in USA, it has resulted in

significant reductions in burden of E. coli O157:H7 In feces At TRM (RAMS, RAJ) On ROPES (an indication of animal to contact-surface transfer) On hides (both at the feedlot and at the packing plant) Canadian study (not Bioniche product) found no effect Van Donkersgoed et al., CVJ 2005 Likely manufacturing issues rather than an efficacy issue Epitopix Product Minnesota, USA E. coli O157 Targets the mechanism by which some

bacteria acquire iron from environment E. coli O157, Salmonella Iron largely acquired via siderophore receptor & porin proteins (SRP) Passive, gradient dependent At low concentrations, energy dependent SRP Technology Research suggests it is possible to restrict iron acquisition via immunity against cell-surface SRP proteins Competitive disadvantage Vaccine consists of purified SRPs as antigens (Epitopix, 2004) Technology developed

for Salmonella in turkeys, then cattle, and now E. coli O157 E. coli O157 Epitopix Product Minnesota, USA Initial field study suffered from low prevalence so hard to make statistical comparisons Significant reductions on hides or cattle positive at any site Subsequent study changed from 2- Feces 39% RAJ 48% dose to 3-dose regimen Hides 70% Epitopix Product Minnesota, USA P=0.28

P=0.13 P<0.01 Epitopix Product Minnesota, USA P=0.28 P=0.13 P=0.01 98% reduction in MPN P<0.01 Epitopix Product Minnesota, USA Available in USA Conditionally licensed

In studies in USA, it has resulted in significant reductions in burden of E. coli O157:H7 In feces On hides In concentration of E. coli O157 Ongoing research in production system in USA Cows > suckling calves > feedlot Opportunity for further research in pastured cattle outside of North America Ft. Dodge Evaluated once (and published) in commercial settings FPT 2006;26:393-400 Only 3 pens per exposure and unusual analysis Unclear what to conclude except appeared beneficial Feces 68% Hides 50% Other vaccines UNM technology

Licensed to a Colorado company Immunomodulation Strong evidence that it is possible to trick an animal into developing immunity to a commensal Bioniches Econiche licensed in Canada Pending conditional license in the USA Epitopix product conditionally licensed in the USA Dose response 3 doses appears to provide greatest response 2 doses may be more effective when whole herds vaccinated Easily administered to cattle in almost all production systems More doses associated with more handling of animals Appears to be slightly more efficacious than LAB Sodium Chlorate

Generated a substantial amount of interest in North America Suitable for confined animal feeding operations Acts as a suicide substrate for bacteria that respire using nitrate reductase Chlorate >>> Chlorite E. coli O157, Salmonella Very effective in challenge studies No slaughter authorization from USAs FDA so not field studies to report Currently within the regulatory process requesting label claim approval Sodium Chlorate Very little field work with E. coli O157

Challenges of making inferences from these challenge models Needs field studies Callaway et al., JAS 2002 Bacteriophage Warrant consideration Potential application In feed On hide On meat surface Example is Gangagens or Ivy Natural Solutions products Applied data of in-field effects is limited Experimental evidence provides some promise Callaway et al., FPD 2008 challenge study, sheep

Sheng et al., AEM 2006 challenge study, cattle Expectations of Interventions? Smith DR, et al. 2001. J Food Prot 64 (12) 1899-1903 None will provide 100% control May not need 100% USA Feedlot Study 73 feedlots visited twice Summer 15.8% positive Winter 5.6% positive Dave Smith, JFP 2001;64:1899 Summer 30% positive Winter 6.1% positive Dr. Smiths project was supported by the National Research Initiative of the USDA Cooperative State Research, Education and Extension Service, grant number #0002501.

Expectations of Interventions? E. coli O157:H7 Summer season prevalence Summer: EVERY EVERY pen pen positive positive but but Summer: highly variable variable prevalence prevalence (1-80%) (1-80%) highly 1 Frequency proportion 0.9 0.8 Very few

few pens pens near near the the mean. mean. Very Mostly far far greater greater or or far far lower lower Mostly prevalence than than the the mean mean prevalence 0.7 0.6 0.5

0.4 Doesnt follow follow expected expected Doesnt distribution distribution 0.3 0.2 0.1 0 0 0.001 0.101 0.201 0.301 0.401 0.501 0.601 0.701 0.801 to 0.1 to 0.2 to 0.3 to 0.4 to 0.5 to 0.6 to 0.7 to 0.8 and more Prevalence Smith et al. J Food Prot.

2001, 64 (12) 1899-1903 Observed Expected Model to Compare Prevalence of Cattle During Summer, Winter, and with intervention (Summer) Data-driven Simulation. Source, Dave Smith, UNL Stochastic simulation model of the prevalence of E. coli O157:H7 in live cattle with vaccination (@risk 4.5) Distribution of prevalence Mean, 5th and 95th Percentile 5,000 pen simulations 1 (500,000+ cattle) Intervention: Summer (vaccine) No intervention: Summer No intervention: Winter Pert(0.5, 0.65, 0.8) 7

0.6 0.4 6 Process Control Pert(50, 65, 80) Prevalence Intervention efficacy 65% 0.8 5 4 0.2

3 2 1 0 5.0% 90.0% 0.5568 5.0% 0.7432 0.850 0.805 0.760

0.715 0.670 0.625 0.580 0.535 0.490 0.445 0 Summer Summer Vaccination Winter Summer Winter No interv. Interv. No interv.

Model to Compare Prevalence of Cattle During Summer, Winter, and with intervention (Summer) Data-driven Simulation. Source, Dave Smith, UNL Summer, no intervention 1 0.9 Frequency proportion 0.8 Lower risk risk Lower 0.7 Higher risk risk Higher

0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.001 to 0.101 to 0.201 to 0.301 to 0.401 to 0.501 to 0.601 to 0.701 to 0.1 0.200 0.300 0.400 0.500 0.600 0.700 0.800 0.801 and more

Prevalence Winter, no intervention 1 0.9 Frequency proportion 0.8 Lower risk risk Lower 0.7 Higher risk risk Higher

0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.001 to 0.101 to 0.201 to 0.301 to 0.401 to 0.501 to 0.601 to 0.701 to 0.801 0.1 0.200 0.300 0.400 0.500 0.600 0.700 0.800 and more Prevalence Model to Compare Prevalence of Cattle During Summer, Winter, and with intervention (Summer)

Data-driven Simulation. Source, Dave Smith, UNL Summer, no intervention 1 0.9 Frequency proportion 0.8 Lower risk risk Lower 0.7 Higher risk risk Higher Predicted intervention distribution (summer) 0.6

1 0.5 0.4 0.9 0.3 0.2 0.8 Lower risk risk Lower 0 0 0.001 to 0.101 to 0.201 to 0.301 to 0.401 to 0.501 to 0.601 to 0.701 to 0.1

0.200 0.300 0.400 0.500 0.600 0.700 0.800 0.801 and more Prevalence Winter, no intervention 1 0.9 Frequency proportion 0.8

Lower risk risk Lower 0.7 Higher risk risk Higher 0.6 Frequency proportion 0.1 0.7 0.6 0.5 0.4 0.3

0.2 0.1 0 0 0.5 0.4 Higher risk risk Higher 0.001 to 0.101 to 0.201 to 0.301 to 0.401 to 0.501 to 0.601 to 0.701 to 0.801 0.1 0.200 0.300 0.400 0.500 0.600 0.700 0.800 and more Prevalence

0.3 0.2 0.1 0 0 0.001 to 0.101 to 0.201 to 0.301 to 0.401 to 0.501 to 0.601 to 0.701 to 0.801 0.1 0.200 0.300 0.400 0.500 0.600 0.700 0.800 and more Prevalence Expectations of Interventions? No intervention is 100% Need not be

Effective turns peak-season shedding patterns into low-season shedding patterns 30 to 40% of cases in North America attributed to beef In low season, observe: Less frequently contaminated beef Fewer recalls Less product diverted to cooking Fewer human illness Expectations of Interventions? Seasonal Occurrence of Human Illnesses Repeatable overrepresentation of cases during May-October 70% of reported cases in 6 months of the year In winter, prevalence in cattle, percentage ground beef samples positive, and human cases are all substantially lower compared to summer Target winter-time burdens. Supported by data.

Expectations of Interventions? In winter, prevalence in cattle, percentage ground beef samples positive, and human cases are all substantially lower compared to summer Target winter-time burdens. Supported by data. Expectations of Interventions? If not 100% effective, then what is its purpose What do we want it to do? What can we expect it to do? Issues: Purpose of pre-harvest interventions Different measures of prevalence in different studies (e.g., feces versus hides versus RAMs) Cross-contamination and interventions What can pre-harvest do for us Expectations of Interventions? Cattle Operations Ca ttle

Packing Plant HACCP/PR Various Interventions Ground Beef Trim for offsite grinding Not a fail-safe system In-coming load can overwhelm the system Expectations of Interventions? Cattle Operations Ca ttle Packing Plant HACCP/PR Various Interventions Ground Beef Trim for offsite grinding Not a fail-safe system

In-coming load can overwhelm the system Expectations of Interventions? Cattle Operations Ca ttle Packing Plant HACCP/PR Various Interventions Ground Beef Trim for offsite grinding Not a fail-safe system In-coming load can overwhelm the system Expectations of Interventions? Purpose not simply more is better To ensure burden of E. coli O157 on cattle presented for harvest is within manageable limits The in-plant series of HACCP/PR interventions effectively mitigate the burden on incoming cattle

Efficacy need not be (or even approach) 100% Desired efficacy depends on the burden within groups of cattle AND on the pathogen-mitigation capacity of the plant Linear of threshold? I suspect the latter. Expectations of Interventions? Product Efficacy: (RR-1)*100 67% 50% 30% Plant HACCP/PR Threshold Incoming burden without intervention Incoming burden with intervention

50% Expectations of Interventions? 3 e es its h t sf f o rio se h c na rpo a E c e pu s e th Product Efficacy: (RR-1)*100 67% 50%

30% Plant HACCP/PR Threshold Incoming burden without intervention Incoming burden with intervention 50% Summary: Immunomodulation as an Intervention for E. coli O157:H7 No intervention will be 100% effective E. coli coli O157 O157 E. Pre-harvest Harvest

Consumers Scientific evidence, simulation models, and empirical data indicate pre-harvest interventions are effective Ultimately should reduce consumer exposure to E. coli O157 Summary: Immunomodulation as an Intervention for E. coli O157:H7 No intervention will be 100% effective E. coli coli O157 O157 E. Pre-harvest Harvest

Consumers Scientific evidence, simulation models, and empirical data indicate pre-harvest interventions are effective Ultimately should reduce consumer exposure to E. coli O157 Summary: Immunomodulation as an Intervention for E. coli O157:H7 No intervention will be 100% effective E. coli coli O157 O157 E. Pre-harvest Harvest Consumers

Scientific evidence, simulation models, and empirical data indicate pre-harvest interventions are effective and fit the purpose Should reduce consumer exposure to E. coli O157 Summary: Immunomodulation as an Intervention for E. coli O157:H7 No intervention will be 100% effective E. coli coli O157 O157 E. Pre-harvest Harvest Consumers Scientific evidence, simulation models, and empirical data indicate pre-harvest interventions

are effective and fit the purpose Should reduce consumer exposure to E. coli O157 Summary: Immunomodulation as an Intervention for E. coli O157:H7 No intervention will be 100% effective E. coli coli O157 O157 E. Pre-harvest Harvest Consumers Scientific evidence, simulation models, and empirical data indicate pre-harvest interventions are effective and fit the purpose Should reduce consumer exposure to E. coli O157

Summary: Immunomodulation as an Intervention for E. coli O157:H7 No intervention will be 100% effective E. coli coli O157 O157 E. Pre-harvest Harvest Consumers Scientific evidence, simulation models, and empirical data indicate pre-harvest interventions are effective and fit the purpose Should reduce consumer exposure to E. coli O157 Issue: How Much is Enough? Provide cattle with burden below threshold Either by:

Reducing burden from some problematic level Preventing groups of cattle reaching problematic burden Threshold is uncertain But empirical data suggest warmer months associated with increased risk Shedding in cattle and human illnesses Summary Pre-harvest control of E. coli O157 (and potentially other non-O157 VTEC) is possible Not 100% but no need to be 100% Purpose is within a system of multiple controls Pre-harvest > harvest > packaging > consumers Effectively turn peak-season shedding patterns into low season shedding patterns Unclear what numerical contribution this has on risk of final product contamination Opportunities to evaluate products in pasture settings

Acknowledgements SteakExpert organizers for invitation Patrice Arbault for co-ordination Your hospitality You for listening to me in English rather than in French Colleagues for sharing data Contact Information: [email protected] +1 806 651-2287

Recently Viewed Presentations