Scientific Perspective on Listeria monocytogenes in Ready-to-Eat Food Régis Pouillot, DVM, PhD FDA / CFSAN / OAO / Division of Risk and Decision Analysis / Risk Analysis Branch Food Advisory Committee, Dec 7-8, 2015

Outline Epidemiology Listeria monocytogenes in RTE food (and elsewhere) Listeria dose-response models New Data

Epidemiology

Few cases, severe Pathogen Estimated annual number of illnesses in the U.S. (domestically acquired, foodborne) Estimated annual number of associated deaths Listeria monocytogenes 1,600 255 (16% of cases) Salmonella, non typhoidal 1.0 million 378 (.03% of cases) Norovirus 5.5 million 149 (.002% of cases) Scallan et al., Emerg Infect Dis 17(1): 7-12.

Most cases are sporadic Outbreak vs. Sporadic Cases 2004-2011: 1,084 cases reported to CDC via FoodNet 56 (5.2%) only associated to outbreaks Little idea about the implicated foods for sporadic cases Some case characteristics (age, gender, illness severity) comparable to outbreaks Comparable implicated foods? Few contaminated products will not lead to an outbreak Interagency Food Safety Analytics Collaboration webinar, Jan 10th, 2014

Outbreaks: shift in the implicated foods since 1998 Year Cases Food 1998 108 Frankfurters 2005 6 Unknown 2011 2 Undetermined 4 3 Grilled chicken Chive cheese and ackawi cheese 1999 13 Deli meat 147 Whole cantaloupe 12 Mexican-style cheese 5 2006 15 Aged, blue-veined cheese 11 Pâté Taco or nacho salad 2012 22 Ricotta cheese Cheese 2013 Soft ripened cheese 2000 2007 Milk 2014 8 30 2008 Tuna salad 2001 28 20 Sprouts 2002 54 35 Caramel apple 2003 2010 Hog head cheese 2015 10 Ice cream Sushi rolls Pre-cut celery CDC (2013). MMWR Morb Mortal Wkly Rep 62(22): 448-452. Cartwright et al., Emerg Infect Dis 19(1): 1-9. CDC website

Listeria Outbreaks and Incidence, 1983-2014 (per million pop) No. outbreaks Decrease: 24% during 1996-2003 and then reach a plateau Era Outbreaks per year Median cases per outbreak Pre-PulseNet 0.3 69 Early PulseNet 2.3 11 Listeria Initiative 2.9 5.5 WGS 8 4.5 Data are preliminary and subject to change Slide: CDC via M. Wiedmann

Overall: Stable number of cases since 2003 Laboratory confirmed cases in FoodNet CDC MMWR 2014 / 63(15);328-332

Conclusions: Epidemiology Few cases, severe, mostly sporadic Sharp decrease in the number of cases, then stable since 2003 Evolving knowledge, linked to evolving tools More, smaller recognized outbreaks Evolving food implicated in outbreaks Sporadic cases?

Listeria in Ready-to-Eat food (and elsewhere)

L. monocytogenes: ubiquitous bacteria Widely distributed in the environment Soil, water, vegetation 15% Lm in preharvest env. (NY state) Manufacturing environment persistent strains that have found a harborage site within the facility where it may reside for years… transient strains… Retail environment idem e.g. 241 retail establishments tested 60% positive (at least one positive sample) 6.3% food-contact surface samples were positive 17% of non food-contact surface samples were positive Milk crates: 34%! Floor drains: 28%! Environment Commodities Processing Retail Food Strawn et al, 2013 Appl Environ Microbiol. 79(24):7618-27 Hoelzer et al., J Food Prot 74(7): 1083-1095 Simmons et al. J Food Prot 77(11): 1929-1939

In Food: FSIS inspected facilities Percentage of RTE meat and poultry products testing positive for L. monocytogenes in FSIS- inspected facilities, compared with incidence of listeriosis per 100,000, from CDC FoodNet surveillance Interagency Risk Assessment: Listeria monocytogenes in Retail Delicatessens Technical Report (2013) Source data: FSIS and CDC

In Food: survey data (0.27%) (4.31%) (0.76%) (4.70%) (0.049%) (1.25%) FDA/ARS Survey* (2013) (0.27%) (4.31%) (0.76%) (4.70%) (0.049%) (1.25%) (1.04%) (2.36%) 2013 prevalence is significantly lower, p<0.0001 * Preliminary results from phase I Source: Food Advisory Committee Meeting Materials, September 29-30, 2014

No-growth food? Known limits Better knowledge of interactions pH less than or equal to 4.4 Water activity less than or equal to 0.92 Better knowledge of interactions Robust predictive microbiology models (most studied bacteria) Remaining issues Heterogeneity within product, lot-to-lot Validation of challenge tests? see Powell, 2009, IJFM 136 10-17 Augustin and Carlier, 2000, IJFM 56: 53-70 Mejlholm et al, 2010, IJFM 141: 137-150

Conclusions: L. monocytogenes in RTE food Ubiquitous bacteria widely distributed in the environment, manufacturing environment, retail environment Frequently found on non food contact surfaces Decrease in the prevalence in food over the last decade(s) FSIS inspected facilities Other food (decrease in the number of cases?) Growth-No growth boundary in practice?

Listeria dose-response

Lots of bacteria ca. 1.8% RTE food contaminated (Gombas et al., 2003) Smoked seafood: prevalence: 4.3%, up to 105-106 CFU/g Lots of bacteria Listeria monocytogenes Levels in Food (per serving) Number of Servings (per year in the United States) (per person per year) 3.3  1011 1300 1 to 1000 4,900,000,000 19 103 to 106 620,000,000 2.4 106 to 109 130,000,000 0.5 > 109 73,000,000 0.3 (FDA/FSIS, 2003) Few cases 2,500 annual cases (Mead et al., 1999) 1,600 annual cases (Scallan et al., 2011)

Listeria dose-response Probability of invasive listeriosis following the ingestion of Listeria is then On average: very low But function of Individual Susceptibility ++++ Strain Virulence ++ But no Listeria monocytogenes is safe Food vehicle ? Dose

Individual Susceptibility “Healthy” (less 65 year old, no known underlying condition) 1.0 65-74 year old 8.0 more 74 year old 20 Pregnant 116 Chronic Lymphocytic Leukemia 1,138 HIV/AIDS 55 Diabetes 1 44 Diabetes 2 5.4 Heart disease Non Hispanic women FoodNet Data, 2004-2009 (per 100,000) Pouillot et al. 2014 Clin Infect Dis. 54 Suppl 5:S405-10. Goulet et al.2012 Clin Infect Dis. 54:652-60.

Virulence heterogeneity Listeria monocytogenes is one of the most studied intra-cellular pathogen Entry of L. monocytogenes into certain human epithelial cells is receptor mediated, Depending on specific interactions between internalins on the bacterial surface and their respective host cell receptors Point mutations in the inlA gene can lead to virulence attenuation Nevertheless, all Listeria monocytogenes should be considered as pathogen

Dose-Response models Models based on animal data Model based on outbreak data Models based on epi data Hoelzer et al, 2011, Risk anal, 33 (9) 1568-81

How far will we go? ‘Average’ Virulent Strain A Less Virulent Strain B 1 0.9 0.8 0.7 0.6 How far will we go? Pr(Illness|dose) 0.5 0.4 0.3 0.2 0.1 13 log10(Dose) ‘Average’ Virulent Strain A Less Virulent Strain B Strain A, Susceptible Individual Strain B, Non Susceptible Individual

FDA 2015 dose-resposne model and butter outbreak in hospitalized patients Primate DR DR for this specific outbreak as estimated from data FDA 2015 DR for transplant organ patients, highly virulent strain, moderately virulent strain (this study) Established DR for Susceptible population, FAO/WHO 2004 Pouillot et al., Risk Analysis, 35(1):90-108, 2015

Conclusions: Dose-Response In order to explain the relatively high frequency of product contamination and the relative rareness of the disease, one must assume that, “on average”, low dose equals low risk FAO/WHO, 2004 Refined models better characterize the variability in the dose-response: “on average”, low dose equals low risk, but in susceptible subpopulations, a relatively low dose may lead to a measurable risk for some individuals Pouillot et al, 2015 Animal models and outbreak data model would predict too many cases if applied as-is The variability in individual susceptibility and strain virulence maybe more important than the dose

Listeria monocytogenes risk assessments

FDA/FSIS 2003 Quantitative Assessment of Relative Risk to Public Health From Foodborne Listeria monocytogenes Among Selected Categories of Ready-to-Eat Foods Dose Response considers strain variability and host susceptibility. Scaled to epidemiological data. Characterized uncertainty Risk: products that support growth

FSIS Risk Assessment for L. monocytogenes in Deli Meats (2003) Plant-to table probabilistic risk assessment built on FDA/FSIS 2003 risk assessment Evaluated the number of illness prevented and lives saved annually according to various processing interventions Led to Interim final rule Alternative 1: Use of post-lethality treatment AND antimicrobial agent/process Alternative 2: Use of post-lethality treatment OR antimicrobial agent/process Alternative 3: Use of sanitation procedures only Source: FSIS

FAO/WHO 2004: Risk assessment of L. monocytogenes in ready-to-eat food Exponential dose-response models One for susceptible population, One for non-susceptible population “The vast majority of cases of listeriosis are associated with the consumption of foods that do not meet current standards for L. monocytogenes in foods, whether that standard is zero tolerance or 100 CFU/g” FAO/WHO (2004). Risk assessment of Listeria monocytogenes in ready to eat foods - Technical report.

FDA/FSIS Interagency Risk Assessment: L FDA/FSIS Interagency Risk Assessment: L. monocytogenes in Retail Delicatessens (2013) Sampling surveys and RA attribute most listeriosis cases from deli-meat to retail-sliced deli meat This RA evaluated the impact of cross-contamination, sanitation, growth in retail deli Complex discrete event model FAO/WHO dose response Key findings Would explain some sporadic cases Control Growth, notably through the use of growth inhibitors Contaminated products that do not support growth can lead to a risk of listeriosis from cross-contamination of product that do support growth Interagency Risk assessment: Listeria monocytogenes in retail delicatessens (2013)

New data

Usual data to inform Listeria monocytogenes dose-response models are difficult to obtain Volunteers Animal Models Validity of the animal model? Hoelzer et al, 2012 Veterinary Research 43(1): 18 Outbreak Data Long incubation period -> Difficult to find leftovers Case report Case control studies Epidemiological Data

FDA Data Collection Efforts Celery Outbreak, TX, 2010 Limited prevalence and contamination level data Data on underlying health issues Growth studies Cantaloupe Outbreak, Multistate, 2011 Caramel Apple, Multistate, 2015 Ice Cream, Multistate, 2015 Extensive prevalence and contamination data Opportunity to assess exposure and underlying health issues

The case of the caramel apples Apple does not support growth (pH 3.2), Caramel does not support growth (low aw) Caramel coated apple with a stick support growth Up to 7 log10 in few days at ambient temperature 35 illness including 11 associated with a pregnancy 3 meningitis among otherwise healthy children aged 5-15 (target population!) CDC website Bouvier, Science Translational Medicine  7 (311) pp. 311ec184 (2015) K. A. Glass et al., mBio 6, e01232-15 (2015).

The case of ice-cream 10 cases … … over 5 years! 4 cases, observed in a single hospital, linked to one product of factory A (through milk shakes) 5 cases linked to a second factory (factory B) … over 5 years! FDA obtained products from factory A Product that does not support growth (frozen) Observed levels of bacteria are relevant to consumed levels

Enumeration data “Product A”, from factory A 2,290 samples of Product A tested all but 13 samples were positive (99.4% positive) Range: <0.03 MPN/g to > 208 MPN/gm Highly consistent low contamination levels 15% below 1 MPN/gm 58% below 5 MPN/gm 77% below 10 MPN/gm 92% below 20 MPN/gm 98% below 50 MPN/gm 99.8% below 100 MPN/gm 4 samples > 100 MPN/gm* *one >208 MPN/g (direct plating: 357 cfu/gm), two = 208 MPN/gm (direct plating: 142 cfu/gm and Non Available), and one 139 MPN/gm (direct plating: 177cfu/gm) Chen et al, submitted; Chen et al, IAFP 2015

Enumeration of L. monocytogenes in Different Lots of Product A Average: 8 MPN/g Chen et al, submitted; Chen et al, IAFP 2015

Exposure data (preliminary) Our estimates from product distribution records 4,000,000 to 33,000,000 contaminated servings sold to the population 62,000 to 471,000 contaminated servings sold to pregnant women: no case identified 16,000 to 56,000 contaminated servings sold to highly susceptible population  4 identified cases Corresponding dose-response model comparable to previous dose-response models developed from outbreaks Other outbreaks: limited diffusion of products contaminated at relatively high level Here: large diffusion of products contaminated at low level (Pouillot et al, in preparation)

Ice cream Outbreak – Tentative Conclusions We’ll probably never get better data A no-growth product, with a very low average level of contamination (8 cfu/g) did cause an “outbreak” It is likely that most patients were exposed to ice cream with < 100 cfu/g A high dose can’t be excluded Growth in milk shake > few logs improbable but “inoculation” of the shaker at each serving. Biofilm? Growth? The large distribution of a no-growth product, with a low average level of contamination (8 cfu/g) didn’t cause a massive outbreak The underlying health of the patient, cell-mediated immune status and medications may be more important than the dose Sufficient for risk management? Black Swan ? or White Swan? Few products that support growth (e.g. cheese) -> Outbreak Very large diffusion of a contaminated product that does not support growth (ice-cream) -> Outbreak

Conclusions: Listeria monocytogenes… The most studied foodborne bacteria A lot of questions remain unanswered Risk management in an uncertain world