1. 1 Estimation of cancer risks and benefits associated with a potential increased consumption of fruits and vegetables Rick Reiss Exponent September 25, 2012

2. 2 Acknowledgments  Partially funded by the Alliance for Food & Farming, which represents farmers and farm groups  Collaborators include Dr. Carl Keen (UC Davis) and Jason Johnston, Kevin Tucker, and Dr. John DeSesso (Exponent)

3. 3 Paper Accepted to Food and Chemical Toxicology Link: http://www.sciencedirect.com/science/article/pii/S027869 1512006394 http://www.sciencedirect.com/science/article/pii/S027869 1512006394 dx.doi.org/10.1016/j.fct.2012.08.055

4. 4 Outline of Presentation  Review of consumer perception of pesticide residues  Methodology of paper:  Estimation of cancer prevention benefits  Estimation of cancer risk from pesticides  Results  Caveats and uncertainties  Conclusions

5. 5 Introduction  Most fruit and vegetables contain small levels of pesticide residues  There is significant consumer concern about the potential effect of these residues  Any concern must be balanced by the substantial health benefits of eating fruits and vegetables

6. 6 Several Public Campaigns Aimed at Pesticide Residues  Environmental Working Group’s “Dirty Dozen”  No consideration of toxicity or risk  Not transparent how the “dirty dozen” is determined  Organic Trade Association’s “Dietary Risk Index”  Throws out non-detects from analysis  Conflates acute and chronic consumption data

7. 7 Residues Are Prevalent but Small – Pesticide Data Program (2008 PDP)

8. 8 Dr. Oz: Scaring Consumers

9. 9 Cancer Risk Perception  Many consumers see pesticide residues on food as a significant risk concern  About 70% of Spanish respondents considered avoiding pesticide-treated fruits and vegetables as a means to prevent cancer (Garcia et al., 1999)  Organic food buyers estimated the risk of mortality from consuming conventionally-grown food to be similar to a 1 pack per day smoker (Hammitt, 1990)  The Alar scare in 1988 resulted in “near hysteria” (American Dietetic Association, 2007).

10. 10 Cancer Risk Perception  Many consumers do not identify fruits and vegetables consumption as a way to prevent cancer:  About ½ of survey respondents did not identify fruit and vegetable consumption as a protective effect against cancer  It is time to correct some of these perceptions, to the extent possible

11. 11 Prevention of Cancer from Fruit and Vegetable Consumption  Hundreds of studies have been conducted  WCRF/AICR meta-analysis of epidemiologic studies found “probable” evidence of a benefit for several cancers  Case for prevention has weakened somewhat over time, but is still strong  “Probable” evidence for mouth, pharynx, larynx, esophageal, stomach, and lung cancer (fruits and/or vegetables)  Relative risks provided for these cancer types

12. 12 Cancer Sites with Decreased Risk for F&V Consumption (“Probable”)

13. 13 Cancer Sites with Decreased Risk for F&V Consumption (“Limited”)

14. 14 Studies for Green- Yellow Vegetables and Stomach Cancer

15. 15 Prevention of Cancer from Fruit and Vegetable Consumption  Assumptions for estimate:  Reasonable estimates for prevented cases can be estimated from the RRs from the WCRF/AICR meta-analysis  Estimate based on half of the population with least fruit and vegetable consumption increasing consumption by 1 serving each of fruits and vegetables (80 gram/day serving)  Avoided cases based on RRs and SEER background cancer rates.  To avoid overestimates, an average of fruit and vegetable estimates were used.  RR from lung cancer case-control was dropped as an outlier

16. 16 Estimation of Cancer Risks  Steps: 1.Assemble PDP data from 2004-2008. 2.Identify pesticides with cancer unit risk estimates from EPA (21) (OPP and IRIS estimates) 3.Used consumption data from the CSFII (94-96, 98). Use surrogates to account for all food items. Estimate average exposure (assume as lifetime exposure). 4.Apply standard cancer risk assessment methods to estimate total cancer risk for pesticides in the diet (i.e., Risk = Exposure X Q 1 *) 5.Adjust to 80 gram serving each of fruits and vegetables. 6.Current U.S. population assumed as 310 million

17. 17 Cancer Risk Benefit Analysis for Fruit and Vegetable Consumption (Annualized)

18. 18 Summary of Avoided Cases (Annualized) Cancer SiteAvoided Cases Pharynx5839 Larynx1946 Esophagus3431 Stomach4191 Lung6112

19. 19 Chemicals Contributing to Cancer Cases for Vegetable Consumption

20. 20 Uncertainties – Cancer Benefit Estimates  Strengths of epidemiologic data:  Based on humans!  Lots of different studies  Central estimates of risk  Includes pesticide risk (essentially a net benefit)  Weaknesses of epidemiologic data:  Confounding is always a concern  Consumption estimates are uncertain  Some researchers think the cumulative evidence is weaker than “probable” (Key, 2011)

21. 21 Uncertainties – Cancer Benefit Estimates  Confirming evidence  Mechanistic evidence for components of F&V including carotenoids, lycopene, folic acid, and Vitamin C (tumor initiation and progression).  Boffetta et al. (2010) study of European population found similar estimates

22. 22 Uncertainties – Cancer Risk Estimates  Risk based on EPA cancer unit risk factors:  Based on high dose rodent bioassays.  Use an extrapolation method to estimate risk at lower, untested doses.  For example, lowest dose in permethrin study was 3 mg/kg/day, whereas average exposure was 1.3x10 -4 mg/kg/day (20,000-fold difference)  EPA on its cancer risk estimates: “while uncertain, are more likely to overstate than understate hazard and/or risk.”

23. 23 Uncertainties – Cancer Risk Estimates  Criticisms of rodent bioassays (Ames and colleagues):  High doses lead to chronic wounding of tissues, cell death, and chronic cell division of neighboring cells to replace damaged tissue.  Causes cancer at high doses that would not occur at lower doses.  99.9% of pesticides in the diet are from natural sources  Of the small number of natural plant pesticides that have been tested with rodents, half are positive.

24. 24 Diuron Cancer Bioassay vs. Exposure Group (mg/kg/day)Tumor Incidence 01/49 3.750/50 37.51/49 37535/48 Average exposure ~ 2x10 -9 mg/kg/day

25. 25 Uncertainties – Cancer Risk Estimates  Evidence contrary to carcinogen classification for dieldrin:  Study of 570 workers at a production plant from 1954-1970 (Van Amelsvoort et al., 2009).  No elevation of cancer incidence.  Results not consistent with Q 1 * (Sielken et al., 1999)  No association with cancer and organochlorine exposure in the Agricultural Health Study (Purdue et al., 2007).

26. 26 Uncertainties – Cancer Risk Estimates  Potential sources of underestimation:  Non-detect residues were counted as zero  No practical way to do otherwise  Some pesticides are considered “possible” carcinogens and do not have Q 1 * values – these were not included  Counterbalanced by not including F&V-cancer associations with “limited” evidence.

27. 27 Reliability of Data Sources FactorCancer Benefit Estimate Cancer Risk Estimate Extrapolation from animals to humans Not necessaryNecessary Extrapolation from high dose to low dose Not necessaryNecessary Central estimate or upper-bound Central estimateUpper-bound Potential confoundingYesNo

28. 28 Conclusions  Benefits of eating fruits and vegetable far outweighs any potential risks from pesticide residues  For cancer residue effects are 1/2000 th, at most, for prevention benefits  Risk estimates for residues are likely overestimated  Estimates are strongest on a comparative basis.  Substantially more attention is given to the pesticide residue effects than on the benefits  Public health would be served by this perception changing.