1. Molecular Epidemiological Evidence for Diabetogenic effects of Dioxin Among Vietnam Veterans By Phillip Fujiyoshi, Joel Michalek and Fumio Matsumura

2. Molecular Epidemiology A new field of epidemiological science that is designed to increase the sensitivity of detection of biological effects, and to provide the logical bases for empirically observed correlations.

3. Brief Background A Molecular Epidemiology study coordinated with the Air Force Health Study (AFHS) design. Ranch Hand group (RH) was compared to the carefully matched dioxin unexposed Air Force veterans, who served in Southeast Asia, designated as Comparison group (C-group). Altogether 313 volunteers agreed to donate adipose samples.

4. Methodologies 1 Dioxin residues in serum lipids analyzed by CDC. Diabetes determined by diagnosis and by abnormally high 2-hour post-prandial blood glucose levels (fasting glucose). Body mass index (BMI) by weight (kg divided by the square of height (m). % Body fat defined as 2.64 x BMI – 13.04.

5. Methodologies 2 Selection of molecular markers was made by conducting preliminary animal tests (mice,chronic). All markers (mRNA expression) were detected through reverse transcription, DNA amplification, namely quantitative polymerase chain reactions (RT-PCR). Markers studied were:GLUT4 (glucose transporter), NFҝB (inflammation),C/EBPα (adiposity index), c-Src (a signal messenger for toxic action of dioxin) and GAPDH (a housekeeping gene, used here as a normalization standard).

6. Initial Study (Phase 1) Strategies To increase the power of statistics by dividing each service group into four quartiles according to the level of dioxin residues. Start the study by looking for statistically significant correlations between molecular marker expressions and dioxin residues. Search for positive correlations between dioxin and not only individual markers but also all combinations of all marker ratios.

7. Figure 1

8. Interim conclusions The ratio of GLUT4:NFҝB (= G:N ratio) appears to be the best marker among all tested ones, detecting the difference between C- and RH-group in terms of their response to dioxin. There appears to be some subgroups which affect the pattern of correlations.

9. Questions raised by the initial analysis Could such a quartile approach introduce any artifacts? Why are the directions of the slope opposite between RH and C ? Are these trends influenced by particular subgroups within each service group?

10. Strategies for Phase 2 analysis Make efforts to answer the above questions raised. Concentrate in analyzing the relationships between GLUT4:NFҝB (G:N ratio) and Dioxin. Try to find the most affected subgroups by dioxin.

11. Figure 2 (C)(RH) C = Control RH = Ranch Hand p = 0.01 p = 0.02 Dioxin

12. Conclusion from Fig 2 analysis The use of the quartile approach is justifiable, since essentially the same results can be obtained by using undivided populations.

13. Figure 3 O, ND, (C)L, D, (RH) O, ND, F (C+RH)L, ND, F (C+RH) C = Control RH = Ranch Hand O = Obese L = Lean F = Family History Dioxin D = Diabetic ND = Non-diabetic p = 0.09 p = 0.001 p = 0.04

14. Conclusion from Fig. 3 analysis 1.Subgroups differ their G:N responses to dioxin 2.Obese subgroups respond positively and Lean subgroups negatively to dioxin

15. Possible hypothesis for the difference in slopes A. The effect of dioxin is biphasic: at low dose suppressing G:N ratio and at high doses elevating. B. Dioxin mimics obesity-type at low doses, and lean-type diabetes at high doses. C. Long-term dioxin exposure at high doses induce a negative counteraction (such as glucocorticoid up-regulation) by the adipocytes.

16. Phase 3 analysis Objective: to test our hypothesis that the diabetogenic effect of TCDD is similar to that of obesity. Approach: study the relationship between G:N ratio and recent changes in body fat.

17. Figure 4 ND (C+RH)D (C+RH) C = Control RH = Ranch Hand D = Diabetic ND = Non-Diabetic p = 0.03Change in Body Fat p < 0.001

18. Conclusion from Fig.4 analysis The G:N ratio among nondiabetic subjects is negatively correlated to obesity, one of the risk factors for diabetes. However, among diabetic subjects the tendency is opposite.

19. Figure 5 Change in Body Fat ND, NF (C)ND, NF (RH) ND, F (C)ND, F (RH) C = Control RH = Ranch Hand ND = Non-Diabetic F = Family History NF = No Family History p = 0.02 p = 0.003 p = 0.03

20. Brief Conclusions from Phase 3 analysis 1. The G:N ratio can detect diabetogenic effects of obesity as well as genetic risk factors. 2. Dioxin exposure makes nondiabetic RH subjects more susceptible to diabetogenic influence of obesity.

21. Phase 4 analysis Objective: to relate the above findings to diabetes. Approach: Use “fasting glucose” as the diabetes marker, and find its relation ships to (a) G:N ratio, (b) % body fat and (c) dioxin residue levels.

22. Figure 6 C = Control RH = Ranch Hand D = Diabetic ND = Non-Diabetic p = 0.009 ND (C+RH)D (C+RH) p = 0.001

23. Conclusion from Fig 6 analysis The G:N ratio is a reliable marker of diabetes, since it is highly correlated to the fasting glucose levels. The G:N ratio is a reliable marker of diabetes, since it is highly correlated to the fasting glucose levels.

24. Figure 7 C = Control RH = Ranch Hand D = Diabetic ND = Non-Diabetic p = 0.09 ND (C+RH)D (C+RH) p < 0.001

25. Conclusion from Fig 7 analysis Fasting glucose levels is a sensitive marker in detecting the diabetogenic effect of obesity

26. Figure 8 (C)(RH) C = Control RH = Ranch Hand p = 0.02

27. Conclusion from Fig. 8 analysis The fasting glucose levels are also directly affected by dioxin

28. Brief mechanistic explanation/ hypothesis Dioxin causes TNFα-type inflammation (NFҝB up) similar to that induced by obesity. The resulting suppression of glucose uptake (GLUT4 down) by adipose tissue and release of lipids into blood, which causes insulin-resistance and suppression of insulin secretion from pancreas. Inflammation also causes glycogen breakdown in liver resulting in massive release of glucose into blood, contributing to the rise in fasting glucose.

29. The main conclusion We could obtain the definite molecular epidemiological evidence, indicating that dioxin is acting as a diabetogenic risk factor among Vietnam veterans even after many years from the time of exposure at such low levels of exposure

30. Major Health Implication The diabetogenic effect of dioxin can be found even in C subjects whose serum levels of dioxin overlap with the medium to high end of the general public. The implication is very clear that we, environmental health scientists, should become concerned about the potential diabetes-related health effect of dioxin to the general public, particularly those people with known risk factors, even at the current level of low dioxin exposure.

31. Acknowledgements This work could not have been accomplished without the tremendous co- operation offered by volunteered veterans, all AFHS personnel, Medical collaborators and funding managers. We would like thank them all. Supported by the US Air Force contract (sub-project #01-813-32-8280-906- 721900).