Epidemiologic evidence for a relationship between NIR and cancer - a controversial issue

1. Epidemiologic evidence for a relationship between NIR and cancer - a controversial issue Michael KundiInstitute of Environmental Health, Medical University of Vienna, Austria

3. Power Frequency EMF and Cancer • Childhood leukemia: 23 studies • Childhood nervous system tumors: 11 studies • Other childhood cancers: 8 studies • Adult leukemia: 15 studies • Adult nervous system tumors: 10 studies • Breast cancer: 19 studies • Other adult cancers: 4 studies

4. Childhood Leukemia

5. Power Frequency EMF and Cancer The most comprehensively studied endpoint is childhood leukemia. Pooled analyses (Ahlbom et al. 2000; Greenland et al. 2000) demonstrate an about two-fold increased risk at average exposure of >0.3 to 0.4 µT. The International Agency for Research on Cancer (IARC) classified power frequency magnetic fields as a possible human carcinogen (Group 2B).

6. Can the relationship between childhood leukemia and magnetic fields be causally interpreted? • Requirements: • Temporal relation: Exposure onset must be in line with what is known about the natural history of the disease • Association: There must be a significant relationship between exposure and disease • Population equivalence: Cohorts or cases as well as controls must be sufficiently similar (except for the exposure) to consider them equivalent • Environmental equivalence: Cohorts or cases as well as controls must be exposed to the same environmental conditions (except for the exposure under study) to consider their environments equivalent

7. Equivocal support from animal studies due to genetic factors? Fedrowitz et al. 2004

8. Measurements in US homes (HVTRC 1993) Only about 1% of homes above 0.4 µT

9. Attributable risk Source: Kheifets et al. 2007

10. There is an increased risk of childhood leukemia associated with power frequency magnetic fields that is consistent with the natural history of the disease. “In spite of the large number data base, some uncertainty remains as to whether magnetic field exposure or some other factor(s) might have accounted for the increased leukaemia incidence.” (WHO Fact Sheet 263, 2001) No confounder, neither selection nor misclassification bias has been shown to explain the relationship. Hence there are no valid counterarguments against a causal interpretation. There is no established mechanism of action and animal as well as in vitro studies provide only equivocal evidence for the carcinogenicity of magnetic fields.

11. Lack of consistency of animal and in vitro studies is rather the rule as the exception (e.g. chlorinated hydrocarbons). Mechanism of action of carcinogens is rarely known even if the agent has been banned (e.g. chrysotile asbestos). But the number of children exposed to levels above 0.3 or 0.4 µT is very low and “in a global context, the impact on public health, if any, would be limited and uncertain.” (WHO EHC 238, 2007) The attributable fraction is only low under the assumption that there is a threshold and that the average exposure level is the correct metric. “Provided that the health, social and economic benefits of electric power are not compromised, implementing very low-cost precautionary procedures to reduce exposure is reasonable and warranted.” (WHO EHC 238, 2007)

12. Can the relationship between childhood leukemia and magnetic fields be causally interpreted? • Requirements: • Temporal relation: Exposure was determined for residence at birth and at diagnosis • Association: There is a monotonous increase of risk for increasing levels of estimated exposure  • Population equivalence: Participation of controls was in some studies low with a potential for differential misclassification, however, comparison with studies based on registry data only showed the same effect size  • Environmental equivalence: Investigated confounders could not explain the observed increased risk 

13. Summary • There is consistent evidence from epidemiological studies • Support from animal and in vitro studies is equivocal • There is no evidence based exposure metric • If there is a threshold at 0.3 to 0.4 µT attributable risk is low, however, data are consistent with a much higher attributable risk depending on the mechanisms of action • There are no valid counter arguments against a causal interpretation of the relationship between magnetic fields and childhood leukemia • Future studies must focus on the mechanism of action and the definition of a suitable evidence based exposure metric

14. Radiotelephones and Human Health - A European Research Initiative A.F. McKinlay Radiation Protection Dosimetry 72:313-320 (1997) Whereas a large database existsfor possible effects on human health from exposure to extremelylow frequency (particularly power frequency) electromagneticfields, there are far fewer data for radiofrequency (includingmicrowave) fields, and very few related to the emissions andexposures specific to personal telecommunications. A comprehensiveassessment of the risk of effects on human health requires suchdata.

15. Epidemiological Studies of Mobile Phones and Brain Tumors

16. COHORT OF MP USERS Incidence of brain tumors 5:5000 5000 Relative Risk 5:5000/3:5000=1.67 COHORT OF NON-USERS 5000 Incidence of brain tumors 3:5000

17. BRAIN TUMOR CASES Odds for MP use 5:20 Odds ratio 5:20/3:22=1.83 CONTROLS Odds for MP use 3:22

18. Overall 25 epidemiological studies published

19. Methodological Difficulties • If correctly applied, analytical epidemiology is able to detect a risk from an exposure provided the following conditions are met: • There is a reliable exposure metric • There is an evidence based selection of a disease • Exposure duration is compatible with the natural history of the disease None of these conditions are at present met for the study of mobile phone use and tumors in the head region.

20. Problem 1: Exposure Metric

21. Fundamental problems of exposure assessment All these patterns have the same cumulative hours of use! Are they equivalent?

22. Problem 1: Exposure Metric It is unknown which aspect of exposure is responsible for the increased risk. Cumulative hours of use, cumulative number of calls, specific absorption rate in the area of the tumor and other parameters may or may not reflect essential aspects of exposure. The mechanism of interaction between the EMF and cells and tissues must be the basis for the definition of exposure!

23. Problem 2: Selection of Disease

24. Intracranial Tumors Exposure for traditional MP use is predominantly to the head. Therefore, studies were almost exclusively about tumors in the head region. Nervous System Tumors Neuroepithelial tissue Astrocytic tumors (11) Oligodendroglial tumors (2) Mixed glioma (2) Ependymal tumors (8) Choroid plexus (2) Neuronal and mixed (12) Embryonal tumors (11) Others (5) Peripheral nerves Schwannoma (4) Neurofibroma Perineuroma MPNST (5) Lymphoma & Haematopoietic Neoplasms (3) Germ cell (8) Sellar region (4) Meningial Tumors Meningioma (16) Mesenchymal (21)

25. Molecular histopathology Pluripotent Stem-cell One and the same histopathologic type may be distinctly different with respect to molecular histopathology Glioblastoma multiforme

26. Problem 2: Selection of Disease Are all types of brain tumors associated with an increased risk from mobile phone use? It is impossible to differentiate all types of tumors with respect to this risk based on epidemiology! It is important to work on potential mechanisms of action of low intensities of EMF on biological systems!

27. Problem 3: Exposure Duration

28. Is there any chance to find an increased risk for even the strongest environmental carcinogen in the short or medium term?

29. Latency of Brain Tumors • Glioma: • 20-30 years average (Kranzinger et al. 2001; Sadetzki et al. 2005) • Meningioma • 20-40 years average (Umansky et al. 2008) • Acoustic Neuroma • Average doubling time ~1.7 years -> 25 years Latency (Mohyuddin et al. 2003)

30. Interphone Study – Exposure Duration Source: Hepworth et al. BMJ 2006; 332(7546):883-887

31. Problem 3: Exposure Duration Average exposure duration in epidemiological studies conducted so far is ~15% of latency period! If exposure duration is only a small proportion of the natural history of the disease even potent carcinogens are undetectable in epidemiological studies! Only those subgroups that use a mobile phone for at least 10 years are relevant!

32. Exposure is predominantly unilateral! 97%-99% of absorbed power from mobile phone use is absorbed in the hemisphere corresponding to the side of the head the mobile phone is held during calls Only the side of the head the phone is held to during calls can be considered exposed! The most important result from epidemiological studies is the risk estimate associated with long term ipsilateral mobile phone use!

33. Summary Glioma(>10 years of MP use)

34. Summary Acoustic Neuroma(>10 years of MP use)

35. Summary Meningioma(>10 years of MP use)

36. Summary Epidemiologic evidence indicates an increased risk for mobile phone users to develop brain tumors. Considering biases that could have had distorted estimates, combined risks are rather underestimates. For the time being epidemiology provides the only consistent evidence for an increased risk. Although support from other areas of research is weak evidence is sufficient to recommend precautionary measures to reduce exposure.