Cristina Fernandez, the President of Argentina, had her thyroid removed in January of this year only to find out the gland wasn’t cancerous after all. Although her supporters whooped with joy at this news, doctors can’t put her thyroid back, and Fernandez will be on thyroid meds for life.1 Were her doctors incompetent or did they act appropriately? As that debate continues to rage over the internet, the Fernandez case has also led to widespread discussion of why thyroid cancer incidence, especially among women, has dramatically increased over the last thirty years.
AN EPIDEMIC OF THYROID CANCER
According to the National Cancer Institute, the incidence of thyroid cancer has nearly doubled since the early 1970s. Thyroid cancer now affects about eleven people per hundred thousand in the United States. In January 2008, there were 458,403 Americans alive with a history of thyroid cancer of which 100,952 were men and 357,451 women. In 2011, 56,460 new cases of thyroid cancer were diagnosed and 1,740 people died of the disease.2,3
Doctors do not know why the numbers of thyroid cancer cases are increasing though some blame increased overweight and obesity, radiation exposure, and diets low in fruits and vegetables.
Certainly exposure to radiation is a known risk factor for thyroid cancer.4 In 2009 epidemiologist Joseph Mangano, PhD, took data on thyroid cancer incidence from the Centers for Disease Control for the years 2001 to 2005, compared it with the proximity of nuclear power stations, and found that the counties with the highest thyroid cancer incidence were located close together in eastern Pennsylvania, New Jersey, and southern New York. He concluded, “Exposure to radioactive iodine emissions from sixteen nuclear power reactors within a ninety-mile radius in this area…[is] likely a cause of rising incidence rates.”5 Pennsylvania has the highest rate of thyroid cancer in the U.S.
In 2010 the Associated Press revealed that 75 percent of U.S. nuclear power plants leak radioactive materials into our air and water.6 And many of the one hundred four commercial nuclear power plants and thirty-four nuclear research stations now operating in the U.S. sit in seismically active locations, including at least four near the “high risk” San Francisco Bay area and three within the San Francisco Bay area itself.7 As might be expected, there is a high incidence of thyroid cancer in the San Francisco Bay area.
Radiation in ground water linked to hydraulic fracturing—or fracking—the process used to extract oil or natural gas deep in the earth—has also been linked to increasing rates of thyroid cancer. Fracking has also led to a 2,400 percent increase in earthquakes compared to the number of quakes that occurred in the years before fracking started in the U.S. 9,10 Geologist Tracy Bank, speaking at the American Geological Society meeting in Denver last November, reported that fracking releases rock-bound uranium, posing a further radiation risk to our groundwater.11
Hormonal factors may also play a significant role, according to the National Cancer Institute. Although NCI arrived at this conclusion due to the preponderance of thyroid cancer cases in women under age forty-five, human estrogens should be regarded as just one piece of the hormonal picture. Xenoestrogens—estrogenic substances found in the diet and the environment— also play a role. Commonly found in plastics, pesticides, cosmetics, personal care products, our water supply, factory-farmed meats and soy foods, xenoestrogens can be significant “endocrine disruptors” and interfere with the functioning of many systems in the body.12
While it’s human nature to try to single out one factor to blame, the causes of thyroid cancer most likely are many and synergistic. Exposure to radiation, mercury, fluoride, 13,14 plastics, pesticides, dioxins, solvents, low iodine intake,15 and estrogens and estrogen mimickers found in commercial meats and produce, plastic and hormone replacement therapies have all been implicated. And so has soy.
Soy is widely marketed as a “health food” although soybeans naturally contain the phytoestrogens (plant estrogens) known as isoflavones. While not true hormones, isoflavones closely resemble estradiol (E2),16 the most potent of the three forms of estrogen found in the human body17 and the form of estrogen that has been implicated in thyroid cancer.18-20 Soy isoflavones cause significant endocrine disruption both directly by binding with estrogen receptors, and indirectly by interfering with the body’s production of estrogen, testosterone and other hormones. The effects are felt throughout the body, especially the thyroid and reproductive system, and are well documented in chapters twenty-six and thirty of The Whole Soy Story: The Dark Side of America’s Favorite Health Food.21
The key isoflavones found in soy, genistein and daidzein, are potent inhibitors of thyroid peroxidase (TPO), an enzyme involved in the synthesis of the thyroid hormones T3 and T4. In vitro experiments carried out at the National Center for Toxicological Research in Jefferson, Arkansas, Rao L. Divi, PhD, and Daniel R. Doerge, PhD, showed soy isoflavones will inhibit TPO and interfere with a critical stage in thyroid hormone production—the iodinization of the amino acid tyrosine. Although many people assume sufficient iodine will solve this problem, this interference occurs whether or not sufficient or extra iodine is present. As a result, the body produces useless mono-, di- and tri-iodoisoflavones and not mono, di and tri and quarto forms of thyroid hormone. In the human body, this interference can cause a drop in thyroid hormone levels, an increase in thyroid stimulating hormone and stress on the thyroid gland. To put it bluntly, this is a prescription for thyroid trouble.22, 23
Drs. Divi and Doerge, top scientists with the National Center for Toxicological Research, pulled no punches in their conclusion: “The possible association between long-term inhibition of thyroid hormone synthesis (goiter) and induction of thyroid follicular cell hyperplasia and neoplasia underscores the significance of these findings.” 24,25 Follicular cell hyperplasia is a precursor to thyroid tumors and neoplasia is an abnormal proliferation of cells and characteristic of cancer.
We also know soy products pose a special risk to hypothyroid patients treated with Synthroid and other thyroid drugs. According to Mike Fitzpatrick, PhD, boosting the thyroid with drugs like Synthroid, then depressing it with thyroid inhibitors like soy foods or isoflavone supplements, can put extreme stress on the thyroid. In fact, this is the classic way that researchers induce thyroid tumor in laboratory animals. The fact that soy is “natural” does not make it safe or weak. The phytoestrogens in a serving of soy food can provide up to three times the goitrogenic potency of the pharmaceutical thyroid-inhibiting drugs methimazole and 6-propylthiouracil. 26
Over the past seventy years, numerous studies have linked soy to thyroid disorders, especially hypothyroidism and the autoimmune thyroid disease Hashimoto’s thyroiditis. These studies are cited and discussed in detail in chapter twenty-seven of The Whole Soy Story. 27 Less evidence links soy to thyroid cancer, though so many studies proving stress on the thyroid would suggest clear and present danger. Soy proponents and industry spokespersons, however, prefer to assert that soy is protective, and the study cited most frequently is the Bay Area Thyroid Cancer Study.28
THE BAY AREA THYROID CANCER STUDY
This study is described in three articles published by Pamela Horn-Ross, PhD, and colleagues, in the journal, Cancer Epidemiology, Biomarkers and Prevention (CEBP), in 2001 and 2002.29-31
In the 2002 CEBP study, Horn-Ross, Hoggatt and Lee attempted to determine how soy phytoestrogen intake relates to thyroid cancer once other factors such as age, race and other known risk factors were taken into account. In the results section they reported, “In general, a reduction in thyroid cancer risk of 35 percent to 55 percent was associated with increased consumption of non-fermented traditional and nontraditional soy-based foods and sprouts.”32
An astonishing 35 to 55 percent reduction in risk with clear cause and effect certainly seems to support the idea of consuming soy—including modern industrial soy products—for thyroid cancer prevention. But what seems to be too good to be true is often just that. A long, hard look at the study—and not just at the headlines publicized by the soy industry—reveals serious flaws in design, methods, and analysis, including:
• This paper describes an observational, casecontrol, matched study. As J.M. Utts and R. Heckard write in their textbook, Mind on Statistics, “The most common mistake made in reporting research studies is to imply that a cause and effect relationship can be concluded from an observational study. With an observational study, it is difficult, perhaps impossible, to separate the effects of confounding variables from the effects of the main explanatory variables of interest.”33
• The study was not a randomized, controlled trial, which is the gold standard for testing an intervention. Cases were not randomized to treatment groups but drawn from a cancer registry, which was a sample of convenience. As Utts and Heckard put it, “If the sample does not represent a larger population for the question of interest, and randomization to treatments was not used, no inferences can be drawn.”34
• The data were analyzed using unconditional logistic regression. When the sample comes from matched pairs—as was the case in this study—conditional logistic regression is the appropriate test, not unconditional logistic regression. As summed up in the Oxford Journal, “A simple rule of thumb is to use conditional logistic regression if matching has been done, and unconditional if no matching has been done. A second rule of thumb is, when in doubt always use conditional because it always gives unbiased results.”35
• Because the study used unconditional logistic regression, the researchers did not include the matching information in the analysis.36 This is most interesting in the light of research from the University Graduate School of Public Health in Kyoto, Japan, which examined 507 studies from 1991-2000 that used case control matched data sets.37 Of these studies, conditional logistic regression was used in 90.5 percent, and unconditional logistic regression in only 9.5 percent of them. Yet Horn-Ross and colleagues chose to use the unconditional method.
• Unconditional logistic regression analysis seriously overestimates the odds ratio when there are matching data—as was the case with Horn-Ross and colleagues—and great caution should be taken in interpreting the results.38 In Statistical Methods in Cancer Research, a classic text in disease epidemiology, Breslow and Day state: “The unconditional analysis of matched pair data results in an estimate of the odds ratio which is the square of the correct, conditional one: a relative risk of 2 will tend to be estimated as 4 by this approach…” (italic emphasis from Breslow and Day).39
• The spotlighted phytoestrogens yet included a large number of potentially interrelated variables that could interact with one another. In a high quality study, the researchers should have addressed the possibility of collinearity and taken care to rule it out. Collinearity is a bias in statistical procedure due to the correlation of multiple independent variables that influence a single dependent variable. Collinearity can lead to unstable and untrustworthy results.40
• All the subjects came from the San Francisco Bay Area and many were of Asian ethnicity. Environmental, climatic and ethnic aspects were not taken into account in the analysis. External validity is always a key question. Can these results be applied or generalized to other people? Given that people from other areas of the United States live under varying conditions and are of many different ethnicities, the results of the study—if valid—would apply only to the group from which they originated.
• Reliance on a Food Frequency Questionnaire (FFQ) to determine dietary intake during the year before the diagnosis of thyroid cancer, or for the year prior to the interview for the controls, is suspect. FFQs require people to remember what they ate, when they ate it, and how much.41 Over-estimation is common, particularly for foods eaten less often or for foods perceived as “healthy,” such as fruit, vegetables—and soy. In her article, Dr. Horn-Ross does not disclose how her FFQ was tested or evaluated prior to use in the San Francisco Bay Area Thyroid Study. She also admits “phytoestrogen consumption was not a hypothesis of this study when this FFQ was developed.”42
• In Table 1 of Horn-Ross’ article, “Selected characteristics of women participating in the multiethnic San Francisco Bay Area Thyroid Study,” we see how the cases and controls are similar on many variables such as age and number of pregnancies, but we do not know how many subjects were actually included or whether the Table represents all subjects or just a cherry-picked sample.
• In Table 2, “Consumption of selected phytoestrogen-rich foods and thyroid cancer risk among women participating in the Bay Area Thyroid Cancer Study,” the researchers make the dramatic pronouncement of reduced risk of 35 to 55 percent. However, this Table reports odds ratios but no actual risk data. Relative Risk, the basis for determinations such as “reduced risk,” cannot be calculated in a case-control study. Odds ratios can be used to represent relative risk if the disease is relatively rare, as is the case with thyroid cancer, but they are usually “bigger in each case” and “around ten percent larger than Relative Risk.” 43,44
• In Table 3, “Phytoestrogen consumption and thyroid cancer risk among women participating in the Bay Area Thyroid Cancer Study,” the researchers report an “increased consumption of four of the seven specific phytoestrogenic compounds as well as three summary measures were associated with a reduced risk of thyroid cancer . . .” Just how much reduced risk is never established or explained.
• The odds ratios in Table 2 and Table 3 show that many are near or around 1.00 which means that there are no (null) effects. Many rows—subgroups—have too few cases and controls to show statistical value. For the other rows with subgroups, we have no indication of significance (p value). P value is given only for “trend across quintiles.”
In conclusion, this paper should not be accepted as a serious study of thyroid cancer risk related to phytoestrogen intake. The researchers failed to provide details concerning the number of models, the parameters included in each of the models, construction of composite variables (Table 3), and trend tests used to produce the statistical results (p values) in Tables 2 and 3. We don’t even know the statistical software used to fit the models. The article’s clearest and most powerful statement—a reduction in thyroid cancer risk of 35 percent to 55 percent was associated with increased consumption of non-fermented traditional and nontraditional soy-based foods and sprouts—comes without explanation out of the blue.
1. CBS News. http://www.cbsnews.com/8301-501715_162-57354512/docs-argentine-leaders-thyroidwasnt- cancerous/
2. American Cancer Society: http://www.cancer.org/acs/groups/cid/documents/webcontent/003144-pdf.pdf
3. National Cancer Institute, SEER-Surveillance Epidemiology and End Results. Stat Facts Sheet. http://seer.cancer.gov/statfacts/html/thyro.html
4. USA Today. http://yourlife.usatoday.com/health/story/2012-01-15/Doctors-try-to-explain-increase-inthyroid-cancers/52584788/1
5. http://www.radiation.org/reading/pubs/091116Thyroidcancer.pdf. Accessed 1-11, 2012
7. Nuclear Reactors in Earthquake Zones in the US. http://www.treehugger.com/corporate-responsibility/nuclear-reactors-in-earthquake-zones-in-the-us-map.html. 1-24- 12.
8. Horn-Ross, P et al. Why Are Thyroid Cancer Rates So High in Southeast Asian Women Living in the United States? The Bay Area Thyroid Cancer Study. Cancer Epidem Biomar. 2003, 12, 144-150.
9. http://www.activistpost.com/2012/01/thyroid-cancer-fracking-and-nuclear.html. Accessed 1-23, 2012.
10. Ananda, Rady. Food Freedom, Thyroid Cancer, Fracking and Nuclear Reactors. http://foodfreedom.wordpress.com/2012/01/18/thyroidcancer-fracking-nuclear-power. Accessed 1-19, 2012.
11. http://www.buffalo.edu/news/11885 Accessed January 24, 2012.
12. Golden RJ, Noller KL, Titus-Ernstoff L, et al. Environmental endocrine modulators and human health: an assessment of the biological evidence. Crit. Rev. Toxicol, 1998,28, 2, 109–227. doi:10.1080/10408449891344191. PMID 9557209.
13. Connett, Paul. Beck, James. Micklemp, HS. The Case Against Fluoride. (White River Junction, VT, Chelsea Green Publishing Company, 2010) 183-185.
14. NRC, Fluoride in Drinking Water: A Scientific Review of EPA’s Standards (National Academies Press, 2006) 266, ch.8, www.actionpa.org/fluoride/nrc/NRC-2006.pdf
15. Iodine Deficiency. http://emedicine.medscape.com/article/122714-overview. 1-23, 2012.
16. US Soyfoods Directory. http://www.soyfoods.com/nutrition/AbsorptionMetabolism.html. Accessed January 23, 2012.
17. Brownstein, D. Iodine: Why You Need It (Medical Alternative Press, 2008). 73-79.
18. Yao R, Chiu CG, Strugnell SS,et al. Gender Differences in Thyroid Cancer. Expert Rev Endocrinol Metab, 2011, 6, 2, 215-243.
19. Manole D, Schildknecht B, Gosnell B, Adams E, et al. Estrogen Promotes Growth of Human Thyroid Tumor Cells by Different Molecular Mechanisms. J Clin Endocr Metab, 2001, 86, 3, 1072-1077.
20. Kumari A, Klinge CM and Goldstein GM. Estradiol-induced proliferation of papillary and follicular thyroid cancer cells is mediated by estrogen receptors a and ß. Int J of Oncology 2010, 36,1067-1080.
21. Daniel, K.T. The Whole Soy Story: the Dark Side of America’s Favorite Health Food. (Washington, DC, New Trends Publishing, 2005).
22. Doerge DR. Inhibition of thyroid peroxidase by dietary flavonoids. Chem Res Toxicol, 1996, 9, 16-23.
23. Divi RL, Chang HC, Doerge DR. Anti-thyroid isoflavones from soybean. Biochem Pharmacol, 1997, 54, 1087-1096.
24. Doerge DR, Inhibition of thyroid peroxidase by dietary flavonoids. Chem Res Toxicol, 1996, 9, 16-23.
25. Divi RL, Chang HC, Doerge DR. Anti-thyroid isoflavones from soybean. Biochem Pharmacol, 1997, 54, 1087-1096.
26. Fitzpatrick Mike. Soy Formulas and the effects of isoflavones on the thyroid. NZ Med J. 2000, 1131-1103 24-26.
27. Daniel, KT, The Whole Soy Story: The Dark Side of America’s Favorite Health Food (Washington, DC, New Trends, 2005).
28. Syd Baumel, at http://eatkind.net/wholesoystory.htm. Accessed January 22, 2012.
29. Sakoda LC and Horn-Ross PL. Reproductive and menstrual history and papillary thyroid cancer risk: the San Francisco Bay Area thyroid cancer study. Cancer Epidem Biomar, 2002, 11, 51-57.
30. Horn-Ross PL, Hoggatt KJ and Lee MM. Phytoestrogens and thyroid cancer risk: the San Francisco Bay Area thyroid cancer study. Cancer Epidem Biomar, 2002, 11, 43-49.
31. Horn-Ross PL, Morris JS, Lee M, West DM, et al. Iodine and thyroid cancer risk among women in a multiethnic population: the Bay Area thyroid cancer study. Cancer Epidem Biomar, 2001,10,979-985.
32. Horn-Ross PL, Hoggatt KJ and Lee MM. Phytoestrogens and thyroid cancer risk: the San Francisco Bay Area thyroid cancer study. Cancer Epidem Biomar, 2002,11, 44.
33. Utts, J. M, Heckard, R. Mind on Statistics, 3rd edition. (Belmont, CA, Thomson-Brooks/ Cole, 2007). 136.
34. Utts, J. M, Heckard, R. Statistical Ideas and Methods (Belmont, CA, Thomas Brooks/ Cole, 2006) 669.
35. Journal of Tropical Pediatrics. Mother and Child Health: Research Methods. Research Methods II. Analysis of Case-control studies. Logistic Regression 11,www.oxfordjournals.org/tropej/online/ma_chap11.pdf 122.
36. Agresti, Alan. Categorical Data Analysis. (New York, Wiley-Interscience, 2002) sections 6.7.1, 10.2.
37. Rahman, M et al. Conditional versus unconditional logistic regression in the medical literature. Letter to the Editor. J Epidem. 56, 2003, 101–102. (Kyoto University Graduate School of Public Health, Kyoto, Japan).
39. Breslow, NE and Day, NE, Statistical Methods in Cancer Research. Volume 1- The analysis of Case-Control Studies (Switzerland, IARC, 1980) 249-251.
40. Analysis of case control studies. Logistic Regression, 121-122. www.oxfordjournals.org/our_journals/tropej/online/ma_chap11.pdf
41. Willett, Walter. Nutritional Epidemiology (Oxford University Press, New York, 1990) 69-126.
42. Horn-Ross PL, Hoggatt KJ and Lee MM. Phytoestrogens and thyroid cancer risk: the San Francisco Bay Area Thyroid Cancer Study. Cancer Epidem Biomar, 2002, 11, 48.
43. Motulsky, Harvey, Intuitive Statistics (New York, Oxford University Press, 1995) 82-84.
44. Jewell, NP. Statistics for Epidemiology (CRC Press, New York, 2003) 31-34, 41.
This article appeared in Wise Traditions in Food, Farming and the Healing Arts, the quarterly journal of the Weston A. Price Foundation, Spring 2012.