Thyroid cancer is currently the 5^th most commonly diagnosed cancer in United States (US) women (1). Substantial increases in incidence have been observed in US women and men across all ethnic groups (1–3). With an average annual percent increase (AAPI) of 6.6% per year between 1996 and 2010 (compared to an AAPI of 2.5% between 1981 and 1996), incidence rates for thyroid cancer are increasing faster than those for any other cancer in women (1). Improvements in diagnostic technology account for only a portion of the observed increase (2, 4–6). Thus identifying new risk factors for thyroid cancer and understanding temporal changes in both established and new risk factors represent an increasingly important public health priority. At this time, the only well-established risk factors for the papillary (including the papillary/follicular variant) form of thyroid cancer (which comprises 80% of all thyroid cancer) are ionizing radiation, history of proliferative benign thyroid disease (BTD) (e.g., goiter and thyroid nodules), and family history of thyroid cancer or proliferative BTD (7–9). However, these exposures have relatively low prevalence in the US (8, 10). More recent studies have found obesity to increase risk (11–13) and the several years following a full-term pregnancy to be a period of high risk (14, 15). Little research has focused on immunologic correlates of thyroid carcinogenesis despite the fact that half of all autoimmune diseases in women involve the thyroid (16). These autoimmune diseases result from hyperactive cell-mediated immune responses against self tissue (17), and women with autoimmune diseases such as systemic lupus erythematosus (SLE), are at significantly higher risk of developing thyroid cancer (18).

Given the rapid increase, environmental exposures relevant to immune function represent an understudied set of possible risk factors for thyroid cancer. During the last century we have seen unprecedented increases in persons living in more sterile environments and less crowded housing conditions, resulting in diminished exposure to a diversity of microbes (19). This reduction in exposure to microbial-rich environments, especially when it occurs in early life, has been linked to hyperactive immune responses to allergens (e.g., atopic disease) in children and is thought to be detrimental to establishing the appropriate immune “calibration” or “priming” that may be needed for lifelong healthy immune function (19–22). This set of circumstances has been termed the “hygiene hypothesis.” Exposure to microbial-rich environments have been associated with lower risk of other cancers (e.g., daycare and childhood leukemia, occupational exposures to endotoxin-rich agricultural and textile environments and lung cancer (23, 24)) and it is plausible that similar exposures could be related to thyroid cancer development. Here we investigate the association between early life self-reported exposures to microbial-rich environments and papillary thyroid cancer risk in a large, prospective cohort of female California teachers.

The California Teachers Study (CTS) cohort, established in 1995–96, includes 133,479 active and retired female public school teachers, administrators, and other professionals (25). Participants initially completed a self-administered baseline questionnaire addressing health and medical history, lifestyle, diet, and other behaviors. The fourth follow-up questionnaire, completed in 2005–06, included questions on exposures related to microbial and infectious exposures throughout the lifespan. The study was approved by the Institutional Review Boards of the Cancer Prevention Institute of California, City of Hope National Medical Center, the University of Southern California, the University of California, Irvine, and the California Health and Human Services Agency.

The average age of women in the analytic cohort was 62 (± 0.1) years at the time they completed the 2005–06 questionnaire; 88% were non-Latina white, 29% were overweight (BMI 25–29.9 kg/m²), and 18% were obese (BMI >30 kg/m²). Only 1.4% reported having a family history of thyroid cancer in a first degree relative, but almost 15% had a family history of BTD, and 10% reported a personal history of benign thyroid disease. Women in the analytic cohort who were diagnosed with papillary thyroid cancer (n=100) were slightly younger, with an average age of 60 (± 10.4) years, more likely to be obese (31%), parous (80%), and live in areas of higher SES (56%).

Among those women eligible for the analytic cohort, 23% reported living in a rural area during early childhood (at age 5 years or younger) and 25% reported living within a half mile of barns or stables housing horses or other hoofed animals. Rural residence during childhood was associated with a 49% reduced risk of developing papillary thyroid cancer as an adult (HR=0.51, 95% CI: 0.28–0.94) while living near hoofed animals reduced risk by 53% (HR=0.47, 95% CI: 0.26–0.84; Table 1). The reduction in risk waned somewhat for exposure at age 12 years (HR=0.66, 95% CI: 0.37–1.17) and even more so for adult and recent exposure. These associations were independent of each other. Having an indoor dog or cat in early childhood was also associated with reduced risk (HR=0.43, 95% CI: 0.26–0.72) and was independent of the association for living near hoofed animals. Living in a rented home and sharing a bedroom were generally not associated with risk of papillary thyroid cancer risk at the ages assessed. We did not find a significant association with attending daycare or kindergarten during early childhood, but further assessment of age-specific exposures suggested increased risks for exposure at ages 6 months and 3 years.

To our knowledge, this is the first study addressing the association between papillary thyroid cancer development and events that could reflect immunologically-relevant microbial exposures during specific periods of life. We examined two types of factors culled from the literature regarding their association with atopic disease: those presumably reflecting exposure to a wide diversity of microbes and microbial byproducts and endotoxins (i.e., living within close proximity to hoofed animals, and with pets), and those presumably reflecting greater exposure to infections from other children (i.e., close contact with siblings or other children). We found that events reflecting early childhood exposures to diverse microbes, but not those reflecting earlier or greater childhood exposure to other children, were inversely associated with papillary thyroid cancer risk. Very early childhood exposure (age ages ≤ 3 years) to other children in daycare, however, increased subsequent thyroid cancer risk. We did not detect important associations with any of these exposures when they occurred during adulthood.

The ‘hygiene hypothesis’ suggests that insufficient microbial exposures in early life can adversely affect immune responsiveness in later life (22). This line of reasoning holds that the immune system evolved to expect early and intense exposure to a diversity of microbes but that the recent, dramatic modification of these exposures associated with the relatively sterile western environment, a reduction in crowded living conditions in modern society, and most recently, widespread use of antimicrobial products, especially in early life, may be detrimental to appropriate priming of immune responses needed for healthy immune function throughout the lifespan (19). It is now thought that the relevant range of microbes extends beyond those responsible for clinically apparent infections, to gastrointestinal microbiome colonization with commensal bacteria and exposures to sterile microbes and bacterial components, including endotoxins (20, 21). To our knowledge, data are lacking to corroborate the associations reported here, as little has been published regarding thyroid cancer risk among those living on farms or in rural areas (27), although occupational exposures to agricultural and textile environments rich in microbial byproducts reduce risk of lung cancer in adults (23), and daycare attendance and early life infections have been shown consistently to reduce risk of childhood lymphoid malignancies (24). However, we did not observe any similar, consistent protection from daycare or kindergarten attendance or crowded households, which suggests that the important exposures are animal-derived, perhaps reflecting a greater overall microbial diversity. Our findings are consistent with the notion that key early-life microbial exposures from animals, likely important to immune ‘calibration’, might be relevant to the etiology of papillary thyroid cancer. Whether microbial exposure patterns are related to worldwide trends of thyroid cancer incidence (4), especially its increasing incidence, is not yet known but warrants further exploration.

Strengths of this study include the novel hypothesis, the large and diverse CTS cohort, and complete and accurate outcome assessment based on linkage of the cohort to the statewide cancer registry and monitoring of residential address and mortality (25). An important limitation of this study is that the exposure questions were answered only by cohort members who completed the 10 year follow-up questionnaire. We opted to limit the analytic cohort to those women who completed this questionnaire and to carry out a prospective analysis, despite the relatively small number of cases even though the exposures of interest occurred during childhood. Alternative analyses including the full cohort and including both prevalent and incident thyroid cancers yielded very similar result s to those shown here. This secondary analysis is reassuring in terms of generalizability given that the prospective cohort is substantially older than the average population which develops thyroid cancer, i.e., women during their reproductive and early menopausal years, whereas the retrospective analysis includes women at a younger average age. The smaller number of cases available for stratified analyses also means that some of our observations are based on small numbers and limited our ability to test formally for statistical interactions. Having nested this study within the prospective cohort did allow us to compare characteristics of women who responded and those that did not respond to the 2005–06 questionnaire. This comparison showed that the two groups of women were generally similar, although responders were somewhat younger than non-responders at baseline (mean age 52 vs. 56 years, respectively). A related limitation is that some incomplete control for confounding may still be present, since while information on BMI was updated on the 2005–06 questionnaire, parity was not. However, given the older average age of the cohort, this may not be a substantial issue as the majority of the women were likely to have completed their childbearing prior to joining the cohort.

Unfortunately, our questionnaire included only a small subset of all possible markers of relevant early-life microbial exposures. Future studies should address a more detailed set of exposures to children, animals, farming and other relevant environments. Future studies should also address alternative exposures that could explain associations of thyroid cancer with early life environments including iodine intake and exposure to diagnostic X-rays or other medical radiation. Studies with sufficient numbers to mutually adjust exposures could potentially tease out their independent effects. We did not have prospective blood samples with which we could have measured antibody profiles to specific antigens. A final limitation is the potential for misclassification of self-reported early life exposures. While few data validating such self-reports are available, Nicholas et al. (28) did find that 18-year olds could recall pet ownership at ages 1 to 6 relatively well (between 77% and 89%) compared to parental reports from 12–18 years earlier, with errors towards underreporting. Svanes et al. also reported that adults age <45 years of age recalled childhood pet ownership with 81% to 86% accuracy (29).

In summary, we found that early childhood exposure to certain household pets and hoofed animals were associated with reduced risk of subsequently developing papillary thyroid cancer in women. This finding is consistent with the broader notion that diverse microbial inputs in early life may affect the priming or calibration of immune responses in later life, which in turn could affect the risk of cancer development. This new area of inquiry deserves further study as a novel pathway for understanding the rapid increases in thyroid cancer occurrence observed in many developed populations worldwide.