Obesity,Physical Activity,and Thyroid Cancer Risk: Disentangling True Associations from Detection Bias

Obesity is an established risk factor for many chronic diseases, including cancer. Evidence from epidemiological and experimental studies suggests that these associations are largely mediated by inflammation and metabolic and hormonal abnormalities, including hyperinsulinemia and increased circulating levels of sex steroid hormones and adipokines. ¹

Whether obesity contributes to the development of thyroid cancer has been a topic of debate, however. Although results from observational studies evaluating associations of body mass index (BMI) and thyroid cancer risk have been largely consistent, ^{2 –4} concerns have been raised about the potential for detection bias, occurring when the exposure of interest is also an indicator of more frequent interaction with the health care system and more diagnostic testing and head/neck imaging, ultimately leading to the detection of indolent disease. ⁵ Indeed, identifying “true” causes of thyroid cancer has been challenging in the current era of thyroid cancer overdiagnosis, responsible for an estimated 50–90% of cases in higher-income countries and the major factor thought to have contributed to rising thyroid cancer incidence rates. ^6,7

In 2016, an expert panel convening at the International Agency for Research on Cancer concluded that there was sufficient evidence in support of a causal association between excess adiposity and 13 types of cancer, including thyroid cancer. ⁸ The report cited findings published earlier that year from a large international pooled analysis of 22 cohort studies with approximately two million study subjects. ³ During follow-up, nearly 3000 of these individuals were diagnosed with thyroid cancer and 104 died from the disease. Baseline BMI was positively associated with thyroid cancer, with a 6% higher risk per 5 kg/m². Similar positive associations were observed for young-adult BMI (based on recalled height/weight), baseline waist circumference, and adulthood weight gain.

Associations were generally stronger in men than women, in cohorts based in the United States versus in other countries/regions, and for White compared with Black or Asian participants (apart from a stronger association of adulthood weight gain in Asians). Positive associations were observed for papillary, follicular, and anaplastic thyroid carcinomas, but not medullary carcinomas (thought to be etiologically distinct). Importantly, stronger associations were observed for thyroid cancer mortality than for thyroid cancer incidence, ruling out detection bias as a major influencing factor.

Findings from three recent publications in Thyroid add to this growing body of evidence in support of a causal association between excess adiposity and thyroid cancer.

Shin et al conducted a large pooled analysis of cohort study data from the Asia Cohort Consortium, including a total of 538,857 men and women across 13 cohorts in China, Korea, Japan, and Singapore. ⁹ During follow-up (mean = 15.1 years), 1132 incident thyroid cancers were diagnosed. A positive linear association was observed for BMI and thyroid cancer in men, with a 25% higher risk for every 5-kg/m² increase. A weaker nonlinear association was observed in women (7% higher risk per 5 kg/m²).

Risk of papillary thyroid cancer was slightly elevated among underweight compared with normal-weight men; however, this association was not seen in women. This study allowed for a more powerful and detailed investigation of these associations in Asian populations than was possible in the international pooled analysis, ³ which included mostly White participants. While information on thyroid cancer stage could have helped to distinguish thyroid cancer cases that were more versus less likely to have been detected incidentally, these data were unfortunately unavailable.

The second cohort study, by Bui et al, examined the association between physical activity and thyroid cancer risk in a cohort of 15,175 South Korean adults who were disease-free at baseline (2007–2014). ¹⁰ Responses to questions on frequency and duration of low, moderate, and vigorous intensity exercise were converted to metabolic equivalent of task (MET) minutes per week for each exercise type. Participants were classified into “low,” “moderate,” or “high” overall physical activity levels using tertiles of total MET minutes per week. During follow-up (median 9.5 years), 234 participants were diagnosed with thyroid cancer. In models adjusted for potential confounders (including BMI), high versus low physical activity level was associated with a 35% lower risk of thyroid cancer.

This association was observed in women and men and was stronger among participants who were overweight or obese, those with higher household income and no first-degree family history of thyroid cancer, and in nondrinkers and nonsmokers, providing reassurance against residual confounding by these factors. Although findings from prior studies were mostly null, ^{11 –14} these generally focused on vigorous-intensity recreational activity rather than total physical activity, and inaccuracies in self-reported physical activity could have driven associations toward the null. In contrast, high physical activity may simply be a predictor of incidental detection. Detection bias is a particular concern in studies conducted in South Korea, where an estimated 90% of thyroid cancer diagnoses are attributable to overdiagnosis. ⁶

The third study by Sørensen et al evaluated prepregnancy BMI and risk of differentiated thyroid cancer in a nationwide registry-based cohort in Denmark. ¹⁵ The study enrolled 443,403 women on the date of first live birth (median age = 30 years) and identified 463 incident thyroid cancers during follow-up. Overall, prepregnancy BMI was positively associated with differentiated thyroid cancer, with a 17% higher risk for every 5 kg/m². All models were adjusted for history of benign thyroid diseases and type 2 diabetes, among several other factors, and similar results were observed after excluding women with a history of benign thyroid disease, providing reassurance against confounding or detection bias.

The findings from this population of mostly younger-to-middle-aged adult women agree with results from the international pooled analysis. ³ Moreover, the distribution of cases by stage at diagnosis was similar across BMI categories, again suggesting that detection bias was unlikely to explain the findings. This latter observation concurs with a study by Zagzag et al, which found no difference in method of thyroid cancer detection for normal-weight, overweight, and obese patients. ¹⁶

Further advancements in our understanding of thyroid cancer etiology will require additional well-designed epidemiological studies with large numbers of participants, accurate exposure measures, and information to help address or reduce the potential for detection bias. Experimental studies are also needed to provide mechanistic insights. If obesity and physical inactivity could serve as effective targets in the primary prevention of thyroid cancer, such interventions could complement ongoing efforts aimed at avoiding thyroid cancer overdiagnosis and help to reduce the overall burden of this disease.