Relationship Between Physical Activity Levels and Thyroid Cancer Risk: A Prospective Cohort Study in Korea

Introduction

In 2020, thyroid cancer (TC) was ranked as the ninth leading type of incident cancer, with 586,000 new cases diagnosed worldwide according to the latest GLOBOCAN cancer statistics. ¹ The global incidence rate among women was threefold higher than that among men, which were 10.1 cases and 3.1 cases per 100,000 persons per year, respectively. ¹ In South Korea, TC was the most frequently diagnosed cancer among both sexes, with 40.2 new cases per 100,000 persons per year. ² Korean women have been reported to have an increased risk of TC compared with men, as the age-standardized incidence rates were 62.2 cases and 18.8 cases per 100,000 persons per year, respectively. ²

To date, the cause of TC is poorly understood. The only substantial evidence is that childhood exposure to ionizing radiation is a risk factor for TC, ³ although overweight, obesity, and radiation exposure may play roles in increasing its risk. ⁴ In contrast, the consumption of vegetables and fruits may protect against the development of TC, ⁵ and the consumption of iodine-rich food such as fish or shellfish might reduce TC risk in populations with iodine insufficiency. ^6,7 However, the association between physical activity and TC remains controversial.

The World Cancer Research Fund/American Institute for Cancer Research report defines physical activity as any movement using skeletal muscles and requiring more energy than resting. ⁸ Physical activity may affect several systems in the human body, including endocrinologic, immunologic, and metabolic processes, and is related to the risk of cancer development. ⁸ Recently, solid and consistent evidence has supported the protective role of increased physical activity in the progression of several types of cancer (esophageal adenocarcinoma, gastric cardia, colon, breast, endometrial, and bladder cancers). ⁹ While physical activity is reported that it has inversely associated with TC risk, ^10,11 some studies have revealed a positive correlation ¹² or a nonsignificant relationship between physical activity and TC. ^{13 –15}

Our aim was to determine the association between quantity of physical activity and the development of TC in a Korean population using prospectively collected cohort data.

Materials and Methods

Study population

The eligible participants in our study were properly selected from the Cancer Screenee Cohort at Korean National Cancer Center (KNCC). Details of the cohort were described elsewhere. ¹⁶ Briefly, the cohort data included 30,435 participants from 20 years of age who underwent evaluation of their health status at the Center for Cancer Prevention and Detection of the KNCC from June 2007 to December 2014. Data were included from participants who completed the baseline questionnaires and were followed until December 2019 for a new diagnosis of TC. The exclusion criteria were subjects with incomplete questionnaire data and those with a previous cancer diagnosis of any type. We selected 15,175 participants with available self-reported physical activity data, including the frequency (days per week) and duration (minutes per day) of their exercises in three intensity levels (walking, moderate, and vigorous) (Fig. 1). Written informed consent was obtained from all participants, and the study protocol was approved by the Institutional Review Board of the KNCC (No. NCCNCS-07-077).

OPEN IN VIEWER

FIG. 1.

Flowchart of collecting participants in the cohort study.

TC incidence was determined through linkage with the 2019 Korea National Cancer Incidence Database of the Korea Central Cancer Registry. Most patients were ultimately diagnosed with TC through a biopsy procedure. The International Statistical Classification of Disease and Related Health Problems 10th revision (ICD-10) (C73), was considered the criterion to identify TC cases in the KCCR database. Patients diagnosed with primary TC after enrollment were identified as incident cases.

Results

Out of 15,175 subjects, we identified 234 TC cases (Fig. 1). The median duration of follow-up was 9.5 years with interquartile range being 4.1 years, and the total person-years followed was 140,844.45. Table 1 shows the demographic characteristics of the study participants. Among the subjects, 5118 reported a low physical activity level, 4983 exercised at a moderate level, and 5074 exercised at a high level. Compared with participants with the lowest physical activity level, participants with higher levels were older (50.4 ± 9.0 vs. 51.6 ± 9.0 and 52.5 ± 8.6 years, p < 0.001), more likely to be current alcohol drinkers (52.8% vs. 54.9% and 59.7%, p < 0.001), and spent more time walking (75.5 ± 71.2 vs. 273.3 ± 171.8 and 585.8 ± 391.0 minutes/week, p < 0.001), as well as participating in moderate- (6.0 ± 21.2 vs. 76.5 ± 109.9 and 306.7 ± 325.4 minutes/week, p < 0.001) and vigorous-intensity activities (1.4 ± 7.8 vs. 38.8 ± 65.2 and 278.7 ± 293.2 minutes/week, p < 0.001).

Participants with higher physical activity levels also included a lower proportion of current smokers than those with the lowest physical activity level (14.9% and 15.6% vs. 18.8%, p < 0.001). Moreover, compared with subjects reporting lower physical activity levels, the most active individuals had a higher BMI (23.7 ± 3.1 and 23.6 ± 3.0 vs. 23.9 ± 2.8 kg/m², p < 0.001), lower education level (45.3% and 51.7% vs. 43.3%, p < 0.001), and spent less time sitting per week (2786.9 ± 1561.5 and 2564.3 ± 1411.2 vs. 2192.0 ± 1303.8 minutes/week, p < 0.001).

The relationship between physical activity level and TC incidence is shown in Table 2. A reduced risk of TC was observed among participants reporting the highest physical activity level (hazard ratio [HR] = 0.65 [confidence interval, CI = 0.44–0.94], p-trend = 0.028) than among participants reporting the lowest level. After stratification by sex, this inverse relationship remained among both sexes; however, it was only significant in women (HR = 0.63 [CI = 0.41–0.95], p-trend = 0.035).

Notably, 12,633 eligible participants had sufficient information on sitting behavior for exploratory analysis. A higher risk of TC was observed among subjects who spent more time sitting, and this trend was constant in the female subgroup but not in the male subgroup. However, this association was not statistically significant (Supplementary Table S1).

In this study, we also clarified the potential relationship between other lifestyle factors and TC risk. While subjects without FHTC were associated with a reduced TC incidence (HR = 0.47 [CI = 0.23–0.96]), a positive association was observed between nonsmokers and the risk of this malignant disease (HR = 3.11 [CI = 1.32–7.35]). However, other risk factors did not show any significant association (Supplementary Table S2). Therefore, in our secondary analysis, we investigated the association between physical activity and TC with stratification by potential confounders. Apart from FHTC and smoking status, we also considered BMI, household income status, and alcohol consumption as confounding factors because of their effects on TC incidence in previous studies. ^{4,19 –25}

For the participants with BMI ≥25 kg/m², the highest physical activity level was significantly associated with TC risk reduction compared with the subjects with the lowest level, but this difference was only observed in women (HR = 0.38 [CI = 0.16–0.93], p-trend = 0.034) (Table 3). Similarly, participants with a household income >4 million won/month reporting the highest physical activity level were found to have a reduced risk of TC (HR = 0.53 [CI = 0.30–0.94], p-trend = 0.034). However, no significant association was observed among participants with a lower monthly income (less than 4 million per month) (Table 4). There was also a significant association between active lifestyle and a reduction in the TC risk among nondrinkers (HR = 0.48 [CI = 0.26–0.88], p-trend = 0.018) (Supplementary Table S3), nonsmokers (HR = 0.61 [CI = 0.40–0.95], p-trend = 0.038) (Supplementary Table S4), and subjects without FHTC (HR = 0.66 [CI = 0.45–0.96], p-trend = 0.040) (Supplementary Table S5).

Discussion

The current study identified an inverse association between physical activity level and development of TC in Korean screening study participants. After stratification by confounding variables, the inverse association between physical activity and TC risk was significant among female subjects with BMI ≥25 kg/m², participants with a household income higher than 4 million won/month, without FHTC, nondrinkers, and nonsmokers.

Previous studies reported the effect of physical activity on the development of TC with inconsistent results. While some studies suggested that the risk of TC was reduced among people engaging in physical activity or daily walking, ^10,11 a study by Kim et al ¹² reported a positive relationship between a higher physical activity level and TC occurrence. However, since their study design was cross-sectional, it was impossible to conclude a temporal relationship of exercise and TC risk. ¹² A potential hypothesis is that after being diagnosed with TC, patients participate in physical activity to improve their health and combat the disease. ¹² No significant association was observed between exercise and the development of TC in studies by Kabat et al, ¹³ Leitzmann et al, ¹⁴ and Schmid et al. ¹⁵ Kabat et al ¹³ reported findings that were limited to postmenopausal women, which may limit generalizability to other groups. Furthermore, physical activity measurement in their study was restricted to recreational physical activity, ¹³ which may limit the statistical power to detect an association between total physical activity and TC.

Several biological mechanisms have already been introduced to underly this association, such as improved metabolic function, the prevention of chronic inflammation, ⁹ and an altered balance between reactive oxygen species and antioxidant defenses. ²⁶ First, physical activity has been demonstrated to protect against cancer progression through contributing to decrease insulin-like growth factor 1 (IGF-1) level. ²⁶ IGFs are mitogens with a critical effect on regulating cell proliferation, differentiation, and apoptosis. ²⁷ IGFs signal through their receptor, which induces cell transformation upon activation by tumor virus proteins and oncogene products. ²⁷ The role of IGFs in cancer has been replicated in several studies, which have shown that circulating IGF-1 level is positively related to the development of the most diagnosed cancers, including lung, colorectal, breast, and prostate cancers. ²⁷ Furthermore, physical activity is related to lower levels of plasma insulin, higher insulin sensitivity, and better glucose metabolism, even in high-risk individuals. ²⁶ Second, regular exercise is suggested to be an anti-inflammatory factor. ²⁸

This effect of physical activity might be regulated through reducing visceral fat mass and generating an anti-inflammatory environment. ²⁸ Clinical studies have reported that by reducing adiposity, physical activity decreases the accumulation of proinflammatory biomarkers and raises the levels of anti-inflammatory factors. ²⁸ In adipocytes present in adipose tissue, bioactive signaling molecules, such as proinflammatory adipokines and cytokines, are secreted that may lead to cancer progression. ⁹ For example, leptin is a hormone predominantly secreted by adipocytes that helps to regulate energy expenditure ²⁹ ; previous studies have revealed that high circulating leptin levels are associated with the development of cancers. ³⁰ Leptin is reported to affect tumor growth by modulating several pathways involved in proliferation, angiogenesis, cell survival, and cancer progression. ²⁹

Last, evidence from recent studies has indicated that physical activity enhances antioxidant defenses. ³¹ An active lifestyle could help enable a response to oxidative stress before damage of cellular structures. ²⁸ Furthermore, the relationship between active lifestyle and TC might be mediated by reducing obesity, which is a well-known risk factor for TC development. ^{32 –34}

In the current study, we further examined the relationship of physical activity and the risk of TC incidence after stratification by confounding variables, including BMI, FHTC, alcohol consumption, and smoking status. First, several studies have suggested that a higher BMI classification is a well-established risk factor for TC development. ^{4,10,19 –22} After controlling for this variable through stratification, we identified a significant role of physical activity in TC risk reduction in female subjects with BMI ≥25 kg/m². Overdiagnosis of TC is a critical issue due to cancer screening in Korea. It was reported that incidence of TC in 2011 was 15 times higher than that in 1993, ³⁵ while, its mortality rate was generally stable. We considered that a high income may be associated with a high TC screening rate. ²³

Recent Korean studies have reported associations of alcohol consumption, ^21,22,24,25 smoking status, ^21,22,25 and FHTC with the risk of developing this disease. ²¹ Thus, when evaluating the effect of exercises against TC occurrence, we stratified the study participants into groups according to these established confounders in the Cox proportional hazards regression models. Interestingly, the protective association was significant among subjects with the highest household income level, without FHTC, nondrinkers, and nonsmokers.

Our study has several strengths. Selection and recall biases were addressed because of a prospective design in our cohort study. We obtained sufficient information on almost all possible confounding variables and were able to control for their effects on the association by adjustment and stratification. TC cases were ascertained from a central cancer registry. However, our study has some limitations. First, there were many missing data. Second, when investigating the association between physical activity and the risk of TC stratified by the confounders, the small sample size of various strata led to a low statistical power to detect a significant relationship. Third, radiation exposure is a well-known risk factor for TC, but this information was not investigated among the subjects included in this study. The limitation may be a source of bias in estimating the association. Fourth, physical activity information was obtained by interviewing the study participants rather than by direct measurement by accelerometer.

Inconsistency in self-reporting of the subjects might lead to the under- or overestimation of physical activity levels. Fifth, new cases of TC in our study could not be verified from the medical records and charts. Information regarding the TC new cases was obtained from linkage with the KCCR database based on the ICD-O. Therefore, the TC subtypes of each patient were not classified, and false negatives may also occur in our data, which might lead to an underreported association between physical activity levels and the risk of TC. Finally, overdiagnosis of TC still occurred in Korea when the study was conducted. Some TC incident cases in our study may have resulted from overdiagnosis and may bias the findings.

Conclusion

The results of this study supported the hypothesis that a high level of physical activity is inversely associated with development of TC in the Korean population. This association was strengthened among female subjects with BMI ≥25 kg/m², participants with a household income ≥4 million won/month, without FHTC, and those who did not drink alcohol or smoke. Future prospective studies are needed to confirm our findings on the association between physical activity levels and the prevention of TC in other populations. Our results may be helpful for future planning and implementation of physical activity programs for TC prevention and control.