Associations between Physical Activity and Incidence of Cancer among Overweight Adults in Korea: Results from the Health Examinees-G Study

Physical activity is a well-known protective behavioral factor that reduces the risk of cancer (1). Among the different types of cancers, colorectal and breast cancers have been most actively studied in relation to physical activity (2, 3). Recent meta-analyses suggest that the risk of gastric (4), lung (5), prostate (6), bladder (7), ovarian (8), kidney (9), endometrial (10), esophageal (11), and pancreatic cancers (12), as well as glioma (13) and meningiomas (13) is reduced by physical activity. However, most of these studies have been conducted in Western countries, and the reports on the association between physical activity and cancer risk in the Asian population have been inconsistent (14, 15). The discrepancies may be due to the different levels and characteristics of physical activity studied or different sensitivities to other factors that affect the reduction of cancer risk via physical activity, such as obesity. Given that the growth rate of cancer incidence in the adult population is high in the Western Pacific region, particularly in East Asia (16), the association between physical activity and cancer incidence in the general Asian population warrants investigation.

Obesity is a well-known risk factor for various types of cancers (17, 18); it may increase the risk of cancer through several biologic pathways, such as chronic inflammation and microbiome alterations (19). As a modifiable factor, physical activity counteracts the mechanism by which obesity increases the risk of cancer (20, 21). In addition, the level of physical activity is reported to be low in obese individuals. Therefore, the risk of cancer may be reduced more significantly in the obese population than in the normal-weight group via physical activity (22). In fact, the effects of physical activity on several cancers were found to be attenuated after adjustment for body mass index (BMI; ref. 23). However, physical activity and obesity have no interactive effect regarding the risk of postmenopausal breast, prostate, colorectal, ovarian, and uterine cancer (24, 25). In addition, studies reporting on the protective effect of physical activity according to obesity status are scarce in Asian populations. Therefore, this study aimed to evaluate the effects of physical activity on the incidence of overall and type-specific cancers in the Korean adult population according to gender and obesity status. In addition, the characteristics of physical activity, such as frequency, intensity, and duration, may have different effects on cancer risk. For example, according to the recommendations for cancer prevention, adults should perform 150 to 300 minutes of moderate-intensity or 75 to 150 minutes of vigorous-intensity physical activity per week, as well as different types of physical activity, such as muscle-strengthening activities or aerobic exercise (26); therefore, we included different characteristics of physical activity in our analysis.

This longitudinal analysis was conducted within the Health Examinees-Gem (HEXA-G) study, which was derived from the Health Examinees study, a component of the Korean Genome and Epidemiology Study (KoGES). The characteristics of this cohort have been described previously (27, 28). Briefly, KoGES-HEXA is a large-scale cohort study that recruited participants ages 40 to 69 years between 2004 and 2013 (N = 173,202) from 38 health examination centers and hospitals across eight regions of Korea. Of these, 139,267 participants were included in the HEXA-G study after excluding participants who met the following criteria (Fig. 1): were recruited from centers that had only performed a pilot study, had undergone different procedures for quality control and biospecimen collection, participated for less than 2 years or did not participate in the cohort study (N = 31,309), and were aged < 40 years or > 69 years or were not the target age cohort (N = 2,626). Subsequently, 23,211 participants were excluded because cancer incidence and death data were not linked, and 3,948 participants were excluded because of a past cancer diagnosis before baseline. Finally, 112,108 participants were included in the analysis.

All the study participants voluntarily signed an informed consent form. This study was approved by the Institutional Review Board (IRB) of Seoul National University Hospital, (Seoul, Korea; IRB no. E-2110–004–1257). The study procedures were designed and conducted in accordance with the tenets of the Declaration of Helsinki

The Korea National Cancer Incidence Database (KNCI DB) is a national population–based cancer incidence database that contains information on patients diagnosed with cancer in hospitals across Korea. The Korea Central Cancer Registry (KCCR) compiled the KNCI DB and reported nationwide statistics from 1999 to 2018. The KCCR database provides information on cancer cases, including the date and site of the cancer diagnosis. Detailed information on the KCCR and KNCI DB has been provided elsewhere (29, 30). Completeness, an important indicator of data quality, was estimated to be 98.2% for the year 2017.

The KoGES-HEXA data were linked to the KCCR data using a resident registration number, which is a unique number assigned to individuals for identification purposes in Korea. The most commonly diagnosed cancers in Korea (31) were selected using the International Classification of Diseases (ICD) 10th Revision codes: gastric (C16), colon/rectum (C18–C21), lung (C33–C34), female breast (C50), prostate (C61), thyroid (C73), and all cancers combined (C00–C96). In addition, the incidence of the 13 obesity-related cancers suggested by a previous study (18) was evaluated and combined: esophageal adenocarcinoma (C15, M814–M857; ref. 32), gastric cardia (C160), colon/rectum (C18–C21), liver (C22), gallbladder (C23), pancreas (C25), postmenopausal breast (C50), corpus uteri (C549), ovary (C569), renal-cell carcinoma (C64), meningioma (C70–C72 and M9530–M9539; ref. 33), thyroid (C73), and multiple myeloma (C421). Because menopausal status at diagnosis was not available, patients with postmenopausal breast cancer were defined as those who were menopausal at baseline. Death was ascertained via linkage to mortality data from 1983 to 2019 from Korean Statistical Information Service 2021, Statistics Korea (accessed May 9, 2023. https://kosis.kr/eng/statisticsList/statisticsListIndex.do?menuId=M_01_01&vwcd=MT_ETITLE&parmTabId=M_01_01, RRID: SCR_023565).

Self-reported information on participation in and duration, intensity, types, and diversity of leisure-time physical activity (LTPA) was assessed at baseline (2004–2013). Participants were asked whether they regularly participated in LTPA to the point of sweating. Those who participated in LTPA were then asked, “How many times per week do you participate in regular exercise sufficient to cause sweating?” and “On average, for how many minutes per session did you participate in regular exercise?” The LTPA duration per week was calculated by multiplying the duration of each session by the frequency per week. This variable was used as the LTPA duration. Participants were then classified according to the recommendations of the International Physical Activity Guidelines (34): nonparticipants, participating for < 150 minutes per week, and participating for ≥ 150 minutes per week. Information regarding participation in each of the 39 types of common activities was collected and classified into six categories on the basis of previous studies (Supplementary Table S1) namely nonparticipants and participating in walking, climbing, aerobics, recreational activities, sports, and strength training (35, 36). Participants were not categorized exclusively by activity category as some participated in more than one activity. The intensity of each activity was defined as moderate [<6 metabolic equivalent of task (MET)] or vigorous (≥6 MET) based on the MET values of the 39 common activities (37). Subsequently, the participants were categorized as nonparticipants, participants in moderate-intensity LTPA only, and participants in vigorous-intensity LTPA. The diversity of physical activity was defined by the number of categories as follows: nonparticipants, participating in one activity, and participating in two or more activities.

The validity and reliability of the KoGES-HEXA physical activity questionnaire have been described previously (38, 39). In brief, LTPA from HEXA was positively correlated with MET-hours per day from the ActiGraph accelerometer (ρ = 0.343; P < 0.05), average energy expenditure per day from the multisensor (ρ = 0.272; P < 0.05), and the level of LTPA from the Global Physical Activity Questionnaire (GPAQ; ρ = 0.222; P < 0.05). Physical activity levels from HEXA also had moderate to high reliability for repeated measures at 3-month test–retest intervals (interclass correlation coefficient = 0.722).

All analyses were performed separately according to the gender of the participants. A Cox proportional hazards analysis with age as the time scale was used to estimate the HRs and 95% confidence intervals (CI) of the association of participation status, duration, intensity, type, and diversity of LTPA with overall, type-specific, and 13 obesity-related cancers. Our models were adjusted for education (≤middle school, high school, and ≥ college), household income in Korean currency (<2 million won, 2–3.9 million won, and ≥4 million won), marital status (living with a spouse or living alone), current occupation (office work, manual labor, and unemployed or housewife), smoking status (never, former, and current), alcohol consumption (never, former, and current), BMI (<18.5, 18.5–24.9, 25.0–29.9, and ≥30 kg/m²), average caloric intake (quartiles according to gender), and self-reported history of hypertension, diabetes, hyperlipidemia, and cardiovascular disease (CVD). Average caloric intake was calculated using a National Rural Nutrition Institute Food Composition Tables, 2012 developed by the National Rural Nutrition Institute (accessed May 9, 2023, http://koreanfood.rda.go.kr/eng/fctFoodSrchEng/engMain). In the analyses of overall cancer and breast cancer incidence in females, menopausal status was additionally adjusted at baseline. Postmenopausal females were defined as those who reported the absence of menstrual periods for > 12 months. In the analysis of gastric cancer, salt intake was additionally adjusted for in quartiles, according to gender, after a significant association between physical activity and sodium intake was confirmed in our study. A category for missing data was created and included in the model for all variables including covariates and physical activity. The rates of missing information for most variables were less than 5%, except for income, for which the rate was 7.2% for males and 10.1% for females. The person-time was calculated from the age at recruitment to the age at cancer diagnosis, death from any cause, or the end of follow-up (December 31, 2018), whichever occurred first. For type-specific cancer analyses, the cancer of interest was defined as the outcome and those diagnosed with other types of cancer were censored. To avoid potential reverse causation, patients diagnosed within the first year of follow-up were excluded. A stratified analysis was performed to assess the effect of differences in obesity status on the association between LTPA and cancer risk. BMI was used as an indicator of obesity. Normal weight and obesity were defined as a BMI ≤ and > 25.0 kg/m², respectively. Participants with a BMI < 18.5 were included in the normal-weight group because their proportion was less than 2%. In addition, the results were confirmed not to be affected when the normal-weight group was defined as those with a BMI between 18.5 and 25.0 kg/m². Because the proportion of participants with a BMI > 30.0 was too small to categorize separately (N = 1,049; 2.7%), the other group was defined as the overweight group with a BMI > 25.0 rather than the obese group.

A series of sensitivity analyses were also performed. First, in the analysis of overall cancer incidence, those diagnosed with cancer within the first 2 years of follow-up (N = 1,926) were excluded. Second, participants with a previous diagnosis of hypertension, diabetes, hyperlipidemia, or CVD at baseline were excluded to examine whether the results could be influenced by participants with poorer health (N = 69,440). In addition, because an over-diagnosis of thyroid cancer has been reported in Korea in the past decade (from 7.2 cases per 100,000 in 1994 to 51.0 per 100,000 individuals in 2016) compared with the United States (from 4.56 cases per 100,000 in 1975 to 14.42 per 100,000 individuals in 2013; refs. 40, 41), a sensitivity analysis was performed by excluding those diagnosed with thyroid cancer (N = 2,446).

The data generated in this study are available on the website of the NIH in Korea (https://nih.go.kr/ko/main/contents.do?menuNo=300566). The NIH permits any researcher who agrees to follow the research ethics to access these data.

Among 38,197 males and 72,820 females, 2,670 males (7.0%) and 4,069 females (5.6%) were diagnosed with cancer during the follow-up period after excluding those diagnosed within the first year of follow-up. The median follow-up duration was 9.12 years [interquartile range (IQR), 7.71–10.37]. The baseline characteristics of the study population are shown in Table 1. The mean and SD of age were 54.4 (8.4) years for males and 53.2 (7.8) years for females. Approximately 70% of males were former or current smokers, but 96% of females had never smoked. Regarding alcohol consumption, 73% of males were current consumers, but 65% of females never drank. Regarding both males and females, more than half of the participants were in the normal-weight group (BMI 18.5–24.9), and less than 3% of the population had a BMI < 18.5 or ≥ 30. Altogether, 57.4% of females were postmenopausal.

The analysis of males showed that participation in LTPA was associated with a 10% lower risk of overall cancer incidence in the crude model (HR, 1.10; 95% CI, 0.84–0.98; Table 2). In addition, participation in LTPA for >150 minutes per week, vigorous LTPA, walking, climbing, and LTPA diversity were associated with a lower risk of overall cancer incidence in the crude model. However, these associations were attenuated in the fully adjusted model, and no significant associations were observed. In females, participation in recreational activities was associated with a 17% higher risk of overall cancer incidence (HR, 1.17; 95% CI, 1.04–1.32). In the sensitivity analyses (Supplementary Tables S2–S4), the significantly increased risk in those who participated in recreational activities was attenuated when thyroid cancer incidence was excluded (Supplementary Table S3). None of the other associations showed significant changes in either males or females.

In the normal-weight group (Table 3), no significant associations were observed in males. In females, the increased risk of cancer incidence among those involved in recreational activities was attenuated when the incidence of thyroid cancer was excluded (Supplementary Table S5). These results did not change when the normal-weight group was defined as those with a BMI between 18.5 and 25.0 (Supplementary Tables S6–S7). In the overweight population (Table 3), participation in LTPA for < 150 minutes per week (HR, 0.76; 95% CI, 0.61–0.94) or > 150 minutes per week (HR, 0.87; 95% CI, 0.76–0.99), vigorous LTPA (HR, 0.84; 95% CI, 0.72–0.97), walking (HR, 0.84; 95% CI, 0.72–0.98), and participation in only one type of LTPA (HR, 0.82; 95% CI, 0.71–0.96) were associated with a lower risk of overall cancer incidence in males. These results did not change when thyroid cancer incidence was excluded (Supplementary Table S5). An increased risk of cancer was also seen among females who participated in recreational activities, but this association was not significant.

Participation in sports was associated with a lower risk of colorectal cancer (HR, 0.46; 95% CI, 1.06–1.40) in males (Table 4; Supplementary Table S8). The stratified analysis (Table 4) indicated that the risk of colorectal cancer was low in males in the normal-weight group who participated in sports (HR, 0.20; 95% CI, 0.05–0.82) and in more than two activities (HR, 0.63; 95% CI, 0.41–0.98), as well as in those in the overweight group who participated in climbing activities (HR, 0.61; 95% CI, 0.37–1.00). Regarding gastric cancer (Supplementary Tables S9–S10), participation in vigorous LTPA (HR, 1.32; 95% CI, 1.02–1.71), climbing (HR, 1.47; 95% CI, 1.10–1.98), recreation (HR, 1.61; 95% CI, 1.11–2.32), and only one type of LTPA (HR, 1.28; 95% CI, 1.01–1.63) were associated with a higher risk of gastric cancer in females (Supplementary Table S9). These significant associations were stronger in females with a normal BMI (Supplementary Table S10). Considering lung cancer (Supplementary Tables S11–S12), several LTPA characteristics were associated with a lower risk of lung cancer in the crude model. However, these significant associations were attenuated when the full set of covariates was included in the model for both normal-weight and overweight males. For breast and prostate cancers (Supplementary Tables S13–S16), most of the results were not significant.

Participation in vigorous-intensity activities (HR, 0.85; 95% CI, 0.72–0.99), recreation (HR, 0.69; 95% CI, 0.53–0.92), and sports (HR, 0.63; 95% CI, 0.44–0.88) and diversity of LTPA (HR, 0.81; 95% CI, 0.68–0.96) were associated with a lower risk of obesity-related cancers in males (Supplementary Table S17). In the stratified analysis (Table 5), participation in LTPA for >150 minutes per week (HR, 0.81; 95% CI, 0.66–0.99), vigorous LTPA (HR, 0.74; 95% CI, 0.58–0.94), climbing (HR, 0.73; 95% CI, 0.54–0.99), aerobics (HR, 0.73; 95% CI, 0.54–0.99), recreation (HR, 0.52; 95% CI, 0.34–0.80), and strength training (HR, 0.71; 95% CI, 0.51–0.98) and participation in more than two types of LTPA (HR, 0.71; 95% CI, 0.55–0.92) were associated with a lower risk of obesity-related cancer incidence in overweight males.

In this population-based study of adults ages 40 to 69 years, LTPA was not associated with overall cancer incidence in males. However, the duration, intensity, type, and diversity of LTPA were associated with a lower risk of overall cancer incidence in overweight males. In females, although an increased risk of cancer was found in those performing recreational activities, it was attenuated when those diagnosed with thyroid cancer were excluded from the analysis in the total and overweight groups. In terms of type-specific cancer risk, different LTPA characteristics were associated with a lower risk of colorectal cancer in overweight males. LTPA increased the risk of gastric cancer in the normal-weight female group. No significant association was observed for lung cancer after adjusting for the full set of covariates. Regarding obesity-related cancers, most LTPA was associated with a lower risk of incidence in males.

Existing literature reporting a reduced risk of cancer from physical activity is mostly derived from studies conducted in Western countries. In previous umbrella reviews of cancer risk in 22 major anatomic sites (15), a subgroup analysis by location showed that the number of studies from Asian countries was insufficient for combination or had a low level of evidence. One reason why physical activity and cancer are understudied in Asia is publication bias due to negative or nonsignificant results (42). Our study showed that physical activity was not significantly associated with overall cancer incidence and that associations varied depending on the type of cancer. This is consistent with a previous study in Asia, which reported insignificant associations (43). Therefore, many studies conducted in Asia may not have been reported because they had negative or insignificant results. A systematic review is needed to evaluate publication bias in Asian studies, and it is important to report the associations observed in studies from Asia even if they are insignificant.

In this study, physical activity was associated with a reduced risk of overall cancer incidence in males, which was attenuated after adjusting for multiple covariates. This phenomenon was even clearer for lung cancer, and thus, the presence of confounding factors that determine cancer risk more strongly than physical activity can be concluded. Smoking may be one of these factors in patients with lung cancer. A recent Mendelian randomized study reported that the direct causality between physical activity and lung cancer was not significant (44). In addition, the effect of physical activity on the risk of lung cancer is reported to be modified according to smoking status (45). Future studies should confirm this finding by comparing the effects of various risk factors for the development of cancer in Asian populations with those of physical activity. In females, the risk of overall cancer incidence in those performing physical activity was attenuated after excluding thyroid cancer. In Korea, the overestimation of thyroid cancer incidence is highly probable (40). This phenomenon combined with evidence that those who participate in physical activity are more likely to participate in health-related activities such as screening (46) indicates that the risk of thyroid cancer may be overestimated in those who participate in physical activity, which may have increased the risk of overall cancer incidence.

In a previous study reporting population-attributable fractions (PAF) for cancer incidence related to excess body weight in Korea, the overall PAF for excess body weight in Korea was lower than that in the United States and the United Kingdom (47). The authors attributed these results to the lower prevalence of obesity in Korea. In addition, the PAF for low LTPA was very low in Korea (0.1% and 1.4% of the overall cancer incidence in males and females, respectively). Considering this evidence together with the fact that the protective effects of physical activity against cancer risk could be partially explained by specific aspects of reducing abdominal adiposity (48), physical activity may have insignificant effects on cancer incidence in the general Asian population. In this study, the participants were stratified by obesity status to determine the effects of physical activity on cancer risk, particularly in the overweight population, rather than simply evaluating the associations in the general population. In addition, the observed associations were strengthened for obesity-related cancers. Therefore, we suggest that physical activity may have a protective effect regarding cancer incidence, especially in the overweight population, which is more susceptible to cancer than the general population.

In our study, we consistently found that physical activity could reduce the risk of colorectal cancer and that climbing was particularly effective in preventing colorectal cancer in overweight males. Previous studies have clearly shown that physical activity is more strongly associated with colorectal cancer than other cancers (15, 49, 50). Gender-based differences in the risk of colorectal cancer associated with physical activity have also been reported. For example, in a multi-ethnic cohort study, reduced risk of colorectal cancer with physical activity was found only in males (51–53). In Korea, the PAF of physical activity for colorectal cancer was reported to be higher in males than in females (54). Considering that colorectal cancer is strongly affected by obesity (54), physical activity may be an effective strategy to suppress or even reduce the increased risk of colorectal cancer due to obesity in Asian populations. Compared with other types of cancer, the association between gastric cancer and physical activity has been studied relatively extensively in Asia, and a “suggestive” level of protective effect has been reported in previous umbrella reviews (15). In this study, conflicting results were observed, as physical activity increased the risk of gastric cancer in normal-weight females. In Korea, gastric cancer is reported to have the highest incidence (31), and sodium intake and Helicobacter pylori infection have been suggested as representative risk factors for gastric cancer (55, 56). We included salt intake in the model after confirming that those who participated in physical activity for >150 minutes per week had a higher level of sodium intake than the inactive group. However, information on Helicobacter pylori infection was not available in this study and we could not control for its confounding effect in the analysis. More reliable results may be obtained if information regarding Helicobacter pylori is included in future studies.

The present study has several limitations. First, LTPA variables were collected using a self-reported questionnaire. Although the validity and reliability of the physical activity questionnaire were confirmed by comparison with objectively measured physical activity and 3-month test–retest measures the possibility of overestimation of physical activity remains. Second, the unmeasured covariates may have had residual confounding effects. For example, we could not adjust for information regarding Helicobacter pylori infection in the analysis of gastric cancer. Further studies with a longitudinal design addressing important confounders are required to obtain more reliable associations. In addition, although we examined the effects of physical activity in the overweight group, we could not determine whether those who participated in LTPA lost enough weight to prevent cancer. If this factor is considered, the level of physical activity that the overweight group should perform to overcome cancer risk could be more accurately determined. Due to the lack of incidence of cancer, it was not possible to evaluate the role of physical activity in carcinomas that have been studied extensively, such as esophageal, bladder, ovarian, and endometrial cancer. Because most studies on these carcinomas have been conducted in Western countries, it is essential to perform more research in Asia. In addition, because our study participants were ≥ 40 years of age, the participation rate in some categories of LTPA, such as team-based sports, was expected to be low. This could be confirmed by future studies with other age groups or life-course analyses. Finally, more than 20,000 participants were excluded because national cancer and death data were not available. Although the included study population had a higher proportion of males and those with higher levels of education and income than the excluded population (Supplementary Table S18), the distribution of BMI and history of chronic diseases did not differ significantly between the two groups. Therefore, we concluded that the possibility of selection bias that could change the results of the present study is low.

In this population-based cohort study in Asia, we found that LTPA was associated with a lower risk of overall cancer incidence in overweight males and a lower risk of colorectal cancer in both normal-weight and overweight males. In addition, we suggest that different characteristics of physical activity have different associations with cancer risk, depending on the type of cancer assessed. Important considerations for the risk assessment of cancer types in Asia, such as the overestimation of thyroid cancer, are also described for the female population. In terms of public health recommendations, our findings suggest that participation in physical activity of moderate duration and intensity can successfully reduce the risk of cancer in overweight males. These findings suggest the need for greater public awareness of the protective effects of physical activity and future consortia studies with adequate power and more detailed information on normal and overweight populations. 

No disclosures were reported.

J. Choi: Conceptualization, data curation, software, formal analysis, investigation, visualization, methodology, and writing–original draft. J. Park: Formal analysis, visualization, methodology, writing–review, and editing. J.-E. Kim: Formal analysis, visualization, methodology, writing–review, and editing. M. Lee: Conceptualization, visualization, methodology, writing–review, and editing. D. Kang: Conceptualization, resources, visualization, methodology, writing–review, and editing. A. Shin: Conceptualization, resources, visualization, methodology, writing–review, and editing. J.-Y. Choi: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, investigation, visualization, methodology, and writing–original draft.

The authors would like to thank all the participants of this study. We also thank the other investigators, staff, and participants of the KoGES study for their valuable contributions. This work was supported by grants from Korea Centers for Disease Control and Prevention (2004-E71004–00, 2005-E71011–00, 2005-E71009–00, 2006-E71001–00, 2006-E71004–00, 2006-E71010–00, 2006-E71003–00, 2007-E71004–00, 2007-E71006–00, 2008-E71006–00, 2008-E71008–00, 2009-E71009–00, 2010-E71006–00, 2011-E71006–00, 2012-E71001–00, 2013-E71009–00, 2018-P7106–00, and 2021-E0608–00; to J. Choi, A. Shin, and D. Kang); Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2021R1I1A1A01052952 and NRF-2022R1A2B5B01002471; to J. Choi).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.