Association Between Metabolic Syndrome and Risk of Esophageal Cancer: a Nationwide Population-Based Study

Esophageal cancer is the seventh most common cancer with more than 600,000 new cases each year (1), and it ranks sixth in terms of cancer-related death. Over 540,000 individuals die from esophageal cancer annually (1). The incidence of esophageal cancer has regional variation with the highest rates and mortality observed in Eastern Asia (2). Although the treatment of esophageal cancer has improved, the survival rate is still low. The 5-year survival rate in Korea was estimated to be below 40% in 2009 to 2013 (3). Therefore, esophageal cancer is an important public health concern.

Efforts have been made to find modifiable risk factors for esophageal cancer. However, the factors have not been identified sufficiently and there are differential associations between risk factors and esophageal cancer by histologic subtypes (esophageal squamous cell carcinoma, ESCC, and esophageal adenocarcinoma, EAC). Smoking, alcohol consumption, and low vegetable/fruit intake are well-established modifiable risk factors of ESCC (4). However, in a study from China where nearly all cases of esophageal cancer are ESCC, the three factors together accounted for only 46% of esophageal cancer by population attributable fraction (5). In EAC, established modifiable risk factors include obesity, gastroesophageal reflux, and smoking (4). In a study conducted in Australia, these factors accounted for 76% of EAC cases (6). Therefore, other factors might explain the increased risk of esophageal cancer, especially ESCC.

Recently, the role of metabolic syndrome (MetS) in esophageal cancer has gained interest. However, there have been only a few studies on this topic. For EAC, one case-control study (7) and one cohort study (8) showed positive association between metabolic syndrome and EAC risk, but another case-control study (9) and two other cohort studies (10, 11) showed null association. In regard of ESCC, only two prospective cohort studies included ESCC risks in MetS (8, 11) and they did not show a significant association. However, both studies were conducted in European countries where ESCC is not prevalent, and the number of ESCC cases was small; only 184 and 64 cases. Previous studies are summarized in Supplementary Table S1.

Worldwide, more than 80% of esophageal cancer is ESCC (12). However, subtypes of esophageal cancer show high regional variation. Whereas EAC is prevalent in Western countries, ESCC is much more prevalent than EAC in Asian countries. About 80% of the ESCC cases occur in Central and South-East Asia, the so-called ‘Asian Esophageal Cancer Belt’ (12, 13). More than 90% of esophageal cancer in China (13) and Korea (3) is ESCC. However, the relationship between MetS and esophageal cancer in Asia is unclear.

Thus, this study aimed to investigate the association between MetS and risk of esophageal cancer in Korea where ESCC accounts for the majority of esophageal cancer.

This study analyzed the Korean National Health Insurance Service (NHIS) data, which includes the health examinations and claims data of the Korean population. The Korean NHIS provides medical insurance to 97%, and subsidizes medical expenses to 3% of the population with lowest income. The NHIS database contains data for identifying insurance qualification (demographics such as age and sex, and income status) and diagnosis codes belonging to the International Classification of Disease 10^th revision (ICD-10) (14). The NHIS also offers regular health examinations to employees and people over 40 years old every other years (15). Many epidemiological studies used the NHIS database, and database details can be found elsewhere (14, 15).

For this study, we included participants who received the health examination provided by the NHIS in 2009 (index year). We excluded participants under 40 years old (n = 3,345,043) considering the eligibility for health check-up program and very low incidence of esophageal cancer under this age of 40 (3, 16). We also excluded participants without data on the variables analyzed (n = 268,300), with cancer diagnosis before the index date (n = 153,458), and participants who were diagnosed with esophageal cancer or died within 1 year of the lag period after the index date (n = 23,305). Finally, 6,795,738 participants were included in this study (Fig. 1). This study received approval from the Institutional Review Board at Samsung Medical Center (IRB File No. SMC 2021–11–003), which waived the requirement for informed consent because we used deidentified data (Fig. 1).

MetS was defined by the harmonized criteria proposed by a joint scientific committee (17). MetS was defined as 3 or more of the following : (i) serum triglycerides ≥ 150 mg/dL or patients with dyslipidemia, (ii) serum high-density lipoprotein (HDL) cholesterol < 40 mg/dL and < 50 mg/dL for men and women, respectively, or patients with dyslipidemia, (iii) systolic blood pressure (SBP) ≥ 130 mm Hg, diastolic blood pressure (DBP) ≥ 85 mm Hg, or patients with hypertension, (iv) fasting blood glucose (FBG) ≥ 100 mg/dL or patients with diabetes, and (v) waist circumference ≥ 90 cm for men and ≥ 85 cm for women. For defining abdominal obesity, the criteria of the Korean Society for the Study of Obesity (18) were used. Dyslipidemia was defined through identification of at least 1 claim per year under the ICD-10 code E78 for prescribing lipid-lowering medication or total cholesterol ≥ 240 mg/dL. Hypertension was when antihypertensive medication was prescribed with ICD-10 code of I10–13, I15, or either SBP ≥ 140 mm Hg or DBP ≥ 90 mm Hg. Diabetes was defined as a prescription of antidiabetic medication with ICD-10 code of E11–14 or FBG ≥ 126 mg/dL

Anthropometric measurements and laboratory tests were conducted at the NHIS health examinations in the index year (2009), following the screening protocol. The waist circumference was measured in between the bottom edge of the last rib and the top border of the iliac crest located at the midaxillary line. For blood pressure measurements, a trained clinician measured the brachial blood pressure after the participants were seated for 5 minutes. Blood samples for FBG, triglyceride, and HDL levels were obtained from the participants after an overnight fast.

In addition, although the body mass index (BMI) is not a component of MetS, it is an important factor of metabolic diseases. Thus, we also included BMI in the analysis. BMI was computed by the weight (kg) divided by the square of height (m) measured at the index year health examination. BMI was categorized into underweight (<18.5), normal (18.5–22.9), overweight (23–24.9), and obese (>24.9), according to the World Health Organization Asia-Pacific criteria for classifying obesity, endorsed by the Korean Society for the Study of Obesity (refs. 18, 19; Table 1)

The primary endpoint was incident esophageal cancer. It was identified through the diagnosis code for esophageal cancer (ICD-10 C15) registered on the Korean national program for copayment of critical illnesses. Through this program, a person diagnosed with cancer pays for only 5% of the work-up and treatment fees for cancer (vs. 20%–30% for other relatively common diseases). Therefore, almost all patients with cancer are registered in the national copayment program. For this reason, cancer incidence is rarely omitted from the claims database in Korea, making it sufficiently reliable (20). We followed the study population beginning at 1 year after the index date (lag period) to the date of new esophageal cancer diagnosis, death, or until December 31, 2019, whichever came first.

Information on lifestyle factors including smoking, alcohol consumption, and physical activity were retrieved from questionnaires performed at the index year health checkup. Smoking history was categorized into non-smoker, ex-smoker, and current smoker. Based on average amount of alcohol per day, alcohol consumption was categorized as: nondrinking, mild-to-moderate drinking (average of <30 g/day), and heavy drinking (average ≥30 g/day; ref. 21). Regular physical activity was defined when participants performed moderate level activity for over 30 minutes ≥ 5 times per week, or strenuous level activity for over 20 minutes ≥ 3 times per week (22). Income levels were divided into quartiles, and the participants receiving medical aid (∼3% of the population) were integrated into the lowest quartile. For multivariable analyses, sex, smoking, alcohol consumption, physical activity, and income level were analyzed as categorical variables (as shown in Table 1), and BMI was entered as a continuous variable.

We performed descriptive statistics to analyze the baseline characteristics of the study population. The incidence rate of esophageal cancer was calculated by the number of incident cases divided by the total follow-up duration, and presented per 100,000 person-years. The incidence probability of esophageal cancer by MetS and its components (increased waist circumference, blood pressure, triglycerides, and FBG, and reduced HDL cholesterol) was obtained using Kaplan–Meier curves; the log-rank test was used to analyze differences between the groups. Multivariable Cox proportional hazard models were used to calculate hazard ratios (HR) and 95% confidence intervals (CI) for esophageal cancer incidence, according to baseline MetS and its components. After a review of the literature, potential confounders were identified a priori and included age, sex, income level, baseline BMI, and health behaviors (smoking, drinking, and physical activity; refs. 8, 11).

Because smoking and alcohol are well-recognized risk factors for esophageal cancer, especially ESCC, we also performed subgroup analyses of the nonsmoker/nondrinker population, to explore the relationship between MetS and esophageal cancer risk without the potential confounding effect of smoking and alcohol. Potential effect modification was evaluated for age, sex, smoking, alcohol, physical activity, BMI, and cardiometabolic comorbidities (any of hypertension, diabetes, and dyslipidemia). Statistical analyses were conducted with SAS version 9.4 (SAS Institute Inc.) and the P values presented are two-sided, with the level of significance at 0.05.

Among the 6,795,738 total study population, 2,124,394 had MetS at baseline and 6,414 had esophageal cancer that occurred during a mean (± SD) follow-up of 8.2 (± 1.1) years. Smoking, heavy drinking, diabetes mellitus, and hypertension were more prevalent in the MetS group at baseline than the non-MetS group and those who developed esophageal cancer than those who did not. Subjects with esophageal cancer had a higher waist circumference and lower BMI than subjects without esophageal cancer (Table 1). Baseline characteristics of non-smoker/non-drinker population (n = 3,222,812) are described in Supplementary Table S2.

Compared with individuals without MetS, those with MetS had a higher risk of esophageal cancer in the unadjusted model (HR, 1.24; 95% CI, 1.18–1.30). Also, in the multivariable-adjusted model, the esophageal cancer risk was higher in the MetS group [adjusted HR (aHR), 1.11; 95% CI, 1.05–1.18]. Among each component of MetS, elevated waist circumference (aHR, 1.24; 95% CI, 1.16–1.33), high blood pressure (aHR, 1.29; 95% CI, 1.22–1.37), and high FBG (aHR, 1.16; 95% CI, 1.11–1.22) were associated with increased esophageal cancer risk. HDL and triglyceride levels did not show a significant association (Table 2; Fig. 2).

For BMI category, compared to the normal BMI group, the underweight group had a higher risk of esophageal cancer (aHR, 1.45; 95% CI, 1.29–1.62). On the contrary, overweight (aHR, 0.75; 95% CI, 0.70–0.79) and obese (aHR, 0.65; 95% CI, 0.61–0.69) groups showed lower risk for esophageal cancer than the normal BMI group (Table 2).

In the nonsmoker/nondrinker population, the association between MetS and esophageal cancer was attenuated and was no longer significant in the multivariable-adjusted model (aHR, 1.06; 95% CI, 0.93–1.21). However, the pattern of the associations among the individual components remained persistent: elevated waist circumference (aHR, 1.26; 95% CI, 1.07–1.49), elevated blood pressure (aHR, 1.14; 95% CI, 1.00–1.30), and high FBG (aHR, 1.18; 95% CI, 1.04–1.33) were associated with increased risk of esophageal cancer; while reduced high-density lipoprotein cholesterol was associated with lower risk (aHR, 0.87; 95% CI, 0.76–0.99) in the multivariable-adjusted model. Also among the nonsmoker/nondrinker population, the underweight group had a higher risk of esophageal cancer than the normal BMI group (aHR, 1.53; 95% CI, 1.15–2.04). The obese group also showed a lower risk for esophageal cancer than the normal group (aHR, 0.82; 95% CI, 0.71–0.95; Table 3).

In stratified analyses, smoking and alcohol consumption affected the association between MetS and esophageal cancer risk. Regarding smoking, the association between MetS and esophageal cancer was slightly stronger in nonsmokers (aHR, 1.13; 95% CI, 1.03–1.25) than past smokers (aHR, 1.08; 95% CI, 0.96–1.20, P_interaction = 0.02), while the trend for a positive association was persistent across the smoking groups. For alcohol consumption, the association was prominent for mild-to-moderate drinkers (aHR, 1.23; 95% CI, 1.13–1.34), while no significant association was found for nondrinkers or heavy drinkers (P_interaction = 0.01). There were no significant interactions by age, sex, BMI, and cardiometabolic comorbidities (Fig. 3).

To the best of our knowledge, this is the first study to investigate the relation between MetS and esophageal cancer risk in Asia, where ESCC is predominant. In this study, MetS and its characteristic components – waist circumference, blood pressure, and FBG – were associated with increased esophageal cancer risk.

In our study of around 6,000 cases of esophageal cancer, patients with MetS had an 11% higher risk of esophageal cancer compared to those without MetS. Although we did not have information on histologic subtypes, our findings suggest that MetS may be associated with increased risk of ESCC because EAC is rare in Korea (3). Although previous studies in European countries concluded that MetS is not associated with ESCC, the HR estimates were generally similar to our study: HRs (95% CI) were 1.09 (0.87–1.36; ref. 8) and 1.08 (0.64–1.83; ref. 11). Therefore, the previous studies showed a slightly positive association. However, we speculate that it was not significant because of the low statistical power of an insufficient number of cases. In contrast, our study had around 6,000 estimated cases of ESCC and had sufficient power, so the relationship has become clearer.

Recently, the presence of MetS and its components showed an association with many types of cancers, especially adenocarcinomas – liver, colorectal, endometrial, pancreatic, postmenopausal breast, gastric, kidney, and prostate, among others (23–25). The association between MetS and the occurrence of cancer may be explained by a common biochemical environment. Visceral adiposity, hyperinsulinemia, hyperglycemia, IGF-I pathway, estrogen signaling, and inflammation, all of which are associated with MetS and also could affect carcinogenesis and cancer progression (25). These mechanisms might also explain the association between MetS and EAC in previous studies (7–9).

Fewer studies have been conducted on the relationship between MetS and squamous cell carcinomas. However, some recent studies suggest an association between metabolic factors and squamous cell carcinomas – head and neck, cervix, skin cancer, etc (26–29). Although the mechanism still needs to be elucidated, some studies suggest potential mechanisms between metabolic factors and squamous cell carcinomas. In a recent experiment with mice, obese/diabetic mice were more susceptible to ESCC than control mice (30). In addition, several proinflammatory cytokines were higher in obese/diabetic mice than in control mice. Thus, inflammation was suggested to be a mechanism of higher ESCC occurrence in obese/diabetic mice. In another study on head and neck squamous cell carcinoma (HNSCC), prolonged glucose incubation in HNSCC cells induced cell malignancy (29). Excessive glucose exposure was considered to modulate cell growth and inhibit cell differentiation. Our analyses of individual components of MetS also showed that visceral obesity and prediabetes/diabetes was associated with an increased risk of esophageal cancer, which may be consistent with such mechanisms.

Another possible mechanism is related with human papillomavirus (HPV) infection. HPV infection increases the risk of cervical squamous cell carcinoma. Although the study results are still inconsistent, a relationship between HPV infection and ESCC has also been suggested (4, 31). In cervical cancer, metabolic risk factors – insulin, insulin-like growth factors, and obesity – were thought to promote HPV infection and persistence (27). Although further studies are needed, it is possible that MetS may increase persistent HPV infection, and therefore increase the risk of ESCC. In the pathogenesis of ESCC, repeated insults to the normal esophageal epithelium result in squamous hyperplasia. With the accumulation of genetic and molecular changes (p53, CDKN2A, and RB), squamous hyperplasia precedes low-grade and high-grade squamous dysplasia, which develops into invasive cancer (32). Proinflammatory cytokines and glucose in MetS may promote this progression. Persistent HPV infection may also exacerbate this process.

In our study, high blood pressure/hypertension was associated with an increased risk of esophageal cancer. This is consistent with a previous study, which showed that hypertension is associated with an increased risk of oral, laryngeal, and esophageal cancer, of which the histology is mainly squamous cell carcinoma (33). A European study also proposed an association between hypertension and ESCC (8). It is suspected that some antihypertensive agents (angiotensin‐converting enzyme inhibitors, calcium channel blockers, beta-blockers, or diuretics) may promote carcinogenesis (34). However, prehypertension was also associated with an increased risk of esophageal cancer, and hypertensive patients with medication had a lower risk than hypertensive patients without medication. In addition, a 10 mm Hg increase of systolic and diastolic pressure was associated with a 7% and 10% higher risk of esophageal cancer in the previous study (33). Therefore, it is also possible that high blood pressure is associated with ESCC.

In our study, neither HDL nor triglyceride was associated with esophageal cancer risk. Although studies on the relationship between lipid profile and esophageal cancer are limited, the results are inconsistent. One study found that dietary cholesterol intake increased the risk of both subtypes of esophageal cancer (35). In another study, triglyceride was associated with an increased risk of esophageal cancer in the low glucose level group, and the relation was not significant in the high glucose level group (36). Meanwhile, it has been suggested that statins lower the esophageal cancer risk (37). HDL and triglycerides may have less effect on the esophageal cancer risk than other MetS components, or statins may offset the effects of HDL and triglycerides. Further research is needed on this topic.

In our study, BMI and esophageal cancer risk showed inverse association, consistent with previous studies (8, 38). The positive association with increasing waist circumference and the negative association with increasing BMI shown in our study are intriguing findings, but the reason is yet unclear. The mechanism underlying the inverse association between increasing BMI and esophageal cancer risk is uncertain. Potential explanations include micronutrient deficiencies or malnutrition in the underweight group, which can aggravate cancer occurrence; low level consumption of fruits and vegetables and deficiency of selenium, zinc or vitamin E are important risk factors of ESCC (4).

Our subgroup analysis of the non-smoker/non-drinker population showed a somewhat attenuated risk, but the association by individual MetS components remained consistent. This further suggests that MetS and its individual components are independently associated with esophageal cancer risk, even when the confounding effects of smoking and alcohol consumption are excluded.

Stratified analysis showed that smoking and alcohol consumption affected the association between MetS and esophageal cancer risk, suggesting more prominent associations in non-smokers and mild-to-moderate drinkers. While smoking and alcohol are well known risk factors for ESCC, their potential interaction with metabolic syndrome in developing ESCC has not been studied yet. Therefore, the reason for these findings is not clear, and needs further investigation.

Although further research is warranted, this study implies that efforts to lower the MetS and its components might contribute to the prevention of esophageal cancer, including ESCC. Lifestyle modification – healthy diet, regular physical activity, low sodium intake, as well as strict management of high blood pressure, and diabetes mellitus – might be helpful in the prevention of esophageal cancer as well as cardiovascular disease and other cancers (25). However, future study is needed to determine whether these efforts are effective in preventing esophageal cancer.

Although this study has strengths including the use of a large nationwide database, it has some limitations. First, this was an observational study; therefore, the association may not be interpreted as a causal relationship. However, to minimize reverse causality, we excluded participants diagnosed with esophageal cancer within 1 year after the index date. Second, we used Korean data, therefore the results are not generalizable to other populations. Third, we could not include information on the mechanism of carcinogenesis by MetS. In the future, studies that include inflammatory markers or HPV infection are needed. Fourth, our data is based on claims data and does not include information on the histologic subtypes. However, according to a previous study, ESCC was over 90% and EAC was only 3.1% among esophageal cancers in Korea; therefore, our results can be largely interpreted as the association between MetS and ESCC (3). Fifth, this study using the NHIS health screening questionnaire did not include information on diet, although some dietary factors such as meat consumption (39) or dairy consumption (40) could be potentially related with the risk of esophageal cancer.

In conclusion, MetS and its components – elevated waist circumference, blood pressure, and FBG – were associated with an 11% higher risk of esophageal cancer in Korea, where esophageal cancer is mostly ESCC. Further studies are needed to establish the relationship between MetS and esophageal cancer.

No disclosures were reported.

J. Lee: Conceptualization, writing-original draft, writing-review and editing. K. Han: Data curation, methodology. J. Yoo: Data curation, formal analysis, methodology. Y. Yeo: Data curation. I.Y. Cho: Writing-review and editing. B. Cho: Writing-review and editing. J.-H. Park: Writing-review and editing. D.W. Shin: Conceptualization, writing-review and editing. J.H. Cho: Conceptualization, writing-review and editing. Y.-M. Park: Writing-review and editing.

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.