Influence of Fasting Glucose Level on Gastric Cancer Incidence in a Prospective Cohort Study

Although the prevalence of gastric cancer is declining, it still ranks fifth and fourth worldwide in terms of cancer incidence and mortality, respectively (1). The trend of gastric cancer incidence varies geographically. Gastric cancer cases in Eastern and Southeastern Asia accounted for more than 60% of total cases in 2018 (2), and estimates revealed Korea to be the country with high incidence rate for both males and females (3).

The risk factors for gastric cancer have been well documented in the literature (4). However, in addition to identifying substantial risk factors such as Helicobacter pylori (H. pylori) infection and smoking status, hyperglycemia may be a contributor to the development and progression of malignancies (5). Diabetes is identified as a fasting plasma glucose ≥126 mg/dL, and prediabetes is characterized by a fasting plasma glucose level from 100 to 125 mg/dL (6). To date, the prevalence of hyperglycemia has increased significantly due to socioeconomic growth (7). In particular, a high proportion of hyperglycemia was recorded among postmenopausal women (8, 9). Similar to other industrialized countries, the prevalence of hyperglycemia in Korea has been reported to have increased (10). In particular, the prevalence of prediabetes has increased over time, from 21.5% in 2006 to 25.0% in 2013, and this prevalence is higher than the prevalence of diabetes (11, 12).

To date, several studies have recognized that not only diabetes mellitus but also prediabetes is associated with an elevated risk of certain types of cancers (13–16). Alterations in insulin signaling, increased insulin-like growth factor-1 (IGF-1), and chronic inflammation (17–19) have been proposed as potential mechanisms. In addition, hyperglycemia may be responsible for promoting the epithelial–mesenchymal transition (6). With regard to gastric cancer, some studies have investigated gastric cancer risk in relation to diabetes mellitus, but the results have been inconsistent. For example, a significant association was found in prospective cohorts conducted in Austria, Norway, and Sweden (20, 21), whereas an association was not observed in other studies (22–24). Importantly, the association between prediabetes and gastric cancer risk has not been documented in the majority of previous studies.

To our knowledge, only a few studies have been conducted to explore gastric cancer risk in relation to hyperglycemia in Korea, and the results remain conflicting (23, 25). Furthermore, epidemiologic studies related to the association of prediabetes and diabetes with gastric cancer risk have not been reported thus far. Due to the high incidence rate of gastric cancer and hyperglycemia, it is necessary to elucidate this association. Thus, our study aimed to identify the influence of fasting glucose levels on incident gastric cancer in the Korean population.

The Cancer Screenee Cohort was designed in 2002 to investigate the risk factors for cancer in Korea. The details of the cohort have been described previously (26). Briefly, the source population included 41,837 participants aged 16 and older who underwent health examinations at the Center for Cancer Prevention and detection at the National Cancer Center in South Korea between August 2002 and December 2014. Participants were asked to complete baseline questionnaires and were followed up until December 2017. For our analysis, participants were excluded as follows: 2,097 participants who had a previous diagnosis of any cancer and 6 participants aged <20 years. We also excluded 51 patients who had a gastric cancer diagnosis within one year after enrollment because gastric cancer has a long latency period. In total, we included 37,213 subjects in the final analysis (Fig. 1). We obtained written informed consent from all participants and approval for the study protocol from the Institutional Review Board of the National Cancer Center (No. NCCNCS-07-077). This study was conducted according to the guidelines of the Declaration of Helsinki.

Gastric cancer incidence was determined through linkage with the 2017 Korea National Cancer Incidence Database of the Korea Central Cancer Registry. The International Classification of Diseases, 10th Revision (C16) was considered a criterion for gastric cancer diagnosis. Patients who were diagnosed with primary gastric cancer more than one year after enrollment were identified as incident cases.

The criteria of Adult Treatment Panel III of the National Cholesterol Education Program (ATPIII NCEP) were used for high fasting glucose diagnosis, including prediabetic and diabetic levels, and included those who had information on fasting glucose and fasting glucose ≥100 mg/dL or a history of diabetes mellitus (27).

A self-administered questionnaire was used to collect general characteristics at baseline. We used InBody 3.0 (Biospace) or automatic height and weight measurement (DS-102, Dong Shin Jenix Co., Ltd.) to measure the weight (kg) and height (m) of participants. The measurements were performed by trained personnel. The ratio of body weight (kg) to height squared (m²) was used to calculate the body mass index (BMI). The fasting glucose test was performed with venous blood samples taken from participants after they had fasted for 8 hours. A chemistry analyzer (TBA-200FR, Toshiba) was used for the measurement (28). H. pylori infection was assessed by using a rapid urease test (29).

We used the χ² test and t test to compare demographic characteristics between the normal fasting glucose group and high fasting glucose group. The participants were followed up for gastric cancer occurrence. The date from baseline to the date of cancer diagnosis, death, or end of follow-up (December 31, 2017), whichever came first, was considered to calculate person-years. We used the Cox proportional hazards regression model adjusted for confounding factors to explore the association of fasting glucose levels with gastric cancer incidence. Based on the literature, the confounding variables included were age, sex, educational level (high school graduate or less, college or higher), income (10,000 won/month; <200, 200–400, >400), smoking status (nonsmoker, ex-smoker, and current smoker), alcohol consumption (nondrinker, former drinker, and current drinker), regular exercise (yes, no), first-degree family history of gastric cancer (yes, no), and BMI (<23, 23–<25, ≥25; refs. 27, 30). We assessed the proportional hazard assumption in the Cox regression model with time-dependent variables, and the assumption was satisfied. With regard to stratification, smoking status was categorized as nonsmoker and ever smoker (ex-smoker and current smoker). SAS software (version 9.4, SAS Institute, Cary, NC) was used for all statistical analyses, and a two-sided P value less than 0.05 was considered statistically significant.

Table 1 represents the general information of subjects at baseline. Among the 37,213 participants, 7,113 participants had high fasting glucose levels. Participants with high fasting glucose levels were older than participants with normal fasting glucose levels, and the age was 53.7 ± 8.7 for the former and 48.7 ± 9.1 for the latter (P < 0.001). In comparison with the normal fasting glucose group, the high fasting glucose group had a higher prevalence of BMI ≥ 25 (48.8% vs. 29.5%, P < 0.001), first-degree family history of gastric cancer (12.9% vs. 12.0%, P = 0.030), current smoking (25.8% vs. 22.7%, P < 0.001), current alcohol consumption (63.8% vs. 61.7%, P < 0.001), and regular exercise (47.7% vs. 45.2%, P = 0.033), whereas they reported lower proportions of high educational level (37.9% vs. 45.5%, P < 0.001) and high income (38.9% vs. 45.6%, P < 0.001). Among men, the participants with high fasting glucose levels were older (53.1 ± 8.7 vs. 48.7 ± 9.3, P < 0.001), exhibited a higher proportion of BMI ≥25 (53.1% vs. 39.8%, P < 0.001), and were more likely to be married or cohabitating (91.6% vs. 90.0%, P < 0.001), ex-smokers (43.4% vs. 37.3%, P < 0.001), and ex-drinkers (7.2% vs. 5.4%, P < 0.001). In comparison with women in the normal fasting glucose group, women with high fasting glucose exhibited a higher proportion of these characteristics, except for marital status and smoking status (Supplementary Table S1).

The mean follow-up was 9.07 years, and the high fasting glucose group exhibited a higher rate of incident gastric cancer than the other group during the follow-up period (Fig. 2). Table 2 represents the hazard ratio (HR) with the 95% confidence interval (CI) of incident gastric cancer related to fasting glucose. We observed a significant association between high fasting glucose and gastric cancer incidence for all subjects in the model adjusted for age and sex (HR = 1.32; 95% CI = 1.01–1.72), but the significant association disappeared in the fully adjusted model (HR = 1.25; 95% CI = 0.95–1.64). Males exhibited a similar pattern; in the fully adjusted model, a nonsignificant association between fasting glucose and incident gastric cancer was found (HR = 1.13; 95% CI = 0.82–1.56). In contrast, high fasting glucose significantly increased gastric cancer risk in the model adjusted for age for postmenopausal women (HR = 1.93; 95% CI = 1.15–3.26), and the association remained constant in the fully adjusted model (HR = 1.88; 95% CI = 1.11–3.20). Furthermore, we performed a sensitivity analysis to compare the gastric cancer risk between participants with high fasting glucose and no history of diabetes and participants with a history of diabetes, but we failed to find a significant association (Supplementary Table S2).

We further performed subgroup analyses based on BMI level, smoking status, and first-degree family history of gastric cancer to observe the association between fasting glucose and gastric cancer risk. Among all participants with BMI < 25 kg/m², high fasting glucose significantly increased gastric cancer risk (HR = 1.45; 95% CI = 1.00–2.12). Heterogeneity was found for the two BMI groups (P_{heterogeneity} = 0.009; Table 3). When stratified by smoking status, the association of fasting glucose with gastric cancer is not presented for women because the majority of women were nonsmokers. Among all participants, we observed a significantly increased risk of gastric cancer for all nonsmokers in the model adjusted for age and sex, and the significant association was unchanged in the fully adjusted model (HR = 2.00; 95% CI = 1.28–3.10 and HR = 1.89; 95% CI = 1.21–2.95, respectively). However, a nonsignificant association was found for ever-smokers (HR = 1.00; 95% CI = 0.70–1.44), and heterogeneity was found for the two smoking statuses (P_{heterogeneity} = 0.027; Table 4). Furthermore, high fasting glucose tended to significantly increase gastric cancer risk in all participants who did not have a first-degree family history of gastric cancer (HR = 1.45; 95% CI = 1.06–1.99). We observed heterogeneity between the two groups of first-degree family histories of gastric cancer (P_{heterogeneity} < 0.001; Table 5). In contrast, we failed to find a significant association of fasting glucose with gastric cancer risk among groups stratified by alcohol consumption and H. pylori infection.

In this study, an elevated risk of gastric cancer was consistently observed for postmenopausal women with high fasting glucose. In addition, we found a significant positive association of fasting glucose with gastric cancer risk for participants who were nonsmokers, had a BMI < 25 kg/m², and did not have a first-degree family history of gastric cancer.

To date, some studies have been conducted to explore the association of fasting glucose with gastric cancer. A meta-analysis suggested that diabetes history was a risk factor for gastric cancer (18). Evidence from previous large cohorts showed that diabetes mellitus significantly increases gastric cancer risk (19–21). Notably, many studies have mainly focused on diabetic participants. Our results emphasized that gastric cancer development might be attributable to high fasting glucose, even if participants have not developed diabetes mellitus (prediabetic level). This association has been observed in large nationwide population-based cohort studies in Korea, where the variability in fasting plasma glucose may play a significant role in gastric cancer development in populations without diabetes mellitus (6), and in Japan, where gastric cancer risk in the group with high fasting plasma glucose was 3.1 times higher than that in the low fasting plasma glucose group (31).

Several tumor-promoting mechanisms can be proposed to explain this association. There is a positive association of high fasting glucose with reactive oxygen species production that leads to oxidative DNA damage (31). In addition, high glucose levels have been reported to contribute to DNA damage (32). The accumulation of glucose-derived advanced glycation end products may act as a potential risk factor for gastric cancer progression and promote the invasion and metastasis of gastric cancer partially through the activation of the RAGE/ERK/Sp1/MMP2 pathway (33). Furthermore, diabetes mellitus may serve as a marker for metabolic dysfunction associated with cancer risk (19). Another possible mechanism is chronic inflammation, which plays a certain role in gastric cancer development and the etiology of diabetes mellitus (18).

However, the significant association observed seemed to be restricted to postmenopausal women in our study. The finding has two potential explanations. First, after menopause, body composition can be changed due to imbalances of hormones related to elevated fasting glucose and insulin sensitivity (8, 34). The hormone-related biological mechanism may be due to increased free testosterone levels (8). Thus, a higher proportion of high fasting glucose was observed in postmenopausal women in our study. Second, estrogen levels tend to decline in postmenopausal women, which can influence the occurrence of gastric cancer (35). High fasting glucose and gastric cancer incidence tended to appear in postmenopausal women, which can explain why the significant association of fasting glucose with gastric cancer risk was clearly observed in postmenopausal women in our study. However, the clear mechanism underlying the association of high fasting glucose with gastric cancer among postmenopausal women is still debated. Increased levels of bioavailable ovarian steroid hormones have been reported to be related to hyperinsulinaemia and/or insulin resistance through reduced sex hormone–binding globulin and increased ovarian steroid hormone production. Furthermore, increased insulin and IGF-1 levels may activate estrogen receptor alpha, which potentiates the estrogen-related proliferation and growth of cancer cells. Taken together, these findings suggest that high fasting glucose may be a contributor to cancer development (36). However, a significant association seemed to be restricted to women, and we failed to find a significant association in men, which may be linked to the effects of sex hormone–binding globulin (37).

Obesity is known to upregulate sex steroid hormones, insulin, mediators of inflammation, and lower adiponectin levels (30). Obesity has been shown to be associated with malignancies. However, when stratified by BMI level, a nonsignificant association in obese participants was observed in our study. Our result is supported by the findings of a cohort study in Japan and another cohort study (20, 38). It should be noted that obesity has been demonstrated to be associated with obesity-related cancers such as colorectal cancer and breast cancer, but less evidence has been recorded for gastric cancer. Importantly, our study emphasized that high fasting glucose was a strong predictor of gastric cancer, and the result was robust even at nonobese BMI levels.

Smoking has been reported to have detrimental effects on gastric cancer risk (39) and may be an important aetiologic factor for diabetes mellitus development (40). However, we failed to find a significant association of high fasting glucose with gastric cancer risk in the subgroup of ever-smokers due to limited sample size. We found that high fasting glucose is an aetiologic factor for gastric cancer in nonsmokers and participants without a first-degree family history of gastric cancer. A study conducted in Finland with 20,720 male smokers reported that a first-degree family history of gastric cancer is related to the risk of specific gastric cancer subtypes (41). Compared with individuals without a family history of gastric cancer, individuals with a family history of gastric cancer developed gastric cancer at a younger age (42). We want to emphasize that high fasting glucose is a predictor of gastric cancer development even in nonsmokers and participants who did not have a first-degree family history of gastric cancer due to the strong associated mechanisms. High fasting glucose was demonstrated to be an independent risk factor for gastric cancer development in our study. This implied a specific role of high fasting glucose in gastric carcinogenesis without the presence of other risk factors. Our result is in accordance with the result of a previous study, where diabetes was found to be a significant independent risk factor for early gastric cancer development (43).

Notably, medications have been demonstrated to be a potential link between diabetes and gastric cancer. Insulin therapy may be suggested as a contributor to the development of cancer, although the association with gastric cancer has been ambiguous (40). A nonsignificant association of insulin use with gastric cancer was found in a previous study (44). In vitro studies indicated high affinity of insulin glargine to the IGF1 receptor. However, insulin glargine is converted to less mitogenic metabolites at the injection site. Thus, the interpretation of the association of insulin use with gastric cancer remains open to discussion (40). In contrast, the anticancer property of metformin against some common cancers is well documented in the literature (45, 46). Similarly, the protective effect of metformin against gastric cancer was suggested in previous studies (47, 48). Biological mechanisms for this association remain unclear. There are several possible mechanisms to explain the protective effect of metformin. First, it disrupts mitochondrial energetics and inhibits mTORC1 kinase. Second, it contributes to reducing gluconeogenesis and promoting the uptake of fat and muscle glucose through AMPK, which inhibits the proliferation and migration of tumor cells. Third, the protective effect of metformin may be linked to inhibition of HIF1α/PKM2 signaling (47). Thus, further studies with additional information on medications regarding diabetes treatment are necessary to elucidate the association of diabetes with gastric cancer risk.

There are several strengths in our study. We considered prediabetic fasting glucose, which was not considered in previous studies. Selection bias and recall bias could be avoided due to the prospective nature of the cohort study. We had sufficient information on possible confounding variables. Outcomes were identified by the national cancer registry using a high-quality database. However, our study has some limitations. First, the measurement of fasting glucose was only performed at baseline, and a history of diabetes was identified based on the baseline questionnaire, which may introduce a risk of nondifferential misclassifications. Second, subgroup analyses for prediabetes and diabetes were not performed due to the limited number of cases. Third, information on diabetes treatment and screening practices of individuals was not available for consideration in our study. Fourth, the findings could have been affected by the small number of gastric cancer cases in subgroups. Fifth, analyses of cancer subtypes were not performed in our study. Sixth, nearly 40% of H. pylori infection data were missing in our study. However, H. pylori infection has not been indicated to affect the association of fasting glucose with gastric cancer risk (24, 31). Finally, the mean follow-up time was only 9.07 years, which may affect statistical inference due to insufficient cases. However, gastric cancer cases were identified through linkage with the national cancer registry using a high-quality database. Therefore, incident cases were not neglected, and thus, our findings accurately reflected the influence of fasting glucose on gastric cancer incidence.

This study showed that high fasting glucose significantly increased the risk of gastric cancer in postmenopausal women. High fasting glucose was also a predictor for gastric cancer development in all participants who were nonsmokers, had a BMI < 25 kg/m², and did not have a first-degree family history of gastric cancer. Our findings may be helpful for future planning and management regarding cancer prevention.

No disclosures were reported.

T.T. Tran: Formal analysis, writing–original draft. J. Lee: Data curation, formal analysis, investigation, methodology. M. Gunathilake: Writing–review and editing. H. Cho: Writing–review and editing. J. Kim: Data curation, supervision, funding acquisition, methodology, project administration, writing–review and editing.

This work was supported by International Cooperation & Education Program (NCCRI•NCCI 52210-52211, 2020) of National Cancer Center, Korea and grants from National Cancer Center, Korea (1910330) to J. Kim.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.