Background
Pancreatic cancer is the third leading cause of cancer-related deaths in the United States, with a 5-year survival rate of 10%.1 The incidence and mortality rates of pancreatic cancer are rapidly increasing worldwide, especially in the young population.2–4 Approximately 80% of patients with pancreatic cancer are diagnosed at an advanced stage; however, screening in the asymptomatic general population has not been shown to be effective.5,6 Therefore, efforts to modify risk factors are crucial for preventing pancreatic cancer, especially in individuals at an increased risk.
Diabetes and prediabetes are both associated with an increased risk of pancreatic cancer.7,8 Diabetes and prediabetes affect 13.0% and 34.5% of adults in the United States, respectively, and the prevalence is increasing worldwide.9 Smoking is also a major risk factor for pancreatic cancer.2 Nevertheless, little is known about the effects of smoking or smoking cessation on pancreatic cancer risk in individuals with diabetes and prediabetes. Elucidation of the effect of smoking and smoking cessation on pancreatic cancer risk in individuals with preexisting risk factors, such as diabetes and prediabetes, may help us establish more effective preventive strategies.
Therefore, we conducted a nationwide cohort study to investigate the association between smoking status (especially smoking cessation) and pancreatic cancer risk according to glycemic status after adjusting for potential confounders. We also analyzed these associations according to smoking pack-years. More than 9 million adults with National Health Screening results and medical claims records from the Korean National Health Insurance Service (KNHIS) were followed for 10 years.
Patients and Methods
Data Source and Study Population
The KNHIS is a mandatory national health insurance system that covers 97% of the South Korean population. The remaining 3% are covered by the Medical Assistance Program, but their medical claims are also recorded in the KNHIS system. In addition, the KNHIS provides a biennial standardized National Health Screening Program for all citizens aged ≥20 years and all employees of any age.10 The participation rate of the target population is approximately 76%. Using these data, we obtained health screening results (anthropometric measurements, responses to a self-administered questionnaire on health-related behavior, and laboratory test results) and claims data (medical treatment, prescription drugs, and disease diagnosis based on ICD-10-CM codes).
Figure 1 shows a flowchart of study enrollment. We initially included 10,585,844 individuals aged ≥20 years who underwent the National Health Screening provided by the KNHIS between January 1 and December 31, 2009. We excluded those with a prior diagnosis of any cancer before cohort entry (n=151,128) and those with missing information on the variables included in this study (n=827,396). To reduce the effect of potential preexisting diseases and reverse causality, we also excluded those who developed any cancer or died within 1 year of cohort entry (n=86,691). In total, 9,520,629 individuals were enrolled and followed until the date of pancreatic cancer development, death, or December 31, 2018, whichever occurred first.
Flowchart of the study population. Smoking status: never-smokers, quitters, and current smokers.
Abbreviation: KNHIS, Korean National Health Insurance Service.
Citation: Journal of the National Comprehensive Cancer Network 21, 11; 10.6004/jnccn.2023.7060
Flowchart of the study population. Smoking status: never-smokers, quitters, and current smokers.
Abbreviation: KNHIS, Korean National Health Insurance Service.
Citation: Journal of the National Comprehensive Cancer Network 21, 11; 10.6004/jnccn.2023.7060
Flowchart of the study population. Smoking status: never-smokers, quitters, and current smokers.
Abbreviation: KNHIS, Korean National Health Insurance Service.
Citation: Journal of the National Comprehensive Cancer Network 21, 11; 10.6004/jnccn.2023.7060
The study protocol was approved by the Institutional Review Board of the Samsung Medical Center (approval No. SMC2019-08-106) and the KNHIS Big Data Steering Department (NHIS-2021-1-718). The requirement for written informed consent was waived because the KNHIS dataset was compiled after anonymization in accordance with their strict confidentiality regulations. This study adhered to the principles outlined in the Declaration of Helsinki.
Definition of Smoking Status
Information on smoking status was obtained using a standardized self-administered questionnaire during the KNHIS National Health Screening (supplemental eTable 1, available with this article at JNCCN.org). The participants were classified as never-smokers, quitters, or current smokers. Never-smokers were defined as those who had smoked <100 cigarettes in their lifetime. To measure cumulative exposure to tobacco, current smokers and quitters were asked about the average daily number of cigarettes smoked and the duration of smoking in years. The pack-years of smoking were calculated by multiplying the number of packs of cigarettes smoked per day by the number of years the person had smoked.
Definition of Glycemic Status
The glycemic status of the participants was categorized as normoglycemia (fasting plasma glucose level <100 mg/dL), prediabetes (fasting plasma glucose level, 100–125 mg/dL), and diabetes. Diabetes was defined as a fasting glucose level ≥126 mg/dL or at least 1 claim per year under ICD-10-CM codes E11–E14 and at least 1 claim per year for the prescription of oral and/or injectable antidiabetic medication.10 The fasting glucose levels of each participant were measured during the National Health Screening.
Definition of Incident Pancreatic Cancer
The primary outcome of this study was newly diagnosed pancreatic cancer. We identified new cases of pancreatic cancer between January 2009 and December 2018 using ICD-10-CM code C25 during hospitalization and the special reimbursement code V193 for cancer. Since 2006, the KNHIS policy has enhanced health coverage for intractable diseases including cancer. To benefit from a reduced copayment rate of 5% for cancer-related examinations and treatments, physicians and medical institutions are required to certify the diagnosis of cancer using a reimbursement code (V193). The KNHIS registers all patients with a confirmed diagnosis of cancer using the V193 code.
Definition of the Clinical Variables
Healthcare professionals record basic health measurements and perform routine blood tests during the National Health Screening in KNHIS-certified hospitals, which are subjected to quality control measures. Blood samples are collected after overnight fasting to measure serum levels of glucose, total cholesterol, triglycerides, high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol. Height, weight, and waist circumference are also measured. Body mass index (BMI) is calculated by dividing weight by height squared (kg/m2). Blood pressure is measured with the participant in a seated position after at least 5 minutes of rest. Lower-income status includes those in the lowest quartile of the required insurance fee or those who received free medical care.
Data on alcohol consumption and physical activity are obtained using standardized self-administered questionnaires. In terms of alcohol consumption, participants were categorized according to the amount of alcohol consumed on an average daily basis as follows: none, mild-to-moderate (<30 g of alcohol per day), or heavy (≥30 g of alcohol per day).11 Regular physical activity was defined as performing ≥30 minutes of moderate-intensity physical activity at least 5 times per week or ≥20 minutes of vigorous-intensity physical activity at least 3 times per week.
Pancreatitis was identified using the ICD-10-CM codes K85, K86.0, and K86.1. Dyslipidemia was defined based on serum total cholesterol levels ≥240 mg/dL or claims for lipid-lowering medications under the ICD-10-CM code E78. Hypertension was defined as a systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, or claims for antihypertensive medications under ICD-10-CM codes I10–I13 and I15.
Statistical Analysis
According to smoking status, baseline characteristics were compared using the chi-test for categorical variables and analysis of variance for continuous variables. The incidence of pancreatic cancer was determined by dividing the number of incident cases by the number of person-years in each group. We used Cox proportional hazards regression models to estimate the hazard ratios (HRs) and 95% confidence intervals for pancreatic cancer. Model 1 was adjusted for age and sex. Model 2 was adjusted for age, sex, BMI, alcohol consumption, physical activity, dyslipidemia, hypertension, income status, and pancreatitis. In addition, the additive interactions were determined using the relative excess risk due to interaction (RERI), which indicates whether the interactive effects of prediabetes and diabetes with smoking were more than the additive effect when present together.12 RERI was calculated for binary variables as the difference between the additive and observed risks for risk factors A and B (RERI = HRAB – HRA – HRB + 1). An RERI of 0 indicates no interaction or exact additivity, <0 indicates a negative or subadditive interaction, and >0 indicates a positive or synergistic interaction.13 All statistical tests were conducted on a 2-sided basis with a significance level set at P<.05. Statistical analyses were conducted using SAS 9.3 (SAS Institute Inc.).
Results
Baseline Characteristics of the Study Population
Among the 9,520,629 participants in this cohort, 15,245 were newly diagnosed with pancreatic cancer during the 78.4 million person-years (median, 8.3 person-years) of follow-up. The mean [SD] age of the participants was 47.0 [13.7] years, and 54.5% were male.
Table 1 shows the baseline characteristics of the study population according to smoking status (never-smokers, quitters, and current smokers). Current smokers were the youngest and had the highest proportion of heavy drinkers (all P<.001). Quitters and current smokers had similar proportions of men. Quitters had the highest BMI, waist circumference, fasting plasma glucose levels, and blood pressure (all P<.001). Quitters also had the highest proportion of individuals who engaged in regular physical activity as well as the highest proportion of those with hypertension, diabetes, and dyslipidemia (all P<.001).
Baseline Characteristics of Study Population
Smoking and Pancreatic Cancer Risk According to Glycemic Status
Table 2 and supplemental eFigure 1 show the risk of pancreatic cancer in current smokers, quitters, and never-smokers according to glycemic status compared with that in never-smokers with normoglycemia (reference group). Among current smokers, those with normoglycemia had a 48% increased risk of pancreatic cancer (HR, 1.48; 95% CI, 1.40–1.58) compared with 11% among quitters (HR, 1.11; 95% CI, 1.03–1.19); those with prediabetes had an 83% increased risk of pancreatic cancer (HR, 1.83; 95% CI, 1.70–1.97) compared with 28% among quitters (HR, 1.28; 95% CI, 1.18–1.39); and those with diabetes had a 172% increased risk of pancreatic cancer (HR, 2.72; 95% CI, 2.52–2.94) compared with 78% among quitters (HR, 1.78; 95% CI, 1.63–1.95). Quitters had a significantly lower risk of pancreatic cancer than current smokers in all glycemic status groups (all P<.001).
Association of Smoking Status With Risk of Pancreatic Cancer
Current smokers with diabetes or prediabetes had a synergistically increased risk of pancreatic cancer above the sum of these individual risks (RERI, 0.69 and 0.20, respectively; all P<.05), but quitters with diabetes or prediabetes did not (RERI, 0.01 and –0.02; respectively; all P>.05). The RERI between current smoking and glycemic status on pancreatic cancer risk increased as the glycemic status worsened (Ptrend<.01).
Smoking and Pancreatic Cancer Risk According to Glycemic Status and Smoking Pack-Years
Figure 2 and supplemental eTable 2 show the risk of pancreatic cancer in current smokers, quitters, and never-smokers according to glycemic status and smoking pack-years. The risk of pancreatic cancer increased as the glycemic status worsened and smoking pack-years increased (Ptrend<.001). As shown in Figure 2A, current smokers with a ≥20 pack-year history and diabetes had the highest risk, with a 3-fold increased risk of pancreatic cancer compared with never-smokers with normoglycemia (HR, 2.98; 95% CI, 2.73–3.26). Significant synergistic interactions between current smoking and diabetes or prediabetes with respect to pancreatic cancer risk were observed, regardless of smoking pack-years (all RERI >0; P<.05) (supplemental eTable 3).
(A, B) Risk of pancreatic cancer in current smokers, quitters, and never-smokers according to glycemic status and smoking PYs.
HRs and 95% confidence intervals were adjusted for age, sex, body mass index, alcohol consumption, physical activity, income status, hypertension, dyslipidemia, and pancreatitis.
Abbreviations: HR, hazard ratio; PY, pack-year.
Citation: Journal of the National Comprehensive Cancer Network 21, 11; 10.6004/jnccn.2023.7060
(A, B) Risk of pancreatic cancer in current smokers, quitters, and never-smokers according to glycemic status and smoking PYs.
HRs and 95% confidence intervals were adjusted for age, sex, body mass index, alcohol consumption, physical activity, income status, hypertension, dyslipidemia, and pancreatitis.
Abbreviations: HR, hazard ratio; PY, pack-year.
Citation: Journal of the National Comprehensive Cancer Network 21, 11; 10.6004/jnccn.2023.7060
(A, B) Risk of pancreatic cancer in current smokers, quitters, and never-smokers according to glycemic status and smoking PYs.
HRs and 95% confidence intervals were adjusted for age, sex, body mass index, alcohol consumption, physical activity, income status, hypertension, dyslipidemia, and pancreatitis.
Abbreviations: HR, hazard ratio; PY, pack-year.
Citation: Journal of the National Comprehensive Cancer Network 21, 11; 10.6004/jnccn.2023.7060
Quitters had a consistently reduced risk of pancreatic cancer when compared with current smokers with a smoking history of ≥20 pack-years (Figure 2A) and <20 pack-years (Figure 2B), regardless of glycemic status (all P<.001). In quitters with a smoking history of <20 pack-years, the risk of pancreatic cancer did not differ from that of never-smokers in any of the glycemic status groups (all P>.05).
Discussion
In this nationwide cohort study of >9 million individuals, we found that current smoking was associated with a synergistically increased risk of pancreatic cancer in individuals with diabetes and prediabetes. However, we also found that smoking cessation was associated with a markedly reduced risk of pancreatic cancer. The risk of pancreatic cancer in quitters was comparable with that in never-smokers, especially when the smoking history was <20 pack-years. Our findings suggest a crucial opportunity to reduce the risk of pancreatic cancer, particularly in individuals with preexisting risk factors, such as diabetes or prediabetes.
Although smoking is a modifiable risk factor for pancreatic cancer,14,15 data on the association between smoking status and pancreatic cancer risk in individuals with prediabetes and diabetes have been limited. Two previous hospital-based case-control studies have examined the interactive effect between diabetes and smoking on pancreatic cancer risk with mixed results, one showing a positive interaction16 and the other showing an interaction only in women.17 Moreover, given the high fatality rate of pancreatic cancer, these case-control studies were limited by survivorship and selection bias. These studies also had recall bias due to the questionnaire-based assessment of diabetes. One cohort study reported that current smoking combined with diabetes increased the risk of pancreatic cancer compared with never-smokers without diabetes (HR, 2.13; 95% CI, 2.00–2.28).18 However, no studies have examined the interactive effect of prediabetes and smoking on the risk of pancreatic cancer. Notably, data on the effect of smoking cessation on pancreatic cancer risk in individuals with prediabetes and diabetes are lacking. We provide new evidence of a synergistic interaction between current smoking and diabetes and even prediabetes on pancreatic cancer risk and show that the synergistically increased risk of pancreatic cancer was decreased by smoking cessation. The effect of smoking status on pancreatic cancer risk may differ according to an individual’s health status, highlighting the need for individualized cancer prevention education.
The underlying mechanism for why the effect of smoking status on pancreatic cancer risk differs according to an individual’s glycemic status is unknown, but there are some potential explanations. We showed that current smoking had a significant synergistic interaction with diabetes or prediabetes on pancreatic cancer risk beyond the sum of these individual risks. According to pathologic studies, tobacco carcinogens act on the pancreas through the blood and bile reflux.19,20 Hyperglycemia causes vascular damage and endothelial dysfunction, which may result in vulnerability to tobacco carcinogens and synergistic pancreatic carcinogenesis.21 Hyperglycemia also can induce a microenvironment that contributes to smoking-induced pancreatic carcinogenesis.22 However, smoking cessation may slow the accumulation of additional damage and lead to the reactivation of normal cells that have not been damaged by tobacco carcinogens.23 In addition, smoking cessation may eliminate the interaction between hyperglycemia and smoking and reduce the risk of pancreatic cancer significantly in individuals with hyperglycemia.
Our study has several strengths. First, this population-based nationwide study is one of the largest cohort studies, using information from >9 million individuals who were followed for up to 10 years. After cohort entry, the KNHIS database maintains an accurate record of the cohort’s clinical course. This large dataset allowed us to investigate the association between smoking status and pancreatic cancer risk according to each individual’s glycemic status and smoking pack-years. Second, we analyzed data systematically collected on lifestyle factors, fasting glucose levels, anthropometric parameters, and extensive medical records. Third, we adjusted for potential confounding factors for pancreatic cancer, including BMI, alcohol consumption, and pancreatitis. Fourth, we used both national registration codes (V codes) and ICD-10-CM diagnostic codes (C codes) to ensure a high diagnostic accuracy for pancreatic cancer.
Our study also has some limitations. First, although smoking status was assessed using standardized questionnaires before pancreatic cancer development, biochemical validation was not performed. Underreporting and misclassification are possible, but self-reported smoking status is generally accurate.24 Second, the participants’ smoking status may have changed since the baseline; some quitters may relapse after the baseline, and some current smokers may quit after the baseline. Thus, the risk of pancreatic cancer may have been overestimated for quitters and underestimated for current smokers in our study. Nonetheless, the risk of pancreatic cancer in quitters was lower than in current smokers in our study. Also, the participants’ glycemic status may have changed since the baseline. Third, we have no information on the time of initiating or quitting smoking. We could not assess the pancreatic cancer risk according to age at initiating or quitting smoking. We could not evaluate the time course of the decrease in pancreatic cancer risk by quitting smoking duration. Fourth, we were unable to consider HbA1c levels or family history of pancreatic cancer owing to a lack of data on these in the KNHIS. Fifth, information on the histopathologic type of pancreatic cancer was unavailable. However, pancreatic ductal adenocarcinoma accounts for >95% of all pancreatic cancers. Sixth, our nationwide cohort consisted of >9 million people from the general population, but they were from a single country. Whether any ethnic differences exist in the association between smoking status and pancreatic cancer risk has not been determined.
Conclusions
We found a synergistic increase in the risk of pancreatic cancer among current smokers with diabetes and prediabetes. However, smoking cessation markedly reduced the increased risk of pancreatic cancer to the level of never-smokers, particularly if smoking history was <20 pack-years. Our findings provide an important opportunity to reduce the risk of pancreatic cancer in individuals with diabetes and prediabetes who are at increased risk for pancreatic cancer, given the rapidly increasing incidence of pancreatic cancer, diabetes, and prediabetes. More individualized and intensive cancer prevention education should be emphasized for those at a high risk of pancreatic cancer beyond the one-size-fits-all approach. Further studies are needed to elucidate the mechanisms underlying the possible synergistic interaction between smoking and hyperglycemia that promotes pancreatic carcinogenesis.
References
- 2.↑
Huang J, Lok V, Ngai CH, et al. Worldwide burden of, risk factors for, and trends in pancreatic cancer. Gastroenterology 2021;160:744–754.
- 3.↑
Pourhams A, Sepanlou SG, Ikuta KS, et al. The global, regional, and national burden of pancreatic cancer and its attributable risk factors in 195 countries and territories, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet Gastroenterol Hepatol 2019;4:934–947.
- 4.↑
Rahib L, Smith BD, Aizenberg R, et al. Projecting cancer incidence and deaths to 2030: the unexpected burden of thyroid, liver, and pancreas cancers in the United States. Cancer Res 2014;74:2913–2921.
- 5.↑
Owens DK, Davidson KW, Krist AH, et al. Screening for pancreatic cancer: US Preventive Services Task Force reaffirmation recommendation statement. JAMA 2019;322:438–444.
- 6.↑
Singhi AD, Koay EJ, Chari ST, et al. Early detection of pancreatic cancer: opportunities and challenges. Gastroenterology 2019;156:2024–2040.
- 7.↑
Koo DH, Han KD, Park CY. The incremental risk of pancreatic cancer according to fasting glucose levels: nationwide population-based cohort study. J Clin Endocrinol Metab 2019;104:4594–4599.
- 8.↑
Park JH, Han K, Hong JY, et al. Changes in metabolic syndrome status are associated with altered risk of pancreatic cancer: a nationwide cohort study. Gastroenterology 2022;162:509–520.e7.
- 9.↑
Centers for Disease Control and Prevention. National diabetes statistics report 2020: estimates of diabetes and its burden in the United States. Accessed May 2, 2023. Available at: https://www.cdc.gov/diabetes/pdfs/data/statistics/national-diabetes-statistics-report.pdf
- 10.↑
Lee YH, Han K, Ko SH, et al. Data analytic process of a nationwide population-based study using National Health Information Database established by National Health Insurance Service. Diabetes Metab J 2016;40:79–82.
- 11.↑
Park JH, Han K, Hong JY, et al. Association between alcohol consumption and pancreatic cancer risk differs by glycaemic status: a nationwide cohort study. Eur J Cancer 2022;163:119–127.
- 12.↑
Knol MJ, VanderWeele TJ. Recommendations for presenting analyses of effect modification and interaction. Int J Epidemiol 2012;41:514–520.
- 13.↑
Knol MJ, VanderWeele TJ, Groenwold RH, et al. Estimating measures of interaction on an additive scale for preventive exposures. Eur J Epidemiol 2011;26:433–438.
- 14.↑
Lugo A, Peveri G, Bosetti C, et al. Strong excess risk of pancreatic cancer for low frequency and duration of cigarette smoking: a comprehensive review and meta-analysis. Eur J Cancer 2018;104:117–126.
- 15.↑
Koyanagi YN, Ito H, Matsuo K, et al. Smoking and pancreatic cancer incidence: a pooled analysis of 10 population-based cohort studies in Japan. Cancer Epidemiol Biomarkers Prev 2019;28:1370–1378.
- 16.↑
Bo X, Shi J, Liu R, et al. Using the risk factors of pancreatic cancer and their interactions in cancer screening: a case-control study in Shanghai, China. Ann Glob Health 2019;85:103.
- 17.↑
Hassan MM, Bondy ML, Wolff RA, et al. Risk factors for pancreatic cancer: case-control study. Am J Gastroenterol 2007;102:2696–2707.
- 18.↑
Park BK, Seo JH, Chung JB, et al. Lifestyle, body mass index, diabetes, and the risk of pancreatic cancer in a nationwide population-based cohort study with 7.4 million Korean subjects. Br J Cancer 2022;127:549–557.
- 19.↑
Hao J, Li G, Pang B. Evidence for cigarette smoke-induced oxidative stress in the rat pancreas. Inhal Toxicol 2009;21:1007–1012.
- 20.↑
Weissman S, Takakura K, Eibl G, et al. The diverse involvement of cigarette smoking in pancreatic cancer development and prognosis. Pancreas 2020;49:612–620.
- 21.↑
Campagna D, Alamo A, Di Pino A, et al. Smoking and diabetes: dangerous liaisons and confusing relationships. Diabetol Metab Syndr 2019;11:85.
- 22.↑
Storz P, Crawford HC. Carcinogenesis of pancreatic ductal adenocarcinoma. Gastroenterology 2020;158:2072–2081.
- 23.↑
Yoshida K, Gowers KH, Lee-Six H, et al. Tobacco smoking and somatic mutations in human bronchial epithelium. Nature 2020;578:266–272.
- 24.↑
Patrick DL, Cheadle A, Thompson DC, et al. The validity of self-reported smoking: a review and meta-analysis. Am J Public Health 1994;84:1086–1093.
