Abstract
Evidence has suggested that aspirin reduces the incidence of several cancers, but these benefits may not occur with pancreatic cancer.
A 12-year nationwide longitudinal cohort merged with the health checkup data was divided into “exposure ascertainment period” and “outcome ascertainment period” to avoid immortal time bias. The daily defined dose system was used to indicate the drug exposure.
We found no significant association between aspirin use and incident pancreatic cancer based on HR.
Aspirin does not prevent pancreatic cancer.
A large Asian cohort study with reliable medication information affirms no impact of aspirin on pancreatic cancer development.
Introduction
As pancreatic cancer is usually advanced and unresectable at diagnosis, early detection and chemoprevention are important (1). Aspirin plays a critical role in preventing vascular disease and has been suggested as a potential cancer-preventive agent. Mechanistically, aspirin may reduce cancer risk via cyclooxygenase-2 inhibition or other proapoptotic effects (2). Epidemiologic studies have suggested that daily aspirin reduces the incidence of colorectal and several other cancers, but these benefits may not occur with pancreatic cancer (3).
Two case–control studies reported that aspirin lowered the risk of pancreatic cancer (4, 5), whereas the Queensland Pancreatic Cancer Study found no benefit (6). However, aspirin usage in these studies was ascertained by self-report, a critical flaw that may have. More recently, a prospective analysis of two large U.S. cohorts determined plasma levels of aspirin metabolites (salicylate levels) and found no association between regular aspirin use and future risk of pancreatic cancer (7).
South Korea has a single-payer health insurance system, the National Health Insurance Services (NHIS), which covers almost the entire population. NHIS also provides health examinations biennially to all individuals ≥40 years old. Prescription data (medication identifier, dosage, and duration) from the NHIS claims database allow accurate determination of the medication exposure. We used this database to evaluate the influence of aspirin on pancreatic cancer risk. We hypothesized that regular aspirin use would reduce the risk of incident pancreatic cancer.
Materials and Methods
The institutional review board of Seoul National University Hospital (Seoul, South Korea) approved the protocol (IRB number: E-1509-004-699). Detailed methodology information was reported previously (8).
We obtained NHIS data from a standardized 12-year longitudinal cohort (the Health Examination Cohort) consisting of individuals who underwent ≥1 biennial examination between January 1, 2002, and December 31, 2003. Data from the first 5 years (January 1, 2002, to December 31, 2006) was used to identify cohort inclusion and exclusion criteria, as well as the exposures and potential confounding factors related to aspirin use and pancreatic cancer. January 1, 2007, was designated the cohort entry date. The follow-up observation period extended through December 31, 2013, to identify newly diagnosed pancreatic cancer cases while avoiding immortal time bias. We excluded the participants diagnosed with any cancer [indicated by International Classification of Diseases code–10th Revision (ICD-10) “C” codes] before December 31, 2006. We included 461,489 participants in our analysis.
Sociodemographic, medical history, cumulative prescription, and health-related habits data during the first 5 years (before cohort entry date) were extracted. Pancreatic cancer cases were defined as individuals who visited the hospital at least once with a C25 ICD-10 code and met any of these criteria during follow-up: three outpatient visits related to the code, admission for ≥3 days with the code diagnosis, receipt of any curative cancer treatments for the Korean Diagnosis-Related Group code “G60-Digestive Malignancy,” or death from a cause related to the code. To determine drug exposure, we used the daily defined dose (DDD) system of the World Health Organization Collaborating Center for Drug Statistics Methodology.
Results
Table 1 shows the characteristics of pancreatic cancer cases and controls. In this cohort study with reliable medication information, we found no significant association between aspirin use and incident pancreatic cancer based on HRs (Table 2). The lack of association persisted when we evaluated subgroups stratified by available variables. This contrasts with a previously reported association between aspirin use and reduced pancreatic cancer incidence in diabetic patients (7).
Comparison of basic characteristics between pancreatic cancer cases and controls
| . | Censored without event . | Pancreatic cancer cases . | P . |
|---|---|---|---|
| Number | 459,399 | 2,090 | |
| Age group, years | |||
| 40–49 | 222,232 (48.4) | 512 (24.5) | <0.001 |
| 50–59 | 129,494 (28.2) | 575 (27.5) | |
| 60–69 | 83,956 (18.3) | 699 (33.4) | |
| ≥70 | 23,717 (5.2) | 304 (14.5) | |
| Sex | |||
| Male | 246,226 (53.6) | 1,172 (56.1) | 0.025 |
| Female | 213,173 (46.4) | 918 (43.9) | |
| Low income levela | 81,534 (17.7) | 367 (17.6) | 0.845 |
| Current smoking | 95,031 (20.9) | 473 (23.0) | 0.025 |
| Excess alcohol consumption | 48,093 (10.5) | 324 (15.6) | <0.001 |
| Regular exercise | 97,704 (21.5) | 445 (21.6) | 0.882 |
| Body mass index, kg/m2 | |||
| <23 | 171,521 (37.3) | 817 (39.1) | 0.258 |
| 23–24.9 | 127,293 (27.7) | 564 (27.0) | |
| ≥25 | 160,451 (34.9) | 709 (33.9) | |
| Charlson comorbidity indexb | 1.55 ± 1.47 | 2.07 ± 1.71 | <0.001 |
| Diabetes | 50,296 (10.9) | 397 (19.0) | <0.001 |
| Hypercholesterolemia | 217,006 (47.2) | 954 (45.6) | 0.152 |
| Drug exposure (≥30-DDD) | |||
| NSAID use | 113,065 (24.6) | 670 (32.1) | <0.001 |
| Aspirin use | 64,448 (14.0) | 436 (20.9) | <0.001 |
| . | Censored without event . | Pancreatic cancer cases . | P . |
|---|---|---|---|
| Number | 459,399 | 2,090 | |
| Age group, years | |||
| 40–49 | 222,232 (48.4) | 512 (24.5) | <0.001 |
| 50–59 | 129,494 (28.2) | 575 (27.5) | |
| 60–69 | 83,956 (18.3) | 699 (33.4) | |
| ≥70 | 23,717 (5.2) | 304 (14.5) | |
| Sex | |||
| Male | 246,226 (53.6) | 1,172 (56.1) | 0.025 |
| Female | 213,173 (46.4) | 918 (43.9) | |
| Low income levela | 81,534 (17.7) | 367 (17.6) | 0.845 |
| Current smoking | 95,031 (20.9) | 473 (23.0) | 0.025 |
| Excess alcohol consumption | 48,093 (10.5) | 324 (15.6) | <0.001 |
| Regular exercise | 97,704 (21.5) | 445 (21.6) | 0.882 |
| Body mass index, kg/m2 | |||
| <23 | 171,521 (37.3) | 817 (39.1) | 0.258 |
| 23–24.9 | 127,293 (27.7) | 564 (27.0) | |
| ≥25 | 160,451 (34.9) | 709 (33.9) | |
| Charlson comorbidity indexb | 1.55 ± 1.47 | 2.07 ± 1.71 | <0.001 |
| Diabetes | 50,296 (10.9) | 397 (19.0) | <0.001 |
| Hypercholesterolemia | 217,006 (47.2) | 954 (45.6) | 0.152 |
| Drug exposure (≥30-DDD) | |||
| NSAID use | 113,065 (24.6) | 670 (32.1) | <0.001 |
| Aspirin use | 64,448 (14.0) | 436 (20.9) | <0.001 |
NOTE: Data are number (%) or mean ± SD.
Abbreviation: NSAID, Nonsteroidal antiinflammatory drug.
aIncome is based on the nationwide income distribution, which determines monthly premium payment for individuals.
bCalculated from acute myocardial infarction, congestive heart failure, peripheral vascular disease, cerebral vascular accident, dementia, pulmonary disease, connective tissue disorder, peptic ulcer, liver disease, diabetes, diabetes complications, paraplegia, renal disease, severe liver disease, and HIV based on ICD-10 codes of hospital visits from 2003 to 2006.
Factors associated with incident pancreatic cancer
| . | HR (95% confidence interval) . | |
|---|---|---|
| . | Age- and sex-adjusted . | Multivariatea . |
| Aspirin use | 1.01 (0.97–1.06) | 0.97 (0.92–1.02) |
| NSAID use | 1.04 (0.99–1.08) | 1.02 (0.96–1.09) |
| Low income | 1.01 (0.90–1.13) | 1.05 (0.94–1.18) |
| Current smoking | 1.25 (1.12–1.40) | 1.21 (1.07–1.35) |
| Excess alcohol consumption | 1.49 (1.32–1.69) | 1.44 (1.27–1.64) |
| Regular exercise | 1.00 (0.90–1.11) | 0.98 (0.88–1.09) |
| BMI (vs. <23 kg/m2) | ||
| 23–24.9 | 0.95 (0.85–1.05) | 0.94 (0.84–1.05) |
| ≥25 | 0.93 (0.84–1.03) | 0.90 (0.81–0.99) |
| CCIb | 1.14 (1.11–1.17) | 1.12 (1.09–1.16) |
| Diabetes | 1.58 (1.41–1.76) | 1.43 (1.27–1.60) |
| Hypercholesterolemia | 0.91 (0.83–0.99) | 0.92 (0.85–1.01) |
| . | HR (95% confidence interval) . | |
|---|---|---|
| . | Age- and sex-adjusted . | Multivariatea . |
| Aspirin use | 1.01 (0.97–1.06) | 0.97 (0.92–1.02) |
| NSAID use | 1.04 (0.99–1.08) | 1.02 (0.96–1.09) |
| Low income | 1.01 (0.90–1.13) | 1.05 (0.94–1.18) |
| Current smoking | 1.25 (1.12–1.40) | 1.21 (1.07–1.35) |
| Excess alcohol consumption | 1.49 (1.32–1.69) | 1.44 (1.27–1.64) |
| Regular exercise | 1.00 (0.90–1.11) | 0.98 (0.88–1.09) |
| BMI (vs. <23 kg/m2) | ||
| 23–24.9 | 0.95 (0.85–1.05) | 0.94 (0.84–1.05) |
| ≥25 | 0.93 (0.84–1.03) | 0.90 (0.81–0.99) |
| CCIb | 1.14 (1.11–1.17) | 1.12 (1.09–1.16) |
| Diabetes | 1.58 (1.41–1.76) | 1.43 (1.27–1.60) |
| Hypercholesterolemia | 0.91 (0.83–0.99) | 0.92 (0.85–1.01) |
Abbreviation: BMI, Body mass index; CCI, Charlson comorbidity index; NSAID, nonsteroidal antiinflammatory drug.
aAdjusted for sociodemographic characteristics (age, sex, economic status), behavioral factors (cigarette smoking, alcohol consumption, physical activity), medical conditions (body mass index, type 2 diabetes, hypercholesterolemia, CCI score), and medication use (aspirin, nonaspirin NSAIDs).
bHRs are per 1 point.
Discussion
Besides avoiding obesity and smoking, few options exist for preventing pancreatic cancer. Although aspirin is appealing for chemoprevention, none of the possible mechanisms has been convincingly established (4), and epidemiologic studies yielded mixed results. The highest-quality previous study was that of Khalaf and colleagues, with its large size, long follow-up, prospective design, and objective ascertainment of aspirin usage (7). However, all participants were health care workers, limiting their ability to represent the general population, and plasma salicylate levels do not definitively confirm aspirin exposure.
As aspirin's cardiovascular protective effects are well established, only individuals without cardiovascular risks may be ethically eligible for long-term clinical trials. To generate unbiased results regarding aspirin chemoprevention of cancer, well-designed cohort studies could be the best alternative.
Our results add to current evidence that aspirin does not prevent incident pancreatic cancer. Although our study is based on observational data, it was designed to estimate the effect size of drug exposure while minimizing bias. We used a randomly selected sample to minimize selection bias, determined aspirin exposure using national prescription database and validated DDD system to minimize recall bias, and choose a study time window to minimize survivor bias and immortal time bias.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: M.-H. Kim, S.M. Park, Y.H. Yun, I.C. Hwang
Development of methodology: M.-H. Kim, S.M. Park, I.C. Hwang
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.M. Park, I.C. Hwang
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.-H. Kim, S.M. Park, I.C. Hwang
Writing, review, and/or revision of the manuscript: M.-H. Kim, S.M. Park, Y.H. Yun, I.C. Hwang
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.-H. Kim, S.M. Park
Study supervision: M.-H. Kim, S.M. Park, Y.H. Yun, I.C. Hwang
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.
RSS Feeds