Abstract
Background: COX-2 overexpression may contribute to colorectal cancer occurrence. Aspirin and nonsteroidal anti-inflammatory drugs (NSAIDs) can reduce colorectal cancer recurrence, but the efficacy of primary prevention in Asian populations is still elusive. Thus, we examined the primary preventive efficacy of aspirin and NSAIDs on colorectal cancer incidence in Taiwan.
Methods: A nested case–control study was conducted using the National Health Insurance Research Database (NHIRD) in Taiwan. We identified patients with diagnosis of colorectal cancer from 2005 to 2013 in the Registry of Catastrophic Illness Patient Database. We selected patients without colorectal cancer from the Longitudinal Health Insurance Database as the controls and matched them with cases. NSAID exposure was defined as at least two prescriptions 13 to 48 months prior to the index date. Conditional logistic regression models were performed to evaluate the association between NSAID use and colorectal cancer.
Results: A total of 65,208 colorectal cancer cases and 65,208 matched controls were identified. Patients with aspirin use had a lower risk of colorectal cancer compared with nonusers [adjusted OR (AOR) = 0.94, 95% confidence interval (CI) = 0.90–0.99]. NSAID use was associated with lower incidence of colorectal cancer (AOR = 0.96; 95% CI = 0.92–1.00). When examining colon or rectal cancer, similar decreased risks were observed. Patients taking more cumulative days of NSAIDs use tended to experience a more protective effect on colorectal cancer, but no dose–response effects were noted.
Conclusions: Aspirin and NSAIDs were associated with a reduced risk of colorectal cancer development among a study cohort in an Asian population.
Impact: This study provided a possible chemoprevention for colorectal cancer in an Asian population. Cancer Epidemiol Biomarkers Prev; 27(7); 737–45. ©2018 AACR.
This article is featured in Highlights of This Issue, p. 717
Introduction
Nonsteroidal anti-inflammatory drugs (NSAIDs) are widely used as symptomatic treatment for conditions such as acute pain, chronic inflammatory, and degenerative joint diseases (1). Generally, NSAIDs are divided into nonselective and selective NSAIDs depending on whether or not the inhibitory target is on COX-2 selectivity. Colorectal cancer is the third deadliest cancer in America (2), and also the second most common cancer in Taiwan (3). The age-standardized rates of colorectal cancer incidence and death are even higher in Taiwan than in the Western world (4).
It is known that the mechanisms by which the enzymes or COX cause inflammation are related to inhibition of the formation of prostaglandins, prostacyclins, and thromboxanes (5). Previous studies have documented that COX-2 plays an important role in the pathogenesis of colorectal cancer (6, 7). Given the association between COX-2 and colorectal cancer, many studies have investigated the efficacy of NSAIDs as chemoprevention for colorectal cancer. For example, several studies have shown that aspirin reduces the recurrence of adenomatous polyps among patients with a history of colorectal cancer (8–10). However, other randomized trials failed to show that low-dose aspirin given as primary prevention lowered the risk of colorectal cancer (11, 12).
Although, numerous studies have examined the role of aspirin or NSAIDs on colorectal cancer chemoprevention, data regarding primary prevention among Asian populations are still unclear. Furthermore, dose and duration response data are also limited. To extend our understanding in this area, we conducted a nested case–control study to examine the primary preventive efficacy of aspirin and NSAIDs on colorectal cancer incidence in a large nationwide population-based cohort in Taiwan.
Materials and Methods
Data source
The data used in this study were derived from the National Health Insurance Research Database (NHIRD) in Taiwan. The National Health Insurance (NHI) program in Taiwan was launched in 1995 and the NHIRD has been released for research purposes since 1998. NHIRD has collected demographic characteristics, outpatient visits and inpatient claims data, and prescription records and disease diagnoses for approximately 99% of the entire population of 23 million people residing in Taiwan. However, the data in NHIRD before 2001 were not comprehensive for research purpose. Therefore, in this study, we set 2005 as index year and evaluated 4-year exposure duration for cases in 2005, starting from 2001.
Cases of colorectal cancer and matched controls
The incident cases of colorectal cancer were defined as patients diagnosed with colorectal cancer from 2005–2013 (ICD-9-CM codes: 153 and 154) who were identified from the Registry of Catastrophic Illness Patient Database (RCIPD). The date of the first diagnosed record of colorectal cancer was defined as the index date. In detail, patients were mandated to have histologic, radiographic, and/or pathologic confirmations for cancer diagnosis before catastrophic illness certificates were issued and included in the RCIPD. Of note, in this study, we did not include those with in situ malignancies because such early-stage diseases do not qualify for catastrophic illness certificates and hence are not captured by the RCIPD. As we intended to analyze the association between NSAID and colorectal cancer incidence, we excluded nearly 80,000 cases before the year 2005.
For each case, we randomly selected one control from the Longitudinal Health Insurance Database for the year 2005 (LHID2005) who did not have a colorectal cancer diagnosis at the time when the matched case was diagnosed with colorectal cancer. The LHID2005 is a subset derived from the NHIRD and includes medical claims data of 1,000,000 individuals randomly sampled from the 2005 NHI Registry for Beneficiaries. Previous studies have reported that there is no difference of distributions in age and gender between LHID 2005 and NHIRD (13). Therefore, the LHID2005 is representative of the original NHIRD. We then applied risk set sampling and randomly selected one control from the LHID2005 without having a colorectal cancer diagnosis at the time of matching colorectal cancer case. The controls were matched with the cases by age, sex, and index year. Controls were assigned the same index date as their matched cases. Of note, the exclusion criteria were as follows: (i) less than 30 years old or over 85 years old on the first diagnostic date of colorectal cancer; (ii) previous diagnosis of cancer (ICD-9 codes: 140–208) or benign lesions (ICD-9 code: 210–239) prior to 2005; (iii) diagnosis of ulcerative colitis (ICD-9 code: 556), familial adenomatous polyposis (ICD-9 code: 211.3) or Crohn disease (ICD-9 code: 555) prior to 2005; and/or (iv) participation in the NHI program less than 4 years before the index date. In addition, for controls, subjects with colorectal cancer diagnosed prior to 2005 were excluded.
Utilization of NSAIDs
We identified information related to NSAID use from prescription records in the NHIRD including types of prescribed medication, time of prescription, and duration of drug supply and dosage. NSAIDs were classified as follows: (i) aspirin; (ii) selective COX-2 inhibitors: celecoxib, etoricoxib, and rofecoxib; (iii) propionic acid derivatives: ibuprofen, naproxen, ketoprofen, fenbufen, fenoprofen, alminoprofen, flurbiprofen, and tiaprofenic; (iv) acetic acid derivatives: indomethacin, alclofenac, aceclofenac, acemetacin, nabumetone, diclofenac, etodolac, ketorolac, sulindac, and tolmetin; (v) enolic acid derivatives: piroxicam, meloxicam, and tenoxicam; (vi) anthranilic: flufenamic, mefenamic, niflumic, and tolfenamic; and (vii) sulphonailide: nimesulide.
NSAID exposure
In this study, we examined the long-term effects of NSAID (including aspirin, selective, and nonselective NSAIDs) use. Specifically, we defined users, recent or former users were based on a previous population-based case–control Demark study (14). We accounted for the duration of exposure to NSAID use that was between 13 months and 48 months prior to the index date. Study participants were categorized on the basis of their NSAID exposure status. First, patients exposed to NSAIDs were classified as users (defined as at least two prescriptions of NSAIDs during the 13–48 months prior to the index date) or nonusers. Patients with only one prescription would be excluded from the subsequent analyses. Next, we classified patients as recent users (defined as at least two prescriptions of NSAIDs in the 13–24 months prior to the index date), former users (defined as at least two prescriptions of NSAIDs in the 25–48 months prior to the index date, but not in the 13–24 months prior to the index date), or nonusers. Furthermore, we grouped patients into continuous users (defined as at least two prescriptions of NSAIDs per months and continuing for over 6 months in the period of 13–48 months prior to the index date), noncontinuous users (defined as not meeting the criteria of continuous use, but with identified prescriptions of aspirin or NSAIDs in the period of 13–48 months prior to the index date), or nonusers (14).
Adjustment of potential confounding factors
We took into account the potential confounding effects caused by comorbid medical disorders, concomitant medication use, and health service utilization; and adjusted for these factors in the subsequent analytical models. In detail, comorbid medical disorders included: type 2 diabetes mellitus, alcoholism, chronic pulmonary disease, asthma, migraines, rheumatic diseases, cardiovascular diseases (e.g., acute myocardial infarction, ischemic heart disease, congestive heart failure, and atrial flutter), and cerebrovascular disease; and concomitant drugs included: bisphosphonates, statins, antidepressants, angiotensin converting enzyme inhibitors/angiotensin receptor blockers, and hormone replacement agents. Health service utilization was computed using the number of ambulatory visits and hospitalizations during 2003 and 2004.
Statistical analysis
Descriptive statistics for the colorectal cancer cases and matched controls are presented as counts and corresponding percentages, or by mean and the corresponding SD for demographic and clinical characteristics, comorbid medical disorders, concomitant medication use, and health service utilization. The Student t test was used for testing continuous variables and the χ2 test was used for testing discrete variables, separately, to compare demographic and clinical characteristics between colorectal cancer cases and matched controls. Conditional logistic regression models (with and without covariate adjustment) were carried out to evaluate the association between NSAID use (aspirin and selective and nonselective NSAIDs) and colorectal cancer, based on NSAID exposure status, various drug classes, number of cumulative use days, and defined daily dose (DDD), which was defined as “the assumed maintenance dose per day for a drug used for its main indication.” If patients took two or more classes of NSAIDs at the same time, we defined those patients as combined users. In this study, most subjects taking aspirin dose less than 100 mg per day, only less than 0.5% of aspirin users took dose over 100 mg per day. For NSAIDs, we used the data in WHO Collaborating Centre for Drug Statistics Methodology (https://www.whocc.no/) as the source to reference of DDD of NSAIDs examined in this study (15). In addition, the definition of “user” might not precisely capture users of aspirin/ NSAIDs, we further grouped participants based on number of prescriptions, that is, nonusers versus participants with more than 12 prescriptions. We then assessed the association between aspirin/ NSAID use and colorectal cancer based on this definition as a sensitivity analysis. The covariates that were adjusted in the analytical models are listed above. P values less than 0.05 were declared to be statistically significant. All analyses were conducted using SAS version 9.2 for Windows (SAS Institute).
Ethics approval
The Institutional Review Board of the National Health Research Institutes, Taiwan approved the study protocol.
Results
A total of 130,416 study participants (65,208 colorectal cancer cases and 65,208 matched controls) with cases diagnosed between January 1, 2005, and December 31, 2013, were included. Figure 1 shows the detailed flow chart for identification of the study participants. Demographic and clinical characteristics of the study participants are presented in Table 1. The mean age was 65.61 ± 12.14 years in both groups, and 43.3% were females. The number of prescriptions for aspirin and NSAIDs, separately, and concomitant drugs and comorbidities are also provided in Table 1.
. | Controls (n = 65,208) . | Cases (n = 65,208) . | P . |
---|---|---|---|
Demographic characteristics | |||
Age (year; mean ± SD) | 65.61 ± 12.14 | 65.61 ± 12.14 | 1.00 |
Gender (n, %) | |||
Males | 36,964 (56.69) | 36,964 (56.69) | 1.00 |
Females | 28,244 (43.31) | 28,244 (43.31) | |
Number of prescriptions for aspirin (n, %) | |||
0–1 | 49,170 (75.40) | 49,095 (75.29) | 0.66 |
2–12 | 6,112 (9.37) | 6,036 (9.26) | |
13–24 | 3,077 (4.72) | 3,167 (4, 86) | |
≥25 | 4,332 (6.64) | 4,338 (6.65) | |
Number of prescriptions for NSAIDs (n, %) | |||
0–1 | 9,783 (15.00) | 10,092 (15.48) | <0.01 |
2–12 | 32,240 (49.44) | 32,758 (50.24) | |
13–24 | 9,452 (14.50) | 8,970 (13.76) | |
≥25 | 7,592 (11.64) | 6,735 (10.33) | |
Clinical characteristics | |||
Concomitant drugs (n, %) | |||
Statins | 8,947 (13.72) | 9,072 (13.91) | 0.32 |
Antidepressants | 5,164 (7.92) | 5,607 (8.60) | <0.01 |
ACEIs or ARBs | 16,733 (25.66) | 18,088 (27.74) | <0.01 |
Bisphosphonates | 93 (0.14) | 70 (0.11) | 0.07 |
Hormone replacement therapy | 1,250 (1.92) | 1,319 (2.02) | 0.17 |
Comorbidity (n, %) | |||
T2DM | 12,141 (18.62) | 15,502 (23.77) | <0.01 |
Alcoholism | 120 (0.18) | 192 (0.29) | <0.01 |
COPD or asthma | 5,888 (9.03) | 6,445 (9.88) | <0.01 |
Migraine | 692 (1.06) | 665 (1.02) | 0.46 |
Rheumatic, soft tissue, or connective tissue disease | 22,503 (34.51) | 21,878 (33.55) | <0.01 |
AMI/ischemic heart disease | 8,475 (13.00) | 9,161 (14.05) | <0.01 |
Congestive heart failure | 852 (1.31) | 1,163 (1.78) | <0.01 |
Atrial flutter | 2,135 (3.27) | 2,430 (3.73) | <0.01 |
Cerebrovascular disease | 6,291 (9.65) | 6,639 (10.18) | <0.01 |
. | Controls (n = 65,208) . | Cases (n = 65,208) . | P . |
---|---|---|---|
Demographic characteristics | |||
Age (year; mean ± SD) | 65.61 ± 12.14 | 65.61 ± 12.14 | 1.00 |
Gender (n, %) | |||
Males | 36,964 (56.69) | 36,964 (56.69) | 1.00 |
Females | 28,244 (43.31) | 28,244 (43.31) | |
Number of prescriptions for aspirin (n, %) | |||
0–1 | 49,170 (75.40) | 49,095 (75.29) | 0.66 |
2–12 | 6,112 (9.37) | 6,036 (9.26) | |
13–24 | 3,077 (4.72) | 3,167 (4, 86) | |
≥25 | 4,332 (6.64) | 4,338 (6.65) | |
Number of prescriptions for NSAIDs (n, %) | |||
0–1 | 9,783 (15.00) | 10,092 (15.48) | <0.01 |
2–12 | 32,240 (49.44) | 32,758 (50.24) | |
13–24 | 9,452 (14.50) | 8,970 (13.76) | |
≥25 | 7,592 (11.64) | 6,735 (10.33) | |
Clinical characteristics | |||
Concomitant drugs (n, %) | |||
Statins | 8,947 (13.72) | 9,072 (13.91) | 0.32 |
Antidepressants | 5,164 (7.92) | 5,607 (8.60) | <0.01 |
ACEIs or ARBs | 16,733 (25.66) | 18,088 (27.74) | <0.01 |
Bisphosphonates | 93 (0.14) | 70 (0.11) | 0.07 |
Hormone replacement therapy | 1,250 (1.92) | 1,319 (2.02) | 0.17 |
Comorbidity (n, %) | |||
T2DM | 12,141 (18.62) | 15,502 (23.77) | <0.01 |
Alcoholism | 120 (0.18) | 192 (0.29) | <0.01 |
COPD or asthma | 5,888 (9.03) | 6,445 (9.88) | <0.01 |
Migraine | 692 (1.06) | 665 (1.02) | 0.46 |
Rheumatic, soft tissue, or connective tissue disease | 22,503 (34.51) | 21,878 (33.55) | <0.01 |
AMI/ischemic heart disease | 8,475 (13.00) | 9,161 (14.05) | <0.01 |
Congestive heart failure | 852 (1.31) | 1,163 (1.78) | <0.01 |
Atrial flutter | 2,135 (3.27) | 2,430 (3.73) | <0.01 |
Cerebrovascular disease | 6,291 (9.65) | 6,639 (10.18) | <0.01 |
NOTE: The significance of bold values is P < 0.05.
Abbreviations: ACEIs/ARBs: angiotensin converting-enzyme inhibitors/angiotensin receptor blockers; AMI: acute myocardial infarction; COPD: chronic obstructive pulmonary disease; T2DM: type 2 diabetes mellitus.
Table 2 presents the association between NSAID use and colorectal cancer. Aspirin use was inversely associated with colorectal cancer compared with its counterparts [AOR = 0.94, 95% confidence interval (CI): 0.90–0.99]. When further stratifying study participants to non-, recent-, and former aspirin users, similar associations were found (adjusted OR = 0.98; 95% CI, 0.93–1.02 for recent users and AOR = 0.85; 95% CI, 0.79–0.92 for former users). Continuous use of aspirin was associated with lower colorectal cancer risk significantly (adjusted OR = 0.89; 95% CI, 0.83–0.97). NSAID use was also associated with a decreased risk of colorectal cancer compared with its counterparts (AOR = 0.96; 95% CI, 0.92–1.00). Similarly, when stratifying participants to nonuser, recent- and former users or to nonuser, and continuous and noncontinuous users, significant beneficial effects were found on colorectal cancer (AOR = 0.94; 95% CI, 0.89–0.98 for recent users and AOR = 0.67; 95% CI, 0.63–0.73 for continuous users). In addition, when we further grouped participants based on number of prescription, no significant association was found for aspirin users with >12 prescriptions (AOR = 1.01; 95% CI, 0.95–1.06), compared with aspirin users with >2 prescriptions (AOR = 0.94; 95% CI, 0.90–0.99). Whereas for NSAID users, more strong significant association was found for NSAID users >12 prescriptions (AOR = 0.78; 95% CI, 0.74–0.83), compared with NSAID users with >2 prescriptions (AOR = 0.96; 95% CI, 0.92–1.00).
. | Controls . | Cases . | Total . | Crude OR (95% CI) . | AORa (95% CI) . |
---|---|---|---|---|---|
Aspirin | |||||
Users vs. nonusers | |||||
Nonuserb | 49,170 (78.43) | 49,095 (78.38) | 98,265 | Reference | Reference |
User | 13,521 (21.57) | 13,541 (21.62) | 27,062 | 1.01 (0.98–1.04) | 0.94c (0.90–0.99) |
Total | 62,691 | 62,636 | 125,327 | ||
Recent or former users vs. nonusers | |||||
Nonuser | 49,170 (78.43) | 49,095 (78.38) | 98,265 | Reference | Reference |
Recent user | 10,480 (16.72) | 10,526 (16.81) | 21,006 | 1.01 (0.98–1.04) | 0.98 (0.93–1.02) |
Former user | 3,041 (4.85) | 3,015 (4.81) | 6,056 | 1.00 (0.95–1.06) | 0.85 (0.79–0.92) |
Total | 62,691 | 62,636 | 125,327 | ||
Continuous or noncontinuous users vs. nonusers | |||||
Nonuser | 49,170 (78.43) | 49,095 (78.38) | 98,265 | Reference | Reference |
Continuous user | 3,129 (4.99) | 3,103 (4.95) | 6,232 | 1.00 (0.95–1.05) | 0.89 (0.83–0.97) |
Noncontinuous user | 10,392 (16.58) | 10,438 (16.66) | 20,830 | 1.01 (0.98–1.04) | 0.96 (0.91–1.00) |
Total | 62,691 | 62,636 | 125,327 | ||
Subjects with >12 prescriptions vs. nonusers | |||||
Nonuser | 49,170 (78.43) | 49,095 (78.38) | 98,265 | Reference | Reference |
>12 prescriptions | 7,409 (11.82) | 7,505 (11.98) | 14,914 | 1.02 (0.98–1.06) | 1.01 (0.95–1.06) |
Total | 56,579 | 56,600 | 113,179 | ||
NSAIDs | |||||
Users vs. nonusers | |||||
Nonuser | 9,783 (16.56) | 10,092 (17.24) | 19,875 | Reference | Reference |
User | 49,284 (83.44) | 48,463 (82.76) | 97,747 | 0.95 (0.92–0.98) | 0.96 (0.92–1.00) |
Total | 59,067 | 58,555 | 117,622 | ||
Recent or former users vs. nonusers | |||||
Nonuser | 9,783 (16.56) | 10,092 (17.24) | 19,875 | Reference | Reference |
Recent user | 39,134 (66.25) | 38,300 (65.41) | 77,434 | 0.95 (0.92–0.98) | 0.94 (0.89–0.98) |
Former user | 10,150 (17.18) | 10,163 (17.36) | 20,313 | 0.97 (0.93–1.01) | 1.02 (0.96–1.07) |
Total | 59,067 | 58,555 | 117,622 | ||
Continuous or noncontinuous users vs. nonusers | |||||
Nonuser | 9,783 (16.56) | 10,092 (17.24) | 19,875 | Reference | Reference |
Continuous user | 5,149 (8.72) | 4,523 (7.72) | 9,672 | 0.85 (0.80–0.89) | 0.67 (0.63–0.73) |
Noncontinuous user | 44,135 (74.72) | 43,940 (75.04) | 88,075 | 0.96 (0.93–1.00) | 0.97 (0.93–1.01) |
Total | 59,067 | 58,555 | 117,622 | ||
Subjects with >12 prescriptions vs. nonusers | |||||
Nonuser | 49,170 (78.43) | 49,095 (78.38) | 98,265 | Reference | Reference |
>12 prescriptions | 17,044 (28.86) | 15,705 (26.82) | 32,749 | 0.88 (0.85–0.92) | 0.78 (0.74–0.83) |
Total | 26,827 | 25,797 | 52,624 |
. | Controls . | Cases . | Total . | Crude OR (95% CI) . | AORa (95% CI) . |
---|---|---|---|---|---|
Aspirin | |||||
Users vs. nonusers | |||||
Nonuserb | 49,170 (78.43) | 49,095 (78.38) | 98,265 | Reference | Reference |
User | 13,521 (21.57) | 13,541 (21.62) | 27,062 | 1.01 (0.98–1.04) | 0.94c (0.90–0.99) |
Total | 62,691 | 62,636 | 125,327 | ||
Recent or former users vs. nonusers | |||||
Nonuser | 49,170 (78.43) | 49,095 (78.38) | 98,265 | Reference | Reference |
Recent user | 10,480 (16.72) | 10,526 (16.81) | 21,006 | 1.01 (0.98–1.04) | 0.98 (0.93–1.02) |
Former user | 3,041 (4.85) | 3,015 (4.81) | 6,056 | 1.00 (0.95–1.06) | 0.85 (0.79–0.92) |
Total | 62,691 | 62,636 | 125,327 | ||
Continuous or noncontinuous users vs. nonusers | |||||
Nonuser | 49,170 (78.43) | 49,095 (78.38) | 98,265 | Reference | Reference |
Continuous user | 3,129 (4.99) | 3,103 (4.95) | 6,232 | 1.00 (0.95–1.05) | 0.89 (0.83–0.97) |
Noncontinuous user | 10,392 (16.58) | 10,438 (16.66) | 20,830 | 1.01 (0.98–1.04) | 0.96 (0.91–1.00) |
Total | 62,691 | 62,636 | 125,327 | ||
Subjects with >12 prescriptions vs. nonusers | |||||
Nonuser | 49,170 (78.43) | 49,095 (78.38) | 98,265 | Reference | Reference |
>12 prescriptions | 7,409 (11.82) | 7,505 (11.98) | 14,914 | 1.02 (0.98–1.06) | 1.01 (0.95–1.06) |
Total | 56,579 | 56,600 | 113,179 | ||
NSAIDs | |||||
Users vs. nonusers | |||||
Nonuser | 9,783 (16.56) | 10,092 (17.24) | 19,875 | Reference | Reference |
User | 49,284 (83.44) | 48,463 (82.76) | 97,747 | 0.95 (0.92–0.98) | 0.96 (0.92–1.00) |
Total | 59,067 | 58,555 | 117,622 | ||
Recent or former users vs. nonusers | |||||
Nonuser | 9,783 (16.56) | 10,092 (17.24) | 19,875 | Reference | Reference |
Recent user | 39,134 (66.25) | 38,300 (65.41) | 77,434 | 0.95 (0.92–0.98) | 0.94 (0.89–0.98) |
Former user | 10,150 (17.18) | 10,163 (17.36) | 20,313 | 0.97 (0.93–1.01) | 1.02 (0.96–1.07) |
Total | 59,067 | 58,555 | 117,622 | ||
Continuous or noncontinuous users vs. nonusers | |||||
Nonuser | 9,783 (16.56) | 10,092 (17.24) | 19,875 | Reference | Reference |
Continuous user | 5,149 (8.72) | 4,523 (7.72) | 9,672 | 0.85 (0.80–0.89) | 0.67 (0.63–0.73) |
Noncontinuous user | 44,135 (74.72) | 43,940 (75.04) | 88,075 | 0.96 (0.93–1.00) | 0.97 (0.93–1.01) |
Total | 59,067 | 58,555 | 117,622 | ||
Subjects with >12 prescriptions vs. nonusers | |||||
Nonuser | 49,170 (78.43) | 49,095 (78.38) | 98,265 | Reference | Reference |
>12 prescriptions | 17,044 (28.86) | 15,705 (26.82) | 32,749 | 0.88 (0.85–0.92) | 0.78 (0.74–0.83) |
Total | 26,827 | 25,797 | 52,624 |
aAdjusted covariates included: comorbid medical disorders (type 2 diabetes mellitus, alcoholism, chronic obstructive pulmonary disease, asthma, migraine, cardiovascular disease, cerebrovascular disease, acute myocardial infarction, ischemic heart disease, congestive heart failure, atrial flutter, and cerebrovascular disease), concomitant drugs (bisphosphonates, statins, antidepressants, angiotensin converting enzyme inhibitors/angiotensin receptor blockers, and hormone replacement agents), and health service utilization (ambulatory visits and inpatient visits).
bStudy participants without aspirin or NSAID use were treated as the reference group.
cSignificant results (P < 0.05) are in bold.
When investigating cumulative use days and DDD, a significantly reduced colorectal cancer risk was observed among NSAID users (Table 3). Specifically, patients who took NSAIDS for more cumulative days tended to have a more reduced colorectal cancer risk (AOR = 0.96; 95% CI, 0.92–1.01 for 1–365 days; AOR = 0.71; 95% CI, 0.64–0.78 for 366–730 days; AOR = 0.60; 95% CI, 0.52–0.70 for more than 730 days; Ptrend < 0−3). However, no dose–response effects were found for the use of NSAIDs (Table 3).
. | Controls . | Cases . | Crude OR (95% CI) . | Adjusted ORa (95% CI) . |
---|---|---|---|---|
Aspirin | ||||
Cumulative days | ||||
Nonuserb | 49,170 (78.43) | 49,095 (78.38) | Reference | Reference |
1–365 days | 6,659 (10.62) | 6,554 (10.46) | 0.99 (0.95–1.03) | 0.87c (0.83–0.92) |
366–730 days | 3,008 (4.80) | 3,113 (4.97) | 1.04 (0.99–1.10) | 1.00 (0.93–1.08) |
>730 days | 3,854 (6.15) | 3,874 (6.18) | 1.02 (0.97–1.07) | 1.06 (0.98–1.13) |
Total | 62,691 | 62,636 | Ptrend = 0.53d | |
NSAIDs | ||||
Cumulative days | ||||
Nonuser | 9,783 (16.56) | 10,092 (17.24) | Reference | Reference |
1–365 days | 46,029 (77.93) | 45,715 (78.07) | 0.96 (0.93–0.99) | 0.96 (0.92–1.01) |
366–730 days | 2,330 (3.94) | 2,028 (3.46) | 0.83 (0.77–0.89) | 0.71 (0.64–0.78) |
>730 days | 925 (1.57) | 720 (1.23) | 0.74 (0.67–0.83) | 0.60 (0.52–0.70) |
Total | 59,067 | 58,555 | Ptrend <10−3 | |
DDD | ||||
Nonuser | 9,783 (16.56) | 10,092 (17.24) | Reference | Reference |
DDD ≤ 0.75 | 11,829 (20.03) | 11,927 (20.37) | 0.97 (0.93–1.01) | 0.96 (0.91–1.02) |
0.75 < DDD ≤ 1 | 14,008 (23.72) | 13,615 (23.25) | 0.95 (0.91–0.98) | 0.93 (0.88–0.98) |
1 < DDD ≤ 1.25 | 12,516 (21.19) | 12,042 (20.57) | 0.93 (0.89–0.97) | 0.95 (0.90–1.00) |
1.25 < DDD | 10,931 (18.51) | 10,879 (18.58) | 0.97 (0.93–1.01) | 0.99 (0.94–1.04) |
Total | 59,067 | 58,555 | Ptrend = 0.006 |
. | Controls . | Cases . | Crude OR (95% CI) . | Adjusted ORa (95% CI) . |
---|---|---|---|---|
Aspirin | ||||
Cumulative days | ||||
Nonuserb | 49,170 (78.43) | 49,095 (78.38) | Reference | Reference |
1–365 days | 6,659 (10.62) | 6,554 (10.46) | 0.99 (0.95–1.03) | 0.87c (0.83–0.92) |
366–730 days | 3,008 (4.80) | 3,113 (4.97) | 1.04 (0.99–1.10) | 1.00 (0.93–1.08) |
>730 days | 3,854 (6.15) | 3,874 (6.18) | 1.02 (0.97–1.07) | 1.06 (0.98–1.13) |
Total | 62,691 | 62,636 | Ptrend = 0.53d | |
NSAIDs | ||||
Cumulative days | ||||
Nonuser | 9,783 (16.56) | 10,092 (17.24) | Reference | Reference |
1–365 days | 46,029 (77.93) | 45,715 (78.07) | 0.96 (0.93–0.99) | 0.96 (0.92–1.01) |
366–730 days | 2,330 (3.94) | 2,028 (3.46) | 0.83 (0.77–0.89) | 0.71 (0.64–0.78) |
>730 days | 925 (1.57) | 720 (1.23) | 0.74 (0.67–0.83) | 0.60 (0.52–0.70) |
Total | 59,067 | 58,555 | Ptrend <10−3 | |
DDD | ||||
Nonuser | 9,783 (16.56) | 10,092 (17.24) | Reference | Reference |
DDD ≤ 0.75 | 11,829 (20.03) | 11,927 (20.37) | 0.97 (0.93–1.01) | 0.96 (0.91–1.02) |
0.75 < DDD ≤ 1 | 14,008 (23.72) | 13,615 (23.25) | 0.95 (0.91–0.98) | 0.93 (0.88–0.98) |
1 < DDD ≤ 1.25 | 12,516 (21.19) | 12,042 (20.57) | 0.93 (0.89–0.97) | 0.95 (0.90–1.00) |
1.25 < DDD | 10,931 (18.51) | 10,879 (18.58) | 0.97 (0.93–1.01) | 0.99 (0.94–1.04) |
Total | 59,067 | 58,555 | Ptrend = 0.006 |
aAdjusted covariates: comorbid medical disorders (type 2 diabetes mellitus, alcoholism, chronic obstructive pulmonary disease, asthma, migraine, cardiovascular disease, cerebrovascular disease, acute myocardial infarction, ischemic heart disease, congestive heart failure, atrial flutter and cerebrovascular disease), concomitant drugs (bisphosphonates, statins, antidepressants, angiotensin converting-enzyme inhibitors/angiotensin receptor blockers and hormone replacement agents), and health service utilization (ambulatory visits and inpatient visits).
bStudy participants without aspirin or NSAID use were treated as the reference group.
cSignificant results (P < 0.05) are in bold.
dP values are obtained from the Cochran-Armitage trend test.
Furthermore, we assessed the associations between different NSAID classes and colorectal cancer. Table 4 shows a significant beneficial effect on colorectal cancer among patients with a combined use of selective and nonselective NSAIDs (AOR = 0.81; 95% CI, 0.76–0.87). In addition, the combined use of various nonselective NSAIDs was also found to significantly reduce the risk of colorectal cancer (AOR = 0.93; 95% CI, 0.89–0.98). No associations were found when examining individual classes of NSAID use (Table 4).
. | Controls . | Cases . | Total . | COR (95% CI) . | AORa (95% CI) . |
---|---|---|---|---|---|
Selective vs. nonselective NSAIDs | |||||
Nonuserb | 9,783 (16.56) | 10,092 (17.24) | 19,875 | Reference | Reference |
Nonselective NSAIDs | 41,249 (69.83) | 41,035 (70.08) | 82,284 | 0.96c (0.93–1.00) | 0.97 (0.93–1.02) |
Selective NSAIDs | 198 (0.34) | 199 (0.34) | 397 | 1.03 (0.84–1.28) | 1.11 (0.83–1.48) |
Combinationd | 7,837 (13.27) | 7,229 (12.35) | 15,066 | 0.89 (0.85–0.93) | 0.81 (0.76–0.87) |
Total | 59,067 | 58,555 | 117,622 | ||
Different classes of nonselective NSAIDs | |||||
Nonuser | 9,783 (16.56) | 10,092 (17.24) | 19,875 | Reference | Reference |
Propionic | 1,484 (2.52) | 1,593 (2.73) | 3,077 | 1.06 (0.98–1.15) | 1.11 (1.00–1.23) |
Acetic | 5,201 (8.83) | 5,296 (9.08) | 10,497 | 0.98 (0.93–1.03) | 1.01 (0.95–1.08) |
Enolic | 409 (0.69) | 410 (0.70) | 819 | 0.96 (0.83–1.12) | 1.00 (0.82–1.22) |
Anthranilic | 1,796 (3.05) | 1,961 (3.36) | 3,757 | 1.05 (0.97–1.13) | 1.05 (0.96–1.16) |
Sulphonailide | 63 (0.11) | 70 (0.12) | 133 | 1.05 (0.74–1.50) | 0.82 (0.51–1.32) |
Combinatione | 40,133 (68.17) | 38,934 (66.72) | 79,067 | 0.94 (0.91–0.97) | 0.93 (0.89–0.98) |
Total | 58,869 | 58,356 | 117,225 |
. | Controls . | Cases . | Total . | COR (95% CI) . | AORa (95% CI) . |
---|---|---|---|---|---|
Selective vs. nonselective NSAIDs | |||||
Nonuserb | 9,783 (16.56) | 10,092 (17.24) | 19,875 | Reference | Reference |
Nonselective NSAIDs | 41,249 (69.83) | 41,035 (70.08) | 82,284 | 0.96c (0.93–1.00) | 0.97 (0.93–1.02) |
Selective NSAIDs | 198 (0.34) | 199 (0.34) | 397 | 1.03 (0.84–1.28) | 1.11 (0.83–1.48) |
Combinationd | 7,837 (13.27) | 7,229 (12.35) | 15,066 | 0.89 (0.85–0.93) | 0.81 (0.76–0.87) |
Total | 59,067 | 58,555 | 117,622 | ||
Different classes of nonselective NSAIDs | |||||
Nonuser | 9,783 (16.56) | 10,092 (17.24) | 19,875 | Reference | Reference |
Propionic | 1,484 (2.52) | 1,593 (2.73) | 3,077 | 1.06 (0.98–1.15) | 1.11 (1.00–1.23) |
Acetic | 5,201 (8.83) | 5,296 (9.08) | 10,497 | 0.98 (0.93–1.03) | 1.01 (0.95–1.08) |
Enolic | 409 (0.69) | 410 (0.70) | 819 | 0.96 (0.83–1.12) | 1.00 (0.82–1.22) |
Anthranilic | 1,796 (3.05) | 1,961 (3.36) | 3,757 | 1.05 (0.97–1.13) | 1.05 (0.96–1.16) |
Sulphonailide | 63 (0.11) | 70 (0.12) | 133 | 1.05 (0.74–1.50) | 0.82 (0.51–1.32) |
Combinatione | 40,133 (68.17) | 38,934 (66.72) | 79,067 | 0.94 (0.91–0.97) | 0.93 (0.89–0.98) |
Total | 58,869 | 58,356 | 117,225 |
aAdjusted covariates: comorbid medical disorders (type 2 diabetes mellitus, alcoholism, chronic obstructive pulmonary disease, asthma, migraine, cardiovascular disease, cerebrovascular disease, acute myocardial infarction, ischemic heart disease, congestive heart failure, atrial flutter, and cerebrovascular disease), concomitant drugs (bisphosphonates, statins, antidepressants, angiotensin converting-enzyme inhibitors/angiotensin receptor blockers, and hormone replacement agents), and health service utilization (ambulatory visits and inpatient visits).
bStudy participants without NSAID use were treated as the reference group.
cSignificant results (P < 0.05) are in bold.
dCombination is defined as combined use of selective and nonselective NSAIDs.
eCombination is defined as combined use of various classes of nonselective NSAIDs.
We further separated cases to patients with colon cancer or rectal cancer. Table 5 shows aspirin use reduced the risk of rectal cancer (AOR = 0.91; 95% CI, 0.85–0.98), but not colon cancer (AOR = 0.97; 95% CI, 0.91–1.03). The use of NSAIDs did not reduce the risk of colon cancer or rectal cancer, however, significant beneficial effects of NSAID use in continuous users were found on both colon and rectal cancers (AOR = 0.61; 95% CI, 0.55–0.67 for colon cancer and AOR = 0.78; 95% CI, 0.69–0.87 for rectal cancer).
(A) Colon cancer . | ||||
---|---|---|---|---|
. | Controls . | Cases . | Total . | Adjusted ORa (95% CI) . |
Aspirin | ||||
Users vs. nonusers | ||||
Nonuserb | 28,211 (78.21) | 29,087 (77.46) | 57,298 | Reference |
User | 7,861 (21.79) | 8,462 (22.64) | 16,323 | 0.97 (0.91–1.03) |
Total | 36,072 | 37,549 | 73,621 | |
Recent or former users vs. nonusers | ||||
Nonuser | 28,211 (78.21) | 29,087 (77.46) | 57,298 | Reference |
Recent user | 6,086 (16.87) | 6,607 (17.60) | 12,693 | 1.01 (0.95–1.08) |
Former user | 1,775 (4.92) | 1,855 (4.94) | 3,630 | 0.85c (0.77–0.94) |
Total | 36,072 | 37,549 | 73,621 | |
Continuous or noncontinuous users vs. nonusers | ||||
Nonuser | 28,211 (78.21) | 29,087 (77.46) | 57,298 | Reference |
Continuous user | 1,779 (4.99) | 1,931 (5.14) | 3,710 | 0.99 (0.89–1.09) |
Noncontinuous user | 6,082 (16.58) | 6,531 (17.39) | 12,613 | 0.96 (0.91–1.03) |
Total | 36,072 | 37,549 | 73,621 | |
NSAIDs | ||||
Users vs. nonusers | ||||
Nonuser | 5,187 (16.30) | 5,963 (16.98) | 11,150 | Reference |
User | 26,639 (83.70) | 29,165 (83.02) | 55,804 | 0.96 (0.91–1.02) |
Total | 31,826 | 35,128 | 66,954 | |
Recent or former users vs. nonusers | ||||
Nonuser | 5,187 (16.30) | 5,963 (16.98) | 11,150 | Reference |
Recent user | 21,226 (66.69) | 23,032 (65.57) | 44,258 | 0.94 (0.88–1.00) |
Former user | 5,413 (17.01) | 6,133 (17.46) | 11,546 | 1.04 (0.97–1.12) |
Total | 31,826 | 35,128 | 66,954 | |
Continuous or noncontinuous users vs. nonusers | ||||
Nonuser | 5,187 (16.30) | 5,963 (16.98) | 11,150 | Reference |
Continuous user | 2,904 (9.12) | 2,624 (7.47) | 5,528 | 0.61 (0.55–0.67) |
Noncontinuous user | 23,735 (74.58) | 26,541 (75.56) | 50,276 | 0.98 (0.92–1.04) |
Total | 31,826 | 35,128 | 66,954 | |
(B) Rectal cancer | ||||
Controls | Cases | Total | Adjusted ORa (95% CI) | |
Aspirin | ||||
Users vs. nonusers | ||||
Nonuserb | 18,348 (79.25) | 19,170 (79.74) | 37,518 | Reference |
User | 4,805 (20.75) | 4,872 (20.26) | 9,677 | 0.91c (0.85–0.98) |
Total | 23,153 | 24,042 | 47,195 | |
Recent or former users vs. nonusers | ||||
Nonuser | 18,348 (79.25) | 19,170 (79.74) | 37,518 | Reference |
Recent user | 3,745 (16.18) | 3,744 (15.57) | 7,489 | 0.93 (0.86–1.00) |
Former user | 1,060 (4.58) | 1,128 (4.69) | 2,188 | 0.87 (0.78–0.98) |
Total | 23,153 | 24,042 | 47,195 | |
Continuous or noncontinuous users vs. nonusers | ||||
Nonuser | 18,348 (79.25) | 19,170 (79.74) | 37,518 | Reference |
Continuous user | 1,125 (4.86) | 1,125 (4.68) | 2,250 | 0.80 (0.71–0.90) |
Noncontinuous user | 3,680 (15.89) | 3,747 (15.59) | 7,427 | 0.95 (0.88–1.02) |
Total | 23,153 | 24,042 | 47,195 | |
NSAIDs | ||||
Users vs. nonusers | ||||
Nonuser | 3,395 (16.74) | 3,960 (17.66) | 7,355 | Reference |
User | 16,890 (83.26) | 18,469 (82.34) | 35,359 | 0.95 (0.88–1.01) |
Total | 20,285 | 22,429 | 42,714 | |
Recent or former users vs. nonusers | ||||
Nonuser | 3,395 (16.74) | 3,960 (17.66) | 7,355 | Reference |
Recent user | 13,359 (65.86) | 14,594 (65.07) | 27,953 | 0.93 (0.87–1.00) |
Former user | 3,531 (17.41) | 3,875 (17.28) | 7,406 | 0.98 (0.91–1.07) |
Total | 20,285 | 22,429 | 42,714 | |
Continuous or noncontinuous users vs. nonusers | ||||
Nonuser | 3,395 (16.74) | 3,960 (17.66) | 7,355 | Reference |
Continuous user | 1,685 (8.31) | 1,829 (8.15) | 3,514 | 0.78 (0.69–0.87) |
Noncontinuous user | 15,205 (74.96) | 16,640 (74.19) | 31,845 | 0.95 (0.89–1.02) |
Total | 20,285 | 22,429 | 42,714 |
(A) Colon cancer . | ||||
---|---|---|---|---|
. | Controls . | Cases . | Total . | Adjusted ORa (95% CI) . |
Aspirin | ||||
Users vs. nonusers | ||||
Nonuserb | 28,211 (78.21) | 29,087 (77.46) | 57,298 | Reference |
User | 7,861 (21.79) | 8,462 (22.64) | 16,323 | 0.97 (0.91–1.03) |
Total | 36,072 | 37,549 | 73,621 | |
Recent or former users vs. nonusers | ||||
Nonuser | 28,211 (78.21) | 29,087 (77.46) | 57,298 | Reference |
Recent user | 6,086 (16.87) | 6,607 (17.60) | 12,693 | 1.01 (0.95–1.08) |
Former user | 1,775 (4.92) | 1,855 (4.94) | 3,630 | 0.85c (0.77–0.94) |
Total | 36,072 | 37,549 | 73,621 | |
Continuous or noncontinuous users vs. nonusers | ||||
Nonuser | 28,211 (78.21) | 29,087 (77.46) | 57,298 | Reference |
Continuous user | 1,779 (4.99) | 1,931 (5.14) | 3,710 | 0.99 (0.89–1.09) |
Noncontinuous user | 6,082 (16.58) | 6,531 (17.39) | 12,613 | 0.96 (0.91–1.03) |
Total | 36,072 | 37,549 | 73,621 | |
NSAIDs | ||||
Users vs. nonusers | ||||
Nonuser | 5,187 (16.30) | 5,963 (16.98) | 11,150 | Reference |
User | 26,639 (83.70) | 29,165 (83.02) | 55,804 | 0.96 (0.91–1.02) |
Total | 31,826 | 35,128 | 66,954 | |
Recent or former users vs. nonusers | ||||
Nonuser | 5,187 (16.30) | 5,963 (16.98) | 11,150 | Reference |
Recent user | 21,226 (66.69) | 23,032 (65.57) | 44,258 | 0.94 (0.88–1.00) |
Former user | 5,413 (17.01) | 6,133 (17.46) | 11,546 | 1.04 (0.97–1.12) |
Total | 31,826 | 35,128 | 66,954 | |
Continuous or noncontinuous users vs. nonusers | ||||
Nonuser | 5,187 (16.30) | 5,963 (16.98) | 11,150 | Reference |
Continuous user | 2,904 (9.12) | 2,624 (7.47) | 5,528 | 0.61 (0.55–0.67) |
Noncontinuous user | 23,735 (74.58) | 26,541 (75.56) | 50,276 | 0.98 (0.92–1.04) |
Total | 31,826 | 35,128 | 66,954 | |
(B) Rectal cancer | ||||
Controls | Cases | Total | Adjusted ORa (95% CI) | |
Aspirin | ||||
Users vs. nonusers | ||||
Nonuserb | 18,348 (79.25) | 19,170 (79.74) | 37,518 | Reference |
User | 4,805 (20.75) | 4,872 (20.26) | 9,677 | 0.91c (0.85–0.98) |
Total | 23,153 | 24,042 | 47,195 | |
Recent or former users vs. nonusers | ||||
Nonuser | 18,348 (79.25) | 19,170 (79.74) | 37,518 | Reference |
Recent user | 3,745 (16.18) | 3,744 (15.57) | 7,489 | 0.93 (0.86–1.00) |
Former user | 1,060 (4.58) | 1,128 (4.69) | 2,188 | 0.87 (0.78–0.98) |
Total | 23,153 | 24,042 | 47,195 | |
Continuous or noncontinuous users vs. nonusers | ||||
Nonuser | 18,348 (79.25) | 19,170 (79.74) | 37,518 | Reference |
Continuous user | 1,125 (4.86) | 1,125 (4.68) | 2,250 | 0.80 (0.71–0.90) |
Noncontinuous user | 3,680 (15.89) | 3,747 (15.59) | 7,427 | 0.95 (0.88–1.02) |
Total | 23,153 | 24,042 | 47,195 | |
NSAIDs | ||||
Users vs. nonusers | ||||
Nonuser | 3,395 (16.74) | 3,960 (17.66) | 7,355 | Reference |
User | 16,890 (83.26) | 18,469 (82.34) | 35,359 | 0.95 (0.88–1.01) |
Total | 20,285 | 22,429 | 42,714 | |
Recent or former users vs. nonusers | ||||
Nonuser | 3,395 (16.74) | 3,960 (17.66) | 7,355 | Reference |
Recent user | 13,359 (65.86) | 14,594 (65.07) | 27,953 | 0.93 (0.87–1.00) |
Former user | 3,531 (17.41) | 3,875 (17.28) | 7,406 | 0.98 (0.91–1.07) |
Total | 20,285 | 22,429 | 42,714 | |
Continuous or noncontinuous users vs. nonusers | ||||
Nonuser | 3,395 (16.74) | 3,960 (17.66) | 7,355 | Reference |
Continuous user | 1,685 (8.31) | 1,829 (8.15) | 3,514 | 0.78 (0.69–0.87) |
Noncontinuous user | 15,205 (74.96) | 16,640 (74.19) | 31,845 | 0.95 (0.89–1.02) |
Total | 20,285 | 22,429 | 42,714 |
aAdjusted covariates included: comorbid medical disorders (type 2 diabetes mellitus, alcoholism, chronic obstructive pulmonary disease, asthma, migraine, cardiovascular disease, cerebrovascular disease, acute myocardial infarction, ischemic heart disease, congestive heart failure, atrial flutter, and cerebrovascular disease), concomitant drugs (bisphosphonates, statins, antidepressants, angiotensin converting-enzyme inhibitors/angiotensin receptor blockers, and hormone replacement agents), and health service utilization (ambulatory visits and inpatient visits).
bStudy participants without aspirin or NSAID use were treated as the reference group.
cSignificant results (P < 0.05) are in bold.
Discussion
From the results of this nationwide population-based case–control study, our data demonstrate that taking aspirin or NSAIDs is associated with a lower incidence of colorectal cancer compared with nonuse in Taiwanese population. Importantly, the protective effect was even stronger when the cumulative duration of NSAID use was longer. However, we did not identify a dose response partially due to moderate protective effect of aspirin or NSAIDs on colorectal cancer or sample size constraint.
The chemopreventive efficacy of aspirin for colorectal cancer has been examined in several previous studies. In randomized controlled studies for high-risk populations, previous studies showed conclusive results that aspirin reduced colorectal cancer risk by 19% to 40% (8–10, 16). The treatment duration ranged from 1 to 2.5 years among these studies. However, these studies could not determine the optimal dose of aspirin. Sandler and colleagues reported that the daily use of 325 mg aspirin was associated with a reduced incidence of colorectal adenomas in individuals with previous colorectal cancer (9). Baron and colleagues suggested that aspirin at a dose of 81 mg rather than 325 mg showed a protective effect (8). In a nurse health study, a lower incidence of colorectal cancer was found among participants who took more than two 325 mg tablets of aspirin weekly (17). Huang and colleagues suggested that the regular use of low-dose aspirin (50–150 mg per day) over at least 3.5 years could reduce the risk of colorectal cancer occurrence in patients with high cardiovascular risks by 50% (18). On the basis of data from the aforementioned studies, it is clear that the long-term use of aspirin may have a preventive effect on colorectal cancer, yet the optimal treatment duration and dose remain unclear. Therefore, further investigation of the protective effects of aspirin as related to colorectal cancer is needed, especially in Asian populations.
Previous studies have also documented the chemopreventive efficacy of NSAID use for colorectal cancer. For example, previous randomized trials have reported that daily use of celecoxib for 3 years could reduce adenoma incidence in patients who had colon adenomas removed 3 years or 5 years prior (19, 20). Baron and colleagues showed that taking rofecoxib for 3 years among patients with a history of colon adenoma could reduce adenoma recurrence, and a protective effect was found at the 1-year follow-up (21). In addition to selective NSAIDs, the efficacy of nonselective NSAIDs was examined in a randomized trial. Takayama and colleagues examined the efficacy of sulindac and found that taking a daily dose of 300 mg of sulindac for 2 months could reduce colorectal cancer risk at a 1-year follow-up (22). In another study, the regular use of more than 2 tablets of NSAIDs weekly for 20 years showed primary preventive efficacy for colorectal cancer, with a more significant risk reduction in the group taking 6–14 tablets per week (23). Yang and colleagues suggested that patients who used COX-2 inhibitors had a decreased risk of colorectal cancer (24). Friis and colleagues reported that the preventive effect was more obvious in the groups that took a dose greater than 0.3 DDD and had consistent use for more than 5 years (14). While most studies have evaluated the chemopreventive efficacy of NSAIDs on colorectal cancer, only a limited number of studies have reported on the primary preventive efficacy of NSAIDs on colorectal cancer. In line with these previous studies, our results showed that patients who used NSAIDs showed reduced colorectal cancer incidence. The efficacy was apparent in groups of recent users and continuous users along with a duration response.
On the other hand, some studies reported no association of aspirin or NSAID use with colorectal cancer. For example, in a randomized controlled study of women taking 100 mg aspirin on alternative days for up to 10 years, no reduction in colorectal cancer incidence was reported (12). In a prospective cohort study, no reduction of colorectal cancer risk was observed in male participants regularly taking NSAIDs for 5 or more years (25). In Limburg study, taking 150 mg of sulindac twice daily for 6 months did not reduce colorectal cancer risk in patients with a previous diagnosis of colorectal cancer (26). Takayama and colleagues found that taking 400 mg etodolac daily for 2 months did not reduce colorectal cancer incidence at a 1-year follow-up (22). These inconsistent findings may be due to different examined doses, treatment durations, study populations, sample sizes, and/or study designs. Our data provide supportive evidence that there are beneficial effects of NSAID use on the primary prevention of colorectal cancer in Asian population. However, due to inconclusive results, our findings should be interpreted with caution. Further studies would be merited to confirm the findings from this study.
This study had several strengths. First, we identified cases from the RCIPD, which is part of the NHIRD. In Taiwan, once patients are diagnosed with a malignancy, they are considered to have catastrophic illness due to the need for intensive medical care. Because of this, we were able to ensure an accurate diagnosis of colorectal cancer by identifying our studied participants from the RCIPD. Second, a total of 130,416 patients were included in this study; this is a large sample size with sufficient power to allow us to address the beneficial effects of NSAID use on the primary prevention of colorectal cancer. Third, we comprehensively and systematically investigated various conditions of NSAID use such as exposure status, cumulative use days, average daily dose, and individual drug structure classes, separately. Moreover, we accounted for disease heterogeneity by examining the efficacy on colon and rectal cancers, separately.
This study had a few limitations. First, our findings were similar to previous observational studies from the western countries, but with smaller effect. The observed smaller effect in this study may be due to the difference in genetic make-up or lifestyle-associated factors, such as smoking, exercise, and body mass index, since majority of previous studies were conducted in the western countries. However, the information on genetic background or some epidemiologic variables such as life style, smoking, alcohol, red meat consumption, obesity, and the familial history of colorectal cancer is not available in the NHIRD. Second, despite adjusting for several key confounders, our results are still likely to be biased by unmeasured residual confounding factors. Third, aspirin and NSAIDs are prescribed medications in Taiwan. According to the data in Taiwan's food and drug administration, 18 of 52 (34.6%) oral aspirin are over-the-counter (OTC) products. Among NSAIDs, 66 of 174 (37.9%) oral ibuprofen are OTC products. A previous study in Taiwan has reported that few people would purchase OTC drugs when they felt sick because people preferred taking medication prescribed by trusted physicians; had unsatisfied experience when purchasing OTC aspirin/NSAIDs from pharmacy stores; or were lack of knowledge to self-purchase which kinds of OTC aspirin/NSAIDs. Therefore, patients generally seek medical help from medical professionals rather than independently purchasing OTC drugs due to low medical costs in Taiwan (27). Hence, the percentage of study participants who were taking OTC aspirin or NSAIDs should have been small. Fourth, information regarding medication adherence or compliance is not available in the NHIRD. However, medication adherence or compliance would most likely leads to an underestimation of the observed effect and would be in favor of a null association. Fifth, we could not examine the effect of aspirin or NSAID use in patients who were in the very early stages of colorectal cancer because they are not included in the RCIPD. We also did not investigate the combined effect of drug-use timing and drug-use cumulative duration due to considerable combination (or complex scenarios). Similarly, this may have underestimated the effect of aspirin or NSAID use on colorectal cancer. Moreover, multiple testing was not taken into account in this study. Sixth, the results for aspirin use observed in this study were inconsistent, as well as the duration of aspirin use, it was likely because the data of aspirin obtained from registry medical claims data might be imperfect, which might be in part explained the observed inconsistent results for aspirin use. Therefore, the results should be interpreted with caution.
Conclusions
The results of our study provide supportive evidence that the use of aspirin or NSAID is associated with a reduced risk of colorectal cancer incidence in Asian population. Unlike most previous studies that have reported chemoprevention efficacy of NSAID use on colorectal cancer, we demonstrate that NSAID use has a chemoprevention efficacy on colorectal cancer. Further investigation of the underlying regulatory mechanisms related to the observed beneficial effect of NSAID use on colorectal cancer would be merited.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed by the authors.
Authors' Contributions
Conception and design: H.-J. Tsai, W.-C. Chang
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.-N. Kuo, J.-J. Pan, Y.-W. Huang, H.-J. Tsai
Writing, review, and/or revision of the manuscript: C.-N. Kuo, J.-J. Pan, H.-J. Tsai, W.-C. Chang
Study supervision: H.-J. Tsai, W.-C. Chang
Acknowledgments
H.-J. Tsai is supported in part by a grant from the National Health Research Institutes (PH-104-PP-14, PH-104-SP-05, PH-104-SP-16, PH-105-SP-05, and PH-105-SP-04 to H.-J. Tsai; MOST105-2628-B-038-001-MY4 to W.-C. Chang). W.-C. Chang is supported by Taipei Medical University (105-5807-002-400). Chun-Nan Kuo is supported by a grant (103-wf-eva-06) from Taipei Municipal Wanfang Hospital (managed by Taipei Medical University).
We thank Tami R. Bartell at the Stanley Manne Children's Research Institute, Ann & Robert H. Lurie Children's Hospital of Chicago, for English editing.
This study is based in part on data from the National Health Insurance Research Database provided by the Bureau of National Health Insurance, Department of Health and managed by the National Health Research Institutes (registered numbers: 99081, 99136, 99287, 101014, NHRID-101-548, and NHRID-105-046). The interpretation and conclusions contained herein do not represent those of the Bureau of National Health Insurance, Department of Health, or National Health Research Institutes.
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.