Epidemiological studies have clarified the potential associations between regular aspirin use and cancers. However, it remains controversial on whether aspirin use decreases the risk of cancers risks. Therefore, we conducted an updated meta-analysis to assess the associations between aspirin use and cancers.
Methods
The PubMed, Embase, and Web of Science databases were systematically searched up to March 2017 to identify relevant studies. Relative risks (RRs) with 95% confidence intervals (CIs) were used to assess the strength of associations.
Results
A total of 218 studies with 309 reports were eligible for this meta-analysis. Aspirin use was associated with a significant decrease in the risk of overall cancer (RR = 0.89, 95% CI: 0.87–0.91), and gastric (RR = 0.75, 95% CI: 0.65–0.86), esophageal (RR = 0.75, 95% CI: 0.62–0.89), colorectal (RR = 0.79, 95% CI: 0.74–0.85), pancreatic (RR = 0.80, 95% CI: 0.68–0.93), ovarian (RR = 0.89, 95% CI: 0.83–0.95), endometrial (RR = 0.92, 95% CI: 0.85–0.99), breast (RR = 0.92, 95% CI: 0.88–0.96), and prostate (RR = 0.94, 95% CI: 0.90–0.99) cancers, as well as small intestine neuroendocrine tumors (RR = 0.17, 95% CI: 0.05–0.58).
Conclusions
These findings suggest that aspirin use is associated with a reduced risk of gastric, esophageal, colorectal, pancreatic, ovarian, endometrial, breast, and prostate cancers, and small intestine neuroendocrine tumors.
Aspirin has been used as an analgesic and in the prevention of cardiovascular diseases events in the past decades and is one of the most commonly used drugs worldwide [1, 2]. Clinical and epidemiological studies reported that the rates of aspirin usage in different populations across different countries ranging from 11% to 54% [3‐5]. Since the 1970s, many researchers started to focus on the effects of aspirin on cancers [6, 7]. However, these original studies were not comprehensive, and the effects on some cancers were controversial [8, 9].
Although several meta-analyses have been conducted to assess the associations between aspirin use and the risk of cancers(e.g., gastric, esophageal, pancreatic, lung, squamous cell carcinoma, breast, ovarian, and prostate cancers) [10‐18], most of these studies were restricted to certain types of cancers, and some types such as hepatobiliary and cervical cancer could not be investigated. In addition, 70 new studies have been published since 2012. Therefore, this comprehensive systematic review and updated meta-analysis was conducted to explore the reliability of risk estimates between aspirin usage and most types of cancers and provide a landscape of aspirin use and cancer incidence.
Anzeige
Methods
Search strategy
This systematic review was conducted in accordance with the checklist proposed by the Meta-analysis of Observational Studies in Epidemiology group [19]. We searched multiple electronic bibliographic databases to identify studies published from database inception till March 2017, including PubMed, Embase, and Web of Science databases, with the following search terms: (“cancer” OR “neoplasm” OR “carcinoma”) AND (“aspirin” OR “acetylsalicylic acid” OR “non-steroidal anti-inflammatory drugs” OR “NSAIDs”). We restricted our search to human studies and published in English. In addition, reference lists from relevant reviews and retrieved articles were searched for qualifying studies.
Inclusion criteria
The inclusion criteria were: 1) case-control or cohort studies; 2) studies that evaluated the relationships between the use of aspirin and the risk of cancers; 3) studies that reported risk estimates with 95% confidence interval (CI) or provided information that enabled us to calculate them. The exclusion criteria were: 1) studies that used other combinations of NSAIDs, which prevented the determination of the specific effect of aspirin, and 2) studies involving patients with specific diseases (e.g., Barrett’s esophagus, Crohn’s disease, or ulcerative colitis). Only the latest or the most informative study was included when multiple studies were published on the same study population.
Data extraction
The following information was obtained from each study: first author name, year of publication, study period, study location, study design, number of cases, number of participants, gender, definition of aspirin exposure, as curtained methods of exposure, odds ratios (ORs), hazard ratios (HRs) or relative risks (RRs) with their corresponding 95% CIs, and confounding factors adjusted in the analysis. The most fully-adjusted risk estimates with its corresponding 95% CIs (when available) were preferentially extracted. Data extraction was conducted independently by two authors (Y.Q. and T.T.Y.), and discrepancies were resolved by discussion with a third investigator (Z.X.L.).
Quality assessment
Quality assessment of eligible studies was performed independently by two reviewers (Y.Q. and T.T.Y.) according to the Newcastle-Ottawa Quality Assessment Scale [20]. This scale allocates a maximum of nine points based on the selection (0–4 points), comparability (0–2 points), and exposure/outcome of the study participants (0–3 points). Scores of 0–3, 4–6, and 7–9 were classified as low, moderate, and high-quality studies respectively.
Anzeige
Statistical analysis
RRs were used as the common measurement of the associations between aspirin use and the risk of cancer. Because cancer is a rare event in general, we could generally ignore the distinctions among the various measures of relative risk (e.g., odds ratios, rate ratios, and risk ratios) [21], and considered that ORs and HRs were similar to RRs. When risk estimates for different durations of aspirin use or different levels of aspirin utilization were available, the study-specific RRs were subsequently recalculated in the primary analysis by pooling the risk estimates compared with the reference group. A random effects model was selected to estimate the pooled RRs (95% CI) for the associations between aspirin use and the risk of cancer if the risk estimates for different subtypes of cancer were available. Summary estimates were derived from meta-analyses using random effects models. Studies involving different populations or different types of cancers were treated as independent studies.
To assess the heterogeneity in results of individual studies, I2 statistic (values of 25%, 50%, and 75% represented cutoff points for low, moderate, and high degrees of heterogeneity, respectively) were used [22]. Publication bias was assessed with Funnel plots, the Begg’s rank correlations and Egger’s regression model. Subgroup analyses for study design, study location, gender, exposure assessment, quality assessment, duration of aspirin use (years), and frequency of aspirin use (tablets/week) were conducted to explore the potential heterogeneity among studies. Subgroup analysis was not conducted for strata with less than five studies. Because time-related biases are common in observational studies of medications and are often responsible for apparent protective effects of drugs, we conducted analyses both including and excluding studies with immortal time bias (bias because of the inclusion of follow-up time during which events cannot occur) [23]. Statistical analyses were performed with Stata version 12.0. (College Station, TX, USA). All reported probabilities (P values) were two-tailed with a significance level of 0.05.
Results
Literature search and study characteristic
Figure 1 shows the process for the identification of eligible studies. A total of 28,683 studies were identified and 298 studies remained in the analysis after assessing the titles and abstracts according to the criteria mentioned above. In total, 307 potentially relevant articles were reviewed in their entirety. Among them, 89 articles were further excluded due to the following reasons: 26 articles were not observational design, 11 articles defined exposure combined with other NSAIDs, 8 articles evaluated cancer mortality, 39 articles were duplicate publications on the same subject population, and 5 articles (1 for Crohn’s disease [24], 1 for ulcerative colitis [25], 3 for Barrett’s esophagus [26‐28]) included patients with specific diseases. Ultimately, 218 studies with 309 independent reports were included in the present meta-analysis.
Fig. 1
Flow chart of study selection
×
The main characteristics of the 218 eligible articles published between 1985 and 2016 are summarized in Tables 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 and 21. Results were presented according to study design. This study altogether included 161 cohort studies and 148 case-control studies. Among them, 135 studies were conducted in North America, 12 in Asia, 61 in Europe, 8 in Oceania, and 2 were multi-country studies. Overall, the summarized RR was 0.89(95%CI: 0.87–0.91), indicating a decreased risk of cancer associated with the use of aspirin. The combined RRs were 0.82 (95% CI: 0.79–0.85) for the case-control studies and 0.94 (95% CI: 0.92–0.97) for the cohort studies. We also observed a apparent beneficial effect of aspirin use when excluding 41 studies deemed to be prone to immortal time bias (RR = 0.87, 95%CI:0.85–0.89) in the meta-analysis.
Table 1
Characteristics of included studies- gastric cancer
Use aspirin during the 30-day period before the interview
1,2
6
1 = age, 2 = sex, 3 = smoking, 4 = family history, 5 = educational level, 6 = alcohol intake, 7 = marriage, 8 = fat distribution, 9 = social status, 10 = BMI, 11 = total cholesterol, 12 = triglyceride, 13 = charlson comorbidity index, 14 = statin, 15 = metformin, 16 = ACE inhibitors, 17 = angiotensin II receptor blockers, 18 = helicobacter pylori, 19 = history of diabetes, 20 = resident district, 21 = percent body fat, 22 = HDL cholesterol, 23 = LDL cholesterol, 24 = period of interview, 25 = race, 26 = gastro-esopageal refiux disease, 27 = antacid use, 28 = upper gastrointestinal tract history, 29 = health plan, 30 = duration of continuous, 31 = calendar year enrollment in the health plan at the date of diagnosis, 32 = interview year, 33 = center, 34 = religion, 35 = comorbidity, 36 = total calorie, fibre and calcium intake, 37 = fruit, vegetable and/or vitamin intake, 38 = physical activity, 39 = processed meat intake
AARP AARP diet and health study, GHC Group Health Cooperative, HFHS Henry Ford health system’s health alliance plan, KNHI Korean National Health Insurance database
aStudy deemed to be prone to immortal time bias
Table 2
Characteristics of included studies- esophagus cancer
1 = age, 2 = sex, 3 = smoking, 4 = family history, 5 = educational level, 6 = alcohol intake, 7 = marriage, 8 = Fat distribution, 9 = social status, 10 = BMI, 11 = race 12 = gastroesopageal refiux disease, 13 = other NSAID 14 = upper gastrointestinal tract history, 15 = antacid use, 16 = birthplace,17 = health plan, 18 = duration of continuous enrollment in the health plan at the date of diagnosis, 19 = calendar year, 20 = total calorie, fibre and calcium intake, 21 = fruit, vegetable and/or vitamin intake, 22 = physical activity, 23 = CHD, 24 = stroke, 25 = COX-2 inhibitors, 26 = duration of observation in the database, 27 = deprivation, 28 = household income, 29 = cumulative and frequency of gastroesophageal reflux symptoms 10 y before diagnosis, 30 = location, 31 = job type, 32 = year of completing the questionnaire, 33 = general practice
AARP AARP diet and health study, FINBAR the factors influencing the Barrett’s adenocarcinoma relationship study, GHC Group Health Cooperative, GPRD General Practitioners research database, HFHS Henry Ford health system’s health alliance plan, PCCIU primary care clinical informatics unit database, RPCI the Roswell park cancer Institute
aStudy deemed to be prone to immortal time bias
Table 3
Characteristics of included studies- colorectal cancer
1 = age, 2 = sex, 3 = smoking, 4 = family history, 5 = educational level, 6 = alcohol intake, 7 = race, 8 = BMI, 9 = marital status, 10 = self-rated health, 11==C-reactive protein level, 12 = cholesterol, 13 = Charlson comorbidity index, 14 = statin, 15 = metformin, 16 = ACE inhibitors, 17 = angiotensin II receptor blockers, 18 = physical activity 19 = fruit, vegetable and/or vitamin intake, 20 = seafood and dairy foods intake, 21 = number of PCP visits in the year before the index date, 22 = other NSAIDs, 23 = paracetamol, 24 = insulin, 25 = oral steroids, 26 = area (county/region), 27 = hormone replacement therapy, 28 = history of diabetes mellitus, 29 = history of cholecystectomy, 30 = history of colonoscopy, 31 = chronic obstructive pulmonary disease or asthma,32 = antidepressants, 33 = migraine,34 = total energy intake, 35 = deprivation index, 36 = hypertension ampling probability, 37 = ever use of calcium supplements in the past 5 years, 38 = primary care practitioner, 39 = dietary fat intake, 40 = sampling probability, 41 = morbidity (diabetes, ischemic heart disease, hypertension, stroke, colitis, rheumatoid arthritis, and osteoarthritis), 42 = former health checkup, 43 = red meat, 44 = Nitro-vasodilator use, 45 = number of drugs, 46 = number of physician encounters, 47 = all-cause hospitalization in prior year, 48 = dietary fiber, 49 = folate, 50 = height, 51 = Alternate Healthy Eating Index-2010, 52 = PSA test in past 2 y, 53 = mammogram in past 2 y, 54 = duration of diabetes, 55 = propensity score at baseline, 56 = unique number of hospitalizations in the year prior to start of follow up, 57 = observational study enrollment, 58 = diet modification trial enrollment, 59 = screening for cancer, 60 = age at menarche, 61 = age at menopause, 62 = gravidity, 63 = age atfirst birth, 64 = duration of estrogen therapy, 65 = duration of combined postmenopausal hormone therapy, 66 = hysterectomy status, 67 = use of antihypertensive medication, 68 = history of coronary heart disease, 69 = use of cholesterol-lowering medication, 70 = history of arthritis, 71 = history of Ulcer, 72 = unique number of dispensing
AARP AARP diet and health study, CHRI Cancer Hospital and Richard J. Solove Research Institute, GPRD General Practitioners Research Database, HPFS Health Professionals follow-up study, IWHS Iowa Women’s Health Study, KPMCP Kaiser Permanente Medical Care Program of Northern California, MEC Multiethnic Cohort Study, MECC the molecular epidemiology of colorectal cancer, NHS nurses’ health study, RAMQ Re′gie de l’Assurance Maladie du Que’bec, SCCS study of colorectal cancer in Scotland, THIN the health improvement Network, VITAL the vitamins and lifestyle, WHI women’s health initiative
aStudy deemed to be prone to immortal time bias
Table 4
Characteristics of included studies- hepato-biliary cancer
1 = age, 2 = sex, 3 = smoking, 4 = family history, 5 = educational level, 6 = alcohol intake, 7 = marriage, 8 = Fat distribution, 9 = social status, 10 = BMI, 11 = race,12 = primary sclerosing cholangitis (PSC), 13 = non-PSC-related cirrhosis, 14 = biliary tract diseases, 15 = diabetes, 16 = hepatitis B or C virus infection, 17 = rare metabolic disorders, 18 = use of paracetamol, antidiabetic medications, and statins,19 = follow–up duration, 20 = the date of the diabetes diagnosis, 21 = cohort (AARP, AHS, USRT, PLCO, HPFS, CPSII, IWHS, BWHS, WHI, NHS), 22 = biliary stone status
AARP AARP diet and health study, AHS Agriculture Health Study, BCDDP the breast cancer detection demonstration project, BWHS black women’s health study, CPRD clinical practice research datalink, CPSII cancer prevention study II, HPFS Health Professionals follow-up study, IWHS Iowa Women’s Health Study, LGH Leicester General Hospital NHS Trust, NHIS-NSC National Health Insurance Service National Sample Cohort, NHS nurses’ health study, NNUH Norfolk and Norwich University Hospital, PLCO prostate, lung, colorectal and ovarian cancer screening trial, USRT United State Radiologic Technologist Study, WHI women’s health initiative
aStudy deemed to be prone to immortal time bias
Table 5
Characteristics of included studies- pancreatic cancer
Use aspirin during the 30-day period before the interview
1,2
6
1 = age, 2 = sex, 3 = smoking, 4 = family history, 5 = educational level, 6 = alcohol intake, 7 = diabetes, 8 = Fat distribution, 9 = social status, 10 = BMI, 11 = race, 12 = folate, 13 = height, 14 = Alternate Healthy Eating Index-2010, 15 = PSA test in past 2 y, 16 = mammogram in past 2 y, 17 = hormone replacement therapy, 18 = physical activity, 19 = fruit, vegetable and/or vitamin intake, 20 = history of colonoscopy, 21 = total energy intake, 22 = ever use of calcium supplements in the past 5 years, 23 = former health checkup, 24 = red meat, 25 = other NSAIDs, 26 = area (county/region), 27 = prior cancer, 28 = migraine, 29 = ever use of calcium supplements in the past 5 years, 30 = red meat, 31 = Nitro-vasodilator use, 32 = height, 33 = unique number of hospitalizations in the year prior to start of follow up, 34 = observational study enrollment, 35 = diet modification trial enrollment, 36 = screening for cancer, 37 = age at menarche, 38 = age at menopause, 39 = gravidity, 40 = age atfirst birth, 41 = duration of estrogen therapy, 42 = duration of combined postmenopausal hormone therapy, 43 = hysterectomy status, 44 = use of antihypertensive medication, 45 = history of coronary heart disease, 46 = use of cholesterol-lowering medication, 47 = history of arthritis, 48 = history of ulcer, 49 = hypertension, 50 = H. pylori CagA seropositivity, 51 = ABO blood group A vs. non-A, 52 = ABO blood group O vs. non-O, 53 = center, 54 = year of interview, 55 = history of chronic pancreatitis
GPRD General Practitioners Research Database, HPFS Health Professionals follow-up study, IWHS Iowa Women’s Health Study, NHS nurses’ health study, QPCS the Queensland Pancreatic Cancer Study, RPCI the Roswell Park Cancer Institute, WHI women’s health initiative
1 = age, 2 = sex, 3 = smoking, 4 = education level, 5 = family history, 6 = alcohol intake, 7 = height, 8 = Alternate Healthy Eating Index-2010, 9 = PSA test in past 2 y, 10 = BMI, 11 = race, 12 = folate, 13 = Charlson comorbidity index, 14 = statin, 15 = metformin, 16 = ACE inhibitors, 17 = Angiotensin II receptor blockers, 18 = physical activity, 19 = fruit, vegetable and/or vitamin intake, 20 = history of colonoscopy, 21 = total energy intake, 22 = ever use of calcium supplements in the past 5 years, 23 = former health checkup, 24 = red meat, 25 = hormone replacement therapy, 26 = unique number of dispensing, 27 = unique number of hospitalizations in the year prior to start of follow up, 28 = mammogram in past 2 y, 29 = other NSAIDs, 30 = area (county/region), 31 = history of coronary heart disease, 32 = diabetes, 33 = hypertension, 34 = housing type, 35 = history of arthritis, 36 = interview year, 37 = history of COLD, 38 = study, 39 = use of acetaminophen, 40 = smoking cessation advice by general practitioner, 41 = smoking cessation treatment, 42 = number of visits to general practitioner, 43 = number of referrals, 44 = use of oral corticosteroids, 45 = antihypertensives and other lipid-lowering drugs, 46 = chronic obstructive pulmonary disease, 47 = cerebrovascular disease, 48 = ischemic heart disease, 49 = menopausal status, 50 = age started and years since quitting smoking, 51 = emphysema, 52 = randomization arm of the DM trial, 53 = history of ulcer, migraine or chronic headache, osteoarthritis or chronic joint pain, 54 = coronary artery disease, 55 = C-reactive protein level, 56 = pack years of smoking, 57 = cholesterol, 58 = any heart disease/heart attack
AHFTS American Health Foundation Tobacco Study, DDCHS Danish Diet Cancer and Health Study, HPFS Health Professionals follow-up study, IWHS Iowa Women’s Health Study, MSKCC Memorial Sloan-Kettering Cancer Center, NELCS New England Lung Cancer study, NHS nurses’ health study, NICCC National Israel Cancer Control Center, NYU New York University, RPCI the Roswell Park Cancer Institute, THIN the Health Improvement Network, VITAL the vitamins and lifestyle, WHI women’s health initiative
aStudy deemed to be prone to immortal time bias
Table 7
Characteristics of included studies- breast cancer
1 = age, 2 = sex, 3 = smoking, 4 = education level, 5 = family history, 6 = alcohol intake, 7 = height, 8 = Alternate Healthy Eating Index-2010, 9 = PSA test in past 2 y, 10 = BMI, 11 = race, 12 = folate, 13 = Charlson comorbidity index, 14 = statin, 15 = metformin, 16 = ACE inhibitors, 17 = Angiotensin II receptor blockers, 18 = physical activity, 19 = fruit, vegetable and/or vitamin intake, 20 = history of colonoscopy, 21 = total energy intake, 22 = ever use of calcium supplements in the past 5 years, 23 = former health checkup, 24 = red meat, 25 = hormone replacement therapy, 26 = unique number of dispensing, 27 = unique number of hospitalizations in the year prior to start of follow up, 28 = mammogram in past 2 y, 29 = other NSAIDs, 30 = area (county/region), 31 = migraine, 32 = Nitro-vasodilator use, 33 = observational study enrollment, 34 = diet modification trial enrollment, 35 = screening for cancer, 36 = age at menarche, 37 = age at menopause, 38 = gravidity, 39 = history of arthritis, 40 = age at first birth, 41 = duration of estrogen therapy, 42 = duration of combined postmenopausal hormone therapy, 43 = hysterectomy status, 44 = use of antihypertensive medication, 45 = history of coronary heart disease, 46 = use of cholesterol-lowering medication, 47 = history of ulcer, 48 = diabetes, 49 = hypertension, 50 = number of deliveries, 51 = menopausal status, 52 = household income, 53 = personal history of benign breast disease, 54 = study center, 55 = referent year, 56 = percentage Native American ancestry, 57 = breast procedure in the prior 3 years, 58 = other breast disease in the prior 3 years, 59 = neighborhood socioeconomic status, 60 = visit to a gynecologist in the prior year, 61 = practice of breast selfexamination, 62 = year of interview, 63 = number of physician visits 2 years before hospitalization, 64 = duration of oral contraceptive use, 65 = paracetamol, 66 = steroid, 67 = time since the last mammogram and duration and frequency of use, 68 = use of oral contraceptives, 69 = relative weight at age 12, 70 = questionnaire cycle, 71 = weight change since age 18 years, 72 = number of breast biopsies, 73 = history of surgical menopause, 74 = years of combined estrogen and progesterone hormone therapy, 75 = C-reactive protein level, 76 = all pain medication use, 77 = parity status before age 30
AARP AARP diet and health study, BWHS Black Women’s Health Study, CHRI Cancer Hospital and Richard J. Solove Research Institute, GPRD General Practitioners Research Database, HPFS Health Professionals follow-up study, IWHS Iowa Women’s Health Study, MCC the Spanish Multi-Case-control study, NHS nurses’ health study, RAMQ Re′gie de l’Assurance Maladie du Que’bec, VITAL the vitamins and lifestyle, WEB Western New York exposures and breast cancer study, WHI women’s health initiative
aStudy deemed to be prone to immortal time bias
Table 8
Characteristics of included studies- ovarian cancer
Use three or more times per week for at least 6 months
4,36,50, 55
8
1 = age, 2 = sex, 3 = smoking, 4 = family history, 5 = educational level, 6 = alcohol intake, 7 = chronic obstructive pulmonary disease or asthma, 8 = Fat distribution, 9 = religion, 10 = BMI, 11 = race, 12 = folate, 13 = Charlson comorbidity index, 14 = statin, 15 = metformin, 16 = ACE inhibitors, 17 = Angiotensin II receptor blockers, 18 = physical activity, 19 = fruit, vegetable and/or vitamin intake, 20 = history of colonoscopy, 21 = total energy intake, 22 = ever use of calcium supplements in the past 5 years, 23 = former health checkup, 24 = red meat, 25 = hormone replacement therapy, 26 = unique number of dispensing, 27 = unique number of hospitalizations in the year prior to start of follow up, 28 = mammogram in past 2 y, 29 = other NSAIDs, 30 = area (county/region), 31 = migraine, 32 = Nitro-vasodilator use, 33 = observational study enrollment, 34 = diet modification trial enrollment, 35 = screening for cancer, 36 = age at menarche, 37 = age at menopause, 38 = gravidity, 39 = history of arthritis, 40 = age at first birth, 41 = duration of estrogen therapy, 42 = duration of combined postmenopausal hormone therapy, 43 = hysterectomy status, 44 = use of antihypertensive medication, 45 = history of coronary heart disease, 46 = use of cholesterol-lowering medication, 47 = history of ulcer, 48 = income, 49 = tubal ligation, 50 = oral contraceptive use, 51 = menopausal status, 52 = endometriosis, 53 = pelvic inflammatory disease, 54 = study site, 55 = parity, 56 = infertility, 57 = diabetes mellitus, 58 = tubal sterilization, 59 = interview year, 60 = breastfeeding, 61 = duration of oral contraceptive use, 62 = talc use, 63 = months of pregnancy, 64 = severe menstrual cramping, 65 = presence of irregular menses, 66 = menstrual, headache, or arthritic pain, 67 = ibuprofen, 68 = paracetamol, 69 = partial oophorectomy
AACES the African American Cancer Epidemiology Study, AARP AARP Diet and Health Study, BCDDP the Breast Cancer Detection Demonstration Project, HOPE hormones and Ovarian cancer prediction study, IWHS Iowa Women’s Health Study, MALOVA Danish MALignant Ovarian cancer study, MEC multiethnic cohort study, NHS nurses’ health study, NYU New York University, RPCI the Roswell Park Cancer Institute, WHI women’s health initiative
aStudy deemed to be prone to immortal time bias
Table 9
Characteristics of included studies- endometrial cancer
Use aspirin during the 30-day period before the interview
1,2
6
1 = age, 2 = sex, 3 = smoking, 4 = family history, 5 = educational level, 6 = alcohol intake, 7 = calendar year, 8 = Fat distribution, 9 = social status, 10 = BMI, 11 = race, 12 = folate, 13 = Charlson comorbidity index, 14 = statin, 15 = metformin, 16 = ACE inhibitors, 17 = Angiotensin II receptor blockers, 18 = physical activity, 19 = fruit, vegetable and/or vitamin intake, 20 = history of colonoscopy, 21 = total energy intake, 22 = ever use of calcium supplements in the past 5 years, 23 = former health checkup, 24 = red meat, 25 = hormone replacement therapy, 26 = unique number of dispensing, 27 = unique number of hospitalizations in the year prior to start of follow up, 28 = mammogram in past 2 y, 29 = other NSAIDs, 30 = area (county/region), 31 = migraine, 32 = Nitro-vasodilator use, 33 = observational study enrollment, 34 = diet modification trial enrollment, 35 = screening for cancer, 36 = age at menarche, 37 = age at menopause, 38 = gravidity, 39 = history of arthritis, 40 = age at first birth, 41 = duration of estrogen therapy, 42 = duration of combined postmenopausal hormone therapy, 43 = hysterectomy status, 44 = use of antihypertensive medication, 45 = history of coronary heart disease, 46 = use of cholesterol-lowering medication, 47 = history of ulcer, 48 = parity, 49 = obesity, 50 = diabetes, 51 = chronic obstructive pulmonary disease, 52 = oral contraceptive use, 53 = study center, 54 = period of interview, 55 = menopausal status, 56 = hypertension, 57 = years of oral contraceptive use, 58 = oophoerectomy, 59 = history of stroke, 60 = waist-hip ratio, 61 = intrauterine device use, 62 = height
AARP AARP diet and health study, ANECS Australian National Endometrial Cancer Study, EDGE Study estrogen, diet, genetics, and endometrial cancer, IWHS Iowa Women’s Health Study, MEC multiethnic cohort study, NHS nurses’ health study, RPCI the Roswell Park Cancer Institute, VITAL the vitamins and lifestyle, WHI women’s health initiative
aStudy deemed to be prone to immortal time bias
Table 10
Characteristics of included studies- cervix uterus
1 = age, 2 = sex, 3 = smoking, 4 = education level, 5 = family history, 6 = alcohol intake, 7 = height, 8 = Alternate Healthy Eating Index-2010, 9 = PSA test in past 2 y, 10 = BMI, 11 = race, 12 = folate, 13 = Charlson comorbidity index, 14 = statin, 15 = metformin, 16 = ACE inhibitors, 17 = Angiotensin II receptor blockers, 18 = physical activity, 19 = fruit, vegetable and/or vitamin intake, 20 = history of colonoscopy, 21 = total energy intake, 22 = ever use of calcium supplements in the past 5 years, 23 = former health checkup, 24 = mammogram in past 2 y, 25 = hormone replacement therapy, 26 = unique number of dispensing, 27 = unique number of hospitalizations in the year prior to start of follow up, 28 = other NSAIDs, 29 = history of heart attack, 30 = diabetes, 31 = hypertension, 32 = simultaneous use of other medications (cholesterol lowering drugs, anti-diabetic drugs, antihypertensive drugs and benign prostatic hyperplasia medication), 33 = the primary care centres from which they were recruited, 34 = study center, 35 = any paracetamol use, 36 = residence (by design), 37 = use of high-dose aspirin, 38 = 5-alpha reductase inhibitors, 39 = income, 40 = selected cardiovascular drugs, and antidepressants or neuroleptics, 41 = ever visited a urologist 1–11 years prior, 42 = SCREENED and volume of family physician visits in the 5 years prior to the index date, 43 = finasteride, 44 = medical institution, 45 = calendar year, 46 = prior BPH history, 47 = number of visits to general practitioners, 48 = referrals, 49 = hospitalizations, 50 = recent medical contacts, 51 = socio-economic status, 52 = total polyunsaturated fat consumption, 53 = a-linolenic acid and ratio of dietary n-6:long-chain n-3 polyunsaturated fatty acids, 54 = presence of obesity, 55 = benign prostatic hypertrophy, 56 = alpha-adrenoreceptor antagonists, 57 = immunosuppressive drugs, 58 = natural log-transformed prostate specific antigen (PSA) concentration, 59 = PSA quotient, 60 = use of antidiabetic medication, 61 = C-reactive protein level, 62 = ibuprofen use, 63 = osteoarthritis, 64 = rheumatoid arthritis,65 = enlarged prostate, 66 = coronary artery disease, 67 = chronic joint pain, chronic headaches, and migraines, 68 = acetaminophen, 69 = and number of health checkups, 70 = red meat
BLSA Baltimore Longitudinal study of Aging, CHRI Cancer Hospital and Richard J. Solove Research Institute, GPRD general practitioners research database, HPFS Health Professionals follow-up study, HSD health search IMS health longitudinal patient database, NHS nurses’ health study, PLCO prostate, lung, colorectal and ovarian cancer screening trial, ProtecT prostate testing for cancer and Treatment, RAMQ Re′gie de l’Assurance Maladie du Que’bec, RPCI the Roswell Park Cancer Institute, SEER surveillance, epidemiology and end results, VITAL the vitamins and lifestyle
1 = age, 2 = sex, 3 = smoking, 4 = family history, 5 = educational level, 6 = alcohol intake, 7 = diabetes, 8 = fat distribution, 9 = social status, 10 = BMI,11 = race, 12 = folate, 13 = height, 14 = Alternate Healthy Eating Index-2010, 15 = PSA test in past 2 y, 16 = mammogram in past 2 y, 17 = hormone replacement therapy, 18 = physical activity, 19 = fruit, vegetable and/or vitamin intake, 20 = history of colonoscopy, 21 = total energy intake, 22 = ever use of calcium supplements in the past 5 years, 23 = former health checkup, 24 = red meat, 25 = other NSAIDs, 26 = area (county/region), 27 = hypertension, 28 = migraine, 29 = ever use of calcium supplements in the past 5 years, 30 = red meat, 31 = Nitro-vasodilator use, 32 = height, 33 = unique number of hospitalizations in the year prior to start of follow up, 34 = observational study enrollment, 35 = diet modification trial enrollment, 36 = screening for cancer, 37 = age at menarche, 38 = age at menopause, 39 = gravidity, 40 = age atfirst birth, 41 = duration of estrogen therapy, 42 = duration of combined postmenopausal hormone therapy, 43 = hysterectomy status, 44 = use of antihypertensive medication, 45 = history of coronary heart disease, 46 = use of cholesterol-lowering medication, 47 = history of arthritis, 48 = history of ulcer, 49 = method of interview, 50 = obesity, 51 = center, 52 = dialysis treatment, 53 = marital status, 54 = total dietary fiber, 55 = study center, 56 = year of interview, 57 = regular use of amphetamines, 58 = method of interview, 59 = phenacetin, 60 = paracetamol, 61 = urological disease
AARP AARP diet and health study, HPFS Health Professionals follow-up study, NHS nurses’ health study, WHI women’s health initiative
aStudy deemed to be prone to immortal time bias
Table 13
Characteristics of included studies- renal pelvis and ureter
1 = age, 2 = sex, 3 = smoking, 4 = family history, 5 = educational level, 6 = alcohol intake, 7 = history of colorectal endoscopy, 8 = Fat distribution, 9 = social status, 10 = BMI,11 = race, 12 = folate, 13 = height, 14 = Alternate Healthy Eating Index-2010, 15 = PSA test in past 2 y, 16 = mammogram in past 2 y, 17 = hormone replacement therapy, 18 = physical activity, 19 = fruit, vegetable and/or vitamin intake, 20 = history of colonoscopy, 21 = total energy intake, 22 = diabetes, 23 = former health checkup, 24 = red meat, 25 = other NSAIDs, 26 = area (county/region), 27 = study, 28 = migraine, 29 = ever use of calcium supplements in the past 5 years, 30 = red meat, 31 = Nitro-vasodilator use, 32 = height, 33 = unique number of hospitalizations in the year prior to start of follow up, 34 = observational study enrollment, 35 = diet modification trial enrollment, 36 = screening for cancer, 37 = age at menarche, 38 = age at menopause, 39 = gravidity, 40 = age atfirst birth, 41 = duration of estrogen therapy, 42 = duration of combined postmenopausal hormone therapy, 43 = hysterectomy status, 44 = use of antihypertensive medication, 45 = history of coronary heart disease, 46 = use of cholesterol-lowering medication, 47 = history of arthritis, 48 = history of ulcer, 49 = indications for NSAID use, 50 = fluid intake, 51 = hispanic status, 52 = Metamizol, 53 = Acetic acids, 54 = number of years employed as hairdresser/barber, 55 = use of phenacetin, 56 = acetaminophen, 57 = Other salicylic acids, 58 = Propionic acids, 59 = Oxicam,60 = Pyrazolon derivatives, 61 = Dextropropoxyphene, 62 = Phenazon, 63 = Other analgesics (codeine, chlormezanone, caffeine), 63 = hypertension
AARP AARP diet and health study, HPFS Health Professionals follow-up study, PLCO prostate, lung, colorectal and ovarian cancer screening trial, SEER surveillance, epidemiology and end results, USRT United State Radiologic Technologist Study, VITAL the vitamins and lifestyle, WHI women’s health initiative
Danish Cancer Registry, Civil Registration System, National Prescription Registry, Danish National Registry of Patients, and Danisheducation and fertility registries within Statistics Denmark
2688
18,848
Glioma
National Prescription Registry
Use aspirin as a ‘low’ (≤ 100 mg) or ‘high’ (150 mg) daily dose of low-dose aspirin
1 = age, 2 = sex, 3 = smoking, 4 = family history, 5 = educational level, 6 = alcohol intake, 7 = race, 8 = state of residence, 9 = center, 10 = anti-asthma medications, 11 = individual NSAIDs, 12 = acetaminophen, 13 = statins, 14 = diabetes, 15 = stroke, 16 = allergy, 17 = asthma, 18 = antihistamines, 19 = history of heart disease using age as time metric
AARP AARP diet and health study, CPRD clinical practice research datalink, CUMC Columbia University Medical Center, UCSF University of California San Francisco
aStudy deemed to be prone to immortal time bias
Table 16
Characteristics of included studies- head and neck cancers
1 = age, 2 = sex, 3 = smoking, 4 = family history, 5 = educational level, 6 = alcohol intake, 7 = race, 8 = other NSAIDs, 9 = social status, 10 = BMI, 11 = area of residence, 12 = period of interview, 13 = occupation, 14 = deprivation, 15 = CHD, 16 = stroke, 17 = COX-2 inhibitors, 18 = fruit consumption, 19 = marital status
ARCAGE the alcohol-related cancers and genetic susceptibility, CPRD clinical practice research datalink, GUMC Georgetown University Medical School, PCCIU primary care clinical informatics unit database, PLCO prostate, lung, colorectal and ovarian cancer screening trial, RPCI the Roswell Park Cancer Institute
aStudy deemed to be prone to immortal time bias
Table 17
Characteristics of included studies- thyroid cancer
1 = age, 2 = sex, 3 = smoking, 4 = education level, 5 = family history, 6 = alcohol intake, 7 = height, 8 = weight, 9 = history of ulcer, 10 = BMI, 11 = race, 12 = duration of estrogen therapy, 13 = duration of combined postmenopausal hormone therapy, 14 = hysterectomy status, 15 = use of antihypertensive medication, 16 = history of coronary heart disease, 17 = use of cholesterol-lowering medication, 18 = physical activity, 19 = fruit, vegetable and/or vitamin intake, 20 = history of colonoscopy, 21 = total energy intake, 22 = ever use of calcium supplements in the past 5 years, 23 = former health checkup, 24 = red meat, 25 = hormone replacement therapy, 26 = gravidity, 27 = history of arthritis, 28 = age at first birth, 29 = other NSAIDs, 30 = area (county/region), 31 = migraine, 32 = Nitro-vasodilator use, 33 = observational study enrollment, 34 = diet modification trial enrollment, 35 = screening for cancer, 36 = age at menarche, 37 = age at menopause
AARP AARP diet and health study, PLCO prostate, lung, colorectal and ovarian cancer screening trial, WHI women’s health initiative
Use aspirin during the 30-day period before the interview
1,2
6
1 = age, 2 = sex, 3 = smoking, 4 = family history, 5 = educational level, 6 = alcohol intake, 7 = skin reaction to the sun, 8 = Fat distribution, 9 = social status, 10 = BMI, 11 = unique number of dispensing, 12 = unique number of hospitalizations in the year prior to start of follow up, 13 = race, 14 = physical activity, 15 = fruit, vegetable and/or vitamin intake, 16 = ever use of calcium supplements in the past 5 years, 17 = red meat, 18 = hormone replacement therapy, 19 = other NSAIDs, 20 = area (county/region), 21 = migraine, 22 = Nitro-vasodilator use, 23 = observational study enrollment, 24 = diet modification trial enrollment, 25 = screening for cancer, 26 = age at menarche, 27 = age at menopause, 28 = gravidity, 29 = history of arthritis, 30 = history of ulcer, 31 = age at first birth, 32 = duration of estrogen therapy, 33 = duration of combined postmenopausal hormone therapy, 34 = hysterectomy status, 35 = use of antihypertensive medication, 36 = history of coronary heart disease, 37 = use of cholesterol-lowering medication, 38 = mammography, 39 = history of colorectal endoscopy, 40 = history of PSA testing, 41 = diabetes, 42 = hypertension,43 = the number of general practitioner visits in the year before the index date, 44 = use of systemic glucocorticoids, cytostatic or immunosuppressive medication, 45 = drugs with pigmenting adverse effects, 46 = Charlson comorbidity index, 47 = photosensitising or phototoxic drugs,48 = inflammatory bowel disease, 49 = ischemic stroke/ transient ischemic attack, 50 = ischemic heart disease, 51 = psoriasis, 52 = systemic glucocorticoids and other immunosuppressants, 53 = skin type, 54 = lifelong number of painful sunburns, 55 = lifelong cumulative number of hours of sun exposure, 56 = number of sunburns of children, 57 = number of moles, 58 = eye color, 59 = natural hair color, 60 = exposure to industrial chemicals, 61 = history of freckling, 62 = outdoor sun exposure, 63 = occupational sun exposure, 64 = tanning bed use, 65 = history of high-risk exposures such as UV light, 66 = burn scar, 67 = radiation treatment, 68 = arsenic exposure, 69 = personal history of nonmelanoma skin cancer, 70 = personal history of melanoma, 71 = current and childhood summer sun exposure, 72 = sunscreen use, 73 = history of cardiovascular disease, 74 = regional solar radiation (Langleys), 75 = menopausal status and use of postmenopausal hormones, 76 = questionnaire cycle, 77 = ability to tan, 78 = UV-B availability at state of residence, 79 = height, 80 = solar UV exposure quartile calculated from summer erythemal UV values weighted by time outdoors, 81 = ever had moles removed, 82 = chronic pain in last year, 83 = kidney disease or ulcer
GEM the genes, environment, and melanoma study, GPRD general practitioners research database, KPNC Kaiser Permanente Northern California population, NHS nurses’ health study, VITAL the vitamins and lifestyle, WHI women’s health initiative
Use aspirin during the 30-day period before the interview
1,2
6
1=age, 2=sex, 3=smoking, 4=family history, 5=educational level, 6=alcohol intake, 7=year of interview, 8=study center, 9=use of other analgesics, 10=BMI, 11=unique number of dispensing, 12=unique number of hospitalizations in the year prior to start of follow up, 13=Charlson comorbidity index, 14= history of connective tissue disorder, 15=auranofin, chlorambucil, cyclophosphamide, cyclosporine, D-penicillamine, and podophyllotoxin, 16=disease activity, 17=residence, 18=menopausal status, 19=race, 20=sitting time, 21=diabetes status, 22=rheumatoid arthritis status, 23=cholesterol-lowering drug use, 24=acetaminophen use, 25=postmenopausal hormone use, 26=self-reported health, 27=history of coronary artery disease, 28=stroke, 29=marital status, 30=transfusion history,31= red meat and fruit intake,32= replacement therapy, 33=history of fatigue/lack of energy
HPFS Health Professionals follow-up study, IWHS Iowa Women’s Health Study, MEC multiethnic cohort study, NHS nurses’ health study, RPCI the Roswell Park Cancer Institute, VITAL the vitamins and lifestyle
Population from Universities of Rome and Bologna and at the European Institute of Oncology
215
860
Small Intestine Neuroendocrine Tumors
Questionnaire
Use at any dose at least twice a week for more than one consecutive year
1,2,
7
1= age, 2=sex
Aspirin use and the risk of cancers
Figures 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17 and 18 and Additional file 1: Table S1 shows the RRs for the 21 separate cancer sites that we assessed and that of the total cancers. The use of aspirin was associated with a reduced cancer risk for ten specific sites: gastric cancer (RR =0.75, 95%CI:0.65–0.86), esophagus cancer (RR = 0.75, 95%CI:0.62–0.89), colorectal cancer(RR = 0.79, 95%CI:0.74–0.85), pancreatic cancer (RR = 0.80, 95%CI:0.68–0.93), breast cancer (RR = 0.92, 95%CI:0.88–0.96), ovarian cancer (RR = 0.89, 95%CI:0.83–0.95), endometrial cancer (RR = 0.92, 95%CI:0.85–0.99), prostate cancer (RR = 0.94, 95%CI:0.90–0.99), and small intestine neuroendocrine tumors (RR = 0.17, 95%CI:0.05–0.58). However, there was no significant association between aspirin use and the risk of some cancers, including hepato-biliary, lung, cervical uterus, renal, renal pelvis and ureter, bladder, brain, head and neck, thyroid, and skin cancers, as well as lymphoma and leukemia.
Fig. 2
Forest plot of aspirin use and the risk of gastric cancer
Fig. 3
Forest plot of aspirin use and the risk of esophagus cancer
Fig. 4
Forest plot of aspirin use and the risk of colorectal cancer
Fig. 5
Forest plot of aspirin use and the risk of hepato-biliary cancer
Fig. 6
Forest plot of aspirin use and the risk of pancreatic cancer
Fig. 7
Forest plot of aspirin use and the risk of lung cancer
Fig. 8
Forest plot of aspirin use and the risk of breast cancer
Fig. 9
Forest plot of aspirin use and the risk of ovarian cancer
Fig. 10
Forest plot of aspirin use and the risk of endometrial cancer
Fig. 11
Forest plot of aspirin use and the risk of prostate cancer
Fig. 12
Forest plot of aspirin use and the risk of renal cancer
Fig. 13
Forest plot of aspirin use and the risk of bladder cancer
Fig. 14
Forest plot of aspirin use and the risk of brain tumors
Fig. 15
Forest plot of aspirin use and the risk of head and neck cancers
Fig. 16
Forest plot of aspirin use and the risk of skin cancer
Fig. 17
Forest plot of aspirin use and the risk of lymphoma
Fig. 18
Forest plot of aspirin use and the risk of leukemia
×
×
×
×
×
×
×
×
×
×
×
×
×
×
×
×
×
Additional file 1: Tables S1–S18 shows the RRs for cancers at 17 sites, in subgroups of studies defined by their design, study location, gender, exposure assessment, quality assessment, duration of aspirin use, and frequency of aspirin use.
We conducted a subgroup analysis stratified by questionnaires and medical records, and found a lower risk in medical records with most cancers (gastric, esophageal, colorectal, hepato-biliary, and pancreatic cancers), however, significant heterogeneity of effects was noted for those subgroups (Additional file 1: Tables S2–S18). As we expected, the decreased risk of colorectal cancer (RRs = 0.76, 95%CI: 0.66–0.87 for ≥5 years), pancreatic cancer (RRs = 0.75, 95%CI: 0.57–0.99 for ≥5 years), ovarian cancer (RRs = 0.77, 95%CI: 0.63–0.93 for ≥5 years), and brain cancer (RRs = 0.65, 95%CI: 0.43–0.97 for ≥5 years) were more pronounced with longer duration of aspirin use. However, the aspirin-associated RR for 21 specific cancers did not vary significantly by other characteristics (gender, quality assessment and frequency of aspirin use).
Publication bias
The funnel plot showed asymmetry (Fig. 19). In addition, the Begg’s test and Egger’s test provided evidence of publication bias among the included studies (Begg’s test Z = 4.34, P < 0.001; Egger’s test Z = − 5.27, P < 0.001).
Fig. 19
Funnel plot of aspirin use and cancer
×
Anzeige
Discussion
The results of our meta-analysis supported the presence of inverse associations between aspirin use and the risk of overall cancer, gastric, esophageal, colorectal, pancreatic, breast, ovarian, endometrial, and prostate cancers, as well as small intestine neuroendocrine tumors. However, no significant associations were observed between the use of aspirin and the risk of other cancers, including hepato-biliary, lung, cervical uterus, renal, renal pelvis and ureter, bladder, brain, head and neck, thyroid, and skin cancers, as well as lymphoma, and leukemia.
There are several potential biological mechanisms through which aspirin could reduce the risk of cancer. First, aspirin and other NSAIDs have been proven to inhibit the activity of the enzyme cyclooxygenase 2 (COX-2), which is responsible for the synthesis of prostaglandins [29]. COX-2 has been reported to be overexpressed in many cancers and participates in key cellular activities, including cell proliferation, apoptosis, angiogenesis, and metastasis [30‐32]. Second, aspirin could activate the NF-kappa B (NF-κB) signaling pathway, which triggers apoptosis in neoplasia [33, 34]. In addition, some studies showed that aspirin might induce gene selection and modulate mitochondrial voltage dependent anion channels (VDACs) to reduce the risk of cancer progression and metastasis [35, 36].
The results of this meta-analysis indicated that utilization of aspirin had different protective effects on the development of cancer. This difference may be attributed to the different expression levels of COX in various cancers [37]. Furthermore, Zumwalt et al. [38] reported that the effectiveness of aspirin was primarily determined by specific genetic variants. Aspirin inhibited cell growth in all cancer cell lines regardless of mutational background, however, the effects were exacerbated in cells with PIK3CA mutations, which might explain the different effects of aspirin on cancers.
The decreased risk of gastric, esophageal, pancreatic, lung, breast, and ovarian cancers was observed in the case-control studies but not in the cohort studies. One possible explanation for the difference might be that cases in the case-control studies might have a recall bias and tended to overestimate the risk of cancer by aspirin use. Another possible explanation is that misclassification or measurement errors for aspirin use in the cohort studies might have distorted the association because most of our analyses were based on baseline data, and there might be a discrepancy between initial recruitment and subsequent aspirin consumption.
Anzeige
The longer those who had used aspirin, the lower their risk of cancer was, with longer duration of use associated with an RR of 0.90 (95% CI 0.89–0.74), based on 118 studies that reported associations with longer (≥5 years) duration of aspirin use and 105 studies that reported associations with shorter (< 5 years) duration of aspirin use. For most cancers (colorectal, pancreatic, ovarian, and brain cancers), risk reductions were more pronounced with longer duration of use, and these results agree with those of previous studies [39‐41]. In addition, the United States Preventive Services Task Force (USPSTF) indicated that cancer prevention was a significant aspect in the overall health benefit of aspirin, but this benefit was not apparent until several years after the initiation of aspirin therapy [42, 43]. It is of note that a significant inverse association with prostate cancer was observed in the patients who took aspirin for less than 5 years. Indeed, after the study that relied on the General Practice Research Database [44] was excluded, the discrepancy disappeared. Considering that aspirin use was off-prescription in the United Kingdom, misclassification was likely to occur in this study because many commonly used aspirins do not require a prescription. Therefore, it can be deduced that the patients who used aspirin for at least 5 years were more likely to realize the potential cancer prevention benefit.
There was no statistically significant difference between the pooled RRs for the frequency of aspirin in most studies. Given that a few studies were included in the subgroup analysis on the basis of the frequency of aspirin use and most studies lacked information on this variable, the results on the risks associated with the frequency of aspirin use should be interpreted with caution. Further studies that explore the associations between the frequency of aspirin use and cancer risk are necessary to elucidate the effects of aspirin.
In addition, our results indicated that the strongest reduction in the risk of most cancers associated with aspirin was found in North American countries. However, two-thirds of the included studies were performed in North America and a few studies were performed in Asian and European countries, which might distort the accuracy of the results. Therefore, more studies are necessary to examine the discrepancies among the different countries and regions.
Comparison with other studies
Bosetti et al. (2011) [45] conducted a meta-analysis on aspirin and 12 selected cancer sites based on 139 observational studies and 187,167 cases. Our study included 218 studies involving 737,409 cases and examined the correlation between aspirin use and the risk of skin, head and neck, hepatobiliary, thyroid, cervical uterus, renal pelvis, ureter, and brain cancers, lymphoma, small intestine neuroendocrine tumors, and leukemia, thereby providing more comprehensive and reliable evidence for this correlation. More importantly, this study was the first meta-analysis to evaluate the association between aspirin use and the risk of hepatobiliary cancer and we found a non-significant effect of aspirin on the risk of hepatobiliary cancer (OR = 0.64, 95% CI: 0.40–1.02).
Anzeige
Algra and Rothwell (2012) [46] conducted a meta-analysis on the association between aspirin use and the risk of cancer based on 195 studies and 215,211 cases. Compared with their review, our meta-analysis have added approximately 70 new articles published since 2012, with a total of 737,409 cases, which significantly enhanced the statistical power to determine this potential association. In addition, the exposure in the previous review was inconsistent, which may mislead the estimation. Many studies defined aspirin as the exposure but only a few studies defined NSAIDs as the exposure, and thus the specific effect of aspirin on cancers was not defined. The exposure to aspirin in our meta-analysis was consistent and ensured the reliability of the findings.
Strengths and limitations
This study is the most up-to-date comprehensive review of the effect of aspirin use on the risk of all types of cancers, and the large sample size provides reliable results with greater precision and power. The potential limitations of this study should be noted. First, there was substantial heterogeneity across the included studies, which was likely due to differences in the definitions of exposure, units, assessment methods, and the adjusted variables across different studies. Second, misclassification or measurement errors for aspirin use might distort the association because our analyses were based on baseline data, and changes in the exposure to aspirin were not updated during the follow-up period. Third, the visual inspection of a funnel plot showed asymmetry, and the Begg’s test and Egger’s test also identified evidence of publication bias among the studies included in our meta-analysis.
Our meta-analysis indicated a beneficial role for aspirin for overall cancers; however, the results should be interpreted with caution. Considering that most evaluated studies were based on secondary prevention rather than on primary prevention, the totality of evidence for the high-risk population was incomplete, and it is appropriate to let the beneficial role remain uncertain. At present, we should accept the uncertainties, and future chemoprevention trials should clarify the extent to which aspirin decreases cancers incidence.
Conclusions and implications
Evidence from observational studies indicates that utilization of aspirin is associated with reduced risk of gastric, colorectal, esophageal, pancreatic, ovarian, endometrial, breast, and prostate cancers, in addition to small intestine neuroendocrine tumors. A stronger protective effect was observed in the North American populations and patients who used aspirin for at least 5 years. It is important to address immortal time bias not only to ensure the integrity of the meta-analysis, but also to ensure the integrity of pharmacoepidemiological studies. Moreover, given the confidence limits of the evaluated studies, adequately powered mechanistic studies should help elucidate the mechanisms underlying this correlation.
Acknowledgements
We would like to thank Xiaoxv Yin and Shiyi Cao for critical review of this manuscript. We are grateful to the members of the Professor Lu for their support and helpful discussions.
Funding
This study was supported by the Fundamental Research Funds for the Central Universities, Huazhong University of Science and Technology, China(2016YXMS215). The funding body contributed to the design of the study and put forward some constructive suggestions for collection, analysis, and interpretation of data and the review and revision of the manuscript.
Availability of data and materials
All data generated or analysed during this study are included in this published article.
Ethics approval and consent to participate
Not applicable
Consent for publication
Not applicable
Competing interests
The authors declare that they have no competing interests.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
