当前位置:首页 >> >>

Effect of daily aspirin on long-term risk of death due to cancer


Articles

E?ect of daily aspirin on long-term risk of death due to cancer: analysis of individual patient data from randomised trials
Peter M Rothwell, F Gerald R Fowkes, Jill F F Belch, Hisao Ogawa, Charles P Warlow, Tom W Meade

Summary
Background Treatment with daily aspirin for 5 years or longer reduces subsequent risk of colorectal cancer. Several lines of evidence suggest that aspirin might also reduce risk of other cancers, particularly of the gastrointestinal tract, but proof in man is lacking. We studied deaths due to cancer during and after randomised trials of daily aspirin versus control done originally for prevention of vascular events. Methods We used individual patient data from all randomised trials of daily aspirin versus no aspirin with mean duration of scheduled trial treatment of 4 years or longer to determine the e?ect of allocation to aspirin on risk of cancer death in relation to scheduled duration of trial treatment for gastrointestinal and non-gastrointestinal cancers. In three large UK trials, long-term post-trial follow-up of individual patients was obtained from death certi?cates and cancer registries. Results In eight eligible trials (25 570 patients, 674 cancer deaths), allocation to aspirin reduced death due to cancer (pooled odds ratio [OR] 0·79, 95% CI 0·68–0·92, p=0·003). On analysis of individual patient data, which were available from seven trials (23 535 patients, 657 cancer deaths), bene?t was apparent only after 5 years’ follow-up (all cancers, hazard ratio [HR] 0·66, 0·50–0·87; gastrointestinal cancers, 0·46, 0·27–0·77; both p=0·003). The 20-year risk of cancer death (1634 deaths in 12 659 patients in three trials) remained lower in the aspirin groups than in the control groups (all solid cancers, HR 0·80, 0·72–0·88, p<0·0001; gastrointestinal cancers, 0·65, 0·54–0·78, p<0·0001), and bene?t increased (interaction p=0·01) with scheduled duration of trial treatment (≥7·5 years: all solid cancers, 0·69, 0·54–0·88, p=0·003; gastrointestinal cancers, 0·41, 0·26–0·66, p=0·0001). The latent period before an e?ect on deaths was about 5 years for oesophageal, pancreatic, brain, and lung cancer, but was more delayed for stomach, colorectal, and prostate cancer. For lung and oesophageal cancer, bene?t was con?ned to adenocarcinomas, and the overall e?ect on 20-year risk of cancer death was greatest for adenocarcinomas (HR 0·66, 0·56–0·77, p<0·0001). Bene?t was unrelated to aspirin dose (75 mg upwards), sex, or smoking, but increased with age—the absolute reduction in 20-year risk of cancer death reaching 7·08% (2·42–11·74) at age 65 years and older. Interpretation Daily aspirin reduced deaths due to several common cancers during and after the trials. Bene?t increased with duration of treatment and was consistent across the di?erent study populations. These ?ndings have implications for guidelines on use of aspirin and for understanding of carcinogenesis and its susceptibility to drug intervention. Funding None.
Published Online December 7, 2010 DOI:10.1016/S01406736(10)62110-1 Stroke Prevention Research Unit, Department of Clinical Neurology, University of Oxford, Oxford, UK (Prof P M Rothwell FMedSci); Wolfson Unit for Prevention of Peripheral Vascular Diseases, Centre for Population Health Sciences, University of Edinburgh, Edinburgh, UK (Prof F G R Fowkes FRCPE); The Institute of Cardiovascular Research, Vascular and In?ammatory Diseases Research Unit, University Division of Medicine and Therapeutics, Ninewells Hospital and Medical School, Dundee, UK (Prof J F F Belch FRCP); Department of Cardiovascular Medicine, Graduate School of Medical Sciences, Kumamoto University, Kumamoto City, Japan (H Ogawa MD); Department of Clinical Neurosciences, University of Edinburgh, Western General Hospital, Edinburgh, UK (Prof C P Warlow FMedSci); and London School of Hygiene and Tropical Medicine, University of London, London, UK (Prof T W Meade FRS) Correspondence to: Prof Peter M Rothwell, Stroke Prevention Research Unit, University Department of Clinical Neurology, John Radcli?e Hospital, Headington, Oxford OX3 9DU, UK peter.rothwell@clneuro.ox. ac.uk

Introduction
In the developed world, the lifetime risk of cancer is about 40%, and rates are increasing in the developing world.1 In Europe, about 3·2 million new cancers present each year, with about 1·7 million deaths,2 and there are more than 1·5 million new cases each year in the USA.3 By contrast with treatment of cancer, there has been little progress in use of drugs in prevention of the disease. However, several lines of evidence suggest that long-term use of aspirin might reduce the risk of some cancers, particularly gastrointestinal tumours. Aspirin reduces incidence or growth rate, or both, of several cancers in animal models,4–6 mediated at least in part by inhibition of the cyclo-oxygenase (COX) enzymes and reduced production of prostaglandins and other in?ammatory mediators, but these ?ndings might not be applicable to humans. Observational studies in humans also suggest that aspirin reduces risk of certain cancers,5–8 but results

have been con?icting, with more rigorous studies yielding weaker associations.8 Moreover, observational studies have proved to be unreliable in determining risks and bene?ts of medications in the past,9,10 and there is trial evidence that one antiplatelet drug might have adverse e?ects on cancer outcomes.11 Nevertheless, long-term follow-up of randomised trials has shown that aspirin does reduce the risk of colorectal cancer after a delay of several years,12,13 probably by reducing precancerous adenomas,14 possibly by inhibition of COX-2.15 However, proof of an e?ect on other cancers is lacking. 10-year follow-up of the Women’s Health Study, a randomised trial of 100 mg of aspirin on alternate days versus control, showed no reduction in incidence of cancer.16 However, aspirin also failed to prevent colorectal adenomas in this study, which is consistent with observational studies suggesting that daily aspirin is required for

www.thelancet.com Published online December 7, 2010 DOI:10.1016/S0140-6736(10)62110-1

1

Articles

Deaths due to cancer Aspirin Control BDAT19 UK-TIA18 ETDRS20 SAPAT22 TPT17 JPAD21 POPADAD23 AAA24 Total 75/3429 21/1621 16/1856 10/1009 87/2545 15/1262 25/638 78/1635 47/1710 23/814 14/1855 19/1026 104/2540 19/1277 31/638 90/1675

Odds ratio (95% CI)

0·79 (0·55–1·14) 0·45 (0·25–0·82) 1·14 (0·56–2·35) 0·53 (0·25–1·15) 0·83 (0·62–1·11) 0·80 (0·40–1·57) 0·80 (0·47–1·37) 0·86 (0·63–1·17)

327/14 035 347/11 535 0·79 (0·68–0·92) 0 1 Odds ratio (95% CI) 2

psig=0·003, phet=0·84

Figure 1: Meta-analysis of the e?ect of aspirin on deaths due to cancer during all eligible randomised trials of aspirin versus control Data are n/N, where n=number of cancer deaths and N=number of trial participants in that treatment group. BDAT=British Doctors Aspirin Trial. UK-TIA=UK transient ischaemic attack trial. ETDRS=Early Treatment Diabetic Retinopathy Study. SAPAT=Swedish Angina Pectoris Aspirin Trial. TPT=Thrombosis Prevention Trial. JPAD=Japanese Primary Prevention of Atherosclerosis With Aspirin for Diabetes. POPADAD=Prevention of Progression of Arterial Disease and Diabetes. AAA=Aspirin for Asymptomatic Atherosclerosis.

aspirin in primary versus secondary prevention of vascular disease. In view of the availability of published data for all trials of antiplatelet agents from the Antithrombotic Trialists’ (ATT) Collaboration, literature searches were con?ned to publications after the last ATT search (2002).20,21 Trials were identi?ed by searching for relevant systematic reviews in the Cochrane Collaboration Database of Systematic Reviews and by searches of PubMed and Embase (both last done on March 12, 2010) using the terms “aspirin” or “salicyl*” or “antiplatelet” with the term “randomised controlled trial”. The searches were restricted to studies done in humans, but there was no restriction on language.

Procedure
The original investigators were contacted to determine whether anonymised data were available for the number of deaths in which cancer had been regarded as the main underlying cause, the time from randomisation to death, and the primary site of cancer. All cancer deaths had been coded according to the ninth or tenth revision of the International Classi?cation of Diseases (ICD) and the designation of death due to cancer that had been made by the original trialists was used, unless speci?ed otherwise. However, in three trials,17–19 we reviewed the paper case-records of all deaths in patients with known incident cancer to check the designation of cause of death, with the aim of identifying any possible bias resulting from an increase in risk of vascular events due to withdrawal of aspirin treatment after diagnosis of cancer, which might reduce the number of deaths attributed to cancer in the aspirin groups. Three eligible trials, all UK-based, had continued to obtain data for deaths due to cancer after completion of the trials via the national death certi?cation and cancer registration systems—the Thrombosis Prevention Trial (TPT),17 the British Doctors Aspirin Trial (BDAT),19 and the UK transient ischaemic attack (UK-TIA) aspirin trial.18 TPT17 was a 2×2 factorial double-blind randomised trial of aspirin versus placebo and warfarin versus placebo in men aged 45–69 years at increased vascular risk. 135 000 patient records were reviewed in 108 UK primary care practices to exclude ineligible subjects, including those with a recent history of possible peptic ulceration or previous myocardial infarction or stroke. 5085 men with high vascular risk-factor scores were recruited from 1989 to 1992. 2545 were allocated to aspirin (75 mg daily controlled release) and 2540 to placebo. Men were reviewed by their family doctor each year and a research nurse searched their medical records. None were lost to follow-up before the trial end date (October, 1997). All trial participants were ?agged in the National Health Service Central Register and noti?cations of cancer or death were obtained until September, 2009.

prevention of cancer.5–8 Observational studies also suggest that use of aspirin for at least 5 years is required before reductions in risk of cancer are observed,5–8 and the e?ect of aspirin on risk of colorectal cancer on follow-up of randomised trials was greatest in patients with duration of trial treatment of 5 years or longer.12,13 We therefore determined the e?ect of aspirin on risk of fatal cancer by analysis of individual patient data for deaths due to cancer during randomised trials of daily aspirin versus control (done originally for primary or secondary prevention of vascular events) in which the median duration of scheduled trial treatment was at least 4 years. We studied fatal cancers only, in the ?rst instance, because cause of non-vascular deaths was reliably determined in most aspirin trials, and we also aimed to determine the e?ect of any reduction in cancer deaths on overall all-cause mortality. In three trials done in the UK,17–19 we also determined any delayed e?ects of aspirin on the 20-year risk of death due to cancer by long-term post-trial follow-up.

Methods
Search strategy and selection criteria
We searched for randomised trials of aspirin versus control that had a mean or median scheduled trial treatment period of at least 4 years and a range extending beyond 5 years. Eligible trials had investigated the e?ects of randomised allocation to: aspirin (any dose) versus no aspirin in the absence of another agent; or aspirin (any dose) versus no aspirin in the presence of another antiplatelet agent or antithrombotic agent, if the other agent was used in the same way in the aspirin and no aspirin groups. Given the focus on cancer outcomes, no distinction was made between trials of
2

www.thelancet.com Published online December 7, 2010 DOI:10.1016/S0140-6736(10)62110-1

Articles

BDAT19 recruited 5139 male doctors (4377 in 1978 and 762 in 1979) who were resident in the UK, born on or after 1900, had no contraindication to aspirin, no regular aspirin use, and no history of peptic ulcer disease, stroke, or myocardial infarction. Randomisation (in a 2:1 ratio) was to daily aspirin (500 mg ordinary, soluble, or e?ervescent aspirin, as desired, or, if subsequently requested, 300 mg enteric coated aspirin) versus no aspirin or products containing aspirin. Placebo tablets were not used. Treatment was continued until 1984. All participants were asked to complete a questionnaire every 6 months about their health and use of aspirin. Participants were ?agged with the National Cancer Registry and the O?ce of the Registrar General, and all noti?cations of cancer and death were collected until 2001.12,13 UK-TIA18 recruited 2435 patients with a recent TIA or minor ischaemic stroke from 33 centres in the UK and Ireland between 1979 and 1985. Participants were older than 40 years, with no aspirin intolerance, alcoholism, chronic renal failure, or peptic ulceration. Randomisation was to 1200 mg aspirin daily versus 300 mg daily versus placebo, and treatment was double-blind. Patients were seen by a physician every 4 months until the end of the trial in 1986, and none were lost. Data for deaths and incident cancers noti?ed during and after the trial until 2006 were obtained from national registries, as reported previously.12,13

6 5 Risk of cancer death (%) 4 3 2 1 0

Control Aspirin

p=0·01

0

1

2

3

4

5 Years to death 9264 7339

6

7

8

9

Number at risk Aspirin 13 026 Control 10 509

12 849 10 351

12 371 10 026

11 919 9720

10 964 8881

7385 5933

3384 3438

1676 1671

977 969

Figure 2: E?ect of allocation to aspirin versus control on risk of death due to cancer during the trial treatment periods in a pooled analysis of the 23 535 patients in seven trials17–21,23,24 n 0–5 years’ follow-up HR (95% CI) Site of primary cancer* Gastrointestinal Oesophagus Pancreas Colorectal Stomach Other All Non-gastrointestinal Lung Prostate Bladder and kidney Other solid All Unknown primary All solid cancers Histological type? Adenocarcinoma Non-adenocarcinoma Unknown Haematological All cancers* All cancers including ETDRS? 247 224 106 50 627 657 0·86 (0·62–1·18) 0·89 (0·65–1·23) 0·91 (0·58–1·44) 0·82 (0·44–1·54) 0·88 (0·72–1·06) 0·86 (0·71–1·04) 0·34 0·48 0·70 0·53 0·17 0·11 0·53 (0·35–0·81) 0·79 (0·50–1·24) 0·69 (0·34–1·43) 0·34 (0·09–1·28) 0·62 (0·47–0·82) 0·66 (0·50–0·87) 0·003 0·30 0·32 0·11 0·001 0·003 23 45 54 36 24 182 198 37 31 93 359 36 577 0·78 (0·27–2·23) 0·88 (0·44–1·77) 0·78 (0·39–1·56) 1·85 (0·81–4·23) 0·67 (0·23–1·99) 0·96 (0·67–1·38) 0·92 (0·65–1·30) 0·70 (0·29–1·73) 1·04 (0·44–2·47) 0·86 (0·52–1·44) 0·90 (0·69–1·16) 0·56 (0·28–1·15) 0·88 (0·72–1·08) 0·64 0·73 0·48 0·14 0·47 0·81 0·65 0·44 0·93 0·57 0·41 0·12 0·22 0·43 (0·11–1·72) 0·25 (0·07–0·92) 0·41 (0·17–1·00) 3·09 (0·64–14·91) 0·20 (0·04–0·91) 0·46 (0·27–0·77) 0·68 (0·42–1·10) 0·52 (0·20–1·34) 1·28 (0·36–4·54) 1·01 (0·51–1·98) 0·76 (0·54–1·08) 0·56 (0·09–3·38) 0·64 (0·49–0·85) 0·23 0·04 0·05 0·16 0·04 0·003 0·11 0·17 0·70 0·98 0·12 0·53 0·002 p value ≥5 years’ follow-up HR (95% CI) p value

Analysis of deaths during the trial period
The e?ects of allocation to aspirin on risk of death due to cancer and all-cause mortality during each trial were expressed as odds ratios (ORs; with 95% CIs). Pooled estimates were obtained by ?xed-e?ects meta-analysis. After we assessed heterogeneity in e?ect of aspirin across trials, individual patient data were pooled. The cumulative e?ect of aspirin on risk of cancer death was estimated with Kaplan-Meier curves and log-rank test (strati?ed by trial) and by hazard ratios (HRs) obtained from a Cox proportional hazards model strati?ed by trial. All analyses were by intention to treat on the basis of treatment allocation in the original trials. The following strati?ed analyses were done: (1) for cancers of the gastrointestinal tract versus other solid cancers versus haematological cancers, given the prior expectation of greatest e?ects on gastrointestinal cancers (de?ned as primary site oesophagus, stomach, small intestine, colon, rectum, pancreas, biliary tract, gallbladder, and liver); (2) for the ?rst 5 years after randomisation versus thereafter, given the expectation of greater e?ects after scheduled treatment and followup for at least 5 years; (3) for common speci?c solid cancers (oesophagus [with histological type], stomach, pancreas and biliary tract, colorectal, liver, lung [with histological type], prostate, bladder and kidney, and metastases with unknown primary [with histological type]).

The numbers of cancer deaths included from each trial are those shown on webappendix p 3. n=number of cancer deaths. HR=hazard ratio. ETDRS=Early Treatment Diabetic Retinopathy Study. *Analysis con?ned to the six trials with site-speci?c cancer data follow-up.17–19,21, 23,24 ?Analysis con?ned to solid (non-haematological) cancers. ?Analysis included cancer deaths in ETDRS,20 in which neither primary site nor histological type was known in any case.

Table 1: Pooled analysis of the e?ect of allocation to aspirin on risk of death due to cancer during the seven trials from which individual patient data were available,17–21,23,24 strati?ed by type of primary tumour and period of follow-up

Analysis of long-term risk of cancer death
For long-term follow-up of the three UK trial cohorts,17–19 all death certi?cate and cancer registration data relating to events occurring after the trials were also coded according to ICD 9 or 10 (masked to treatment
3

www.thelancet.com Published online December 7, 2010 DOI:10.1016/S0140-6736(10)62110-1

Articles

1–4·9 years 25 Risk of cancer death (%) 20 15 10 5 0 0 Number at risk Aspirin 1337 Control 820 5 1151 733 10 Years to death 942 622 15 732 497 20 347 199 p=0·62 Control Aspirin

5–7·4 years

the e?ect of aspirin versus placebo was compared with that of warfarin versus placebo.

Role of the funding source
p=0·0003

The study was unfunded and was independent of any pharmaceutical company or other commercial interest. The corresponding author had full access to all the data in the study and had ?nal responsibility for the decision to submit for publication.
15 3871 2390 20

0 5426 3383

5 5028 3135

10 Years to death 4528 2814

Results
2274 1134

Trials
Of eight eligible randomised trials of aspirin versus control (webappendix p 1) with a mean duration of scheduled treatment before the end of the trial of 4 years or more, two had been done in primary prevention of vascular disease,17,19 one in secondary prevention after recent vascular events,18 and ?ve in groups with increased vascular risk without previous vascular events (type 1 or 2 diabetes;20 type 2 diabetes;21 stable angina;22 diabetes with asymptomatic peripheral arterial disease;23 low ankle brachial index24). Data for the number of deaths due to cancer were available from all eight trials. Individual patient data were available from seven trials, but all records of one trial had been destroyed (Juul-Moller S, University Hospital, Malmo, Sweden, personal communication)..22

≥7·5 years 25 Risk of cancer death (%) 20 15 p=0·003 10 5 0 0 Number at risk Aspirin 832 Control 861 5 10 Years to death 715 731 15 20

All patients

p<0·0001

0

5

10 Years to death 6185 4167

15

20

788 813

614 616

360 359

7595 5064

6967 4681

5217 3503

2981 1692

Figure 3: E?ect of allocation to aspirin versus control on 20-year risk of death due to any solid cancer strati?ed by scheduled duration of trial treatment in three trials with long-term follow-up17–19 Continuous variable interaction, p=0·01.
See Online for webappendix

In-trial deaths
During the eight trials there were 674 deaths due to cancer among 25 570 patients. The proportion of all deaths that were due to cancer varied (p<0·0001), ranging from 4·2% in the young diabetic population in the Early Treatment Diabetic Retinopathy Study (ETDRS),20 12·8% and 15·4% in the patients with symptomatic vascular disease in the 1980s in UK-TIA18 and the Swedish Angina Pectoris Aspirin Trial,22 to 28·7% (Prevention of Progression of Arterial Disease and Diabetes [POPADAD] study),23 29·2% (BDAT),19 45·4% (TPT),17 46·4% (Aspirin for Asymptomatic Atherosclerosis [AAA] trial),24 and 47·9% (Japanese Primary Prevention of Atherosclerosis With Aspirin for Diabetes [JPAD] study)21 in lower vascular risk or more recently recruited cohorts. However, there was very little heterogeneity between trials (?gure 1; phet=0·84) in the e?ect of allocation to aspirin on risk of death due to cancer (OR 0·79, 95% CI 0·68–0·92, p=0·003, overall; 0·81, 0·68–0·97, p=0·03, in trials of aspirin 75–100 mg daily). Reclassi?cation of cause of death in a small number of cases in UK-TIA and TPT had little e?ect on the pooled estimate (686 deaths; OR 0·80, 0·69–0·93, p=0·004, webappendix p 3). The reduction in cancer deaths on aspirin during the trials resulted in lowered in-trial all-cause mortality (10·2% vs 11·1%, OR 0·92, 0·85–1·00, p=0·047, webappendix p 4), even though other deaths were not reduced (0·98, 0·89–1·07, p=0·63). In our analysis of individual patient data for time to death, which were available for seven trials (657 cancer

allocation). Fatal cancers were de?ned as those in which the cancer had been recorded as the primary underlying cause of death on the death certi?cate. After checking for heterogeneity between the trials in the absolute risk of cancer death during the trial and posttrial follow-up, which might confound a pooled analysis, and in the e?ect of allocation to aspirin on cancer death, we pooled individual patient data. Scheduled duration of trial treatment was 5 or 6 years in BDAT and ranged from 1 to 7 years in UK-TIA and from 4 to 9 years in TPT. The e?ect of scheduled duration of trial treatment on 20-year risk of cancer death was explored within each trial and in the pooled data with an interaction term in a Cox model with duration modelled as a continuous variable, and subsequent analyses were strati?ed accordingly (≤5 vs 5–7·4 vs ≥7·5 years). All analyses were done on an intention-to-treat basis, with scheduled duration of treatment simply de?ned as date of randomisation to date of the end of the trial, irrespective of compliance with treatment. The e?ect of aspirin on 20-year risk of cancer death was also strati?ed by category of cancers (gastrointestinal tract vs other solid vs haematological), by period of follow-up (0–10 vs 10–20 years), and was determined for deaths due to speci?c cancers (as de?ned above) in the 10 502 patients with more than 5 years’ scheduled duration of trial treatment. In TPT,

4

www.thelancet.com Published online December 7, 2010 DOI:10.1016/S0140-6736(10)62110-1

Articles

n

0–10 years’ follow-up HR (95% CI) p value

10–20 years’ follow-up HR (95% CI) p value

0–20 years’ follow-up HR (95% CI) p value

Solid cancers Gastrointestinal Oesophagus Pancreas Colorectal Stomach Other All Non-gastrointestinal Lung Prostate Bladder and kidney Other solid All Unknown primary All solid cancers Histological type* Adenocarcinoma Non-adenocarcinoma Unknown Haematological cancers All cancers 648 302 331 126 1378 0·70 (0·54–0·91) 1·04 (0·72–1·52) 0·66 (0·49–0·90) 1·31 (0·69–2·50) 0·79 (0·66–0·93) 0·008 0·83 0·01 0·41 0·005 0·64 (0·53–0·77) 0·74 (0·55–0·98) 1·12 (0·83–1·52) 1·00 (0·65–1·54) 0·77 (0·67–0·89) <0·0001 0·04 0·46 0·99 0·0002 0·66 (0·56–0·77) 0·87 (0·70–1·08) 0·84 (0·67–1·05) 1·09 (0·76–1·56) 0·78 (0·70–0·87) <0·0001 0·21 0·13 0·65 <0·0001 326 210 94 128 757 89 1251 0·68 (0·50–0·92) 0·83 (0·47–1·46) 0·75 (0·41–1·37) 0·68 (0·39–1·17) 0·71 (0·56–0·88) 1·19 (0·58–2·42) 0·76 (0·63–0·90) 0·01 0·52 0·35 0·16 0·002 0·63 0·002 0·75 (0·55–1·02) 0·80 (0·58–1·09) 0·90 (0·52–1·57) 1·28 (0·80–2·05) 0·85 (0·71–1·03) 0·95 (0·56–1·61) 0·75 (0·65–0·87) 0·07 0·15 0·72 0·31 0·10 0·84 0·0001 0·71 (0·58–0·89) 0·81 (0·61–1·06) 0·83 (0·55–1·25) 0·98 (0·69–1·39) 0·79 (0·69–0·91) 1·03 (0·67–1·57) 0·75 (0·67–0·84) 0·002 0·12 0·37 0·91 0·001 0·90 <0·0001 62 77 179 71 18 409 0·53 (0·24–1·18) 0·82 (0·41–1·67) 0·79 (0·49–1·26) 1·36 (0·64–2·90) 0·68 (0·14–3·36) 0·80 (0·59–1·08) 0·12 0·59 0·32 0·43 0·64 0·14 0·36 (0·18–0·71) 0·79 (0·44–1·42) 0·51 (0·35–0·74) 0·42 (0·23–0·79) 1·97 (0·53–7·27) 0·56 (0·44–0·72) 0·003 0·43 0·0005 0·007 0·31 <0·0001 0·42 (0·25–0·71) 0·81 (0·51–1·26) 0·60 (0·45–0·81) 0·69 (0·43–1·10) 1·33 (0·50–3·54) 0·65 (0·53–0·78) 0·001 0·34 0·0007 0·11 0·57 <0·0001

Analysis limited to patients with scheduled duration of trial treatment or 5 years or longer. n=number of cancer deaths. HR=hazard ratio. *Analysis con?ned to solid (non-haematological) cancers.

Table 2: Pooled analysis of the e?ect of allocation to aspirin on the 20-year risk of death due to cancer during and after the trial treatment periods in the 10 502 patients with scheduled treatment duration of 5 years or longer in the three trials with long-term follow-up,17–19 strati?ed by type of primary tumour and period of follow-up

deaths in 23 535 patients based on data in webappendix p 3),17–21,23,24 aspirin reduced deaths due to cancer (HR 0·82, 0·70–0·95, p=0·01, ?gure 2), due mainly to fewer deaths after ?ve years (0·66, 0·50–0·87, p=0·003; table 1), but had no e?ect on other deaths (n=1871; 1·03, 0·94–1·13, p=0·54). Data were available for the site of the primary cancer in six trials (627 cancer deaths in 19 824 patients).17–19,21, 23,24 Aspirin reduced deaths due to gastrointestinal cancers and deaths due to nongastrointestinal solid cancers (table 1), with most bene?t again seen after 5 years of scheduled trial treatment (gastrointestinal cancers, HR 0·46, 0·27–0·77, p=0·003; non-gastrointestinal solid cancers, 0·76, 0·54–1·08, p=0·12), and including signi?cant reductions in colorectal and pancreatic cancer deaths (table 1).

Post-trial follow-up
Follow-up was obtained to 20 years in TPT, BDAT, and UK-TIA (1634 cancer deaths in 12 659 patients, webappendix p 2).17–19 Aspirin reduced the 20-year risk of death due to all solid cancers (HR 0·80, 0·72–0·88, p<0·0001) and gastrointestinal cancer (0·65, 0·54–0·78, p<0·0001), but not haematological cancer (1·03, 0·74–1·43, p=0·87). However, in both TPT and UK-TIA, in which duration of trial treatment varied, the e?ect on 20-year risk

of solid cancer increased with duration of scheduled treatment (interaction: p=0·016 in TPT, p=0·08 in UKTIA). This interaction remained (p=0·01) in the pooled analysis with BDAT (?gure 3), with no reduction in solid cancers in patients with scheduled treatment for 1–4·9 years (HR 1·06, 0·82–1·39, p=0·62), signi?cant bene?t with 5–7·4 years (0·79, 0·70–0·90, p=0·0003), and greatest bene?t with 7·5 years or longer (solid cancers, 0·69, 0·54–0·88, p=0·003; gastrointestinal cancers, 0·41, 0·26–0·66, p=0·0001). Results given below therefore refer to the 10 502 patients (1378 cancer deaths) with scheduled duration of 5 years or longer unless otherwise speci?ed. In patients with scheduled duration of trial treatment of 5 years or more (table 2), allocation to aspirin reduced the 20-year risk of death due to both gastrointestinal (HR 0·65, 0·53–0·78, p<0·0001) and non-gastrointestinal solid (0·79, 0·69–0·91, p=0·001) cancers. There was no signi?cant heterogeneity in e?ect of aspirin across the di?erent gastrointestinal cancers (p=0·26), but e?ects were greatest for oesophageal and colorectal cancers (table 2, ?gure 4). As expected, there was a latent period before any e?ect was observed, with reductions in risk of death due to oesophageal and pancreatic cancer evident from 5 years onwards and reductions in deaths due to stomach and colorectal cancer not evident until about
5

www.thelancet.com Published online December 7, 2010 DOI:10.1016/S0140-6736(10)62110-1

Articles

Oesophagus 2 Risk of cancer death (%) 1·5 1 0·5 0 Pancreas 2 Risk of cancer death (%) 1·5 1 0·5 0 Lung 5 Risk of cancer death (%) 4 3 p=0·002 2 1 0 Kidney/bladder 2 Risk of cancer death (%) 1·5 1 0·5 0 0 Number at risk Aspirin 6258 Control 4244 5 10 Years to death 5243 3545 15 20 p=0·37 2 1·5 1 0·5 0 1 0 2 4 3 p=0·34 4 3·5 3 2·5 2 1·5 1 0·5 0 p=0·001 Control Aspirin 2 1·5 1 0·5 0

Stomach

p=0·11

Colorectal

p=0·0007

Prostate

p=0·12

Haematological

p=0·65

0

5

10 Years to death 5243 3545

15

20

5816 3948

4485 3006

2634 1493

6258 4244

5816 3948

4485 3006

2634 1493

Figure 4: E?ect of allocation to aspirin versus control on the 20-year risk of death due to the most common fatal cancers in the 10 502 patients with scheduled treatment duration of 5 years or longer in the three trials with long-term follow-up17–19 The eight most common cancer types are shown.

10 years (tables 1 and 2). The e?ect of aspirin on death due to pancreatic cancer was only signi?cant at 20-year follow-up in patients with scheduled duration of trial treatment longer than 7·5 years (HR 0·28, 0·08–1·00, p=0·04). Overall, the absolute risk of death due to gastrointestinal cancer was reduced by 2·18% (1·14–3·22) at 20-year follow-up. The e?ect of aspirin on 20-year risk of death due to non-gastrointestinal solid cancer (table 2) was attributable mainly to a reduction in deaths due to lung cancer and a non-signi?cant late reduction in deaths
6

due to prostate cancer (?gure 4), particularly in patients with scheduled duration of trial treatment of 7·5 years or longer (HR 0·52, 0·24–1·10, p=0·08). Aspirin also reduced deaths due to primary brain tumours during the ?rst 10 years of follow-up (5/6258 in the aspirin groups vs 12/4244 in the control groups; HR 0·31, 0·11–0·89, p=0·03) and mean time from randomisation to death from brain tumour remained longer in the aspirin group than in the control group at 20 years (p=0·018, webappendix p 5). Overall, the absolute risk of death due to non-gastrointestinal solid cancer was reduced by 1·88% (0·57–3·19) at 20-year follow-up. The e?ect of aspirin on risk of death due to gastrointestinal cancer did not di?er by age at randomisation (?gure 5; interaction: relative e?ect, p=0·44; absolute e?ect, p=0·96), but the e?ect on death due to non-gastrointestinal solid cancers increased with age (relative e?ect, p=0·056; absolute e?ect, p=0·001). For the 20-year risk of death due to any cancer, the reductions in absolute risk in the aspirin groups were 1·41% (–0·74 to 3·56) at age less than 55 years, 4·53% (2·06–6·99) at age 55–64 years, 7·08% (2·42–11·74) at age 65 years or older, and 3·49% (1·85–5·13) at all ages combined. Relative and absolute e?ects were similar in smokers and non-smokers (data not shown). Where data for histological type were available, aspirin had no e?ect on the 20-year risk of death due to smallcell (HR 0·85, 0·52–1·39, p=0·56) or squamouscell (1·26, 0·73–2·18, p=0·49) lung cancers, but reduced the risk of death due to adenocarcinoma of lung (0·55, 0·33–0·94, p=0·04). The reduction in deaths due to oesophageal cancer was also con?ned to adenocarcinoma (HR 0·36, 0·21–0·63, p=0·0001), although the number of squamous-cell cancers was small (9/6258 in the aspirin groups vs 2/4244 in the control groups). Indeed, across all cancers (tables 1 and 2, webappendix p 6), aspirin only reduced deaths due to either histologically proven adenocarcinomas or primary cancers in which adenocarcinoma predominates (stomach, small bowel, pancreas, bile duct, colon, rectum, breast, uterus, ovary, and prostate). This e?ect on adenocarcinoma was consistent across the three trials (webappendix p 7) and for di?erent doses of aspirin (webappendix p 6), but was not seen in the comparison of warfarin versus placebo in TPT (?gure 6). In patients with scheduled duration of trial treatment of 5 years or longer, all-cause mortality was reduced at 15 years’ follow-up (HR 0·92, 0·86–0·99, p=0·03), due entirely to fewer cancer deaths, but this e?ect was no longer seen at 20 years (0·96, 0·90–1·02, p=0·37). However, the e?ect on post-trial deaths was diluted by a transient increase in risk of vascular death in the aspirin groups during the ?rst year after completion of the trials (75 observed vs 46 expected, OR 1·69, 1·08–2·62, p=0·02), presumably due to withdrawal of trial aspirin.

www.thelancet.com Published online December 7, 2010 DOI:10.1016/S0140-6736(10)62110-1

Articles

Discussion
We showed previously that treatment with aspirin for longer than 5 years reduced the long-term risk of colorectal cancer.12,13 In analyses of nearly 2000 cancer deaths, we now show that aspirin also reduces deaths due to several other common cancers (panel). First, we showed by meta-analysis that aspirin reduced the risk of death due to cancer by about 20% during the trials. Second, by analysis of individual patient data we showed that this bene?t was due mainly to a delayed reduction of about 30–40% in deaths after 5 years of treatment. Third, by long-term follow-up of three large trials we showed that the reduction in deaths due to solid cancers was maintained for 20 years, only becoming apparent for some cancers after completion of the trials. Fourth, these e?ects were consistent across trials, despite the very di?erent populations, suggesting that the ?ndings will be generalisable. Fifth, as shown for colorectal cancer,12,13 the e?ect of aspirin increased with duration of scheduled trial treatment. Sixth, the e?ect was limited to certain cancers, most particularly adenocarcinomas. Seventh, the e?ect did not appear to increase at aspirin doses greater than 75 mg daily. Eighth, the absolute reduction in death due to cancer increased with age, within the range of patients entered into the trials. Finally, the e?ect of aspirin on risk of fatal cancers resulted in a small reduction in allcause mortality. Our analyses were conservative in several respects. First, although all but one of the trials we studied were double-blind, there were high rates of drop-outs from randomised treatment. In the trials in which we obtained long-term follow-up, about 40% of patients in the aspirin groups had stopped treatment by the end of the trial periods.17–19 Nevertheless, to reduce bias we restricted our analyses to intention to treat. Second, since the e?ect of aspirin increased with scheduled duration of trial treatment, but the trials were of ?nite length, it is likely that we underestimated the bene?t of long-term treatment on deaths due to cancer. The di?erence in aspirin use between the treatment groups was already limited by the end of BDAT, many patients in the TPT control group went on to aspirin after the trial,25 and posttrial aspirin use would not have di?ered much between the treatment groups in UK-TIA because trial treatment allocation was never revealed. The trials that we studied were randomised, but could our ?ndings have been due to bias? First, the trials were not designed to study cancer. However, cancer deaths were recorded during the trials, and long-term follow-up via UK cancer registration achieves high rates of ascertainment and accuracy,26–28 as we found previously for colorectal cancer.12,13 Attribution of cause of death during the trials was masked to treatment allocation, as was coding of the cause of post-trial deaths. Attribution was usually based on death certi?cation, supported by any previous cancer registration, which has been shown previously to agree
12 Risk of cancer death (%) 10 8 6 4

Gastrointestinal cancers <55 years Control Aspirin

20 15

Non-gastrointestinal cancers <55 years

p=0·001

10 5

p=0·58

2 0 0 5 10 Years to death 1681 1259 15 20 0 0 5 10 Years to death 1681 1259 15 20

Number at risk Aspirin 1796 Control 1352 12 Risk of cancer death (%) 10 8 6 4

1743 1314

1566 1179

944 596

1796 1352 20 15

1743 1314

1566 1179

944 596

55–64 years

55–64 years

p=0·003

10 5

p=0·001

2 0 0 5 10 Years to death 2468 1617 15 20 0 0 5 10 Years to death 2468 1617 15 20

Number at risk Aspirin 2850 Control 1924 12 Risk of cancer death (%) 10 8 6 4 2 0 0 Number at risk Aspirin 1612 Control 968

2680 1808

2152 1378

1333 721

2850 1924 20 15

2680 1808

2152 1378

1333 721

≥65 years

≥65 years

p=0·03

10 5 0

p=0·004

5

10 Years to death 1094 669

15

20

0

5

10 Years to death 1094 669

15

20

1393 826

767 449

357 176

1612 968

1393 826

767 449

357 176

Figure 5: E?ect of allocation to aspirin versus control on risk of death due to cancers of the gastrointestinal tract and other solid cancers in the 10 502 patients with scheduled treatment duration of 5 years or longer, strati?ed by age at randomisation in three trials with long-term follow-up17–19

well with expert committee review.29–31 Second, lack of knowledge among the trial investigators that data might later be used to study the e?ect of aspirin on risk of cancer will have limited any potential investigator bias. Third, investigation of side-e?ects of aspirin, such as anaemia and bleeding, might have resulted in earlier diagnosis of cancers and hence a reduction in later deaths. However, analysis of time to incidence of colorectal cancer showed no evidence of earlier diagnosis,13 and the very low cure rates of cancers such as oesophageal cancer would limit any bias due to earlier diagnosis. The only evidence of a possible e?ect of increased investigation in the aspirin groups was a transient increase in risk of deaths attributed to stomach cancer during the trials, and a transient reduction in deaths attributed to cancers with unknown primary site (table 1). Moreover, the complete lack of any e?ect of warfarin on cancer deaths in TPT suggests that
7

www.thelancet.com Published online December 7, 2010 DOI:10.1016/S0140-6736(10)62110-1

Articles

Aspirin vs placebo 2·5 Risk of cancer death (%) 2 1·5 1 0·5 0 0 5 10 Years to death 2197 2189 15 20 p=0·008 Lung adenocarcinoma Control Aspirin 2·5 2 1·5 1 0·5 0 0 5 p=0·99

Warfarin vs placebo Lung adenocarcinoma

10 Years to death 2205 2181

15

20

Number at risk Aspirin 2545 Control 2540

2402 2404

1913 1883

410 412

Number at risk Warfarin 2545 Control 2540

2409 2397

1896 1900

407 415

8 Gastrointestinal adenocarcinoma Risk of cancer death (%) 6 4 2 0 p=0·0002

8 6 4 2 0

Gastrointestinal adenocarcinoma

p=0·91

0

5

10 Years to death 2197 2189

15

20

0

5

10 Years to death 2205 2181

15

20

Number at risk Aspirin 2545 Control 2540

2402 2404

1913 1883

410 412

Number at risk Warfarin 2545 Control 2540

2409 2397

1896 1900

407 415

8 Prostate adenocarcinoma Risk of cancer (%) 6 4 2 0 p=0·09

8 6 4 2 0

Prostate adenocarcinoma

p=0·69

0

5

10 Years to noti?cation 2173 2164

15

20

0

5

10 Years to noti?cation 2179 2158

15

20

Number at risk Aspirin 2545 Control 2540

2388 2387

1859 1835

400 397

Number at risk Warfarin 2545 Control 2540

2394 2381

1845 1849

397 400

14 Risk of cancer death (%) 12 10 8 6 4 2 0 0

All fatal adenocarcinoma

14 12 10 8 6 4 2 0

All fatal adenocarcinoma

p<0·0001

p=0·84

5

10 Years to death

15

20

0

5

10 Years to death 2205 2181

15

20

Number at risk Aspirin 2545 Control 2540

2402 2404

2197 2189

1913 1883

410 412

Number at risk Warfarin 2545 Control 2540

2409 2397

1896 1900

407 415

Figure 6: Comparison of e?ect of allocation to aspirin or warfarin versus placebo on risk of death due to adenocarcinoma during long-term follow-up of the Thrombosis Prevention Trial17 Analysis includes all patients, irrespective of scheduled duration of trial treatment. Analysis of prostate cancer also includes non-fatal cancers because of the small numbers of fatal cancers in the single trial.

8

www.thelancet.com Published online December 7, 2010 DOI:10.1016/S0140-6736(10)62110-1

Articles

Panel: Research in context Findings Using individual patient data from all randomised trials of daily aspirin versus no aspirin with mean duration of scheduled trial treatment longer than 4 years, we showed that aspirin reduced risk of death due to cancer by about 20% in the trials, due mainly to a 34% reduction in cancer deaths after 5 years. By long-term post-trial follow-up of patients in three of these trials, we showed that the 20-year risk of cancer death remained about 20% lower in the aspirin groups, and that bene?t increased with scheduled duration of treatment in the original trial. The latent period before an e?ect on deaths was about 5 years for oesophageal, pancreatic, brain, and lung cancer, but was more delayed for stomach, colorectal, and prostate cancer. For lung and oesophageal cancer, bene?t was con?ned to adenocarcinomas. Interpretation These ?ndings provide the ?rst proof in man that aspirin reduces deaths due to several common cancers. Bene?t was consistent across the di?erent trial populations, suggesting that the ?ndings are likely to be generalisable.

early diagnosis due to bleeding is unlikely to have been a major source of bias. Fourth, many patients would probably have stopped taking their trial drug if they developed cancer. Stopping aspirin could in some cases have triggered a fatal vascular event that might have resulted in an underlying cancer not being diagnosed or at least not being listed on the death certi?cate as the underlying cause of death. However, there was no evidence of any excess of non-fatal vascular events during the year before death due to cancer in the aspirin groups (data not shown), and most of the reduction in cancer deaths occurred after the trials. Finally, we had long-term posttrial follow-up from only three of the trials, but this factor was determined simply by the country in which the trials were done and the era. Moreover, the e?ect of aspirin on in-trial cancer deaths was no greater in these trials than in the others (?gure 1). These results therefore provide the ?rst reliable evidence that aspirin prevents non-colorectal cancer in humans, which is consistent with previous predictions of e?ects on cancers of the oesophagus, stomach, pancreas, lung, prostate,7 and possibly brain.32,33 However, more work is required. E?ects of aspirin on incidence of cancer must be determined, both for cancers that are less commonly fatal and to determine whether the latent period before an e?ect is shorter than for death. More trial data are required for the e?ect of aspirin on risk of breast and other cancers of women. Follow-up beyond 20 years is necessary to identify any late rebound in cancer deaths. The estimate of e?ect of aspirin on death due to cancer in the ?rst 5 years of the trials does not exclude a clinically important short-term bene?t for cancers that would probably have already been present at

randomisation, and so pooled analysis of trials with shorter follow-up is also required. To address each of these issues, the Non-Vascular Outcomes on Aspirin Collaboration is collating all available data from trials of aspirin (Rothwell, personal communication) and will report further results in 2011. Our study does have several potential limitations. First, we included only trials of daily aspirin. Alternate-day aspirin was used in other trials in prevention of vascular events because aspirin irreversibly inhibits COX-1 in platelets, but this e?ect would not be irreversible in other tissues, and observational studies have highlighted the importance of daily aspirin in associations with reduced incidence of cancer.5–8,12 10-year follow-up of the Women’s Health Study, a randomised trial of aspirin 100 mg on alternate days versus control, did show a possible reduction in incidence of lung cancer, but there was no reduction in other cancers or in overall cancer incidence.16 Second, although there was no evidence of any sex-related di?erence in the e?ect of aspirin on deaths due to cancer during the trials (data not shown), or in previous observational studies, we had too few women in the trials with long-term follow-up to allow us to determine the e?ects of aspirin on breast or gynaecological cancers. Third, analysis of e?ect of aspirin on adenocarcinoma overall was data-dependent, although analysis of histological subtype of lung and oesophageal cancers was prespeci?ed. Fourth, we were unable to determine the e?ect of long-term (eg, 20–30 years) continued aspirin use on cancer death or all-cause mortality because of the ?nite duration of the trials. The transient increase in risk of vascular deaths in the aspirin groups after the trials, consistent with studies of aspirin withdrawal,34,35 also diluted the e?ect that we did observe on long-term mortality. Finally, the bene?ts of aspirin may be less in populations with a high dietary intake of salicylates. Our results have implications for clinical practice. Since other antiplatelet drugs do not reduce risk of cancer death in randomised trials (Rothwell, unpublished data), patients with an indication for long-term antiplatelet treatment are likely, on average, to bene?t most from aspirin. Although the reduction in risk of ischaemic vascular events on aspirin in healthy individuals is partly o?set by a small increase in risk of non-fatal bleeding complications,36–38 the balance of risk and bene?t will now be altered by the reduction in cancer deaths after 5 years’ treatment. Our analyses show that taking aspirin daily for 5–10 years would reduce all-cause mortality (including any fatal bleeds) during that time by about 10% (relative risk reduction). Subsequently, there would be further delayed reductions in risk of cancer death, but no continuing excess risk of bleeding. In terms of cost-e?ectiveness,39 such bene?t would exceed that of established initiatives such as screening for breast or prostate cancer, potentially justifying added costs to reduce bleeding complications, such as co-prescription of a proton-pump inhibitor,40,41
9

www.thelancet.com Published online December 7, 2010 DOI:10.1016/S0140-6736(10)62110-1

Articles

treatment to eradicate Helicobacter pylori infection,42 and further development of potentially more e?ective derivatives of aspirin.43 Moreover, since the e?ect of aspirin on risk of cancer death increased with scheduled duration of trial treatment, the roughly 30% reduction in 20-year risk of cancer deaths observed in patients with scheduled trial treatment of 7·5–10 years may well underestimate the bene?t that would result from longerterm treatment (eg, from age 50–75 years). Indeed, a late rebound in cancer deaths in the aspirin group at 10–20 years’ follow-up is clearly present for some cancers (?gures 4 and 6, webappendix p 5). Finally, our results have implications for understanding of carcinogenesis, particularly for adenocarcinoma, and they demonstrate the potential for drug intervention in the prevention of cancer. Although the e?ect of aspirin may be mediated in part by inhibition of COX-2, more research is required, other pro-apoptotic e?ects early in the development of tumours perhaps also being important.43,44
Contributors PMR conceived and coordinated the project, obtained long-term follow-up of the UK-TIA trial, collated all data, planned and performed all analyses, and wrote the report. FGRF was principal investigator on the AAA trial. JFFB was principal investigator on the POPADAD trial. HO was principal investigator on the JPAD trial. CPW was principal investigator on the UK-TIA aspirin trial. TWM was principal investigator on the TPT and obtained long-term follow-up data. All authors commented on drafts of the report. Con?icts of interest This study was completely independent of any pharmaceutical company or other commercial interest. However, PMR has received honoraria for talks, advisory boards, and clinical trial committees from several pharmaceutical companies with an interest in antiplatelet agents, including AstraZeneca, Bayer, Boehringer Ingelheim, Sano?-Aventis/Bristol-Myers Squibb, and Servier. FGRF has had research support, honoraria, and travel expenses from Bayer and Sano?-Aventis/Bristol-Myers Squibb. JFFB has received payment for board membership from Roche Pharmaceuticals and Sano?Aventis. HO has received speakers’ fees from Astellas, AstraZeneca, Banyu, Bayer, Boehringer Ingelheim, Chugai, Daiichi Sankyo, Eisai, Guidant Japan, Japan Lifeline, Kowa, Kyowa Hakko Kirin, Novartis, Otsuka, P?zer, Sano?-Aventis, Schering-Plough, and Takeda. CPW has received speakers’ fees from Bayer for a talk about aspirin. TWM has received an honorarium and travel expenses from Bayer. Collaborators Emily Chew (ETDRS; National Eye Institute, Bethesda, MD, USA); Takeshi Morimoto (JPAD trial; Kyoto University Graduate School of Medicine, Kyoto, Japan); Richard Peto (BDAT; University of Oxford, Oxford, UK). Acknowledgments The study received no speci?c funding. Funding of the original trials was as reported previously.17–19,22–26 The cost of coordination of the project and collation and analysis of data was met by unrestricted research funds from the Stroke Prevention Research Unit, Oxford. PMR is in receipt of an NIHR Senior Investigator Award. We thank Jill Boreham for help with access to data from the BDAT; Christine Knottenbelt and Marilyn Goulding for their help in accessing long-term follow-up data from TPT and Michelle Wilson for help in coding these data; Izzy Butcher for help with data from AAA; and Robert Lee for help with data from POPADAD. We thank Ziyah Mehta for help with analysis and production of graphs. References 1 Jemal A, Center MM, DeSantis C, Ward EM. Global patterns of cancer incidence and mortality rates and trends. Cancer Epidemiol Biomarkers Prev 2010; 19: 1893–907.

2

3 4

5

6 7

8 9

10

11

12

13

14

15 16

17

18

19 20

21

22

23

Ferlay J, Autier P, Boniol M, Heanue M, Colombet M, Boyle P. Estimates of the cancer incidence and mortality in Europe in 2006. Ann Oncol 2007; 18: 581–92. The Lancet. Preventable cancer in the USA. Lancet 2010; 375: 1665. Dannenberg AJ, DuBois RD, eds. Progress in experimental tumour research vol. 37: COX-2—a new target for cancer prevention and treatment. Basel, Switzerland: Karger, 2003. Thun MJ, Henley SJ, Patrono C. Nonsteroidal anti-in?ammatory drugs as anticancer agents: mechanistic, pharmacologic, and clinical issues. J Natl Cancer Inst 2002; 94: 252–66. Elwood PC, Gallagher AM, Duthie GG, Mur LAJ, Morgan G. Aspirin, salicylates, and cancer. Lancet 2009; 373: 1301–09. Cuzick J, Otto F, Baron JA, et al. Aspirin and non-steroidal anti-in?ammatory drugs for cancer prevention: an international consensus statement. Lancet Oncol 2009; 10: 501–07. Bosetti C, Gallus S, La Vecchia C. Aspirin and cancer risk: a summary review to 2007. Recent Results Cancer Res 2009; 181: 231–51. Farquhar C, Marjoribanks J, Lethaby A, Suckling JA, Lamberts Q. Long term hormone therapy for perimenopausal and postmenopausal women. Cochrane Database Syst Rev 2005; 3: CD004143. Lawlor DA, Smith GD, Bruckdorfer KR, Kundu D, Ebrahim S. Those confounded vitamins: what can we learn from the di?erences between observational versus randomised trial evidence? Lancet 2004; 363: 1724–27. Serebruany V, Floyd J, Chew D. Excess of solid cancers after prasugrel: the Food and Drug Administration outlook. Am J Ther 2010; published online July 10. DOI:10.1097/MJT.0b013e3181e9b675. Flossmann E, Rothwell PM, on behalf of the British Doctors Aspirin Trial and the UK-TIA Aspirin Trial. E?ect of aspirin on long-term risk of colorectal cancer: consistent evidence from randomised and observational studies. Lancet 2007; 369: 1603–13. Rothwell PM, Wilson M, Elwin C-E, et al. Long-term e?ect of aspirin on colorectal cancer incidence and mortality: 20-year follow-up of ?ve randomised trials. Lancet 2010; 376: 1741–50. Cole BF, Logan RF, Halabi S, et al. Aspirin for chemoprevention of colorectal adenomas: meta-analysis of the randomised trials. J Natl Cancer Inst 2009; 101: 256–66. Chan AT, Ogino S, Fuchs CS. Aspirin and risk of colorectal cancer in relation to expression of COX-2. N Engl J Med 2007; 356: 2131–42. Cook NR, Lee IM, Gaziano JM, et al. Low-dose aspirin in the primary prevention of cancer: the Women’s Health Study: a randomized controlled trial. JAMA 2005; 294: 47–55. The Medical Research Council’s General Practice Research Framework. Thrombosis prevention trial: randomised trial of low-intensity oral anticoagulation with warfarin and low-dose aspirin in the primary prevention of ischaemic heart disease in men at increased risk. Lancet 1998; 351: 233–41. Farrell B, Godwin J, Richards S, Warlow C. The United Kingdom transient ischaemic attack (UK-TIA) aspirin trial: ?nal results. J Neurol Neurosurg Psychiatry 1991; 54: 1044–54. Peto R, Gray R, Collins R, et al. Randomised trial of prophylactic daily aspirin in British male doctors. BMJ 1988; 296: 313–16. ETDRS Investigators. Aspirin e?ects on mortality and morbidity in patients with diabetes mellitus. Early Treatment Diabetic Retinopathy Study report 14. JAMA 1992; 268: 1292–300. Ogawa H, Nakayama M, Morimoto T, et al. Japanese Primary Prevention of Atherosclerosis With Aspirin for Diabetes (JPAD) Trial Investigators. Low-dose aspirin for primary prevention of atherosclerotic events in patients with type 2 diabetes: a randomized controlled trial. JAMA 2008; 300: 2134–41. Juul-M?ller S, Edvardsson N, Jahnmatz B, Rosén A, S?rensen S, ?mblus R, for the Swedish Angina Pectoris Aspirin Trial (SAPAT) Group. Double-blind trial of aspirin in primary prevention of myocardial infarction in patients with stable chronic angina pectoris. Lancet 1992; 340: 1421–25. Belch J, MacCuish A, Campbell I, et al, Prevention of Progression of Arterial Disease and Diabetes Study Group; Diabetes Registry Group; Royal College of Physicians Edinburgh. The prevention of progression of arterial disease and diabetes (POPADAD) trial: factorial randomised placebo controlled trial of aspirin and antioxidants in patients with diabetes and asymptomatic peripheral arterial disease. BMJ 2008; 337: a1840.

10

www.thelancet.com Published online December 7, 2010 DOI:10.1016/S0140-6736(10)62110-1

Articles

24

25

26 27

28

29

30

31

32

33

34

Fowkes FG, Price JF, Stewart MC, et al. Aspirin for Asymptomatic Atherosclerosis Trialists. Aspirin for prevention of cardiovascular events in a general population screened for a low ankle brachial index: a randomized controlled trial. JAMA 2010; 303: 841–48. Fasey N, Brennan PJ, Meade TW, MRC General Practice Research Framework. Medical Research Council. Thrombosis prevention trial: follow-up study of practical implications. Br J Gen Pract 2002; 52: 208–09. Brewster DH, Stockton D, Harvey J, Mackay M. Reliability of cancer registration data in Scotland, 1997. Eur J Cancer 2002; 38: 414–17. Hawkins MM, Swerdlow AJ. Completeness of cancer and death follow-up obtained through the National health Service Central Register for England and Wales. Br J Cancer 1992; 66: 408–13. Pollock AM, Vickers N. Reliability of data of the Thames cancer registry on 673 cases of colorectal cancer: e?ect of the registration process. Qual Health Care 1995; 4: 184–89. Robinson MH, Rodrigues VC, Hardcastle JD, Chamberlain JO, Mangham CM, Moss SM. Faecal occult blood screening for colorectal cancer at Nottingham: details of the veri?cation process. J Med Screen 2000; 7: 97–98. Doria-Rose VP, Marcus PM. Death certi?cates provide an adequate source of cause of death information when evaluating lung cancer mortality: an example from the Mayo Lung Project. Lung Cancer 2009; 63: 295–300. Ederer F, Geisser M, Mongin S, Church T, Mandel J. Colorectal cancer deaths as determined by expert committee and from death certi?cate. The Minnesota Study. J Clin Epidemiol 1999; 52: 447–52. Sivak-Sears NR, Schwartzbaum JA, Miike R, Moghadassi M, Wrensch M. Case-control study of use of nonsteroidal antiin?ammatory drugs and glioblastoma multiforme. Am J Epidemiol 2004; 159: 1131–39. Amin R, Kamitani H, Sultana H, et al. Aspirin and indomethacin exhibit antiproliferative e?ects and induce apoptosis in T98G human glioblastoma cells. Neurol Res 2003; 25: 370–76. Sibon I, Orgogozo JM. Antiplatelet drug discontinuation is a risk factor for ischemic stroke. Neurology 2004; 62: 1187–89.

35

36

37

38 39

40

41

42

43

44

Maulaz AB, Bezerra DC, Michel P, Bogousslavsky J. E?ect of discontinuing aspirin therapy on the risk of brain ischemic stroke. Arch Neurol 2005; 62: 1217–20. Antithrombotic Trialists’ (ATT) Collaboration. Aspirin in the primary and secondary prevention of vascular disease: collaborative meta-analysis of individual participant data from randomised trials. Lancet 2009; 373: 1849–60. Greving JP, Buskens E, Ko?jberg H, Algra A. Cost-e?ectiveness of aspirin treatment in the primary prevention of cardiovascular disease events in subgroups based on age, gender, and varying cardiovascular risk. Circulation 2008; 117: 2875–83. Morgan G. Aspirin for the primary prevention of vascular events? Public Health 2009; 123: 787–88. Annemans L, Wittrup-Jensen K, Bueno H. A review of international pharmacoeconomic models assessing the use of aspirin in primary prevention. J Med Econ 2010; 13: 418–27. Lai KC, Lam SK, Chu KM, et al. Lansoprazole for prevention of recurrences of ulcer complications from long-term low-dose aspirin use. N Engl J Med 2002; 346: 2033–38. Chan FK, Ching JY, Hung LC, et al. Clopidogrel versus aspirin and esomeprazole to prevent recurrent ulcer bleeding. N Engl J Med 2005; 352: 238–44. Lanas A, Fuentes J, Benito R, Serrano P, Bajador E, Sainz R. Helicobacter pylori increases the risk of upper gastrointestinal bleeding in patients taking low-dose aspirin. Aliment Pharmacol Ther 2002; 16: 779–86. McIlhatton MA, Tyler J, Burkholder S, Ruscho? J, Rigas B, Kopelovich L, Fishel R. Nitric oxide-donating aspirin derivatives suppress microsatellite instability in mismatch repair-de?cient and hereditary nonpolyposis colorectal cancer cells. Cancer Res 2007; 67: 10966–75. Ruscho? J, Wallinger S, Dietmaier W, et al. Aspirin suppresses mutator phenotype associated with hereditary nonpolyposis colorectal cancer by genetic selection. Proc Natl Acad Sci USA 1998; 95: 11301–06.

www.thelancet.com Published online December 7, 2010 DOI:10.1016/S0140-6736(10)62110-1

11


相关文章:
更多相关标签: