Aspirin Use and Risk of Fatal Cancer · possible relationship between aspirin use and risk of fatal cancer in the cohort studied previously for colon cancer (76). SUBJECTS AND METHODS

(CANCER RESEARCH 53. 1.122-1327. March 15. 1993]

Aspirin Use and Risk of Fatal Cancer

Michael J. Thun, Mohan M. Namboodiri, Eugenia E. Calle, W. Dana Flanders, and Clark W. Heath, Jr.

American Cancer Society. Allanta, Georgia Ã•032V /M. J. T., M. M. N.. E. E. C., C. W. //./, ami Epidemiology Division, Emory Universin School of Public Health. Atlanta.Georgia .10.122 I W. I). F.I

ABSTRACT

Aspirin and other nonsleroidal antiinflammatory drugs inhibit pros-i.r.'l.imlm synthesis and tumor growth in many experimental systems, but

it is unclear which of these tumor models are relevant to humans. We havereported reduced risk or fatal colon cancer among persons who usedaspirin in a large prospective study. This analysis examines other fatalcancers in relation to aspirin among 635,031 adults in that study whoprovided information in 1982 on the frequency and duration of theiraspirin use and did not report cancer. Death rates were measured through1988. Death rates decreased with more frequent aspirin use for cancers ofthe esophagus, stomach, colon, and rectum but not generally for othercancers. For each digestive tract cancer, death rates were approximately40% lower among persons who used aspirin 16 times/month or more forat least 1 year compared to those who used no aspirin. The trend ofdecreasing risk with more frequent aspirin use was strongest among persons who had used aspirin for 10 years or more; it remained statisticallysignificant, except for esophagcal cancer, in multivariate analyses thatadjusted for other known risk factors. Biases such as early detection oraspirin avoidance among cases do not appear to explain the results. Ourdata suggest that regular, prolonged use of aspirin may reduce the risk offatal cancer of the esophagus, stomach, colon, and rectum. Future epidemiolÃ³gica) and basic research should examine all digestive tract cancers inconsidering the chemopreventive or therapeutic potential of nonsteroidalantiinflammatory drugs.

INTRODUCTION

NSAIDs1 inhibit prostuglanclin synthesis and tumor growth in a

wide variety of experimental systems (1). NSAIDs such as aspirin,indomethacin, piroxicam. and sulindac inhibit chemically inducedtumors of the colon (2-22), tongue (23), esophagus (24-25), pancreas(26), bladder (27-30), breast (31-35), liver (36, 37), and varioussarcomas (38-40). as well as numerous transplantable tumors (18,41-59). The experimental evidence is strongest for the rodent colon,

which most closely resembles human colon cancer (60). In rodents,NSAIDs consistently reduce both the number and size of tumorsinduced by chemical carcinogens (2-21 ) or radiation (61 ), even when

administered months after exposure to the carcinogen and when microscopic tumors are already present. Experimental evidence is limited for sites other than the colon. Some models show no inhibition(62-64) or even enhancement of tumor growth (65. 66) or epithelialcell proliferation (67-70), and it is unclear which findings are relevant

to humans.Human data on NSAIDs and cancer consist of four epidemiological

studies of aspirin and colorectal cancer (71-76) and two small clinicaltrials (77. 78). We have reported a 40-50% lower risk of fatal colon

cancer in persons regularly using aspirin compared to those who useno aspirin in a prospective study of over 6()(),()()()United States adults(76). A similar reduction in the incidence of colorectal cancer wasfound by two previous case-control studies (71. 73, 75) but not in a

prospective study of 14.000 elderly Californians (72, 74).A randomi/ed trial of ten patients with familial polyposis found

regression of rectal polyps associated with the NSAID sulindac (77),confirming earlier case reports (79, 80). In contrast, aspirin adminis-

Received S/27/92; accepted I/I 1/93.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby marked ath'i'i'lisenient in accordance with

IX U.S.C. Section 1734 solely to indicate this fact.1The abbreviation used is: NSAIDs, nonsteroidal antiinflammatory druys.

tered as adjuvant therapy did not improve survival among 66 patientswith invasive colorectal cancer (78).

Based upon experimental evidence that NSAIDs may affect theincidence or growth of a variety of tumors, we have examined thepossible relationship between aspirin use and risk of fatal cancer in thecohort studied previously for colon cancer (76).

SUBJECTS AND METHODS

The main analyses included 635.031 men and women drawn from CancerPrevention Study II. an ongoing prospective mortality study of 1.185,124persons enrolled by 77.000 volunteers of the American Cancer Society in thefull of 1982 (81). Participants in the study were usually friends, neighbors, oracquaintances of the volunteers. Enrollment took place in all 50 slates, theDistrict of Columbia, and Puerto Rico. Families were enrolled if at least onehousehold member was 45 years old or older and all enrolled members were30 years old or older. The average age of the participants at enrollment was 57years.

The participants' vital status was determined at 2-year intervals through

personal enquiry by the volunteers. Mortality was assessed through August 31.1988. By then. 79.820 participants (6.7%) had died, 1,083,600 (91.4%) werealive, and 21.704 (1.8%) were lost to follow-up. Death certificates were ob

tained for 94.1 % of the persons known to have died. A nosologist coded cancerdeaths according to the first three digits of the International Classification ofDiseases, 9th revision (ICD-9) (82).

At enrollment, participants completed a four page questionnaire on medication use. smoking history, diet, alcohol intake, physical activity, occupation,height, weight, current medical illnesses, family history of cancer, and othercharacteristics. For each of 15 categories of medications, the questionnaireasked two questions: "How many times in the last month have you used thefollowing [medications]'.'" and "How long (in years) have you used them?"Thefirst category listed was "Aspirin, Bufferin, and Anacin" and the second was"Tylenol (acetaminophen)." The other categories included vitamins, trunquil-

izers. and medications for specific medical conditions.We excluded from the analyses 82.382 persons with cancer (except non-

melanoma skin cancer) at the time of enrollment; 16 who died before September 1, 1982; 20,198 who were lost to follow-up; and 2.439 who used aspirin

for less than 1 year. Of the remaining 1.080.089 participants, 635,031 (58.8%)specified both the frequency and (if their use was more than occasional)duration of their aspirin use. 226.715 (21.0%) specified neither frequency norduration, and 218.343 (20.2%) specified frequency but not duration.

Rate Analyses. Age-adjusted death rates from all cancer sites were com

puted based upon underlying cause as coded from death certificates. Thenumber of person-years at risk was determined from September I. 1982,

through August 31. 1988. In adjusting for age. we used the age distribution ofthe overall cohort for direct standardization of men and women separately.Combined death rates were weighted by the total number of person-years in a

given age stratum multiplied by 42.4% (for men), or 57.6% (for women).These fractions correspond to the proportion of person-years contributed by

each sex. The rate analyses for men and women combined were thereforeadjusted for both age and sex.

Rates of death from various cancers were calculated for each level ofreported monthly aspirin use: no use; "occasionally"; 1-15 times; or 16 times/

month or more. We also examined death rates from each cancer among personswith incomplete or missing aspirin data. Similar calculations were made withrespect to acetaminophen. Relative risks were computed with the subjects whodid not use aspirin serving as the referent group, and 95% confidence intervalswere calculated using approximate variance formulas (83, 84).

Multivariate Analyses. We used Cox proportional hazards modelling (85)to adjust for other potential risk factors for digestive tract cancers in which therisk decreased with more frequent aspirin use in men and women. Additionalexclusions from the multivariate analyses were 84,494 persons with inadequatedata on height, weight, tobacco and alcohol use. or diet. Multivariate analyses

1322

Research. on January 21, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

ASPIRIN AND FATAL CANCER

matched on age at enrollment, race, and sex and included six categories ofspecified or incomplete data on aspirin use (described above), body mass index(continuous), family history of specific cancer subsite (yes/no), number ofcigarettes smoked per day currently and formerly (both continuous), ever pipeor cigar (yes/no), drinks of alcohol per day (continuous), sum of the frequency(in days per week) of consuming selected fruits/vegetables/grains (continuous),and index of total fat consumption (continuous). Calculation of the nutritionalmeasures is described elsewhere (86). Parallel analyses were performed foraspirin and acetaminophen. We tested the proportional hazards assumption byincluding time varying covariates which represented the interactions betweeneach level of aspirin use and time. We assessed effect modification by including all two-way interaction (product) terms between level of aspirin use and

other covariates and using likelihood ratio tests of their combined statisticalsignificance with a = 0.05. Multivariate analyses used standard statistical

programs (PHREG; SAS. Cary, NC).

RESULTS

Rate Analyses. Table I presents the number of deaths from allcancers combined, person-years at risk, and death rates at each level

of aspirin use. Table 2 shows the relative risks and 95% confidenceintervals for all fatal cancers combined and for cancers of major organsystems. The risk of all fatal cancers combined was somewhat loweramong persons who used aspirin than in nonusers; this difference wasof borderline statistical significance except in women in the highestcategory of use. No dose-response trend was seen for all cancers

combined.Among organ systems, only digestive system cancers (ICD-9. 150-

159) showed a consistent pattern of decreasing relative risks withmore frequent aspirin use in both sexes. Other organ systems showedeither no statistically significant differences between aspirin users andnonusers or a difference in one sex only with no dose-response trend

(e.g., increased risk of fatal brain cancer in men). Subsequent analysesare restricted to the digestive system.

When digestive system cancers were examined separately (Table3), death rates from cancers of the esophagus, stomach, colon, andrectum were generally inversely associated with aspirin use. No associations were seen between aspirin use and other gastrointestinalcancers including pancreas, liver and gallbladder, and peritoneum(data on these cancers not shown). There were few deaths fromcancers of the esophagus, stomach, and rectum in women. Cancer ofthe small intestine could not be analyzed.

Multivariate Analyses. Table 4 shows the relative risk estimatesfor the digestive tract cancers after controlling for other potential riskfactors. The trends of decreasing risk seen for aspirin alone (in the rateanalyses) were unchanged or somewhat stronger when controlling forobesity, tobacco and alcohol usage, and dietary consumption of fat andvegetables/fruit/grains. Persons who used aspirin 16 times/month ormore for at least 1 year experienced approximately 40% lower riskthan those who used no aspirin. Tests of trend were statisticallysignificant for cancers of the stomach, colon, and rectum and border-

TableI Mortalityfromall cancerscombinedbyfrequencyofaspirinuse

Aspirin use(limes/mo)Men

0Occasionally

1-1516+SubtotalWomen

0Occasionally

1-1516+SubtotalDeaths

fromcancer2,2991,358

I.IXM7415.4021.4541.040

7185343,746Person-years

atrisk644,145478.969

381.890183.8671,688,87168

1,677641.060

485,067227,3212.035.125Death

rates"/

100,000366.7343.5

337.0341.3225.5201.5

196.9217.7

" Age-adjusted death rates/MXMXX)person-years at risk are directly standardised to age

distribution of the Cancer Prevention Study II total population.

line for esophageal cancer. The association between aspirin and rectum cancer was weaker in women than in men.

In Table 5 we further stratified participants according to the duration of their aspirin use. The trends of decreasing risk with morefrequent aspirin use were stronger and more linear among those whohad used aspirin for 10 years or more than in those with shorterduration, even though the number of deaths was smaller than in theanalysis that did not stratify on duration. After controlling for thefrequency of aspirin use, age, race, sex, and other known risk factors,a duration of 10 years or more (compared to 1-9 years) was signifi

cantly associated with lower risk only for stomach cancer.We also assessed whether cases could have differentially discon

tinued aspirin use for any reason, biasing our results by artifactuallyincreasing risk in the unexposed. We repeated the multivariate analyses in Table 4, restricting the referent group to 5047 persons whostated that they had used no aspirin for 5 years or more. The trend testsremained statistically significant for men and women combined forcancers of the esophagus, stomach, and colon, suggesting that forthese cancers aspirin avoidance by cases did not explain the relationships. Only for rectum cancer was the trend weakened by restrictingthe referent group to long-term nonusers. We also tested whether the

relative risk of fatal digestive tract cancers was higher among personswho had recently discontinued aspirin use (<5 years) than amonglong-term nonusers (> 5 years). No consistent pattern was seen for thefour sites. The risk was slightly higher among the short-term nonusers

for rectum cancer (relative risk, 1.28) but this was not statisticallysignificant (P = 0.8) and was not seen for other digestive tract sites.

Therefore, changes in medication status due to undiagnosed cancer donot explain our findings.

None of the interaction terms between aspirin and time varyingcovariates (month of follow-up) or two-way product terms between

aspirin and other risk factors were statistically significant for anydigestive tract cancer.

Effect of Missing Data. We tested the possibility that the exclusion of persons who did not specify both frequency and duration ofaspirin use might bias our findings. In the rate analyses, cancer mortality among persons who gave no information on aspirin use wassimilar to or slightly below that of nonusers; while among personswho specified frequency but not duration of use, risk resembled thatin users. Exceptions were cancer of the esophagus and lung in men.Men who did not specify the frequency of their aspirin use had arelative risk of esophagus cancer of 1.10 (95% confidence interval,0.82-1.46) and of respiratory cancers of 1.20 (95% confidence interval, 1.10-1.30). In multivariate analyses, the relative risk estimates for

digestive tract cancers in persons who specified their aspirin usagebecame somewhat stronger ( i.e., further below 1.0) when persons withmissing or incomplete data were excluded from the models. Wepresent in Tables 4 and 5 only the relative risk estimates for aspirinfrom models which include the missing data categories, since in ourdata, these underestimate the association with aspirin.

Acetaminophen. In contrast to aspirin, acetaminophen was notconsistently associated with decreased risk of any digestive tract cancer. For colon cancer a statistically insignificant trend of decreasingrisk with more frequent acetaminophen use was seen in univariateanalyses in men only; this disappeared in multivariate analyses andwhen both sexes were combined. Rectum cancer mortality was significantly higher in men and women who used acetaminophen 16times/month or more compared to those who used neither aspirin noracetaminophen (relative risk, 3.08; 95% confidence interval, 1.11-8.

54). The trend of increasing risk with more frequent acetaminophenuse was not statistically significant (P = 0.31). and the relative risk

estimate in the most frequent usage categories was based on only fourdeaths from rectum cancer.

1323




Table 2 Relative risk (and 957? confluence mien-aÃ¯s)of fatal cancer by frequency of aspirin use"

Aspirinfrequency/moSiteAll

cancerscombinedBuccal

cavity andpharynxDigestive

systemRespiratory

systemBreast

(female)Genital

system(female)Genital

system(male)Urinary

systemBrain

and other nervoussystemLymphatic

andhematopoieticOther

and unspecifiedICD-9140-149150-159160-165174-175179-184185-187188-189191-192200-208170-173,190193-194,199Occasionally0.94(0.87-1.01)0.70(0.31-1.59)0.89(0.77-1.02)0.94(0.83-1.07)0.97(0.75-1.25)0.77(0.52-1.14)1.35(1.03-1.77)0.95(0.75-1.19)1.02(0.82-1.28)Men1-150.92(0.85-1.00)0.99(0.54-1.82)0.79(0.66-0.94)1.00(0.88-1.13)1.03(0.79-1.33)0.77(0.50-1.19)1.42(1.08-1.87)0.76(0.54-1.06)1.09(0.87-1.36)16+0.93(0.85-1.02)1.15(0.65-2.05)0.71(0.56-0.90)1.11(0.98-1.25)0.82(0.56-1.19)0.70(0.39-1.23)1.40(1.03-1.92)0.89(0.66-1.20)1.10(0.86-1.41)Occasionally0.91(0.83-0.99)0.68(0.20-2.29)0.90(0.76-1.07)0.83(0.68-1.03)0.93(0.73-1.19)0.93(0.73-1.20)0.97(0.60-1.58)1.33(0.99-1.78)0.91(0.68-1.23)0.92(0.67-1.26)Women1-150.88(0.80-0.98)0.55(0.09-3.44)0.84(0.68-1.03)0.73(0.56-0.97)0.98(0.76-1.26)0.84(0.61-1.15)1.16(0.77-1.77)1.12(0.76-1.66)1.06(0.80-1.39)1.04(0.77-1.39)16+0.98(0.88-1.08)1.28(0.66-2.49)0.82(0.65-1.04)1.07(0.88-1.30)0.88(0.62-1.24)1.21(0.97-1.51)1.04(0.62-1.77)0.86(0.44-1.68)0.93(0.65-1.33)1.08(0.79-1.48)

" Groups with no aspirin use are the referent categoriesNumbers in parentheses, 95*7rconfidence intervals.

Relative risks are based on age-adjusted death rates, directly standardized to Cancer Prevention Study 11total population.

DISCUSSION

Our principal finding was that aspirin use was associated withreduced risk of fatal cancers of the esophagus, stomach, colon, andrectum. The decreased risk of digestive tract cancers was similar inmen and women, it was stronger in those with more frequent andprolonged use of aspirin, and it could not be attributed to other riskfactors for these cancers. In contrast, we saw no consistent dose-

response relationships in men and women for cancers outside thedigestive tract. The inverse association of aspirin with fatal digestivetract cancers extends the hypothesis regarding aspirin and the prevention of colon cancer. However, until the findings can be replicated,they should be interpreted cautiously, given the observational natureof our study and the important differences in the biology and epidemiology of these cancers.

Some limitations of our study are its dependence upon a single,brief, self-administered questionnaire, the absence of data on the

actual dose and number of aspirin or other NSAIDs, the lack ofinformation on the indications for aspirin use, the reliance on mortality from cancer rather than incidence rates to identify disease, and therelatively short follow-up. Our use of deaths rather than incident cases

of cancer to define disease status prevents us from separating an effecton survival from incidence.

Several potential biases should be considered that might introduce

a spurious association between aspirin use and reduced risk of digestive tract cancer. First, persons who take aspirin might develop bleeding or receive closer medical scrutiny, leading to earlier diagnosis ofpolyps or tumors at a stage which improves survival. Under thishypothesis, the benefits of aspirin might arise indirectly from its sideeffects rather than through chemoprevention. Such a detection biasmight affect colon and rectum cancer in our study, but it cannotexplain the reduced risk of esophagus or stomach cancer. Five-year

survival is under 20% for these upper gastrointestinal cancers regardless of the stage of diagnosis. Early detection also cannot account forthe experimental evidence for chemoprevention in rodent studies orthe regression of rectal polyps in humans with familial polyposistreated with sulindac.

A second potential bias might involve aspirin avoidance by patientswith undiagnosed gastrointestinal cancer or premalignant lesions dueto side effects such as pain or bleeding. By this hypothesis, undiagnosed patients would change their medication status, stop taking as-

Table 3 Digestive tract cancer mortality by frequency of aspirin use

CancersiteEsophagusStomachColonRectumAspirinuse

(times/mo)0Occasionally1-1516+0Occasionally1-1516+0Occasionally1-1516+0Occasionally1-1516+No.of

deaths663528161055837202411349354461997MenRate"10.437.568.857.9116.8714.3012.339.1138.8734.8730.3224.647.044.243.013.08Relativerisk"1.00.720.850.761.00.850.730.541.00.900.780.631.00.600.430.44No.of

deaths1565551218818412673452014711WomenRate"2.311.291.771.957.904.432.673.3228.3525.7420.8817.563.073.731.874.25TotalRelativerisk*1.00.560.770.841.00.560.340.421.00.910.740.621.01.210.611.38Relativerisk"1.00.690.830.781.00.740.580.491.00.900.760.631.00.830.490.79(95%(0.39-(0.51-(0.42-(0.51-(0.32-(0.22-CI).19).36).44).07).03).12)(0.76-1.07)(0.60-0.96)(0.44-0.89)(0.51-1.35)(0.16-1.50)(0.41-.53)

" Age-adjusted death rate/1 (M),(MK)person-years at risk, directly standardized to Cancer Prevention Study II total population.''Groups with no aspirin use are the referent categories; relative risk in men and women combined is adjusted for both age and sex. None of the separate relative risk estimates

for men and women is statistically significant at P < 0.05. Numbers in parentheses, 95c/c confidence intervals.

1324




Table 4 Multivariate relative risk estimates for fatal digestive tract cancers

CancersiteEsophagusStomachColonRectumAspirinuse

(times/mo)0Occasionally1-1516+0Occasionally1-1516+0Occasionally1-1516+0Occasionally1-1516+No.of

deaths62332613895431162161188947421877MenAdjusted"relativerisks1.00.770.830.621.00.930.710.52*1.00.840.850.62*1.00.630.34*0.48WomenNo.of

deaths12542411988ISO1055631171477Adjusted"relativerisks1.00.600.710.441.00.620.36*0.541.00.890.68*0.52*1.00.980.711.07Adjusted"relativerisks1.00.740.810.591.00.830.600.531.00.860.780.581.00.740.460.66Total(95%CI)(0.50-1.10)(0.53-1.24)(0.34-1.03)(0.62-1.10)(0.42-0.85)(0.34-0.81)(0.73-1.02)(0.64-0.95)(0.45-0.74)(0.48-1.14)(0.25-0.82)(0.37-1.19)Ptrend0.0540.00010.000

10.016

a Multivariate analyses match on age, race, and sex. Models include (as continuous variables) body mass index, number of cigarette smoked currently or formerly, drinks of alcohol

per day, an index of total dietary fai, and fruit/vegetable/grain consumption. Dichotomous variables are family history of the specific cancer, any tobacco use, and ever pipe or cigar.b 95% confidence interval (CI) excludes 1.0 where CI not shown.

Table 5 Multivariate relative risk estimates of fatal digestive tract cancersin men and women combined by duration and frequency of aspirin use.

compared to all nonusers

Duration of aspirinuseCancersiteEsophagusStomachColonRectumAspirinfrequency(times/mo)0Occasionally1-1516+0Occasionally1-1516+0Occasionally1-1516+0Occasionally1-1516+1-9

yrNo.

ofdeaths7451210130141415366184259592211Relativerisk"1.0*1.071.200.621.0*1.590.750.481.0*0.660.800.711.0*0.500.250.8210+

yrNo.

ofdeaths743318513059259366205103195930123Relativerisk"1.0*0.730.670.541.0*0.770.550.521.0*0.880.770.361.0*0.770.500.39

" Relative risks are matched on age, race, and sex and adjusted for smoking (cigarettes

per day), alcohol (drinks per day), family history, and dietary consumption of vegetables/fruits/grains and total fat.

* Referen! comprised of all nonusers of aspirin, regardless of duration.

pirin, and be classified by us as nonusers, artifactually increasing therisk among the unexposed. To minimize this possibility, we repeatedthe multivariate analyses restricting the referent group to persons whostated they used no aspirin for 5 years or more, reasoning that thiswould exclude cancer patients who had changed their medicationhabits. The statistical power of our analyses diminished, but the pattern of lower risk among aspirin users persisted, reassuring us that thisbias was unlikely to explain our findings.

Yet another potential bias, which is of less concern in prospectivestudies than in retrospective ones, might arise from our exclusion ofmany cohort members who did not specify both the frequency andduration of their aspirin use. We found little evidence that persons forwhom data were incomplete or missing were at higher risk of digestive tract cancer than were nonusers of aspirin. The one exception inthe digestive tract was cancer of the esophagus in men. Men whofailed to report their aspirin use were more likely to have missing dataon tobacco use and may have been smokers. Their exclusion does not

bias our findings with regard to aspirin and digestive tract cancers,since smoking is not generally associated with digestive tract cancersother than the esophagus, and since we controlled for smoking in ouranalyses.

Previous epidemiological studies of aspirin and cancer have focused primarily on cancers of the large bowel (71-72, 76), and there

are few published data on other human cancer sites. Two studiesreported finding no statistically significant associations between aspirin use and cancers of the lung, breast, or bladder (72, 73) but didnot mention esophagus or stomach cancer. Record linkage studies inFinland and Sweden have found a lower incidence of cancers of thestomach, colon, and rectum among patients with rheumatoid arthritiscompared to the general population (87, 88). These patients have highand prolonged use of aspirin and other NSAIDs. Although rheumatoidarthritis patients do take other medications and can have severe underlying disease, the lower risk of digestive tract cancers providessome external support for our findings.

Unlike aspirin, acetaminophen was not associated with decreasedrisk of any digestive tract cancer in our study, suggesting that ourfindings are relatively specific to aspirin. The risk of rectum canceractually increased among frequent users of acetaminophen, althoughthis finding was based on few cases. If two or three persons with rectalcancer had changed their medication from aspirin to acetaminophen,this shift could create an artifactual association between rectum cancerand acetaminophen but cannot account for the aspirin findings, whichare based on larger numbers and are statistically more stable. Ourstudy included relatively few long-term users of acetaminophen and

thus cannot evaluate whether the association with rectal cancer iscausal or an artifact of changing medication status.

If aspirin does prevent fatal digestive tract cancer in humans, it isunclear which of several possible mechanisms may be causal. Experimentally, NSAIDs inhibit cell proliferation (89, 90), tumor growth(2-25), tumor promotion (3, 28-31), and metastasis (91) and enhance

host immune responses (48, 92, 93) in various experimental models.The results of these exploratory analyses provide a bridge between

the extensive experimental evidence that NSAIDs inhibit the growthand metastasis of a variety of animal tumor models and observationsin humans. They suggest that aspirin, and perhaps other NSAIDs, mayreduce the risk of fatal digestive cancer throughout the gastrointestinaltract. The findings should stimulate further epidemiological study ofNSAIDs and cancer, as well as basic research on the mechanisms by

1325




which NSAIDs may influence carcinogenesis, particularly within the 29digestive tract. Even a modest reduction in the over 81,000 deaths/year (95) from digestive tract cancers in the United States would havesubstantial public health importance. 30.

REFERENCES 31.

1. MarneÂ». L. J. Aspirin and the role of prostaglandins in cancer. Cancer Res., 52:5575-5589. 1992.

2. Pollard. M., and LÃ¼cken.P. H. Indomethacin treatmenl of rats with dimethylhydra- 32.zine-induced intestinal tumors. Cancer Treat. Rep.. 64: 1323-1327, 1980.

3. Kudo. T.. Narisawa, T.. and Abo. S. Antitumor activity of Â¡ndomethacin on methy- 33.lazoxymethanol-induced large bowel tumors in rats. Gann. 71: 260-264. 1980.

4. Pollard. M.. and LÃ¼cken.P. H. Effect of indomethacin on intestinal tumors inducedin rats by the acetate derivative of dimethylnitrosamine. Science (Washington DC), 34214: 558-559, 1981.

5. Pollard M.. and LÃ¼cken,P. H. Treatment of chemically-induced intestinal cancerswith Â¡ndomeihacin. Proc. Soc. Exp. Biol. Med.. 167: 161-164, 1981.

6. Narisawa, T., Sato, M., Tani. M.. Kudo, T., Takahashi, T.. and Goto, A. Inhibition of 35development of methylnitrosourea-induced rat colon tumors by indomethacin treatment. Cancer Res.. 41: 1954-1957, 1981.

7. Pollard. M.. Luckert, P. H.. and Schmidt, M. A. The suppressive effect of piroxicamon autochthonous intestinal tumors in the rat. Cancer Lett., 21: 57-61, 1983. 36.

8. Pollard, M.. and Luckert, P. H. Prolonged antitumor effect of indomethacin onautochthonous intestinal tumors in rats. J. Nati. Cancer Inst., 70: 1103-1105, 1983.

9. Narisawa, T., Satoh, M., Sano, M., and Takahashi. T. Inhibition of initiation and 37.promotion by /V-methylnitrosourea-induced colon carcinogenesis in rats by non-steroid anti-inflammatory agent indomethacin. Carcinogenesis (Lond.), 4: 1225-

1227. 1983. 38.10. Pollard, M.. and LÃ¼cken.P. H. Effect of piroxicam on primary intestinal tumors

induced in rats by /V-methylnitrosourea. Cancer Lett., 25: 117-121, 1984. 39.11. Narisawa, T., Hermanek. P.. Habs. M., and Schmahl. D. Reduction of carcinogenicity

of /V-nitrosomethylurea by Â¡ndomethacin and failure of resuming effect of prostag-landin E2 (PGE2) against indomethacin. J. Cancer Res. Clin. Oncol., 108: 239-242. 40.

1984.12. Metzger, U.. Meier, J., Uhlschmid. G., and Weihe. H. Influence of various prostag- 41.

landin synthesis inhibitors on DMH-induced rat colon cancer. Dis. Colon Rectum, 27:366-369, 1984.

13. Pollard, M. Antitumor effect of indomethacin in rats with autochthonous intestinal 42.tumors. In: B. Levin and R. H. Riddell (eds.). Frontiers in Gastrointestinal Cancer, pp.91-103. New York: Elsevier Publishing Co.. 1984.

14. Nigro. D. N.. Bull. A. W., and Boyd, M. E. Inhibition of intestinal carcinogenesis in 43.rats: effect of difluoromethylomithine with piroxicam or fish oil. J. Nati. Cancer Inst.,77: 1309-1313, 1986. 44.

15. Reddy. B. S.. Maruyama, H., and Kelloff, G. Dose-related inhibition of colon carcinogenesis by dietary piroxicam, a nonsteroidal anti-inflammatory drug, duringdifferent stages of rat colon tumor development. Cancer Res.. 47: 5340-5346. 1987. 45.

16. Moorghen. M., Ince. P.. Finney, K., Sunter, J. P.. Appleton. D. R., and Watson, A. J.A protective effect of sulindac against chemically-induced primary colonie tumors inmice. J. Pathol., 156: 341-347, 1988. 46.

17. Pollard. M.. and LÃ¼cken.P. H. Prevention and treatment of primary intestinal tumorsin rats by piroxicam. Cancer Res., 49: 6471-6473, 1989. 47.

18. Rubio. C. A.. Wallin. B., Ware, J.. Sveander. M.. and Duvander, A. Effect of indomethacin in autotransplanted colonie tumors. Dis. Colon Rectum, 32: 488â€”491, 48.1989.

19. Reddy. B. S.. Nayini, J.. Tokumo. K., Rigotty. J., Zang. E., and Kelloff. G. Chemo- 49.

prevention of colon carcinogenesis by concurrent administration of piroxicam. anonsteroidal anti-inflammatory drug with D.i.-a-difluoromethylornithine, an ornithinedecarboxylase inhibitor, in diet. Cancer Res.. 50: 2562-2568, 1990. 50.

20. Rao. C. V., Tokumo, K., Rigotty, J., Zang, E., Kelloff, G.. and Reddy. B. S. Chemo-prevention of colon carcinogenesis by dietary administration of piroxicam. a-diflu-oromethylornilhine, l6a-fluoro-5-androsten-17-one. and ellagic acid individually and 51.in combination. Cancer Res., 51: 4528^1534, 1991.

21. Skinner, S. A., Penney. A. G., and O'Brian. P. E. Sulindac inhibits the rate of growth

and appearance of colon tumors in the rat. Arch. Surg.. 126: 1094-10%. 1991. 52.22. Craven, P. A., and Derubenis, F. R. Effects of aspirin on 1,-dimethylhydrazine-

induced colonie carcinogenesis. Carcinogenesis (Lond.), 13: 541-546. 1992.23. Tanaka. T.. Nishikawa, A., Mori. Y., Morishita, M.. and Mori, H. Inhibitory effects of 53.

non-steroidal anti-inflammatory drugs, piroxicam and indomethacin on 4-nitroquin-oline I-oxide-induced tongue carcinogenesis in male ACI/N rats. Cancer Lett., 48:177-182, 1989. 54.

24. Rubio. C. A. Antitumoral activity of indomethacin on experimental esophageal tumors. J. Nail. Cancer Inst., 72: 705-707, 1984.

25. Rubio. C. A. Further studies on the therapeutic effect of indomethacin on esophageal 55.tumors. Cancer (Phila.l. 58: 1029-1031. 1986.

26. Takahashi. M.. Furukawa, F.. Toyoda, K.. Sato. H.. Hasegawa. R.. Imaida. K., and 56.Hayashi, Y. Effects of various prostaglandin synthesis inhibitors on pancreatic carcinogenesis in hamsters after initiation with ALnitrosobis(2-oxopropyl)amine. Carcinogenesis (Lond.), //: 393-395. 1990. 57.

27. Murasaki. G.. Zenser. T. V., Davis, B. B.. and Cohen. S. M. Inhibition by aspirin ofN-\4-t5 nitro-2-furyl)-2-thiazolyl]formamide-induced bladder carcinogenesis and enhancement of forestomach carcinogenesis. Carcinogenesis (Lond.), 5: 53-55, 1984. 58.

28. Hasegawa. R., St. John. M.. Murasaki. G., Fukushima. S.. and Cohen. S. Effect ofaspirin on W-|4-(5-nitro-2-furyl)-2-thiazolyl]formamide-induced epithelial proliferation in the urinary bladder and forestomach of the rat. Cancer Lett., 21: 269-275. 59.

1984.

1326

Sakata. T., Hasegawa. R.. Johansson. S. L., Zenser, T. V.. and Cohen. S. M. Inhibitionby aspirin of A'-[4-<5-Nitro-2-furyl)-2-thiazolyl|formamide initiation and sodium sac

charin promotion of urinary bladder carcinogenesis in male F344 rats. Cancer Res.,46: 3903-3906, 1986.

Cohen. S. M., Hasegawa. R., Sakata, T., and Johansson, S. L. Effect of aspirin onurinary bladder carcinogenesis initiated with N-|4-(5-nitro-2-furyl)-2-thiazolyl]for-mamide in rats. Cancer Res., 49: 372-377. 1989.

Carter, C. A., Milholland, R. J., Shea, W., and Ip, M. M. Effect of the prostaglandinsynthetase inhibitor indomethacin on 7,12-dimethylbenz(a)anthracene-induced mammary tumorigenesis in rats fed different levels of fat. Cancer Res.. 43: 3559-3562.1983.McCormick, D. L., Madigan, M. J.. and Moon, R. C. Modulation of rat mammarycarcinogenesis by indomethacin. Cancer Res., 45: 1803-1808, 1985.McCormick. D. L.. and Wilson, A. M. Combination chemoprevention of rat mammarycarcinogenesis by indomethacin and butylated hydroxytoluene. Cancer Res., 46:3907-3911. 1986.

Carter, C. A., Ip, M. M.. and Ip. C. A comparison of the effects of the prostaglandinsynthesis inhibitors indomethacin and carprofen on 7.12 dimethylbenz(a)anthracene-induced mammary tumorigenesis in rats fed different amounts of essential fatty acid.Carcinogenesis (Lond.), 10: 1369-1374, 1989.Abou-el-ela. S. H., Prasse, K. W., Farrell. R. L., Carroll, R. Wâ€žWade, A. E., andBunce. O. R. Effects of D,L-2-difluoromethylomithine and indomethacin on mammarytumor promotion in rats fed high n-3 and/or n-6 fat diets. Cancer Res., 49: 1434-1440.

1989.Servais, P., Taton. G., and Galand. P. Modulation by indomethacin of diethylnitro-samine-induced early changes in c-myr and c-ras expression and late incidence ofpreneoplastic lesions in rat liver. Carcinogenesis (Lond.), //: 1433-1436. 1990.Taton, G.. Servais, P., and Galand. P. Modulation by indomethacin or prostaglandin E2of the incidence of diethylnitrosamine-induced glutamyltranspeptidase positive foci inrat liver. Jpn. J. Cancer Res., 81: 747-753. 1990.

Lynch, N. R., Castes, M., Astoin, M.. and Salomon. J. C. Mechanism of inhibition oftumor growth by aspirin and indomethacin. Br. J. Cancer, 38: 503-512, 1978.

Sandstrom, R., Gelin, J., and Lundholm, K. The effect of indomethacin on food andwater intake, motor activity and survival in tumour-bearing rats. Eur. J. Cancer, 26:811-814. 1990.Wlodarski. K. H. [ndomethacin influences Moloney's sarcoma and associated peri-

osteal osteogenesis in the mouse. Acta Orthop. Scand., 61: 427-430. 1990.Hial. V., Horakova, Z.. Shaff, R. E., and Beaven. M. A. Alteration of tumor growth byaspirin and indomethacin: studies with two transptantable tumors in mouse. Eur. J.Pharmacol., 37: 367-376, 1976.Lala, P. K., Parhar. R. S.. Singh, P., and Lala. P. K. Cure of murine Ehrlich ascitestumors with chronic oral indomethacin therapy combined with intraperitoneal administration of LAK cells and IL-2. Cancer Lett., 5/: 27-35, 1990.Farrell, C. L., Megyesi, J., and Del Maestro, R. F. Effect of ibuprofen on tumor growthin the C6 spheroid implantation glioma model. J. Neurosurg., 68: 925-930, 1988.Tanaka, Y.. Tanaka. T., and Ishitsuka, H. Antitumor activity of indomethacin in micebearing advanced colon 26 carcinoma compared with those with early transplants.Cancer Res., 49: 5935-5939, 1989.Tashjian. A. H., Voelkel, E. F.. Goldhaber. P.. and Levine, L. Successful treatment ofhypercalcemia by indomethecin in mice bearing a prostaglandin-producing fibrosarcoma. Prostaglandins, 3: 515-524. 1973.Tashjian, A. H.. Voelkel, E. F.. Goldhaber, P., and Levine. L. Prostaglandins, calciummetabolism, and cancer. Prostaglandins, 33: 81-86. 1974.Plescia, O. J., Smith, A. H.. and Grinwich, K. Subversion of immune system by tumorcells and role of prostaglandins. Proc. Nati. Acad. Sci. USA, 72: 1848-1851. 1975.

Lynch. N. R., Castes, M.. Astoin, M.. and Salomon. J. C. Mechanism of inhibition oftumour growth by aspirin and indomethacin. Br. J. Cancer, 38: 503-512. 1978.

Shenoy, M. A., and Singh, B. B. Studies on indomethacin as a potentiator forhyperthermic and radiation responses in a mouse fibrosarcoma. Cancer Lett., 45:227-232, 1989.Mamytbekova, A., Rezabek, K., Kacerovska, H., Grimova, J., and Svobodova, J.Antimetastatic effect of fiurbiprofen and other platelet aggregation inhibitors. Neoplasma (Bratisl.), 33: 417^121, 1986.Mamytbekova. A., Rezabek. K., Grimova, J.. and Svobodova. J. Inhibitory action offlurbiprofen and mopidamole on tumor lung metastasis formation in mice. CancerTreat. Rep.. 71: 1089-1090, 1987.

Bennett, A., Houghton. J.. Leaper, D. J.. and Stamford, I. F. Cancer growth, responseto treatmenl and survival time in mice: beneficial effect of the prostaglandin synthesisinhibitor flurbiprofen. Prostaglandins, 17: 179-191, 1979.

Hillyard, L. A., and Abraham. S. Effect of dietary polyunsaturaled fatty acids ongrowth of mammary adenocarcinomas in mice and rats. Cancer Res., 39: 4430-4437.1979.Kollmorgen. G. M., King, M. M.. Kosanke, S. D.. and Cuong. D. Influence of dietaryfat and indomethacin on the growth of transplantable mammary tumors in rats. CancerRes., Â«.-4714-4719, 1983.

Fulton. A. M. Inhibition of experimental metastasis with indomethacin: role of macrophages and natural killer cells. Prostaglandins, 35: 413â€”425.1988.Hubbard. N. E., Chapkin, R. S., and Erickson. K. L. Inhibition of growth andlinoleate-enhanced metastasis of a transplantable mouse mammary tumor by indomethacin. Cancer Lett.. 43: 111-120, 1988.Bennett. A.. Gaffen, J. D., Melhuish. P. B., and Stamford, I. F. Indomethacin increasesthe cytotoxicity of methotrexate in vitro and in vivo. Adv. Prostaglandin ThromboxaneLeukotriene Res.. 17: 986-988. 1987.Gelin. J., Andersson. C.. and Lundholm. K. Effects of indomethacin. cytokines. andcyclosporin A on tumor growth and the subsequent development of cancer cachexia.Cancer Res., 51: 880-885, 1991.

Lynch. N. R.. and Salomon. J. C. Tumor growth inhibition and potentiation ofimmunotherapy by indomethacin in mice. J. Nati. Cancer Inst., 62: 117-121. 1979.




60. Wcisburger, J. H.. Reddy. B. S., and Joftes, D. L. Colorectal cancer. U1CC TechnicalReport 19. Geneva: Union linlernational Contre le Cancre, 1975.

61. Northway, M. G., Scobey. M. W.. Cassidy, K. T.. and Geisinger. K. R. Piroxicamdecreases postirradiation colonie neoplasia in thÃ©rat. Cancer (Phila.), 66: 2300-2305,1990.

62. Franklin. C. D., and Craig. G. T. The effect of indomethacin on tumor regression inDMBA-induced epithelial neoplasia of hamster cheek pouch mucosa. Oral Surg., 63:335-339, 1987.

63. Feldman, J. M., and Hilf. R. Failure of indomethacin to inhibit growth of theR3230AC mammary tumor in rats. J. Nati. Cancer Inst., 75: 751-756, 1985.

64. Ellis, L. M., Copeland. E. M., Bland. K. I., and Sitren, H. S. Inhibition of tumorgrowth and metastasis by chronic intravenous infusion of prostaglandin E. Ann Surg..212: 45-50. 1990.

65. Lehnen, T., Deschner, E. F.. Karmali. R. A., and Decosse, J. J. Effect of flurbiprofenand 16,16 dimethyl-prostaglandin Ei on gastrointestinal tumorigenesis induced byA/-Methyl-A''-nitro-;V-nitrosoguanidine in rats. I. Squamous epithelium and mesen-

chymal tissue. J. Nati. Cancer Inst., 78: 923-929, 1987.66. Lehnen. T.. Deschner. E. F. Karmali. R. A., and Decosse. J. J. Effect of flurbiprofen

and 16.16-dimethyl prostaglandin E2 on gastrointestinal tumorigenesis induced byA'-methyl-W-nitro-A'-nitrosoguanidine in rats: glandular epithelium of stomach andduodenum. Cancer Res.. 50; 381-384, 1990.

67. Craven, P. A., Saito, R., and Derubertis. F. R. Role of local prostaglandin synthesis inthe modulation of proliferative activity of rat colonie epithelium. J. Clin. Invest., 72:1365-1375, 1983.

68. Craven, P. A., Thornburg. K.. and Derubertis, F. R. Sustained increase in the proliferation of rat colonie mucosa during chronic treatment with aspirin. Gastroenterology,94: 567-575. 1988.

69. Eastwood. G. L.. and Quimby. G. F. Effects of chronic aspirin ingestion on epithelialproliferation in rat fundus. antrum. and duodenum. Gastroenterology, 82: 852-856.1982.

70. Kuwayama. H.. and Eastwood. G. L. Effects of water immersion restraint stress andchronic indomethacin ingestion on gastric antrat and fundic epithelial proliferation.Gastroenterology, 88: 362-365, 1985.

71. Kune, G., Kune, S., and Watson, L. Colorectal cancer risk, chronic illnesses, operations, and medications: case control results from the Melborne Colorectal CancerStudy. Cancer Res.. 48: 4399-4404, 1988.

72. Paganini-Hill. A.. Chao. A.. Ross, R. K., and Henderson. B. E. Aspirin use and chronic-diseases: a cohort study of the elderly. Br. Med. J.. 229: 1247-1250. 1989.

73. Rosenberg. L.. Palmer, J. R.. Zauber, A. G., Warshauer, M. E.. Stolley, P. D.. andShapiro, S. A hypothesis: nonsteroidal anti-inflammatory drugs reduce the incidenceof large-bowel cancer. J. Nati. Cancer Inst., 83: 355-358, 1991.

74. Paganini-Hill. A., Hsu. G., Ross, R. K., and Henderson, B. E. Aspirin use andincidence of large-bowel cancer in a California retirement community. J. Nail. CancerInst., 83: 1182-1183, 1991.

75. Rosenberg, L.. Palmer. J. R.. and Shapiro. S. Response. J. Nati. Cancer Inst.. 83: 1183.1991.

76. Thun. M.. Namboodiri, M., and Heath, C. Aspirin use and reduced risk of fatal colon

cancer. N. Engl. J. Med., 325: 1593-1596, 1991.

77. Labayle. D., Fischer, D., Vielh, P.. et al. Sulindac causes regression of rectal polypsin familial adenomatous polyposis. Gaslroenterology, 101: 635-639, 1991.

78. Lipton, A.. Scialla, S., Harvey, H., et al. Adjuvant antiplatelet therapy with aspirin incolo-rectal cancer. J. Med., 23: 419-429, 1982.

79. Waddell. W. R., and Loughry, R. W. Sulindac for polyposis of the colon. J. Surg.Oncol., 24: 83-87, 1983.

80. Rigau, J., PiguÃ©,J. M.. Rubio, E., Planas, R.. Tarrech, J. M., and Bordas. J. M. Effectsof long-term sulindac therapy on colonie polyposis. Ann. Intern. Med., 115: 952-954,

1991.81. Stellman, S. D., and Garfinkel, L. Smoking habits and tar levels in a new American

Cancer Society prospective study of 1.2 million men and women. J. Nati. Cancer Inst.,76: 1057-1063. 1986.

82. World Health Organization. International Classification of Diseases. Manual of theInternational Statistical Classification of Diseases, Injuries and Causes of Death, Vol.1, 9th Revision. Geneva: World Health Organization, 1977.

83. Rothman. K. J Modern Epidemiology, pp. 230-232. Boston: Little. Brown. 1986.84. Flanders. W. D. Approximate variance formulas for standardized rate ratios. J.

Chronic Dis., 37: 449^t53, 1984.85. Cox, D. R. Regression models and life tables. J. R. Stat. Soc. Ser. B, 34: 187-220,

1972.86. Thun. M. J., Calle, E. E., Namboodiri, M. M., et al. Risk factors for fatal colon cancer

in a large prospective study. J. Nati. Cancer Inst., 84: 1491-1500, 1992.87. Isomaki, H. A., Hakulinen. T.. and Joutsenlahti, U. Excess risk of lymphomas.

leukemia, and myeloma in patients with rheumatoid arthritis, J. Chronic Dis., 31:691-696, 1978.

88. Gridley. G., McLaughlin, J. K.. Ekbom, A., et al. Incidence of cancer among patientswith rheumatoid arthritis. J. Nati. Cancer Inst.. 85: 307-311. 1993.

89. Hial, V.. Demello. M. C., Horakova, Z.. and Beaven. M. Antiproliferative activity ofanti-inflammatory drugs in two mammalian cell culture lines. J. Pharmacol. Exp.Ther, 202: 446-454. 1977.

90. Bayer, B. M., Kruth. H. S., Vaughn, M., and Beaven, M. A. Arrest of cultured cellsin the G, phase of the cell cycle by Â¡ndomethacin. J. Pharmacol. Exp. Then, 2/0:106-111, 1979.

91. Honn, K. V. Bockman. R. S.. and Marnett. L. J. Prostaglandins and cancer: a reviewof tumor initiation through tumor metastasis. Prostaglandins, 21: 833-851, 1981.

92. Manning. L. S.. Bowman. R. V.. Davis. M. R., Musk. A. W., and Robinson. B. W.Indomethacin augments lymphokine-activated killer cell generation by patients withmalignant mesothelioma. Clin. Immunol. Immunopathol.. 53: 68-77, 1989.

93. Chao, T. Y., and Chu, T. M. Effect of indomethacin on tumor-infiltrating lymphocytesof a spontaneously developed murine mammary adenocarcinoma. Cancer Immunol.Immunother.. 30: 158-164. 1989.

94. Honn. K. V. Busse. W. D.. and Sloane. B. F. Prostacyclin and thromboxanes, implications for their role in tumor cell metastasis. Biochem. Pharmacol.. 3: 1-11, 1983.

95. Boring. C. C.. Squires, T. S.. and Tong. T. Cancer statistics, 1992. CA Cancer J. Clin.,42: 19-38, 1992.

1327



1993;53:1322-1327. Cancer Res Michael J. Thun, Mohan M. Namboodiri, Eugenia E. Calle, et al. Aspirin Use and Risk of Fatal Cancer

Updated version

http://cancerres.aacrjournals.org/content/53/6/1322

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/53/6/1322To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



Documents

Aspirin Use and Risk of Fatal Cancer · possible relationship between aspirin use and risk of fatal cancer in the cohort studied previously for colon cancer (76). SUBJECTS AND METHODS