risk of esophageal and gastric adenocarcinomas in relation to … · vol. 7, 749-756, september...

Vol. 7, 749-756, September /998 Cancer Epidemiology, Biomarkers & Prevention 749

3 The abbreviations used are: LES. lower esophageal sphincter: OR. odds ratio:Cl. confidence interval.

Risk of Esophageal and Gastric Adenocarcinomas in Relation to Use of

Calcium Channel Blockers, Asthma Drugs, and Other Medications thatPromote Gastroesophageal Reflux

Thomas L. Vaughan,’ Diana C. Farrow,Philip D. Hansten, Wong-Ho Chow, Marilie D. Gammon,Harvey A. Risch, Janet L. Stanford, Janet B. Schoenberg,Susan T. Mayne, Heidi Rotterdam, Robert Dubrow,Habibul Ahsan, A. Brian West,2 William J. Blot, andJoseph F. Fraumeni Jr.

Program in Epidemiology, Fred Hutchinson Cancer Research Center, and

Department of Epidemiology, School of Public Health and Community

Medicine, University of Washington, Seattle, Washington 98109 [T. L. V.,

D. C. F.. J. L. Si: Department of Pharmacy. University of Washington. Seattle.Washington 98195 [P. D. H.]: Division of Cancer Epidemiology and Genetics,

National Cancer Institute, Bethesda, Maryland 20852 [W-H. C.. J. F. F.J;

Division of Epidemiology, Columbia School of Public Health, New York,

New York 10032 EM. D. G., H. Al: Departments of Epidemiology and Public

Health [H. A. R., S. 1. M., R. D.J and Pathology [A. B. WI, School of

Medicine. Yale University, New Haven, Connecticut 06510; Applied CancerEpidemiology Program, New Jersey Department of Health and Senior

Services, Trenton. New Jersey 08625 [J. B. 5.1; Department of Pathology,College of Physicians and Surgeons of Columbia University. New York, New

York 10032 [H. RI: and Intemational Epidemiology Institute, Rockville,

Maryland 20852 [W. J. B.J

Abstract

Incidence of adenocarcinomas of the esophagus andgastric cardia has risen dramatically over the past 2decades in the U. S., for reasons that are not yet clear. Anumber of common medications (e.g., calcium channelblockers, tricyclic antidepressants, and certain asthmamedications) promote gastroesophageal reflux by relaxingthe lower esophageal sphincter (LES). Reflux is thoughtto increase cancer risk by promoting cellularproliferation, and by exposing the esophageal epitheliumto potentially genotoxic gastric and intestinal contents.Recent studies have suggested that calcium channelblockers may also increase cancer risk by inhibitingapoptosis. Using personal interview data from amulticenter, population-based case-control studyconducted between 1993 and 1995 in three areas of theU. S., we evaluated whether the use of LES-relaxingdrugs was associated with increased risk ofadenocarcinomas of the esophagus and gastric cardia.Cases of esophageal adenocarcinoma (n 293) andgastric cardia adenocarcinoma (n = 261) were compared

Received 1/26/98: revised 5/29/98: accepted 6/17/98.

The costs of publication of this article were defrayed in part by the payment of

page charges. This article must therefore be hereby marked ads’ertisen,en: in

accordance with 18 U.S.C. Section 1734 solely to indicate this fact.I To whom requests for reprints should be addressed, at Program in Epidemiol-

ogy. Fred Hutchinson Cancer Research Center. (MP-474). P. 0. Box 19024,

Seattle, WA 98109.

2 Current address: Department of Pathology. University of Texas. Galveston, TX

77555.

with general population controls (n 695). Informationon additional case groups of esophageal squamous cellcarcinoma (n = 221) and noncardia gastric cancer (n =

368) were also available for comparison. Overall, 27.4%of controls had used one or more of these drugs for atleast 6 months, compared with 30.2% of esophagealadenocarcinoma and 23.8% of gastric cardiaadenocarcinoma cases. The adjusted odds ratios (ORs)for ever use were 1.0 [95% confidence interval (CI) =

0.7-1.5] and 0.8 (95% CI 0.5-1.1), respectively. Therewas little evidence of increasing risk with increasingduration of use of all LES-relaxing drugs together. Wefound an increased risk of esophageal adenocarcinomaamong persons reporting use of asthma drugs containingtheophylline (OR 2.5; 95% CI 1.1-5.6) or �3 agonists(OR = 1.7; 95% CI 0.8-3.8). Risks were higher amonglong-term users (>5 years) of these drugs (OR 3.1;

95% CI = 0.9-10.3 and OR 2.3; 95% CI 0.8-7.0,respectively). In contrast, there was no evidence that theuse of calcium channel blockers or other specific groupsof drugs increased the risk of any of the cancers studied.These results provide reassuring evidence that theincreases in incidence of adenocarcinomas of theesophagus and gastric cardia are not likely to be relatedto the use of LES-relaxing drugs as a group, or calciumchannel blockers in particular, but they do suggest thatpersons treated for long-standing asthma may be atincreased risk of esophageal adenocarcinoma.

Introduction

Esophageal and gastric cardia adenocarcinomas have risen dra-

matically in incidence over the past 2 decades in the U. S. anda number of western countries ( 1-5). The underlying reasonsare not yet clear. In contrast, the incidence of esophageal

squamous cell carcinoma and noncardia gastric adenocarci-noma in the U. S. has remained constant or decreased ( 1).

A large proportion of esophageal and gastric cardia ade-nocarcinomas arise in persons with long-standing gastroesoph-ageal reflux (6-8). Reflux is thought to increase the risk ofesophageal adenocarcinoma by promoting cellular prolifera-

tion, and by exposing the esophageal epithelium to potentially

genotoxic gastric and intestinal contents. A number of commonmedications can promote gastroesophageal reflux by decreas-ing LES3 pressure or impairing esophageal motility. Examples

include calcium channel blockers, certain tricyclic antidepres-sants, asthma medications containing theophylline or /3 ago-fists, nitroglycerin, antihistamines, and antispasmodics (9-14).

Association for Cancer Research. by guest on September 9, 2020. Copyright 1998 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

https://bloodcancerdiscov.aacrjournals.org

750 MedIcation Use and Esophageal and Gastric Cancer

It has been hypothesized that calcium channel blockers mayincrease cancer risk by another mechanism as well, namely the

capacity of these drugs to inhibit apoptosis ( 15-17). Theseeffects, together with their relatively high and increasing prey-

alence of use ( 1 8), raise the possibility that LES-relaxing med-ications may be partially responsible for the increasing mci-

dence of esophageal and gastric cardia adenocarcinomas.Recent reports have linked excessive weight with the risk

of esophageal and gastric adenocarcinomas, presumably be-cause of the increased reflux resulting from obesity (19-21).The positive association with excessive weight is more pro-nounced for esophageal adenocarcinoma than for gastric cardia

adenocarcinoma, and is not seen for esophageal squamous cellcarcinoma or noncardia gastric cancer (20, 2 1 ). Because we

postulate that LES-relaxing medications would act via the samemechanism, our a priori hypothesis was that a similar pattern of

altered risk would be observed with the LES-relaxing medica-tions. For calcium channel blockers, we postulated increasedrisk for each of the four cancer types. We evaluated thesehypotheses using data from a large, multicentered, population-

based case-control study of esophageal and gastric cancers

conducted between 1993 and 1995 at three areas in the U. S.

Materials and Methods

A detailed description of the methods has been reported previ-ously (22). Briefly, we conducted a population-based case-

control study of incident esophageal and gastric cancers diag-nosed during 1993-1995 in three areas of the U. S.: New

Jersey, Connecticut, and western Washington. Rapid reportingprocedures were used to identify cases soon after diagnosis. An

expert panel reviewed histological material and reports fromsurgery, endoscopy, and radiology to classify each cancer as tosite of origin (esophagus, gastric cardia including gastroesoph-

ageal junction, or noncardia gastric sites), and histology (squa-mous cell or adenocarcinoma).

We attempted to interview all subjects with adenocarci-noma of the esophagus or gastric cardia (target cases), and

approximately equal numbers of persons with esophageal squa-mous cell carcinoma or noncardia gastric adenocarcinoma

(comparison cases), frequency matched to the target cases bygeographic area, 5-year age group, gender (in New Jersey and

Washington), and race (white or other; in New Jersey). Popu-lation-based controls were similarly frequency matched to the

target cases. Controls under 65 years of age were selected fromthe general population of the catchment area of each registry byrandom digit dialing (23). Controls 65 years and older wereselected from Health Care Financing Administration rosters

(24).Overall, personal interviews were conducted for 554 target

cases (80.6% of those eligible), comprised of 293 esophagealadenocarcinomas and 26 1 gastric cardia adenocarcinomas; 589

comparison cases (74. 1% of those eligible), comprised of 221esophageal squamous cell carcinomas and 368 noncardia gas-tric adenocarcinomas; and 695 controls (73.7% of those eligi-ble). The overall control response rate was 70.2% after taking

into account telephone screening refusals for those identified byrandom digit dialing. For 29.6% of the target cases, 32.2% of

the comparison cases, and 3.4% of the controls, interviews werecompleted with the closest next-of-kin (usually the spouse)rather than the subject.

The structured questionnaire inquired about history of

tobacco and alcohol use, diet, occupation, height and weight,and demographic background. A major goal of the study was toinvestigate the possible relationship between use of medica-

tions that promote gastroesophageal reflux and risk of esoph-

ageal and gastric cardia adenocarcinomas. On the basis of a

review of the literature, we identified the following groups of

medications for study: calcium channel blockers; asthma mcd-

ications containing theophylline or 13-adrenergic agonists; ni-

trates; disopyramide; and several groups of drugs with antimus-

carinic (anticholinergic) effects, including certain tricyclic

antidepressants, antispasmodics, and over-the-counter asthma

and antihistamine medications (9-14, 25-30). Because the in-

tensity of antimuscarinic effects within a medication class can

vary from one drug to another, only those with moderate to

potent antimuscarinic effects were included. A detailed list of

the medications we considered is included in the appendix.

We inquired about the regular use (defined as at least once

weekly for 6 months or longer) of each group of medication. A

“show card” was given to the respondent listing trade and

generic names of each drug in the group. An initial screening

question asked whether the respondent had ever used any of the

drugs in the group at least weekly for 6 months or more. If the

respondent indicated that he or she had, we inquired about the

years the medications were started and stopped, total duration

of use, frequency of use, and which particular medications the

respondent had used. All information collected in the interview

referred to exposures occurring before a reference date. This

was defined as 1 year before interview for controls, and the

earlier of 1 year before interview or the diagnosis date for cases.

For each group of drugs, we calculated variables indicat-

ing ever versus never use, current (as of the reference date)

versus former use, and total duration of use. To allow for the

possibility that an effect of medication use on cancer risk might

take 5 or more years to manifest, we also calculated the duration

after excluding any use within 5 years of the reference date

(5-year lag; 3 1 ). Using the above variables calculated for eachgroup of drugs, we created a set of summary variables con-

cerning the use of any LES-relaxing medication. Subjects withmissing information for a particular drug group (3. 1% or less

for cases, and 1 .3% or less for controls) were excluded from

analyses of that drug group.

Sixty-two persons whom we directly interviewed (36

cases and 26 controls) had been members of a health mainte-

nance organization that maintained computerized prescription

medication records for the period of time covered by the inter-

view. As a crude measure of validity of self-reports, we com-

pared medication use reported on the questionnaire with the

number of filled prescriptions for the specific calcium channel

blockers, antidepressants, and asthma medications under study

(32).

ORs and 95% CIs were used to estimate relative risks.They were calculated using unconditional logistic regression

(33) after adjusting for the potential confounding effects of

geographic center, age (seven categories), race (white, oth-

er), gender, body mass index (quartiles), cigarette smoking

(five categories of pack-years), income (five categories), and

proxy status. For analyses of esophageal squamous cell

carcinoma, we also controlled for alcohol intake (four cat-

egories of drinks/week.) Effect modification by the above

factors was assessed by examination of stratum-specific ORs

and by adding interaction terms to the logistic models. Most

analyses included all regular users of the medications. Re-

sults of repeat analyses restricted to those subjects reporting

daily medication use were essentially identical (data not

shown).



Cancer Epidemiology, Biomarkers & Prevention 751

Table I Sele cted demographic characteristics of cases of esophageal and gastric cardia adenocarcinoma and controls

Esophageal Gastric cardiaControls .

adenocarcinoma adenocarcinoma

No. % No. % No. ‘3-

Age

<40 33 4.7 8 2.7 12 4.6

40-49 84 12.1 33 11.3 32 12.3

50-59 171 24.6 59 20.1 51 19.5

60-69 245 35.3 97 33.1 94 36.0

70-79 162 23.3 96 32.8 72 27.6

Gender

Male 555 79.9 245 83.6 223 85.4

Female 140 20.0 48 16.4 38 14.6

Race

Caucasian 646 92.9 289 98.6 252 96.6

African-American 34 4.9 2 0.7 4 1.5

Other I 5 2.2 2 0.7 5 I .9

Table 2 Use of any LES-relaxing medications and risk of esophageal and gastric cardia adenocarcinoma

Controls Esophageal adenocarcinoma Gastric cardia adenocarcinoma

No.” No.” OR5 (95% CI) No.” OR5 (95% Cl)

Never 497 192 1.0 - 189 1.0 -

Ever 188 83 1.0 (0.7-1.5) 59 0.8 (0.5-1.1)

Use at reference date

Former 43 22 1 .3 (0.7-2.4) 8 0.5 (0.2- 1 . I)

Current 144 61 1.0 (0.6-1.4) 51 0.9 (0.6-1.3)

Duration of use (yr)

<5 81 34 1.1 (0.7-1.8) 21 0.6 (0.4-1.1)

5-9 39 14 0.9 (0.4-1.8) 17 1.3 (0.6-2.5)

10-19 42 22 1.1 (0.6-2.0) 12 0.7 (0.3-1.4)

20+ 24 12 0.9 (0.4-2.0) 9 0.6 (0.3-1.6)

Duration of use (yr) - 5-yr lag

<5 39 13 0.9 (0.4-1.8) 16 1.1 (0.5-2.1)

5-9 25 13 0.9 (0.4-2.2) 10 1.0 (0.5-2.3)

10-19 29 12 0.8 (0.3-1.8) 5 0.4 (0.1-1.0)

20+ 1 1 9 1.9 (0.7-5.3) 4 0.7 (0.2-2.7)

‘, Totals vary due to missing yalues.b ORs and 95% CIs adjusted for age. gender, center, race, income, body mass index, cigarette use, and proxy status.

Results

Selected demographic characteristics of target cases and con-

trols are given in Table 1. A more detailed description ofsubject characteristics has been published previously (22).

Table 2 shows the associations between the use of any

LES-relaxing medications and risk of esophageal and gastriccardia adenocarcinomas. Overall, 27.4% of controls had used

one or more of these drugs for at least 6 months, compared with30.2% of esophageal and 23.8% of gastric cardia adenocarci-noma cases. The adjusted ORs associated with ever use were1.0 (95% CI = 0.7-1.5) and 0.8 (95% CI = 0.5-1.1), respec-

tively. For esophageal adenocarcinoma, there was no evidence

of increasing risk with increasing duration of use, either whenincluding all years of use or when excluding the most recent 5

years of use; the only exception being a somewhat elevated riskfor users of 20 or more years with a 5-year lag period (OR =

1.9; 95% CI = 0.7-5.3). For gastric cardia adenocarcinoma,most of the ORs were below unity. There was also no evidenceof an association with esophageal squamous cell carcinoma ornoncardia gastric adenocarcinoma (data not shown).

The associations between the use of specific groups ofdrugs and adenocarcinomas of the esophagus and gastric cardiaare given in Table 3 for the most commonly reported drugs that

we investigated. Regular use of calcium channel blockers for 6

months or longer was reported by 16.2% of controls, 17.1% of

esophageal adenocarcinoma cases, and 16.6% of gastric cardiaadenocarcinoma cases. For these sites, the adjusted ORs were

0.8 (95% CI = 0.5-1.3) and 0.9 (95% CI = 0.6-1.5), respec-tively. We found no consistent pattern of increasing risk withincreasing duration of intake for either site when all years of use

before the reference date were examined, or with a 5-year lag.

Similarly, there was no evidence that the use of calcium channelblockers was associated with risk of esophageal squamous cell

carcinoma (OR 0.8; 95% CI = 0.4-1 .3) or noncardia gastricadenocarcinoma (OR = 0.8; 95% CI 0.6-I .3).

Use of nitroglycerin medications was less frequently re-ported among esophageal adenocarcinoma cases (OR = 0.4;95% CI = 0.2-0.9) and gastric cardia adenocarcinoma cases

(OR = 0.8 (95% CI = 0.4-1 .6) than controls, although there

was no evidence of a trend in ORs with increasing duration of

use.There was a suggestion of an increased risk of adenocar-

cinoma of the esophagus (OR = 1.6; 95% CI = 0.7-3.7), but

not of the gastric cardia (OR 0.3; 95% CI = 0.1-1.2),associated with regular use of certain tricyclic antidepressants.

The association seemed to be strongest among short-term users



Table 3 Use of LES-relaxing medications and risk of esophageal and gastric cardia adenocarcinoma

Controls

No.”

Esophageal adenocarcinoma Gastric cardia adenocarcinoma

No.” OR” (95% CI) No.” OR” (95% CI)

575 238 1.0

111 49 0.8

65 28 0.9 (0.5-1.5)

46 20 0.7 (0.4-1.3)

275 1.0 -

13 0.4 (0.2-0.9)

675 272 1.0

18 15 1.6

678 276 1.0

17 15 2.5 (0.2-I .8)

(0.2-20.6)

(0.2-1.7)

4

2

2

673

19

18

10

9

273 1.0

13 1.7

2 7.7

11 1.4

5+


“ Totals vary due to missing values.b ORs and 95% CIs adjusted for age, gender, center, race, income, body mass index, cigarette use, and proxy status.

Calcium channel blockers

Never

Ever


Former

Current


<5

5+

Duration of use (yr): 5-yr lag

<5

5+

Nitroglycerin

Never

Ever


Former

Current


<5

5+


<5

5+

Tncyclic antidepressants

Never

Ever


Former

Current


<5

5+


<5

5+

Theophylline

Never

Ever


Former

Current


<5

5+


<5

5+

f3 agonists

Never

Ever


Former

Current


<5

5+

20

91

23

22

652

42

13

29

16

26

11

18

9

9

11

7

7

6

3

14

10

7

4

2

(0.5-1.3)

6 0.6 (0.2-1.7)

43 0.9 (0.6-1.4)

11 0.7 (0.3-1.6)

9 0.7 (0.3-1.8)

4 0.2 (0.0-1.0)

9 0.5 (0.2-1.2)

4 0.2 (0.1-0.9)

9 0.5 (0.2-1.3)

3 0.2 (0.0-0.9)

7 0.7 (0.2-1.9)

(0.7-3.7)

7 1.7 (0.5-5.5)

8 1 .5 (0.4-4.9)

I 1 2.5 (0.9-6.7)

4 0.5 (0.1-2.7)

5 1.7 (0.5-6.3)

2 0.3 (0.0-2.0)

(1.1-5.6)

3 4.9 (0.7-33.4)

12 2.2 (0.9-5.3)

8 2.2 (0.8-6.2)

7 3.1 (0.9-10.3)

3 2.8 (0.4-8.3)

4 6.6 (1.1-40.2)

(0.8-3.8)

(0.5-108.0)

(0.6-3.4)

5 1.2 (0.4-3.8)

8 2.3 (0.8-7.0)

211 1.0 -

42 0.9 (0.6-1.5)

6 0.6 (0.2-1.7)

36 1.0 (0.6-1.7)

17 0.6 (0.3-1.1)

25 1.5 (0.8-2.7)

14 1.5 (0.7-3.4)

9 1.1 (0.5-2.5)

240 1.0 -

14 0.8 (0.4-1.6)

I 0.1 (0.0-1.2)

13 1.1 (0.5-2.3)

7 0.9 (0.3-2.5)

7 0.7 (0.3-1.8)

2 0.3 (0.1-1.7)

4 0.6 (0.2-1.9)

257 1.0 -

3 0.3 (0.1-1.2)

I 0.3 (0.0-2.3)

2 0.3 (0.1-1.9)

1 0.3 (0.0-2.3)

2 0.3 (0.1-1.9)

1 0.3 (0.0-2.8)

2 0.4 (0.1-2.5)

254 1.0

6 0.6

1 1.9

5 0.5

1.0 (0.3-3.4)

0.2 (0.0-1.7)

0.8 (0.1-6.2)

0.3 (0.0-5.9)

253 1.0 -

6 0.6 (0.2-1.7)

I 0.6 (0.0-19.7)

5 0.6 (0.2-1.8)

3 0.7 (0.2-2.6)

3 0.6 (0.1-2.5)


<5 3

5

1 0.7 (0.1-8.4)

7 4.1 (1.1-15.3)

I 0.8 (0.1-8.4)

2 0.5 (0.1-3.7)




and was reduced when recent users (within S years) were

excluded.We found an increased risk of esophageal adenocarcinoma

among persons reporting use of medications containing theo-phylline (OR = 2.5; 95% CI = 1.1-5.6) and �-agonists (OR =

1.7; 95% CI 0.8-3.8). Relative risk estimates were higheramong long-term users, particularly those in the more distant

past (OR = 6.6; 95% CI = 1 . 1-40.2 for theophylline, andOR = 4. 1 ; 95% CI = 1 . 1-15.3 for �3 agonists). Tests fortrend-relating duration of use of theophylline-containing med-ications to esophageal adenocarcinoma risk were P 0.022

and P = 0.019 for all years and a 5-year lag, respectively. For/3 agonists, they were P = 0.14 and P = 0.051 . ORs for gastriccardia adenocarcinoma among users of these medications, how-

ever, were all below 1.0.

There was no evidence linking regular use of other, lessfrequently reported medications, including anticholinergics,

disopyramide, and over-the-counter antihistamines and asthmadrugs, to the risk of esophageal or gastric cardia adenocarci-nomas, although results are based on small numbers (data notshown). Furthermore, none of the drugs investigated was asso-

ciated with the risk of esophageal squamous cell carcinoma ornoncardia gastric adenocarcinoma. Results were similar whenanalyses were restricted to direct (nonproxy) interviews.

Relative risk estimates did not differ materially by age,

body mass index, or cigarette use. When analyses were strati-fled by gender, however, relative risk estimates tended to behigher among women than men (Table 4). For use of any

LES-relaxing medication, the ORs associated with ever usewere 0.9 (95% CI 0.6-1.3) for men and 1.9 (95% CI

0.9-4. 1) for women. No single drug group seemed to accountfor the observed differences in relative risk by gender.

To compare the subjects’ computerized pharmacy recordswith self-reports, we dichotomized the number of filled pre-

scriptions for calcium channel blockers, asthma medications,and tricyclic antidepressants into those with six or more versus

less than six, corresponding roughly to the screening question

in the interview. Self-reported use of the medications understudy corresponded closely to the pharmacy records. Of seven

persons with six or more prescriptions for a drug group in the

pharmacy records, six (85.7%) reported using a correspondingdrug for at least 6 months. Among persons for whom there werezero to five prescriptions for a drug group in the pharmacy

records, the overall proportion who also reported no regular useof the medications was 97.9%. For individual drug groups, theproportions ranged from 96.5% for calcium channel blockers to100% for theophylline.

Discussion

Overall, we found little evidence that commonly used LES-relaxing medications, as a group, increase the risk of esopha-

geal or gastric cardia adenocarcinomas. There are several pos-sible explanations. First, although each medication included inthe study was selected on the basis of its reported ability toinduce gastroesophageal reflux, the extent to which the use of

these drugs contributes to reflux disease in the general popu-lation is not firmly established. There is probably substantialvariability in the amount of induced reflux both within and

among drug groups, as well as variability in individual suscep-tibility. Thus, it seems likely that drug-related reflux, at least in

most users, is not substantial enough for a carcinogenic effect,especially compared with other conditions that promote gas-troesophageal reflux, such as hiatal hernia and obesity (7, 21,34, 35).

Methodological limitations should also be considered.Chief among them is the reliance on self-reports to assess pastmedication use, which probably contributed to at least some

degree of misclassification. Most likely, the misclassificationwould be primarily nondifferential, tending to bias the resulting

relative risk estimates toward unity. We attempted to minimizeproblems with recall by conducting structured personal inter-views, supplying “show cards” to each respondent with lists of

trade and generic names to improve recognition, and by focus-ing on relatively long-term use (6 months or greater), for which

recollection is likely to be more accurate. The validity substudy,in which we compared prescription records with self-reported

medication use, indicated reasonably high levels of agreement,although the sample size was too small for definitive conclu-

sions. When we carried out analyses excluding subjects withproxy interviews, in whom the reporting accuracy of medica-

tions is likely to be lower, the results changed very little.

Although our overall findings for LES-relaxing drugswere negative, the heterogeneity of results by medication type

and gender suggests the importance of evaluating each group ofmedications separately. Our results are reassuring with regard

to calcium channel blocking medications, as we found nosubstantial evidence for increased risk of any of the esophagealor gastric cancers under study. Relative risk estimates weregenerally close to unity, with upper CIs of 1 .5 or less associatedwith ever use. Calcium channel blockers are of interest because

of their potential not only to promote gastroesophageal reflux,

but also to inhibit apoptosis (15, 16), and some epidemiologicalstudies have suggested an increased cancer risk in users of the

drugs. In particular, Pahor et al. (17), reporting on a cohort of

elderly persons who used calcium channel blockers and werefollowed for a short time for cancer incidence, noted an in-creased relative risk for all cancers (1.72; 95% CI 1.27-

2.34), with one of the highest relative risks observed for gastriccancer (3.63; 95% CI 0.96-13.76; 17). Olsen et al. (36)reported no statistically significant association between the useof calcium channel blockers and digestive system cancers in a

short-term cohort study, although the risk for esophageal cancer(all histologies together) was slightly elevated (standardizedincidence ratio 1.6; 95% CI 0.7-3.3), based on sevenobserved cases versus four expected cases. Zhang et al. (37, 38)

reported hypertension to be a risk factor for esophageal andgastric cardia adenocarcinomas (OR = 2.4; 95% CI 1.1-5.1)and suggested that antihypertensive medications (which include

calcium channel blockers) may contribute to the rising mci-dence of these tumors in the general population. Because in-

formation on the use of antihypertensive drugs was not avail-able in that study, and information on the history of

hypertension was not available in our study, a direct compar-ison of results is not possible. Most recently, in a nestedcase-control study of hypertensive persons in the United King-dom, Jick et al. (39) reported a small excess risk of all cancers(OR = 1.17; 95% CI = 0.98-1.63) and no association with

esophageal and gastric cancer combined (OR = 0.98; 95%CI = 0.41-2.38) among those using calcium channel blockersrelative to those using /3-blockers.

Regular use of nitroglycerin-containing medications wasreported less frequently among cases, particularly those with

esophageal adenocarcinoma, than among controls. We are notaware of a biological reason for this negative association, andchance seems the most likely explanation.

Ever use of certain tricyclic antidepressants was associatedwith a nonsignificant 60% increase in risk of adenocarcinomaof the esophagus, but not of the gastric cardia. The elevated riskseemed limited to short-term users and those who took the




Table 4 Associatio ns by gen der bet ween sd ected LES-relax ing drugs and risk of can phageal adenocarcinoma

Males FemalesTest for

homogeneityNo.

controls

No.

casesOR” (95% Cl)

No.

controls

No.cases

OR5 (95% CI)

Any LES drug

Never 397 168 1.0 - 100 24 1.0 -

Ever 151 63 0.9 (0.6-1.3) 37 20 1.9 (0.9-4.1) P 0.073

<5 years 60 30 1.0 (0.6-1.7) 21 4 0.8 (0.2-2.6)

5+ years 89 32 0.8 (0.5-1.2) 16 16 3.3 (1.3-7.9) P 0.009

Calcium channel blockers

Never 456 203 1 .0 - I I 9 35 1 .0 -

Ever 93 37 0.7 (0.5-1.2) 18 12 1.8 (0.7-4.4) P 0.097

Nitroglycerin

Never 515 232 1.0 - 137 43 1.0 -

Ever 40 10 0.4 (0.2-0.9) 2 3 3.0 (0.4-20.7) P 0.082

Tricyclic antidepressants

Never 542 229 1 .0 - 1 33 43 1 .0 -

Ever I I 10 2.0 (0.8-5.0) 7 5 1.7 (0.5-6.3) P 0.549

Theophylline

Never 54 1 234 1 .0 - I 37 42 1 .0 -

Ever 14 9 1.3 (0.5-3.3) 3 6 8.0 (1.7-38.3) P = 0.063

13 agonistsNever 539 231 1.0 - 134 42 1.0 -

Ever 14 8 1.1 (0.4-2.7) 5 5 3.4 (0.9-13.3) P 0.240

,, Totals vary due to missing values.

S ORs and 95% CIs adjusted for age. center. race, income. body mass index, and cigarette use.

drugs more recently. Future studies should help distinguishassociations with the underlying condition or related life-style

changes versus the use of various types of antidepressants.We observed an excess risk of esophageal adenocarcinoma

among persons who regularly used prescription asthma drugscontaining theophylline or /3 agonists. The increase in risk wasgreatest among those who used the drugs for S or more years, andafter excluding recent use. The associations tended to be strongerand more often statistically significant for theophylline-containingmedications than for /3 agonists. No excess risk was seen for

gastric cardia adenocarcinoma or the other tumors under study.Bronchodilators, first including theophylline and then various

/3 agonists, have been mainstays in the treatment of asthma since

the l950s, although steroidal anti-inflammatory agents havetended to replace theophylline for control of chronic asthma overthe past decade (40, 41). Our results, however, do not necessarily

indicate that these drugs, per se, are causally related to increasedcancer risk, particularly because gastroesophageal reflux and

asthma often occur together (42). It has been estimated that >80%of adult asthmatics have gastroesophageal reflux, regardless oftheir use ofbronchodilators (43), with 39% having esophagitis and13% having Barrett’s esophagus (44). However, the causal con-

nection between the two conditions is complex and poorly under-stood, with each seeming to contribute to the other (42, 44-46).Although it is possible that chronic use of the asthma medicationsmay increase esophageal adenocarcinoma risk by exacerbating the

reflux commonly experienced by persons with asthma, our studydesign did not allow us to separate the effects of asthma from itstreatment.

Regardless of the mechanism involved, however, our

study suggests that persons treated for long-standing asthma areat an elevated risk of esophageal adenocarcinoma. This findingis provocative in light of the substantial increase in asthmaprevalence over the past 2 decades that has been documented inmany western countries (47, 48). This interpretation is sup-ported by the association reported between asthma and reflux

disease, the specificity of our findings to esophageal adenocar-

cinoma, our ability to control for major known risk factors forthe cancer in statistical analyses, the modest rise in risk we

observed with increasing duration of use of asthma drugs, thesimilar results for theophylline and /3 agonist-containing med-ications, and the information from pharmacy records suggestingreasonably accurate reporting of asthma and other drug use. Onthe other hand, the decreased risk for gastric cardia adenocar-

cinoma, the multiple comparisons made, and the fact that onlysome of the comparisons were statistically significant suggest

caution in interpreting the positive findings for esophageal

adenocarcinoma. Information about the dates of diagnosis andseverity of asthma, as well as a history of other respiratory

conditions, would have been helpful in analyzing and interpret-ing these data, but were not available. Although asthma has

been associated with an increased risk of lung cancer in some(49, 50), but not all (5 1 ) studies, none have specifically eval-

uated the risk of esophageal adenocarcinoma.We found no clear evidence that use of the other groups of

LES-relaxing medications, including antispasmodics, disopy-ramide, and over-the-counter asthma and antihistamine medi-

cations were associated with increased cancer risk, although thenumber of regular users of the other specific groups of drugs

was generally quite low. One previous medical record-basedstudy of selected antispasmodic drugs (propantheline, dicyclo-

mine, Donnatal, and Librax) also revealed no association with

the risk of esophageal and gastric cardia adenocarcinomascombined (OR = 0.7; 95% CI = 0.4-1.3; Ref. 7).

The gender difference in the relation between LES-relax-

ing drugs and esophageal adenocarcinoma is puzzling. The roleof chance must be considered, given the borderline statisticalsignificance of the overall finding, the lack of statistical sig-nificance for individual drugs, and the multiple comparisonsthat are inherent in such subgroup analyses. However, theconsistency of the gender differences for various drugs is in-triguing, especially in light of the much higher incidence ofesophageal adenocarcinoma in men than women (2). It is pos-sible that a subtle effect of LES-relaxing medications may be



Group

�-adrenergic agonists

Generic name

most easily observed in a low-risk population, such as women,whereas the greater prevalence of other risk factors in men may

tend to obscure the effect of medication-induced reflux. Argu-ing against this idea, however, are the findings from our study

that suggest little effect modification by other known riskfactors such as age, body mass index, or tobacco use. Theconsistency of the gender differences across drug types alsosuggests the possibility of a gender-specific bias in reporting

medication use among cases or controls.In summary, our results provide reassuring evidence that

recent increases in the incidence of esophageal and gastric cardiaadenocarcinomas are not likely to be related to use of LES-

relaxing medications as a group, or calcium channel blockers inparticular. However, elevated risks of esophageal adenocarcinomawere associated with long-term use of asthma drugs containingtheophylline or /3-agonists. Additional studies are needed to clarifythe role of asthma and specific bronchodilators in the developmentof esophageal adenocarcinoma and its precursor states, and todetermine whether endoscopic screening for early-stage esopha-

geal adenocarcinoma or Barrett’s metaplasia is justified among

individuals with long-standing asthma.

Acknowledgments

We thank Carol Sweeney and Preet Dhillon (Washington). and Shelley Niwa (Wes-

tat) for assistance in data processing and analyses; Sarah Greene and Linda Lannom(Westat), BerEt Nicol-Blades (Washington), Tom English (New Jersey), and Patricia

Owens (Connecticut) for help in study management; and the interviewers at each

center for invaluable efforts in interviewing participants in this study. We especially

appreciate the contributions ofthe study participants, and also thank the 178 hospitals

in Connecticut, New Jersey, and Washington for assistance.

Appendix

Drugs Included in ESICh Drug Group (continued)

Brand name

Alupent metaproterenol

Metaprel metaproterenol

Brethaire terbutaline

Brethine terbutaline

Bricanyl terbutaline

Bronkometer isoetharine

Bronkosol isoetharine

Isuprel isoproterenol

Maxair pitbuterol

Proventil albuterol

Ventolin albuterol

Tomalate bitolterol

Theophylline-containing Acerolate theophylline

Asbron theophylline

Constant-T theophylline

Marax theophylline

Quadrinal theophylline

Quibron theophylline

Theodur theophylline

Aminophyllin aminophylline

Atrovent ipratropium bromide

Choledyl oxtriphylline

Dilor dyphylline

Lufyllin dyphyline

Respbid anhydrous theophylline

Sb-Bid anhydrous thcophylline

T-PHYL anhydrous theophylline

Tedral anhydrous theophylline

Theobid anhydrous theophylline

Theoair anhydrous theophylline

Theo-Organdin anhydrous theophylline

Uniphyl anhydrous theophylline

Antispasmodics Bentyl dicyclomine

Donnatal -

Over-the-counter asthma

Prescription and over-

-the-counter allergy

Drugs Included in Each Drug Group

Group Brand name Generic name

Calcium channel

blockers

Nitrates

Tricyclic antidepressants

Adalat

Procardia

Calan

Isoptin

Verelan

Cardene

Cardizem

DynaCirc

Nimotop

Norvasc

Plendil

Vascor

Isordil

Sorbitrate

Nitro-Bid

Nitrospan

Nitrostat

Nitrodisc

Nitro-Dur

Transderm-Nitro

Asendin

Aventyl

Pamelor

Elavil

Endep

Ludiomil

Sinequan

Surmontil

Tofranil

Vivactil

nifedipine

nifedipine

verapamil

verapamil

verapamil

nicardipine

diltiazem

israpidine

nimodipine

amlopidine

felodipine

bepridil

isosorbide dinitrate

isosorbide dinitrate

nitroglycerin tablets/capsules

nitroglycerin tablets/capsules

nitrogylcerin tablets/capsules

nitroglycerin patches



amoxipine

nortriptyline

nortriptyline

amtriptyline

amtnptyline

maproteline

doxepin

trimipramine

imipramine

protriptyline

Librax

Pro-Banthine

Robinul

Asthmahaler Mist

Asthmanefrin

Bronkaid Mist

Bronkaid Tablets

Bronkolixir

Bronkotabs

Primatene Mist

Primatene Tablets

Actidil

Atarax

Vistanl

Benadryl

Chlor-Trimeton

Teldrin

Dimetane

Phenergan

Ploaramine

Tavist

Temaril

propantheline

glycopyrrolate

triprolidine

hydroxyzine

hydroxyzine

diphenhydramine

chlorpheniramine

chlorpheniramine

brompheniraminc

promethazine

dexchlorpheniramine

clemastine

trimeprazine

References

1. Blot, W. J., Devesa, S. S., Kneller, R. W., and Fraumeni, J. F., Jr. Rising

incidence of adenocarcinoma of the esophagus and gastric cardia. J. Am. Med.

Assoc., 265: 1287-1289, 1991.

2. Blot, W. J., Devesa, S. S., and Fraumeni. J. F., Jr. Continuing climb in ratesof esophageal adenocarcinoma: an update. J. Am. Med. Assoc.. 270: 1320, 1993.

3. Hansson, L. E., Sparen, P., and Nyren, 0. Increasing incidence of both majorhistological types of esophageal carcinomas among men in Sweden. lnt. J.

Cancer, 54: 402-407, 1993.




756 MedicatIon Use and Esophageal and Gastric Cancer

4. Armstrong. R. W.. and Borman, B. Trends in incidence rates of adenocarci-noma of the ocsophagus and gastric cardia in New Zealand. 1978-1992. Int. J.

Epidemiol.. 25: 941-947. 1996.

5. Hansen. S., Wiig. J. N., Giercksky. K. E., and Tretli, S. Esophageal and gastric

carcinoma in Norway 1958-1992: incidence time trend variability according tomorphological subtypes and organ subsites. Int. J. Cancer, 71: 340-344, 1997.

6. MacDonald, W. C., and MacDonald, J. B. Adenocarcinoma of the esophagus

and/or gastric cardia. Cancer (Phila.), 60: 1094-1098, 1987.

7. Chow, W. H., Finkle, W. D., McLaughlin, J. K., Frankl, H., Ziel, H. K., and

Fraumeni. J. F., Jr. The relation of gastrocsophageal reflux disease and its

treatment to adenocarcinomas of the esophagus and gastric cardia. J. Am. Med.

Assoc., 274: 474-477, 1995.

8. Farrow. D. C., Vaughan, T. L., Hansten, P. D., Stanford, J. L., Risch, H. A.,

Gammon, M. D., Chow, W. H., Dubrow. R., Ahsan. H.. Mayne, S. T.,Schoenberg, J. B., West, A. B., Rotterdam, H., Fraumeni, J. F. Jr., and Blot,W. J. Use of aspirin and other non-steroidal anti-inflammatory drugs and risk of

esophageal and gastric cancer. Cancer Epidemiol. Biomark. Prey., 7: 97-102.

I 998.

9. Christensen, J. Effects of drugs on esophageal motility. Arch. Intem. Med..

136: 532-537, 1976.

10. Gelfond, M., Rozen, P.. Keren, S., and Gilat, T. Effect of nitrates on LOS

pressure in achalasia: a potential therapeutic aid. Gut, 22: 3 12-3 1 8, 1981.

I I . Berquist. W. E.. Rachelefsky, G. S.. Rowshan, N., Siegel, S., Katz, R., and

Welch, M. Quantitative gastroesophageal reflux and pulmonary function in asth-

matic children and normal adults receiving placebo. theophylline, and metapro-

terenol sulfate therapy. J. Allergy Clin. Immunol., 73: 253-258, 1984.

12. Hongo. M.. Traube, M., and McCallum, R. W. Comparison of effects of

nifedipine, propantheline bromide, and the combination on esophageal motor

function in normal volunteers. Dig. Dis. Sci., 29: 300-304, 1984.

13. Konrad-Dalhoff, I., Baunack, A. R., Ramsch, K. D., Ahr, G., Kraft, H., and

Schmitz, H. Effect of the calcium antagonists nifedipine, netrendipine, nimodip-

inc and nisoldipine on ocsophageal motility in man. Eur. J. Clin. Pharmacol., 41:

313-316, 1991.

14. Brown, H. J., and Taylor, P. Muscarinic receptor agonists and antagonists. In:J. G. Hardman and L. E. Limberd (eds.), Goodman and Gilman’s The Pharma-cological Basis of Therapeutics. Ed. 9. pp. 141-160. New York: McGraw-Hill,

1996.

15. Wright, S. C., Zhong, J., and Larrick. J. W. Inhibition of apoptosis as a

mechanism of tumor promotion. FASEB J.. 8: 654-660, 1994.

16. Connor. J.. Sawczuk. I. S., Benson. M. C., Tomashefsky, P., O’Toole, K. M.,

Olsson, C. A.. and Buttyan. R. Calcium channel antagonists delay regression of

androgen-dependent tissues and suppress gene activity associated with cell death.Prostate, 13: 119-130, 1988.

17. Pahor, M., Guralnik, J. M., Ferrucci, L., Corti, M. C., Salive, M. E., Cerhan,

J. R., Wallace, R. B., and Havlik, R. J. Calcium-channel blockade and incidence

of cancer in aged populations. Lancet, 348: 493-497, 1996.

18. Wang. H. H., Hsieh, C. C., and Antonioli, D. A. Rising incidence rate of

esophageal adenocarcinoma and use of pharmaceutical agents that relax the lower

esophageal sphincter (United States). Cancer Causes Control, 5: 573-578, 1994.

19. Brown, L. M.. Swanson. C. A., Gndley, G., Swanson, G. M.. Schoenberg,

J. B.. Greenberg, R. S.. Silverman, D. T.. Pottem, L. M., Hayes. R. B., Schwartz,

A. G.. Liff. J. M., Fraumeni, J. F., Jr., and Hoover, R. N. Adenocarcinoma of the

esophagus: role of obesity and diet. J. NatI. Cancer Inst., 87: 104-109. 1995.

20. Vaughan. T. L., Davis, S., Kristal. A.. and Thomas, D. B. Obesity, alcohol.and tobacco as risk factors for cancers of the esophagus and gastric cardia:

adenocarcinoma versus squamous cell carcinoma. Cancer Epidemiol. Biomark.

Prey.. 4: 85-92, 1995.

21. Chow. W. H., Blot, W. J., Vaughan. T. L., Risch, H. A.. Gammon. M. D..

Stanford, J. S., Dubrow, R., Schoenberg, J. B., Mayne, S. T., Farrow, D. C.,

Ahsan, H.. West. A. B.. Rotterdam, H., Niwa, S., and Fraumeni, J. F., Jr. Body

mass index and risk of adenocarcinomas of the esophagus and gastric cardia.

J. NatI. Cancer Inst.. 90: 150-155, 1998.

22. Gammon, M. D., Schoenberg, J. B., Ahsan, H., Risch, H. A., Vaughan, T. L.,

Chow, W. H., Rotterdam. H.. West, A. B.. Dubrow, R., Stanford, J. L.. Mayne,

S. T.. Fan-ow, D. C., Niwa, S.. Blot, W. J., and Fraumeni, J. F.. Jr. Tobacco.alcohol and socioeconomic status and adenocarcinomas of the esophagus and

gastric cardia. J. NatI. Cancer Inst.. 89: 1277-1284, 1997.

23. Waksberg, J. Sampling methods for random digit dialing. J. Am. Stat. Assoc.,73: 40-46, 1978.

24. Hatten. J. Medicare’s common denominator: the covered population. Health

Care Finance Rev., 2: 53-64, 1980.

25. Foster, L. J., Trudeau, W. L.. and Goldman, A. L. Bronchodilator effects ongastric acid secretion. J. Am. Med. Assoc., 241: 2613-2615, 1979.

26. Berquist, W. E., Rachelefshy, G. S., Kadden, M., Siegel. S. C., Katz, R. M.,

Mickey, M. R., and Ament, M. E. Effect of theophylline on gastroesophagealreflux in normal adults. J. Allergy Clin. Immunol.. 67: 407-41 1, 1981.

27. Traube, M., and McCallum, R. W. Calcium-channel blockers and the gas-

trointestinal tract. Am. J. Gastroenterol., 79: 892-896. 1984.

28. Traube, M., Hongo, M., Magyar, L.. and McCallum, R. W. Effects of

nifedipine in achalasia and in patients with high-amplitude penstaltic esophageal

contractions. J. Am. Med. Assoc., 252: 1733-1736, 1984.

29. Short, T. P., and Thomas, E. An overview of the role of calcium antagonists

in the treatment of achalasia and diffuse oesophageal spasm. Drugs. 43: 177-184.

1992.

30. Ruzkowski, C. J.. Sanowski, R. A., Austin, J., Rohwedder, J. J., and Waring,

J. P. The effects of inhaled albuterol and oral theophylline on gastroesophageal

reflux in patients with gastroesophageal reflux disease and obstructive lungdisease. Arch. Intem. Med., 152: 783-785, 1992.

31 . Checkoway, H., Pearce, N., Hickey, J. L., and Dement, J. M. Latency analysis

in occupational epidemiology. Arch. Environ. Health.. 45: 95-100. 1989.

32. Stergachis, A. S. Evaluating the quality of linked automated databases for use

in pharmacoepidemiology. In: A. G. Hartzema, M. S. Porta, and H. H. Tilson

(eds.), Pharmacoepidemiology: An Introduction, Ed. 2, pp. 222-234. Cincinnati:

Harvey Whitney Books, 1991.

33. Breslow, N. E., and Day, N. E. Statistical Methods in Cancer Research. I. The

Analysis of Case-Control Studies. IARC Scientific PubI. No. 32. Lyon, France:

IARC, 1980.

34. Wang. H. H., Antonioli, D. A., and Goldman, H. Comparative features of

esophageal and gastric adenocarcinomas: recent changes in type and frequency.

Human Pathol., 17: 482-487, 1986.

35. Hogan, W. J., and Dodds, W. J. Gastroesophageal reflux disease (reflux

esophagitis). In: M. H. Sleisenger and J. S. Fordstan (eds.), GastrointestinalDisease, Ed. 4, pp. 594-619. Philadelphia: W. B. Sauners, 1989.

36. Olsen. J. H., Sorensen, H. T., Friis, S., Mclaughlin, J. K., Steffensen. F. H.,

Nielsen. G. L., Andersen, M., Fraumeni, J. F., Jr., and Olsen, J. Cancer risk in

users of calcium channel blockers. Hypertension, 29: 1091-1094, 1997.F. H.

37. Zhang, Z. F., Kurtz, R. C., Sun, M., Karpeh, M., Jr., Yu, G. P., Gargon, N.,

Fein, J. S., Georgopoulos, S. K., and Harlap. S. Adenocarcinomas of the esoph-

agus and gastric cardia: medical conditions, tobacco, alcohol, and socioeconomic

factors. Cancer Epidemiol. Biomark. Prey., 5: 761-768, 1996.

38. Zhang, Z. F., Kurtz, R. C.. Yu, G. P., Sun, M., and Harlap, S. Calcium-channel blockers and cancer. Lancet, 348: 1 166-I 167. 1996.

39. Jick, H., Jick. S., Derby. L. E., Vasilakis, C., Myers, M. W.. Meier. C. R.

Calcium-channel blockers and risk of cancer. Lancet, 349: 525-528, 1997.

40. Weinberger, M., and Hendeles, L. Drug therapy: theophylline in asthma.

N. EngI. J. Med., 334: 1380-1388, 1996.

41. Barnes, P. J. Current therapies for asthma. Promise and limitations. Chest,

Ill: 517-526, 1997.

42. Richter, J. E. Typical and atypical presentations of gastroesophageal reflux

disease. The role of esophageal testing in diagnosis and management. Gastroen-

terol. Clin. N. Am., 25: 75-102, 1996.

43. Sontag, S. J., O’Connell, S., Khandelwal, S., Miller, T., Nemchausky, B.,

Schnell, T. G., and Serlovsky, R. Most asthmatics have gastroesophageal refluxwith or without bronchodilator therapy. Gastroenterology, 99: 613-620, 1990.

44. Sontag, S. J., Schnell, T. G.. Miller, T. Q., Khandelwal, S., O’Connell, S.,

Chejfec, G., Greenlee, H., Seidel, U. J., and Brand. L. Prevalence of oesophagitis

in asthmatics. Gut, 33: 872-876, 1992.

45. Richter, J. E. Gastroesophageal reflux disease: an overlooked cause of

asthma. Cleveland Clin. J. Med., 62: 146-147, 1995.

46. Sontag, S. J. Gastroesophageal reflux and asthma. Am. J. Med., 103: 584-

590, 1997.

47. Sears, M. R. Descriptive epidemiology of asthma. Lancet, 350 (Suppl. 2):

1-4, 1997.

48. Sly. R. M. Changing asthma mortality. Ann. Allergy, 73: 259-268, 1994.

49. Vesterinen, E., Pukkala, E., Timonen, T.. and Aromas, A. Cancer incidence

among 78.000 asthmatic patients. mt. J. Epidemiol., 22: 976-82, 1993.

50. Huovinen. E.. Kaprio. J.. Vesterinen, E.. and Koskenyuo, M. Mortality of

adults with asthma: a prospective cohort study. Thorax. 52: 49-54, 1997.

5 1 . Osann, K. E. Lung cancer in women: the importance of smoking, family

history of cancer, and medical history of respiratory disea.se. Cancer Res., 51:

4893-4897, 1991.



risk of esophageal and gastric adenocarcinomas in relation to … · vol. 7, 749-756, september...

Documents