Antacid Drug Use and Risk of Esophageal and Gastric Adenocarcinomas in Los Angeles County

526

Lei Duan,1 Anna H. Wu,2 Jane Sullivan-Halley,1 and Leslie Bernstein1,2 1City of Hope National Medical Center, Duarte, California and 2Department of Preventive Medicine, Keck School of Medicine, University of Southern California/Norris Comprehensive Cancer Center, Los Angeles, California

Abstract

Objectives: Concern has been expressed that antacid among persons with no UGI disorder than among drugs increase the risk of esophageal and gastric those with an UGI disorder (homogeneity of trends adenocarcinomas. P = 0.07). Regular use of nonprescription acid Methods: This population-based case-control study neutralizing agents was not associated with risk of recruited patients with incident esophageal adeno- adenocarcinomas of the gastric cardia or distal carcinoma (n = 220), gastric cardiac adenocarcinoma stomach. Regular use of prescription acid suppressive (n = 277), or distal gastric adenocarcinoma (n = 441) drugs was not associated with risk for any of these diagnosed between 1992 and 1997, and 1,356 control cancers. participants in Los Angeles County. Unconditional Conclusion: We found risk of esophageal adenocarci- polychotomous multivariable logistic regression anal- noma was greater among long-term nonprescription yses were done to evaluate the association between acid neutralizing drugs in participants without physi- antacid drug use and these cancers. cian-diagnosed UGI conditions than among those with Results: Among participants who took nonprescription these conditions; this may represent self medication for acid neutralizing agents for >3 years, the odds ratio for undiagnosed precursor conditions or it may be that esophageal adenocarcinoma was 6.32 compared with nonprescription acid neutralizing drugs, taken without P never users (95% confidence interval, 3.14-12.69; trend limitation on amount used when symptoms are most < 0.01). Analyses stratified by history of physician intense, may permit alkaline bile reflux into the lower diagnosed upper gastrointestinal (UGI) disorders esophagus, thereby increasing esophageal adenocarci- revealed a greater increase in esophageal adenocarcinoma risk. (Cancer Epidemiol Biomarkers Prev noma risk associated with nonprescription antacid use 2009;18(2):526–33)

Introduction

Histamine2-receptor antagonists and proton pump are significantly more effective than H2-receptor antag- inhibitors (PPI) were introduced to the market, in the onists in reducing gastric acid secretion. These drugs late 1970s and 1980s, correspondingly, and became do not permanently change the acid milieu; although widely used in the treatment of acid-peptic disorders they raise gastric pH and provide rapid relief of such as peptic ulcer disease, gastresophageal reflux symptoms; evidence of this impermanence is that disease (GERD), and acute stress ulcers. H2-receptor erosive esophagitis recurs in most patients within 6 antagonists are competitive inhibitors of histamine months after discontinuing the drug therapy (3). binding at the parietal cell H2 receptor. Gastric acid Therefore, chronic maintenance therapy is generally secretion is suppressed if the parietal cell H2 receptor is required. Although today, some doses of these drugs not bound with histamine in the stomach (1). PPIs are available without prescription, initially all required suppress gastric acid secretion by irreversibly blocking prescription. the hydrogen/potassium adenosine triphosphatase en- Some concerns have been expressed that long term zyme system of the gastric parietal cell (2). Because gastric acid suppression, particularly by H2-receptor PPIs target the terminal-step in acid production, they antagonists, may be associated with increased risk of esophageal and gastric adenocarcinomas (4). Some epidemiologic studies, both prospective studies and retrospective studies, have attempted to evaluate associ- Received 8/18/08; revised 10/3/08; accepted 11/4/08; published OnlineFirst 02/03/2009. ations of H2-receptor antagonists and PPIs with esoph- Grant support: 3RT-0122 and 10RT-0251 from the California Tobacco Related ageal and gastric adenocarcinoma risk (5-15), but the Research Program, grant CA59636 from the National Cancer Institute, and NIEHS results have not been consistent. Although some inves- Grant 5P30 ES07048. Incident cancer cases for this study were collected by the University of Southern California Cancer Surveillance Program, which is supported tigators have reported positive associations between under subcontract by the California Department of Health. The Cancer Surveillance antacid drug use and increased risk of gastric cancer Program is also part of the National Cancer Institute’s Division of Cancer Prevention and Control Surveillance, Epidemiology, and End Results Program, under contract (8, 11, 14) or esophageal cancer (12, 15), others find no number N01CN25403. effect (5, 7, 9, 13). Only one study stratified participants Requests for reprints: Leslie Bernstein, Division of Cancer Etiology, Department of according to whether or not they had severe GERD (7). Population Sciences, City of Hope National Medical Center, 1500 East Duarte Road, Duarte, CA 91010. Phone: 626-471-7315; Fax: 626-471-7308. E-mail: [email protected] It is important to consider the effect of GERD when Copyright D 2009 American Association for Cancer Research. examining the association between antacid use and risk doi:10.1158/1055-9965.EPI-08-0764 of esophageal and gastric adenocarcinomas because

Cancer Epidemiol Biomarkers Prev 2009;18(2). February 2009

GERD itself is a strong independent risk factor for interview queried general background information adenocarcinomas of the esophagus and gastric cardia such as ethnicity, marital status, birthplace, highest (5, 7, 9, 16). level of education, smoking history, lifetime use of all We have explored the association between use of types of alcoholic beverages, usual diet, weight at ages antaciddrugsandriskofesophagealandgastric 20 and 40 y and on the reference date, and height. In adenocarcinomas using data from a large population- addition, we asked detailed questions regarding per- based case-control study conducted among residents of sonal and family history of various nonmalignant Los Angeles County. To understand any observed diseases. association better and to assess confounding of results We explicitly asked about 26 over-the-counter (OTC) by indication for use of the drug, we have investigated and 19 prescription brand name antacid and related the relationship separately in groups with a history of drugs in the questionnaire. For each of the listed upper gastrointestinal (UGI) disorders and those without medications, we first asked the participants whether such a history. they had ever used the drug before their reference date. If the answer was ‘‘no’’, the participant was classified as a nonuser. If the answer was ‘‘yes’’, the participant was Materials and Methods asked if he or she had ever taken the drug two or more times a week for one month or longer. If the answer was Study Population. The details of the study popula- no, the participant was classified as an ‘‘irregular user’’. tion and study design have been described elsewhere Otherwise, the participant was defined as a ‘‘regular (16-21). Basically, incident cancer cases for this study user’’ and was asked further about ages at first and last were identified by the University of Southern Califor- use, duration of use, usual frequency and dosage of use, nia Cancer Surveillance Program. This cancer registry and the primary reason for each use. We also asked the covers the ethnically diverse Los Angeles County participants if they had used any medications that were population of >9.5 million people. Case patients who not on our list and recorded the drug name and details of were newly diagnosed with first incident esophageal use if the participant had used the medication ‘‘regular- adenocarcinoma (GCA; International Classification of ly’’. All of the antacid medications in the study were Disease for Oncology code C15.0-C15.9), gastric cardiac categorized as requiring a prescription or as nonpre- adenocarcinoma (GCA; International Classification of scription (OTC). At the time the study was conducted, Disease for Oncology code C16.0), or distal gastric nonprescription (OTC) antacid drugs were limited to adenocarcinoma (DGA; International Classification of acid neutralizing agents (e.g., Tums, Pepto-Bismol, and Disease for Oncology codes C16.1-C16.6 and C16.8- Maalox); the majority of prescription antacids at that time C16.9) between 1992 and 1997 were identified and were H2-receptor antagonists. Only a very small number contacted for participation. Control participants were of participants reported use of the PPI, which also matched individually to each case patient on sex, race, required a prescription. age (F5 y), and neighborhood of residence. To increase To assess medical history, we provided a list of the statistical power, we sought two control partic- diseases, including heart disease, gastrointestinal tract ipants for each case patient whenever possible. Of the disorders, and diabetes, and asked if the participant 938 case patients interviewed, 523 had one control had any of those conditions diagnosed by a physician participants, 381 had 2 or more control participants, before the reference date. UGI-related conditions that and 34 had no eligible control participants identified we queried included gastric ulcer, duodenal ulcer, for an overall total of 1,356 control participants. An gastritis, hiatal or diaphragmatic hernia, esophagitis, experienced interviewer conducted the in-person inter- Barrett’s esophagus, gastresophageal reflux disease, views at participants’ homes or other convenient excess acid, or gastric hyperacidity. If the response locations. Next-of-kin were interviewed when case was yes to any of the conditions, the participant was patients were unable to be interviewed due to death then asked the age at first diagnosis with the condition. or illness. Although it was not feasible to blind the In addition, participants were asked if they had interviewers to case (or next-of-kin) or control status, experienced various symptoms including gas pain interviewers and study participants were not aware of (in the stomach), sour stomach including acid indiges- the study hypotheses. Written informed consent was tion and regurgitation, heartburn after meals, and obtained from each study participant before interview. trouble swallowing (feeling that solid food was sticking A total of 938 case patients (220 incident EA/277 to their throats as it went down) before their reference GCA/441 DGA; numbers for the three tumor sites are date. If the response was positive, the frequency (daily, shown separately with these abbreviations) and 1,356 weekly, monthly, less than monthly) with which each control participants are included in the current study. symptom occurred and the age that the participant first Next-of-kin interviews accounted for 269 of the 938 experienced the symptom on a regular basis were interviews with case patients (65 EA/85 GCA/119 recorded. DGA). Statistical Analysis. Participants were grouped as Measures. We developed a structured questionnaire never users, irregular users, or regular users of nonpre- specifically for this study that was administered during scription acid neutralizing agents and of prescription the in-person interview. The questionnaires obtained acid suppressive drugs, separately. Among participants data up to a specified reference date, defined as the who reported regular use, duration of use variables were date that was 1 y before the date of diagnosis of the created for nonprescription acid neutralizing agents and case patient; this same reference date was used for for prescription acid suppressive drugs by summing all each case patient’s matched control participants. The durations of use of the same class of medication

Cancer Epidemiol Biomarkers Prev 2009;18(2). February 2009

(prescription or nonprescription) for each person (All Results antacid drugs with any reported use are listed in Appendix A). Categories for duration of regular antacid Demographic characteristics of the study sample have use (prescription or nonprescription) were <1 y, 1 to 3 y, been described previously (17). Briefly, the mean ages of and >3 y of use. Because early symptoms of an UGI cases at the reference date (1 year before diagnosis) were cancer may influence the use of antacid drugs, we have 60.0 years (SD, 9.4) for patients with EA, 59.8 years excluded from the calculation of total duration of use any (SD, 10.2) for patients with GCA, and 59.4 years (SD, drugs that were first taken in the year before a case or 11.5) for DGA patients; control participants were, on control participant’s reference date (for case patients, this average, 58.7 years (SD, 11.5) at their reference dates. therefore excludes drug use that was initiated in the 2 y Among control participants, 74% were male, compared before the date of diagnosis). Additional analyses were with 91% of patients with EA, 83% of patients with GCA, conducted extending this exclusionary period to 3 y and and 59% of the DGA patients. Non-Latino Whites to 5 y, respectively. represented 77% of those with EA, 76% of those with Polychotomous logistic regression was used to GCA, and 30% of those with DGA. For all three tumor compute the odds ratios (OR), as estimates of the sites, case patients tended to have lower education and a relative risk, and corresponding 95% confidence inter- higher percentage were current smokers than control vals (CI) for adenocarcinomas of the esophagus, gastric participants. Forty percent of case patients reported a cardia, and distal stomach simultaneously in relation history of UGI disorders diagnosed by physicians, to use of nonprescription acid neutralizing agents and compared with 27% of control participants. These proportions became larger when self-reported UGI to use of prescription acid suppressive drugs. In a symptoms were considered. Participants who reported previous study (19), this approach provided more UGI symptoms accounted for 56% of case patients precise estimates of the ORs, and the magnitude of and 40% of control participants in this study sample the ORs was consistent with those obtained in separate (Table 1). conditional logistic regression analyses that preserved Regular use of nonprescription acid neutralizing the original case-control match within each cancer site. agents for 1 to 3 years was associated with increased Therefore, for the purpose of maximizing statistical risk of EA (adjusted OR, 3.16; 95% CI, 1.39-7.20) after power, we report results based on unconditional adjusting for prescription acid suppressive drug use polychotomous logistic regression models with adjust- and other confounding factors (Table 2). Among those ments made for the matching variables in this analysis; who took nonprescription acid neutralizing agents for this allowed us to use all control participants in a single >3 years, the OR for EA increased to 6.32 (95% CI, analysis. 3.14-12.69; P <0.01).Withdurationofnonprescrip- Forward stepwise logistic regression was used to trend V tion acid neutralizing agent use in the statistical model, select confounders from a list of variables: age ( 39, use of prescription acid suppressive drugs was un- 40-49, 50-59, 60-69, 70+ y), sex (male/female), race related to risk of EA. No strong effect on risk of (Non-Latino White versus other), birthplace (US born, cardiac or DGAs was observed for duration of use of non-US born), smoking status (never smoker, ex-smoker, either prescription acid suppressive drugs or nonpre- current smoker), body mass index at reference age 2 scription acid neutralizing agents. Conversely, irregular (<25, 25-29.9, 30+ kg/m ), history (no/yes) of UGI use of nonprescription acid neutralizing agents was disorders diagnosed by physician (including gastric associated with lower risk of GCA (adjusted OR, 0.47; ulcer, duodenal ulcer, unspecified type of ulcer, gastritis, 95% CI, 0.30-0.72) and DGAs (adjusted OR, 0.48; 95% hiatal hernia, esophagitis, Barrett’s esophagus, gastreso- CI, 0.34-0.68). phageal reflux disease, excess acid or gastric hyperacid- Because prior diagnosis of UGI disorders was also ity, and other diseases of the stomach). A test for trend strongly associated with increased risk of adenocarci- across ordinal categories of duration of antacid use was nomas for esophagus and gastric cardia (16, 17), and done for each type of cancer. because presence of UGI disorders might influence We constructed a 1 degree of freedom likelihood ratio frequency of antacid drug use, we conducted analyses test to assess homogeneity of trends in duration of separately among participants with and without a nonprescription acid neutralizing agent use among history of physician-diagnosed UGI disorders (Table 3). individuals who had a history of physician-diagnosed After stratification of UGI disorders, we observed a UGI disorders and those who did not. With adjustment modest difference in the dose-response effects for for prescription acid suppressive drug use and other nonprescription drug use between persons with UGI confounding factors, we conducted the analyses sepa- disorders and those without UGI disorders (homoge- rately for each type of cancer using unconditional logistic neity of trends P = 0.07), although both trends for P regression comparing a multivariable model that fit two duration of use were statistically significant ( trend trend variables (one for each category of UGI disorder <0.01). The increase in EA risk among long-term history) with a multivariable model that fit a single trend (>3 year) users of nonprescription acid neutralizing variable for all participants. The P values reported for agents was limited to participants without a history of trend tests and for the test for homogeneity of trends are physician-diagnosed UGI disorders (adjusted OR, two sided. 10.89; 95% CI, 4.73-25.05). Although the adjusted ORs For validity purposes, we repeated all statistical for 1 to 3 years and >3 years of nonprescription antacid analyses excluding data collected from next-of-kin use among participants with a positive history of participants. Risk estimates were not materially different physician-diagnosed UGI disorders were also >1, the from the results based on all participants combined CIs were wide and did not exclude 1.0 (adjusted OR, (i.e., self-respondents and next-of-kin participants). 2.01; 95% CI, 0.35-11.40 and adjusted OR, 3.19; 95% CI,

Cancer Epidemiol Biomarkers Prev 2009;18(2). February 2009

Table 1. Demographic characteristics of esophageal and gastric adenocarcinoma case patients and control participants, Los Angeles County

Control (n = 1,356) Esophageal (n = 220) Gastric cardia (n = 277) Gastric distal (n = 441) Age (y) V39 98 (7.2) 9 (4.1) 13 (4.7) 29 (6.6) 40-49 194 (14.3) 14 (6.4) 23 (8.3) 50 (11.3) 50-59 345 (25.4) 56 (25.5) 73 (26.4) 91 (20.6) 60-69 463 (34.1) 95 (43.2) 108 (39.0) 153 (34.7) 70+ 256 (18.9) 46 (20.9) 60 (21.7) 118 (26.7) Sex Men 999 (73.7) 200 (90.9) 231 (83.4) 260 (59.0) Women 357 (26.3) 20 (9.1) 46 (16.6) 181 (41.0) Race Non-Hispanic White 841 (62.0) 171 (77.3) 210 (75.8) 133 (30.2) Other 515 (38.0) 49 (22.7) 67 (24.2) 308 (69.8) Education 30) 218 (16.1) 55 (25.0) 68 (24.6) 78 (17.7) Physician-diagnosed UGI disorder history* Yes 376 (27.33) 104 (47.27) 118 (42.60) 157 (35.60) No 976 (71.98) 111 (50.45) 154 (55.60) 267 (60.54) Self-report UGI symptoms* Yes 546 (40.27) 151 (68.64) 168 (60.65) 204 (46.26) No 316 (23.30) 35 (15.91) 77 (27.80) 147 (33.33) Nonsteroid anti-inflammatory drug use Never use 160 (11.80) 36 (16.44) 50 (18.18) 145 (33.03) No regular use 767 (56.56) 119 (54.34) 126 (45.82) 208 (47.38) <5 y 219 (16.15) 38 (17.35) 55 (20.00) 52 (11.85) z5 y 210 (15.49) 26 (11.87) 44 (16.00) 34 (7.74)

*Percentage of missing data not shown.

0.62-16.48). No significant increases in risk or dose- ed with reduced ORs for GCA and DGAs, but the risk response effects were observed for adenocarcinomas of reductions were statistically significant for persons the gastric cardia or distal stomach. Irregular use of who had no history of physician-diagnosed UGI nonprescription acid neutralizing agents was associat- disorders.

Table 2. Multivariable ORs and 95% CIs for duration of antacid use in relation to esophageal, gastric cardia, and distal gastric adenocarcinomas

Control Adenocarcinoma Adenocarcinoma Adenocarcinoma of esophagus of gastric cardia of distal stomach

nnOR (95%CI) n OR (95%CI) n OR (95%CI) Nonprescription antacid use Never use 139 13 1.0 - 40 1.0 - 91 1.0 - Irregular use 893 69 0.89 (0.47-1.71) 116 0.47 (0.30-0.72) 213 0.48 (0.34-0.68) <1 y 80 11 1.48 (0.59-3.73) 21 0.82 (0.42-1.59) 32 0.62 (0.34-1.12) 1-3 y 77 20 3.16 (1.39-7.2) 19 0.82 (0.41-1.61) 32 0.78 (0.45-1.37) >3 y 163 103 6.32 (3.14-12.69) 78 1.28 (0.76-2.18) 59 0.71 (0.43-1.18) P trend <0.01 0.07 0.61 Prescription antacid use Never use 1,137 146 1.0 - 211 1.0 - 326 1.0 - Irregular use 54 14 1.24 (0.6-2.58) 12 0.74 (0.36-1.51) 14 0.89 (0.44-1.79) <1 y 49 9 0.77 (0.34-1.75) 10 0.65 (0.31-1.37) 27 1.75 (1.00-3.08) 1-3 y 44 20 1.72 (0.88-3.35) 11 0.70 (0.32-1.49) 28 1.76 (0.99-3.13) >3 y 45 21 1.27 (0.65-2.51) 23 1.29 (0.70-2.36) 16 1.15 (0.58-2.29) P trend 0.17 0.18 0.17

NOTE: Model includes both nonprescription and prescription antacid use and, additionally, is adjusted for age, gender, race, body mass index, smoke, and history of UGI disorders.

Cancer Epidemiol Biomarkers Prev 2009;18(2). February 2009

All of the results presented in the current study

exclude exposures that were initiated during the 1 year . interval before each participant’s reference date. We also 0.38 evaluated exposure using a longer lag time, excluding drugs that were first used 2 years before the reference OR (95%CI) date (3 years before diagnosis for a case) and 5 years UGI + before the reference date. The results of the 1, 2, and 5 year lags did not differ in any meaningful way (results for the 2 and 5 year lag times not shown). Similarly, control 0.33

restricting the analyses to self-reported data only, that Distal is, excluding data collected through next-of-kin interviews, did not differ from those presented, which used 0.65

all participant data, although among those who partic- À

ipated in the interview directly, the ORs increased OR (95%CI) Case/

slightly when next-of-kin interviews were excluded (data UGI not shown). control Discussion

In this large population-based case-control study, risk of

EA increased with increasing duration of regular use of 0.30 OTC acid neutralizing agents but not with increasing duration of regular use of prescription acid suppressive OR (95%CI) Case/ drugs. No association with either type of drug was UGI + observed for adenocarcinomas of the gastric cardia or distal stomach. The substantially lower risks of GCA and DGAs observed among persons who used acid neutral- control 0.82

izing agents irregularly was not explained by the number Cardia of such drugs that they used irregularly (data not shown). 0.08

Our findings of increased EA risk among regular users À

of nonprescription acid neutralizing agents and reduced OR (95%CI) Case/

risk of adenocarcinomas of the gastric cardia and distal UGI stomach among irregular users of OTC acid neutralizing agents were restricted to persons who did not report

any physician-diagnosed UGI disorders. None of the control tests for homogeneity of trends in duration of regular use comparing those with and without UGI disorders were statistically significant; these analyses excluded persons with irregular use. We did not observe that persons with UGI disorders had greater risk of EA with increasing <0.01

duration of risk, which would have suggested that our OR (95%CI) Case/ observed association reflected confounding by indica- UGI + tion. Nevertheless, it is likely that persons who did not have a physician-diagnosed UGI disorder were taking

nonprescription acid neutralizing agents for various UGI control

symptoms. 0.07 In the absence of acid neutralization or acid Esophageal ) Model adjusted for age, gender, race, body mass index, smoke, nonsteroidal anti-inflammatory drug use, and prescription acid suppressive drug use

suppression, the degree of esophageal damage has b been found to correlate with the degree of esophageal <0.01

acid exposure through gastro-esophageal reflux À

(GERD) with esophageal 24-hour pH monitoring in OR (95%CI) Case/

humans (22, 23). However, gastric acid is not the only UGI substance that refluxes from the stomach into the esophagus. Animal studies have suggested that expo- 3/34 1.11 (0.27-4.52) 8/46 1.19 (0.19-7.26) 4/34 0.45 (0.14-1.40) 16/46 1.52 (0.33-6.92) 8/24 0.55 (0.22-1.37) 22/46 0.73 (0.22-2.38) 7/29 2.09 (0.64-6.83) 13/48 2.01 (0.35-11.4) 10/29 1.27 (0.54-2.99) 9/48 0.71 (0.14-3.55) 15/29 1.49 (0.70-3.18) 17/48 0.52 (0.16-1.71) 11/127 1.0- 2/12 1.0- 35/127 1.0- 4/12 1.0- 75/127 1.0- 11/12 1.0- 40/48 10.89 (4.73-25.05) 60/114 3.19 (0.62-16.48) 24/48 1.61 (0.82-3.18) 54/114 1.60 (0.38-6.71) 14/48 1.09 (0.51-2.36) 42/114 0.59 (0.20-1.79) Case/

sure to bile acid, a major component of duodenal control juice that originates in duodenum and may reflux into the esophagus through duodenogastro-esophageal reflux (DGER), increases the risk of EA (24-26). In a rat model, duodenal juice refluxed into esophagus

although DGER induced histologic changes, from ) UGI disorders (including gastric ulcer, duodenal ulcer, unspecified type of ulcer, gastritis, hiatal hernia, esophagitis, Barrett’s esophagus, gastresophageal reflux disease, excess acid or gastric a squamous-cell epithelium into columnar-cell epitheli-

um, in the lining of the distal esophagus (25). for heterogeneity trend Irregular use<1 y 1-3 y P 50/737 0.84 (0.41-1.74) 17/154 0.76 (0.14-4.06) 81/737 0.41 (0.26-0.66) 34/154 0.91 (0.22-3.82) 145/737 0.47 (0.33-0.69) 62/154 0.52 (0.18-1.53) Never use >3 y hyperacidity, and other diseases of stomach). ( Table 3. ORsdiagnosed for UGI disorders OTC acid neutralizingExposure drugs in relation to esophageal, gastric cardia, and distal gastric adenocarcinomas, by history of physician- NOTE: ( Moreover, another animal study showed that the P

Cancer Epidemiol Biomarkers Prev 2009;18(2). February 2009

prevalence of EA increased as the amount of acid limited to long-term current users of H2-recepor gastric juice that was permitted to reflux with duodenal antagonists in the multivariable analysis that included juice into the esophagus decreased (26), suggesting that variables for GERD, hiatal hernia, peptic ulcer, and acid suppression therapy may be detrimental by dyspepsia in the model (12). This study was, however, encouraging esophageal metaplasia and tumorigenesis based on a small number of cases and did not consider in participants with DGER reflux. In humans, it has the use of OTC acid neutralizing agents or acid been shown that the degree of esophageal damage suppressive drugs that were available without pre- increases with increasing amount of DGER, with the scription during the study period. In Sweden, Lagerg- highest DGER levels found in patients with Barrett’s ren et al. (9) reported a nearly 3-fold higher risk of EA esophagus (27). With 24-hour ambulatory esophageal among persons who used medications for reflux pH and bilirubin and gastric pH monitoring, Marshall symptoms at least 5 years before study than among et al. (28) have observed that most DGER takes place at those who took no medication. In a U.S. study, Chow night when the pH is between 4 and 7. Based on et al. (5) observed a 4-fold higher risk of adeno- previous laboratory findings, Todd et al. (27) concluded carcinomas of the esophagus and gastric cardia in that at a pH of 6 or greater, deconjugated bile acids, if individuals who had four or more prescriptions for refluxed into the esophagus, cause not only short-term H2-receptor antagonists, but the association disappeared damage to the esophageal mucosa but also long-term after adjusting for the predisposing conditions. One of damage to the cellular DNA. the drawbacks of that study, as well as other earlier Although acid suppressive drugs in the form of studies (6, 15), is that it did not separate EAs from H2-receptor antagonists or PPIs allow symptom relief adenocarcinomas of the gastric cardia. Farrow and her and promote endoscopic mucosa healing in GERD colleagues (7) investigated the separate associations patients, they do not guarantee normalization of between antacid use and adenocarcinomas of the pH level at the gastro-esophageal junction nor do esophagus, gastric cardia, and distal stomach. In this they prevent GERD or DGER (29). There is some study, long-term use of OTC antacids was associated disagreement about how well OTC acid neutralizing with increased risk of EA, but no statistically significant agents work in changing gastric pH. On the one risk increase was observed for H2-receptor antagonist hand, OTC acid neutralizing agents are believed to users, which is consistent with the results we report cause an immediate short-term increase in the pH, here. which differs from the long-lasting and profound Our study represents one of two large population- effect of H2-receptor antagonists. On the other hand, based epidemiologic studies within the United States OTC acid neutralizing agents are taken at the time that has been designed specifically to investigate the of maximum symptoms when reflux is probably etiologies of these tumor types. One strength of this most intense in the patients, and OTC drugs may be study is the definition of the reference group. Instead of taken at relatively large doses to provide the most combining never users with irregular users as many rapid relief (30, 31), causing the gastric pH temporarily other studies have done, our reference group is com- to increase to levels that are much higher than posed of persons who have never used the antacids of those produced by H2-receptor antagonists (32). Accord- interests. In addition, rather than simply adjusting for ing to the first hypothesis, the observed increase in GERD and other UGI symptoms in the analyses as most risk of EA among regular users of long-term OTC acid of other studies have done, we have evaluated the risk neutralizing agents rather than among H2-receptor estimates for antacid use separately among participants antagonist users in this study would be explained by with and without a history of physician-diagnosed UGI the fact that OTC acid neutralizing agents provide less disorders. Furthermore, our study allows comparison of successful acid suppression than H2-receptor antago- risk estimates between OTC acid neutralizing agents and nists. Alternatively, if the second hypothesis is true, the acid suppressive drugs; the former group has been rarely observed association suggests that OTC acid neutraliz- studied separately. ing agents create a more alkaline condition in the Of interest is the reduced risk of DGA in individuals stomach than H2-receptor antagonists at the time of who irregularly used OTC acid–neutralizing agents. maximum symptoms and reflux, and therefore more After carefully studying the reported OTC acid–neutral- bile reflux may reach the distal esophagus. As we izing agents in this group used irregularly, we noticed discussed above, the direct contact of alkaline duodenal that one of the most frequently used OTC acid– bile that enters the lower esophagus can cause cell neutralizing agents was Pepto-Bismol, which is used to turnover and is potentially carcinogenic (33); further- protect the stomach lining from acid, but also kills more, it may promote progression of Barrett’s esopha- Helicobacter pylori (35). Our analyses comparing irregular gus (29, 34). users of Pepto-Bismol to persons with no OTC antacid Although a number of epidemiologic studies have use found a reduction in risk of distal gastric cancer that investigated the possible positive association between was similar to that observed for other OTC antacid the use of H2-receptor antagonists and the risk of preparations (data not shown). gastric adenocarcinomas (7, 8, 10-12, 14), research to Limitations of this study include the potential determine whether the risk of EA is increased among for recall bias, a modest response rate, and the potential users of H2-receptor antagonists or other antacids has for less accurate reporting of exposures by next-of-kin; been limited. In a recent case-control study, which was these have been described elsewhere (17). Next-of-kin nested within a large prospective study in the United interviews accounted for 29% of the patient reports Kingdom, H2-recepor antagonist and PPI users were in this study. Our results did not seem to be affected both at increased risk of EA; the association was by next-of-kin interviews and were similar when

Cancer Epidemiol Biomarkers Prev 2009;18(2). February 2009

next-of-kin interviews were excluded from the Acknowledgments analyses. Our reference group had limited numbers The costs of publication of this article were defrayed in part by for some analyses because of our restricted definition the payment of page charges. This article must therefore be for the reference group, excluding irregular users; in hereby marked advertisement in accordance with 18 U.S.C. fact, we only had two patients with EA with a history of Section 1734 solely to indicate this fact. UGI disorders who reported no antacid use. This We thank all the study participants for their contributions limited the statistical power in our stratified analyses. and Annie Fung, Isaura Rivera, Timothy Stirton, Chiu-Chen In addition, we have few participants who reported use Tseng, and June Yashiki for their help with data collection and of PPIs because the study was conducted before data management. widespread use of these drugs. Further research is required to fully assess the effect of H2-receptor antagonists and PPIs, which are now available as OTC References drugs and to define the mechanisms by which the 1. Bochner F, Rossi S. Australian Medicines Handbook 2005. Adelaide: Pharmaceutical Society of Australia; 2005. common OTC drugs used by our participants affect 2. Chong E, Ensom MHH. Pharmacogenetics of the proton pump risk of EA. inhibitors: a systematic review. Pharmacotherapy 2003;23:460 – 71. In summary, this study observed increases in risk of 3. Hetzel DJ, Dent J, Reed WD, et al. Healing and relapse of severe EA in persons who used nonprescription acid neutraliz- peptic esophagitis after treatment with omeprazole. Gastroenterolo- gy 1988;95:903 – 12. ing agents regularly. This may be due to confounding by 4. Bernstein L. What are the epidemiologic similarities and differences indication with the conditions for which these drugs are between carcinomas arising in Barrett’s mucosa and in the gastric used etiologically related to disease risk, or the increase cardia? In: Guili RGJ, Jamieson GG, Scarpignato C, editors. Barrett’s in risk associated with OTC drugs may be the result of Esophagus; Columnar Lined Esophagus. Montrouge. France: John- Libbey Eurotext Ltd; 2003. p. 702 – 10. alkaline reflux resulting from the acid neutralization 5. Chow WH, Finkle WD, McLaughlin JK, Frankl H, Ziel HK, Fraumeni entering the esophagus. JF, Jr. The relation of gastroesophageal reflux disease and its treatment to adenocarcinomas of the esophagus and gastric cardia. JAMA 1995;274:474 – 7. 6. Colin-Jones DG, Langman MJ, Lawson DH, Logan RF, Paterson KR, Disclosure of Potential Conflicts of Interest Vessey MP. Post-cimetidine surveillance for up to ten years: incidence of carcinoma of the stomach and oesophagus. Q J Med No potential conflicts of interest were disclosed. 1991;78:13 – 9. 7. Farrow DC, Vaughan TL, Sweeney C, et al. Gastroesophageal reflux disease, use of H2 receptor antagonists, and risk of esophageal and gastric cancer. Cancer Causes Control 2000;11:231 – 8. 8. Johnson AG, Jick SS, Perera DR, Jick H. Histamine-2 receptor Appendix A. Drugs included in this study antagonists and gastric cancer. Epidemiology 1996;7:434 – 6. 9. Lagergren J, Bergstrom R, Lindgren A, Nyren O. Symptomatic OTC acid neutralizing drugs Prescription drugs gastroesophageal reflux as a risk factor for esophageal adenocarci- Alka, mints chewable antacid Axid Pulvules noma. N Engl J Med 1999;340:825 – 31. Alka Seltzer, all types Bentyl 10. Moller H, Lindvig K, Klefter R, Mosbech J, Moller Jensen O. Cancer AlternaGEL Liquid Bentyl W/PB occurrence in a cohort of patients treated with cimetidine. Gut 1989; Aludrox Oral suspension Bicitra 30:1558 – 62. Aluminum hyroxide gel Bismorex 11. Moller H, Nissen A, Mosbech J. Use of cimetidine and other peptic Amitone Espaven ulcer drugs in Denmark 1977 – 1990 with analysis of the risk of gastric Amphojel Mylicon/Mylacon cancer among cimetidine users. Gut 1992;33:1166 – 9. Antacid Mint by Riker Laboratory Pepcid 12. Rodriguez GLA, Lagergren J, Lindblad M. Gastric acid suppression Antacid Peru Prevacid and risk of oesophageal and gastric adenocarcinoma: a nested case Antacid Prilosec or losec control study in the UK. Gut 2006;55:1538 – 44. Bromo seltzer Robinul, robinul forte 13. Beresford J, Colin-Jones DG, Flind AC, et al. Postmarketing Camalox Tagamet surveillance of the safety of cimetidine: 15-year mortality report. Pharmacoepidemiol Drug Saf 1998;7:319 – 22. Citrocarbonate Zantac 14. La Vecchia C, Negri E, D Avanzo B, Franceschi S. Histamine-2- Dicarbosil receptor antagonists and gastric cancer risk. Lancet 1990;336:355 – 7. Di-Gel Antacid/Anti-Gas 15. Suleiman UL, Harrison M, Britton A, McPherson K, Bates T. Eno fruit salts H2-receptor antagonists may increase the risk of cardio-oesophageal Gaviscon adenocarcinoma: a case-control study. Eur J Cancer Prev 2000;9: Gelusil 185 – 91. Glygel powder 16. Wu AH, Tseng CC, Bernstein L. Hiatal hernia, reflux symptoms, Maalox body size, and risk of esophageal and gastric adenocarcinoma. Magonate Cancer 2003;98:940 – 8. Mi-Acid 17. Duan L, Wu AH, Sullivan-Halley J, Bernstein L. Nonsteroidal anti- Mylanta inflammatory drugs and risk of esophageal and gastric adenocarci- Mylicon/Mylacon nomas in Los Angeles County. Cancer Epidemiol Biomarkers Prev Pepto-Bismol 2008;17:126 – 34. Philip’s milk of Magnesia 18. Wu AH, Crabtree JE, Bernstein L, et al. Role of Helicobacter pylori Polycrol CagA+ strains and risk of adenocarcinoma of the stomach and Pure Baking Soda esophagus. Int J Cancer 2003;103:815 – 21. 19. Wu AH, Tseng CC, Hankin J, Bernstein L. Fiber intake and risk of Riopan Antacid adenocarcinomas of the esophagus and stomach. Cancer Causes Rolaids Control 2007;18:713 – 22. Sal de uvas 20. Wu AH, Wan P, Bernstein L. A multiethnic population-based study Tempo of smoking, alcohol and body size and risk of adenocarcinomas of Titralac antacid the stomach and esophagus (United States). Cancer Causes Control Titralac Extra Strength 2001;12:721 – 32. Tums Antacid 21. Wu YL, Fu SL, Zhang YP, Qiao MM, Chen Y. Cyclooxygenase-2 WinGel liquid & tablets inhibitors suppress angiogenesis and growth of gastric cancer xenografts. Biomed Pharmacother 2005;59 Suppl 2:S289 – 92.

Cancer Epidemiol Biomarkers Prev 2009;18(2). February 2009

22. Neumann CS, Cooper BT. 24 hour ambulatory oesophageal pH 29. Chandrasoma PT, Demeester TR. GERD Reflux to Esophageal monitoring in uncomplicated Barrett’s oesophagus. Gut 1994;35: Adenocarcinoma. Amsterdam: Academic Press; 2006. 1352 – 5. 30. Fordtran JS, Collyns JA. Antacid pharmacology in duodenal ulcer. 23. Robertson D, Aldersley M, Shepherd H, Smith CL. Patterns of acid Effect of antacids on postcibal gastric acidity and peptic activity. reflux in complicated oesophagitis. Gut 1987;28:1484 – 8. N Engl J Med 1966;274:921 – 7. 24. Attwood SE, Smyrk TC, DeMeester TR, Mirvish SS, Stein HJ, Hinder 31. Fordtran JS, Morawski SG, Richardson CT. In vivo and in vitro RA. Duodenoesophageal reflux and the development of esophageal evaluation of liquid antacids. N Engl J Med 1973;288:923 – 8. adenocarcinoma in rats. Surgery 1992;111:503 – 10. 32. Engelhardt D, Karl R, Kolb HJ, et al. Comparison between 25. Fein M, Peters JH, Chandrasoma P, et al. Duodenoesophageal reflux cimetidine-pirenzepine and antacids for the prevention of stress induces esophageal adenocarcinoma without exogenous carcinogen. hemorrhage in intensive care patients. A controlled clinical study on J Gastrointest Surg 1998;2:260 – 8. 125 patients. Dtsch Med Wochenschr 1985;110:908 – 14. 26. Ireland AP, Peters JH, Smyrk TC, et al. Gastric juice protects against 33. Stamp DH. Bile acids aided by acid suppression therapy may be the development of esophageal adenocarcinoma in the rat. Ann Surg associated with the development of esophageal cancers in western- 1996;224:358 – 70; discussion 370 – 1. ized societies. Med Hypotheses 2006;66:154 – 7. 27. Todd JA, de Caestecker J, Jankowski J. Gastro-esophageal reflux 34. Stein HJ, Kauer WK, Feussner H, Siewert JR. Bile reflux in benign disease and bile acids. J Pediatr Gastroenterol Nutr 2003;36:172 – 4. and malignant Barrett’s esophagus: effect of medical acid suppres- 28. Marshall RE, Anggiansah A, Owen WA, Owen WJ. The temporal sion and nissen fundoplication. J Gastrointest Surg 1998;2:333 – 41. relationship between oesophageal bile reflux and pH in gastro- 35. H. pylori and Peptic Ulcer. Department of Health and Human oesophageal reflux disease. Eur J Gastroenterol Hepatol 1998;10: Services. NIH Web site. Available from: http://digestive.niddk.nih. 385 – 92. gov/ddiseases/pubs/hpylori/. Accessed July 29, 2008.

Cancer Epidemiol Biomarkers Prev 2009;18(2). February 2009

Lei Duan, Anna H. Wu, Jane Sullivan-Halley, et al.

Cancer Epidemiol Biomarkers Prev 2009;18:526-533.

Updated version Access the most recent version of this article at: http://cebp.aacrjournals.org/content/18/2/526

Cited articles This article cites 31 articles, 6 of which you can access for free at: http://cebp.aacrjournals.org/content/18/2/526.full#ref-list-1

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cebp.aacrjournals.org/content/18/2/526. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.