Author + information
- Received June 3, 1997
- Revision received November 11, 1997
- Accepted December 22, 1997
- Published online March 15, 1998.
- Shimon Braun, MDA,* (, )
- Valentina Boyko, MScB,
- Solomon Behar, MDB,
- Henrietta Reicher-Reiss, MD, FACCB,
- Shlomo Laniado, MD, FACCA,
- Elieser Kaplinsky, MD, FACCB,
- Uri Goldbourt, PhDBC,
- on Behalf of the Benzafibrate Infarction Prevention (BIP) Study Research Group2
- ↵*Dr. Shimon Braun, Department of Cardiology, Tel-Aviv Medical Center, 6 Weizman Street, Tel Aviv 64239 Israel.
Objectives. This analysis sought to estimate the risk ratio for cancer incidence and cancer-related mortality associated with the use of calcium channel blocking agents (CCBs) in a large group of patients with chronic coronary heart disease (CHD).
Background. Recent publications contend that the use of short-acting CCBs may double the risk of cancer incidence and possibly increase mortality in hypertensive patients.
Methods. Cancer incidence data were obtained for 11,575 patients screened for the Bezafibrate Infarction Prevention (BIP) study, one-half of whom were treated at the time of screening with CCBs, over a mean follow-up period of 2.8 years. Cause-specific mortality was available through September 1996 (mean follow-up 5.2 years). The statistical power of detecting an odds ratio ≥1.5 (given the cancer incidence rate of 2.1 in the nonusers of CCBs) was 0.91. The power declined to 0.77, 0.54 and 0.41, with declining odds ratios of 1.4, 1.3 and 1.25, respectively.
Results. Of 246 incident cancer cases, 129 occurred among the users (2.3%) and 117 among nonusers of CCBs (2.1%). After adjustment for age, gender and smoking, the odds ratio estimates for all cancers combined was 1.07 (95% confidence interval [CI] 0.83 to 1.37) for CCB users relative to nonusers. The adjusted risk ratio for all-cause mortality for age, gender and smoking and pertinent prognostic clinical characteristics was estimated at 0.94 (95% CI 0.85 to 1.04). The adjusted risk ratio for cancer-related mortality was 1.03 (95% CI 0.75 to 1.41).
Conclusions. Patients with CHD treated with CCBs exhibited a similar risk of cancer incidence and total and cancer-related mortality compared with nonusers of CCBs. This analysis provides a certain assurance that CCB use in middle-aged and elderly patients with CHD is not associated with a meaningful difference in cancer incidence and related mortality.
In two recent reports, Pahor et al. [1, 2]raised the possibility that calcium channel blocking agents (CCBs) may increase the risk of developing cancer. They presented an analysis of participants of the Established Population of Epidemiological Studies of the Elderly (EPESE). An increased risk of cancer associated with the use of CCB therapy postulated by those investigators provoked apprehension among many physicians and patients. Because long-term safety data in patients with coronary heart disease (CHD) are lacking for most CCBs, the aim of the present analysis was to compare the risk of cancer incidence and cancer-related mortality between users and nonusers of CCBs in a large sample of patients with chronic CHD, 70% of whom had had at least one myocardial infarction. Data were collected for patients screened for participation in the Bezafibrate Infarction Prevention (BIP) study .
Between February 1, 1990 and October 30, 1992, clinical and biochemical data for >20,000 male and female patients 45 to 74 years old with suspected CHD were recorded in the logbooks of 18 departments of cardiology in Israel. Patients with an established diagnosis of chronic CHD (n = 15,502) were screened for inclusion in the BIP study, and they comprised the population in the BIP registry. The institutional committees on human research approved the study protocol. The study was conducted according to the principles of the Declaration of Helsinki, and written informed consent was obtained for all patients.
All patients screened for the BIP study underwent a complete medical examination and a biochemical blood test. Detailed medical history and use of medication data were recorded. The diagnosis of CHD was based on documented myocardial infarction or typical angina pectoris (within 2 years preceding the screening) accompanied by a positive exercise test, a positive radionuclear study or at least 60% stenosis of one major coronary artery. The major exclusion criteria were an acute coronary event or coronary intervention within 6 months before screening, severe congestive heart failure, cancer or chronic liver or renal disease. The complete list has been published elsewhere .
The current analysis was restricted to patients who had been screened but not included in the BIP study (n = 11,575). We linked this cohort to the National Cancer Registry and National Death Registry, using the unique identification number of each Israeli resident. Each matched record was checked for correct identification by a computer program that provided a core for the certainty of match. We verified the identification numbers of the entire cohort with the National Population Registry. Death certificate diagnoses were coded using the system described in the ninth edition of the International Classification of Disease (ICD-9-CM) in which cancer is denoted by codes 140 to 209. We classified patients according to the underlying cause of death.
During the follow-up period, for cancer incidence, ending December 31, 1993 (mean 34 months, range 14 to 46) 246 new cases of cancer were identified. We excluded 420 patients with a diagnosis of cancer dated before the screening visit from the analysis of cancer incidence and cancer-related mortality. The follow-up period for death ended on September 30, 1996 (mean 5.2 years, range 4.0 to 6.7). Total mortality rates for a follow-up period of 3.2 years have been reported previously . We excluded 805 patients because we could not verify their vital status from the analysis of total mortality. The proportions of CCB users among the 11,575 patients with known vital status and the 805 for whom it was not known was 50% and 47%, respectively. The mean age was 59.8 and 60.6 years, and 78% and 74% were men in these two groups, respectively. The frequency of previous myocardial infarction (71%) and current smoking (11%) was equal in the two groups.
1.2 Statistical Analysis
The pooled risk ratio of cancer incidence over 2.8 years was calculated by the Cochran-Mantel-Haenszel method . The strata used for risk ratio estimates were three age groups (<55, 55 to 64 and >65 years), gender and smoking (current, past or never). The odds ratio for cancer incidence controlling for potential confounding variables was also examined using logistic regression. The Cox proportional hazards model was used to control for potential confounding baseline clinical characteristics when calculating the risk ratio estimates and confidence intervals during the longer follow-up available for total and cancer-related mortality. Cumulative cancer-related mortality rates in CCB users and nonusers, adjusted for age and smoking habits by the same method, were plotted.
Assuming that a clinically meaningful odds ratio of cancer incidence between users and nonusers of CCBs is ≥1.5, the statistical power of detecting an odds ratio of at least this magnitude (given the cancer incidence rate of 2.1 in the nonusers of CCBs) over the given follow-up period was 0.91. The statistical power declined to 0.77, 0.54 and 0.41, with declining odds ratios of 1.4, 1.3 and 1.25, respectively.
At the screening visit, 5,843 patients were treated with either nifedipine, diltiazem or verapamil. The reference group consisted of 5,732 patients who did not receive a CCB. The clinical characteristics of both groups have been previously reported . We found concurrent use of beta-adrenergic blocking agents in 29% and 39% and angiotensin-converting enzyme (ACE) inhibitors in 9% and 12% of CCB users and nonusers, respectively. Diuretic drugs (16%) were reportedly used with equal frequency in both groups. After exclusion of 420 patients with a diagnosis of cancer before the screening visit, 5,611 cancer-free patients reported CCB treatment and 5,543 did not.
We analyzed a total of 246 cases of cancer in both groups (129 cases in users, 117 in nonusers of CCBs). Table 1shows the most frequent cancer sites.
After stratification within age, gender and smoking, the pooled risk ratio estimate for all cancers combined was 1.07 (95% confidence interval [CI] 0.83 to 1.37) for users relative to nonusers of CCBs. In a similar analysis applying the logistic regression and adjusting for exact age, gender and current and past smoking, the estimated adjusted risk was virtually identical: 1.06 (95% CI 0.82 to 1.37). The risk ratio estimates for the specific CCBs are shown in Table 2. Further stratification within additional baseline characteristics, including age, gender, past myocardial infarction, hypertension, New York Heart Association functional class, peripheral vascular disease, chronic obstructive pulmonary disease, diabetes mellitus and smoking, did not change the risk ratio estimates meaningfully, reducing it slightly to 1.00 (95% CI 0.76 to 1.32). Adding the use of diuretic drugs, beta-blockade and ACE inhibitors in the stratifying variables also did not change the risk ratio estimates, yielding 1.04 (95% CI 0.81 to 1.34).
Overall mortality analysis revealed 840 deaths (14.4%) in the CCB users compared with 730 in the nonusers (12.7%). On multivariate adjustment for the differences between the groups, the adjusted risk ratio was estimated at 0.94 (95% CI 0.85 to 1.04). Adjustment was made by the Cox proportional hazard model by age, gender, current and past smoking, diabetes, past myocardial infarction, angina pectoris, functional class, hypertension, peripheral vascular disease and chronic obstructive pulmonary disease, all of which were significant predictors of all-cause mortality in these patients. The adjusted risk ratio estimates for the specific CCBs are shown in Table 3.
Among patients free of cancer at the screening visit, there were 83 deaths (1.5%) due to cancer in the CCB users, compared with 75 in the nonusers (1.4%). Fig. 1shows similar curves for cancer-related mortality among users and nonusers of CCBs. The risk ratio estimate for cancer mortality, adjusted for age, gender and smoking (the only variables closely associated with cancer mortality) was 1.03 (95% CI 0.75 to 1.41). The adjusted risk ratio estimates for specific CCBs are shown in Table 4.
CCBs are among the most commonly used drugs for patients with cardiovascular disease. One-half of the patients screened for the BIP study were treated by one of the three short-acting CCBs approved in Israel for the treatment of angina pectoris or hypertension. We found no increased risk for all cancers among CCB users in relation to nonusers. Similar findings were reported recently by Jick et al. , who found no evidence of a material increase in the risk of any cancer causally associated with use of CCBs relative to use of beta-blockers. The risk of cancer was unrelated to the duration of CCB use. The cancer-related mortality data over an average period of 5.2 years in our study showed a minor mortality difference between users and nonusers of CCBs, which could easily be related to sampling variation. Because of a different prevalence of clinical characteristics between the groups, an analysis was performed to adjust statistically for characteristics that might potentially be linked to cancer and confound the association. This analysis is consistent with similar risks of cancer-related mortality in CCB users and nonusers.
3.1 Comparison with other Studies
In a recent investigation of 5,052 EPESE patients, Pahor et al. reported that CCB users exhibited a 1.42-fold increase in cancer incidence over nonusers. CCB users differed to a great extent from those who were not using them. Two dramatic differences were the prevalence of CHD in 71% and heart failure in 29% among the users compared with only 22% and 10%, respectively, among the nonusers. Disease severity was not recorded. There was also a sizable excess of diabetes prevalence among users of CCBs. Concomitant use of nitrates, diuretic drugs, digoxin, beta-blockade and ACE inhibitors among users, in correspondence to the large excess of CHD, was also highly increased. Physical disability was recorded for 63% of CCB users and 45% of nonusers. Multivariate analysis, adjusted for age, gender, ethnicity, current smoking and drinking, the presence of diagnosed heart failure and number of hospital admissions, increased the estimated relative cancer incidence risk to 1.72. Pahor et al. failed to adjust for lifetime smoking and drinking habits, two potential harbingers of cancer in these elderly patients. These and other vital exposures over time may well have been related to both CHD (hence, a higher likelihood of CCB prescription) and subsequent cancer. Our cohort included only patients with CHD who were considerably younger than the subjects in the EPESE cohort. Our analysis, including an adjustment encompassing past smoking, failed to indict CCBs as increasing the risk of cancer incidence and cancer-related mortality. In the study by Jick et al. , after adjustment for smoking body mass index, change in medication and duration and dose of CCBs, the absence of a duration effect renders a causal explanation of a small association of cancer risk with CCBs unlikely.
A significant point in assessing the safety of a drug by comparing the statistics of subsequent events or death is the “bias by indication,” in that doctors prescribe medication to a patient on the basis of a decision that is not fully documented in terms of all clinical consideration and pertinent variables. Users and nonusers of a certain drug are therefore certain to differ in many important ways. Attempts to compare their fate by observation serves as a preliminary substitute to the appropriate examination of safety through randomized trials and should be conducted and interpreted with the utmost caution. Thus, the postulated hypothesis that inhibition of apoptosis by CCBs may promote cancer remains hypothetical and speculative.
3.2 Study Limitations
There are several potential limitations of cohort studies. Similar to other studies, we conducted an observational analysis in groups with different characteristics and adjusted statistically for the differences, rather than judging experimentally an effect of medication on morbidity and mortality in a randomized clinical trial. In the present study, we relied on a single report of therapy for each patient made available to us during a screening examination. Therapy may undergo several changes within a given follow-up period and is a shortcoming of most observational studies, including recent ones [1, 2, 10]. Although we do not know who of the actual patients in the cohort presently analyzed discontinued CCBs or, conversely, was not taking CCBs initially but started therapy during follow-up, we have, unlike other studies, an indirect estimate for the extent of such changes. These estimates were derived from the experience of the patients participating in the clinical trial (BIP) proper, not included in this cohort. Among 1,593 patients taking CCBs at the beginning of the trial, 83.9% of those alive and evaluated 1 year later (n = 1,544) were still taking CCBs. This proportion dropped to 70.9% after 4 years. Conversely, in 1,529 patients not using CCBs at baseline, 8.8% of those surviving and evaluated 1 year later (n = 1,474) started CCB therapy within the first year, increasing to 18.2% within 4 years.
The wide confidence intervals, resulting from the limited number of cancer incidence cases and cancer deaths, provides (given the low rates) no grounds for labeling CCBs as promoting cancer in patients with CHD. We are proceeding to follow up this cohort to obtain more stable estimates of the risk of cancer incidence and mortality and to learn more about long-term associations.
These limitations notwithstanding, the results were derived from a large number of patients with chronic CHD, after adjusting for comorbidity, disease severity and past smoking habits, potentially related to the likelihood of cancer. Thus, the analysis is of significance in assessing the safety of CCB use in patients with CHD until decisive results from ongoing randomized clinical trials are available.
Patients with CHD treated with CCBs exhibited a similar risk for the incidence of cancer, overall mortality and cancer-related mortality as did nonusers of CCBs. Our analysis provides a certain assurance that CCB use in middle-aged and elderly patients with CHD is not associated with a meaningful difference in mortality and cancer incidence.
A.1 Participating Centers and Investigators for the Benzafibrate Infarction Prevention Study Group
Assaf Harofe Hospital, Zrifin:Zwi Schlesinger, MD, Aharon Fridensohn, MD. Barzilai Medical Center, Ashkelon:Leonardo Reisin, MD, Jamal Jafari, MD. Beilinson Medical Center, Petach Tikva:Samuel Sclarovsky, MD, Yaakov Friedman, MD, Bruno Ostfeld, MD. Bnei-Zion Hospital, Haifa:Edward Abinader, MD, Shmuel Rochfleish, MD. Carmel Hospital, Haifa:Abraham Palant, MD, Hanan Schneider, MD. Central HaEmek Hospital, Afula:Tiberio Rosenfeld, MD, Suleiman Khalid, MD. Edith Wolfson Medical Center, Holon:Yehezkiel Kishon, MD Rene Rotzak, MD. Hasharon Hospital, Petach-Tikva:Izhar Zahavi, MD, Janash Vitrai, MD. Hillel Yaffe Hospital, Hadera:Benyamin Pelled, MD, Joseph Pardu, MD. Tel Aviv Medical Center, Tel-Aviv:Shlomo Laniado, MD, Libi Sherf, MD, Shimon Braun, MD, Yemima Eschar, MD. Kaplan Hospital, Rehovot:Avraham Caspi, MD, Alexander Arditi, MD, Shulamit Botwin, MD. Meir Hospital, Sapir Medical Center, Kfar Saba:Daniel David, MD, Daniel Weisnenberg, MD; Naharia Hospital, Naharia:Nathan Roguin, MD, Alicia Glusman, MD; Rambam Medical Center, Haifa:Walter Markiewicz, MD, Diav Motlak, MD; Rivka Ziv Hospital, Tzfad:Alon Marmour, MD, Michael Flich, MD; Shaare Zedek Medical Center, Jerusalem:Monty Zion, MD, Jonathan Balkin, MD; Sheba Medical Center Heart Institute, Tel Hashomer:Babeth Rabinowitz, MD, Eddy Barasch, MD; Soroka Medical Center, Be’er Sheba:Natalio Kristal, MD, Noa Liel, MD.
- angiotensin-converting enzyme
- Bezafibrate Infarction Prevention study
- calcium channel blocking agent
- coronary heart disease
- confidence interval
- Established Population of Epidemiological Studies of the Elderly
- Received June 3, 1997.
- Revision received November 11, 1997.
- Accepted December 22, 1997.
- The American College of Cardiology
- Pahor M,
- Guralnik JM,
- Salive ME,
- Corti MC,
- Carbonin P,
- Havlik RJ
- Goldbourt U,
- Behar S,
- Reicher-Reiss H,
- et al.
- Braun S,
- Boyko V,
- Behar S,
- et al.
- Mantel S,
- Haenszel W
- Cox DR
- Daling JR
- Lenfant C
- Koenig W,
- Lowel H,
- Lewis M,
- Hormann A