Fax: (011) +49-6221-421302
Beta blocker use and colorectal cancer risk†
Population-based case-control study
Version of Record online: 14 MAY 2012
Copyright © 2012 American Cancer Society
Volume 118, Issue 16, pages 3911–3919, 15 August 2012
How to Cite
Jansen, L., Below, J., Chang-Claude, J., Brenner, H. and Hoffmeister, M. (2012), Beta blocker use and colorectal cancer risk. Cancer, 118: 3911–3919. doi: 10.1002/cncr.26727
We thank Ute Handte-Daub for her excellent technical assistance; the study participants; the interviewers who collected the baseline data; the Cancer Registry of Rhineland-Palatinate; and the following hospitals that recruited patients for this study: Chirurgische Universitätsklinik Heidelberg, Klinik am Gesundbrunnen Heilbronn, St. Vincentiuskrankenhaus Speyer, St Josefskrankenhaus Heidelberg, Chirurgische Universitätsklinik Mannheim, Diakonissenkrankenhaus Speyer, Krankenhaus Salem Heidelberg, Kreiskrankenhaus Schwetzingen, St Marien- und St. Annastiftkrankenhaus Ludwigshafen, Klinikum Ludwigshafen, Stadtklinik Frankenthal, Diakoniekrankenhaus Mannheim, Kreiskrankenhaus Sinsheim, Klinikum am Plattenwald Bad Friedrichshall, Kreiskrankenhaus Weinheim, Kreiskrankenhaus Eberbach, Kreiskrankenhaus Buchen, Kreiskrankenhaus Mosbach, Enddarmzentrum Mannheim, Kreiskrankenhaus Brackenheim.
- Issue online: 3 AUG 2012
- Version of Record online: 14 MAY 2012
- Manuscript Accepted: 7 NOV 2011
- Manuscript Revised: 11 OCT 2011
- Manuscript Received: 26 AUG 2011
- colorectal cancer;
- antihypertensive agents;
- beta blocker;
- case-control studies
Recently, it has been postulated that long-term use of beta blockers might decrease the risk of certain types of cancer because of weakening of norepinephrine signaling. Previous studies on colorectal cancer (CRC) yielded inconsistent results, but lacked information on covariates. Thus, the authors investigated the association of beta blocker use and CRC risk in a large population-based case-control study (DACHS study).
Between 2003 and 2007, information on beta blocker use and potential confounders was collected by personal interviews for 1762 CRC cases and 1708 control individuals from Germany. The association of CRC risk and beta blocker use and subclasses of beta blockers was estimated by multiple logistic regression. In addition, site- and stage-specific analyses were performed.
After adjustment for covariates, no association was observed with beta blocker use (odds ratio [OR], 1.05; 95% confidence interval [CI], 0.86-1.29) or with duration of beta blocker use. Also, the analysis by subclasses of beta blockers (cardioselectivity) and active ingredients (metoprolol, bisoprolol, carvedilol, and atenolol) or by CRC subsite showed no associations. In stage-specific analyses, long-term beta blocker use (6+ years) was associated with a significantly higher risk of stage IV CRC (OR, 2.02; 95% CI, 1.25-3.27).
Our adjusted results do not support the hypothesis that beta blocker use is associated with decreased risk of CRC. In contrast, we found a positive association of long-term beta blocker use and risk of stage IV CRC. The latter result should be further evaluated in future studies. Cancer 2012.© 2012 American Cancer Society.
For >3 decades, the association between cancer risk and the use of antihypertensive drugs has been discussed.1 As many older adults use these drugs regularly, an association with cancer risk or incidence would exhibit a significant public health issue. Antihypertensive drugs like angiotensin-converting enzyme inhibitors,2 beta blockers,3, 4 calcium channel blockers,5, 6 diuretics,5, 7 and rauwolfia derivates8 have at least once been reported to be associated with an increased risk of cancer.
Notwithstanding these results, it has recently been hypothesized that the use of beta and alpha blockers may reduce the risk of different types of cancer, as they weaken norepinephrine signaling systematically.9, 10 This hypothesis is based on animal and cell line studies suggesting that norepinephrine is an etiological factor in some cancer types, such as breast or colon cancer. Supporting this hypothesis, studies have shown that use of the norepinephrine level-lowering alpha blockers was associated with a decreased risk of prostate cancer11 and bladder cancer12 and that use of beta blockers was associated with a decreased risk of prostate cancer13 and all cancers combined.14 As these studies were based on prescription databases, they lacked detailed adjustment for confounding by potentially associated cancer risk factors, such as obesity and smoking, and protective factors, such as participation in screening programs, which may all be associated with the use of beta blockers. A recently published meta-analysis including only randomized trials to minimize the effect of confounding did not find any association between beta blocker use and cancer or cancer-related death,15 but no cancer-specific results were reported and follow-up times were rather short.
With respect to colorectal cancer (CRC) risk, 1 study reported a risk reduction of 21% when comparing beta blocker users with users of diuretics.16 Another large study that screened pharmaceuticals for possible carcinogenic effects found no association of colon or rectal cancer risk with the use of beta blockers.17, 18 But again, detailed adjustment for confounding was not possible, as both studies were based on prescription databases.
To our knowledge, no large study including an assessment of long-term use and a detailed adjustment for relevant confounding variables has investigated the association of beta blockers and CRC risk. Thus, the aim of this study was to investigate use of beta blockers with respect to risk of CRC in a large population-based case-control study with detailed assessment of putative and established CRC risk and preventive factors including medication use.
MATERIALS AND METHODS
Study Design and Study Population
The investigation was based on data from the DACHS study, a population-based case-control study in the Rhine-Neckar-Odenwald region in Germany. Cases were patients with a first histologically confirmed diagnosis of CRC (International Classification of Diseases and Related Health Problems, 10th Edition codes C18-C20) between 2003 and 2007. All 22 hospitals in the study area in which patients with CRC receive treatment were involved in recruitment of the cases. Control individuals were randomly selected from lists of population registries, frequency-matched to cases by 5-year age groups, sex, and county of residence, and contacted by mail and follow-up calls. The recruited patients constitute about 50% of all eligible patients in the study area, and the participation rate among eligible controls was 50.3%. Details of the study design have been reported elsewhere.19 In brief, cases and controls were eligible for study participation if they were at least 30 years of age, German speaking, and mentally and physically able to participate in a personal interview of about 1 hour. Controls with a history of CRC were excluded.
Eligible patients were informed about the study by their clinicians, usually a few days after surgery, and the study center was notified upon receipt of informed consent. In addition, patients who could not be recruited during their in-patient stay were contacted by mail after discharge by clinicians and clinical cancer registries. Controls were contacted by the study center through mail and follow-up calls to ask for participation. The study was approved by the ethics committee of the University of Heidelberg and the state medical boards of Baden-Wuerttemberg and Rhineland-Palatinate. All participants gave written informed consent.
Information on sociodemographic factors and a detailed medical and lifestyle history with regard to known or suspected protective factors or risk factors of CRC were collected by trained interviewers using a standardized questionnaire. Interviews with patients were either conducted during hospitalization or, if patients had already left hospital, at their homes. Controls were generally interviewed at their homes. Information on risk or protective factors was collected for the time prior and up to the index date, which was the date of diagnosis for cases and the date of interview for controls.
Assessment of Medication Use
Information on current regular use more than once per week of drugs (name of the drug, year of initiation) was collected for a variety of indications, including cardiovascular disease and hypertension. If the participant was unsure about the name of the drug or about the indication, a list with the most common drug names for each of the indications was provided. Currently and regularly used drugs that did not match any of the predefined indications in the questionnaire were also assessed. Drugs were coded according to the Anatomical Therapeutic Chemical/Defined Daily Dose classification.20
Participants were classified as current regular users of beta blockers if they reported the use of beta blockers (Anatomical Therapeutic Chemical code starting with C07) in any medication field of the questionnaire. If information on drugs for cardiovascular disease (CVD) or hypertension was missing, participants were excluded from the analysis. As the year of initiation was collected only, we assumed that the use of medication started at the middle of the year (July 1) to compute the duration of use as the difference between this date and the index date. Participants with missing information on duration or with duration of use of <2 years were excluded to minimize bias by drug use that was initiated in the context of symptoms possibly related to CRC. As dose and frequency of beta blocker use were not systematically assessed, we were not able to perform analyses on dose-response relations.
The distribution of potential protective factors and risk factors was compared between cases and controls. The use of beta blockers was described with respect to the cardioselectivity of the beta blocker and the active ingredient. The association of regular use of beta blockers and duration (no use, 2-5 years, 6-10 years, 11+ years) of beta blocker use with CRC risk was estimated with 2 unconditional multiple logistic regression models. In the first model (basic model), the association was estimated with adjustment for the matching variables age, sex, and county of residence only. The second model (fully adjusted model) included additional adjustment for known or suspected protective factors or risk factors of CRC: years of schooling (<10, 10-11, 12-13), body mass index computed from self-reported weight and height of the participants (<25, 25-29.9, 30+ kg/m2), lifetime pack-years of active smoking (never active, <10, 10-19, 20-29, 30+), alcohol consumption (quartiles of lifetime ethanol in grams per day defined by the distribution among controls), physical activity (quartiles of lifetime metabolic equivalents in hours per week defined by the distribution among controls), hypertension (yes/no), CVD (yes/no), history of CRC in a first-degree relative (yes/no), former colorectal endoscopy (yes/no), regular use (at least twice per week for at least 1 year) of nonsteroidal anti-inflammatory drugs (NSAIDs) including aspirin (ever/never), current regular use (at least twice per week) of statins (yes/no), and use of hormone replacement therapy (ever/never).
These analyses were repeated for subclasses of beta blockers according to their cardioselectivity and according to their active ingredient (metoprolol, bisoprolol, carvedilol, and atenolol). For these specific analyses, only results for the fully adjusted models are reported, and the duration of use was categorized in 3 groups (no use, 2-5 years, 6+ years). Participants who used other beta blockers than those included in the subclass of interest were excluded from the respective analyses. To investigate potential site- and stage-specific associations, analyses were repeated for the subgroups colon and rectal cancer, proximal (cecum to transverse colon) and distal (left flexure to rectum) CRC, and stage I to stage IV CRC. Analyses were also repeated separately for men and women to investigate potential effect modification by sex. In sensitivity analysis, analyses were repeated without excluding participants with duration of use of <2 years to test whether this restriction influenced the study results. In addition, the analysis was repeated within the subgroup of participants who reported having hypertension and of patients who reported having hypertension or a history of CVD, as it had previously been hypothesized that the association between beta blocker use and CRC risk may be different in persons who have the same disorder but take different drugs.10
All analyses were performed with SAS software, version 9.2 (SAS Institute Inc., Cary, NC). Statistical significance was defined by a 2-sided P < .05. No multiple comparison corrections were made.
Between January 2003 and December 2007, 1945 patients and 1891 controls participated in the interview. Participants who did not provide information on drugs for hypertension or CVD (n = 138), who did not specify the year of initiation of beta blocker use (n = 39), or who had a duration of beta blocker use <2 years (n = 189) were excluded.
After the exclusions, 1708 patients and 1762 controls remained for the analyses (Table 1). The mean age of patients and controls was 68 and 69 years, respectively, and 40% were women. Of the patients, 59% had colon and 41% rectal cancer; 32% had proximal and 69% distal CRC. Tumor stages I to IV were diagnosed in 24%, 30%, 31%, and 14% of patients, respectively. Patients and controls were comparable in respect to physical activity and history of CVD. Patients had fewer years of schooling, a higher body mass index, and more lifetime pack-years of smoking, were more likely to consume higher amounts of alcohol, and more often had a first-degree relative with CRC. Control individuals were more likely to have hypertension, to have had a former colorectal endoscopy, to be regular NSAID and statin users, and to have conducted hormone replacement therapy.
|Characteristics||Cases, n = 1708||Controls, n = 1762||P for Difference|
|Age at index year|
|30-49 years||79 (4.6%)||67 (3.8%)|
|50-59 years||278 (16.3%)||269 (15.3%)|
|60-69 years||591 (34.6%)||553 (31.4%)|
|70-79 years||529 (31.0%)||578 (32.8%)|
|80+ years||231 (13.5%)||295 (16.7%)|
|Mean [SD]||67.7 [10.5]||68.5 [10.4]||N/Aa|
|Female||689 (40.3%)||715 (40.6%)||N/Aa|
|Years of schoolingb|
|<10 years||1153 (67.7%)||1046 (59.5%)|
|10 or 11 years||277 (16.3%)||351 (20.0%)|
|12 or 13 years||274 (16.0%)||361 (20.5%)||<.0001|
|Body mass index, kg/m2c|
|<25.0||569 (33.7%)||678 (38.7%)|
|25.0-29.9||783 (46.4%)||816 (46.6%)|
|30+||335 (19.9%)||257 (14.7%)||<.0001|
|Active smoking, lifetime pack-yearsd|
|Never||680 (40.2%)||787 (45.1%)|
|<10||378 (22.4%)||430 (24.6%)|
|10-19||213 (12.6%)||196 (11.2%)|
|20-29||183 (10.8%)||155 (8.9%)|
|30+||236 (14.0%)||177 (10.1%)||.0002|
|Alcohol, quartiles, lifetime ethanol, g/de|
|None||260 (15.3%)||254 (14.4%)|
|0.1-5.1||317 (18.7%)||381 (21.6%)|
|5.2-12.0||343 (20.2%)||374 (21.2%)|
|12.1-24.3||339 (20.0%)||377 (21.4%)|
|24.4+||440 (25.9%)||376 (21.3%)||.0105|
|Physical activity, quartiles, lifetime METs, h/wkf|
|0-136.5||400 (24.0%)||440 (25.1%)|
|136.6-203.2||392 (23.5%)||392 (25.0%)|
|203.3-298.7||427 (25.6%)||427 (25.0%)|
|298.8+||451 (27.0%)||451 (25.0%)||.4398|
|No/unknown||926 (54.3%)||890 (50.6%)|
|Yes||781 (45.8%)||870 (49.4%)||.0301|
|No||1299 (76.3%)||1327 (75.5%)|
|Yes||404 (23.7%)||431 (24.5%)||.5854|
|First-degree family history of colorectal canceri|
|No||1415 (85.1%)||1511 (88.7%)|
|Yes||247 (14.9%)||193 (11.3%)||.0023|
|Former colorectal endoscopyj|
|Never/unknown||1349 (79.0%)||842 (47.8%)|
|Ever||358 (21.0%)||920 (52.2%)||<.0001|
|Regular use of NSAIDsk|
|Never||1326 (78.1%)||1228 (69.9%)|
|Ever||372 (21.9%)||528 (30.1%)||<.0001|
|Current regular use of statinsl|
|No||1501 (88.0%)||1444 (82.1%)|
|Yes||205 (12.0%)||315 (17.9%)||<.0001|
|Use of HRT, women onlym|
|Never||331 (65.9%)||243 (52.3%)|
|Ever||171 (34.1%)||222 (47.7%)||<.0001|
As shown in Table 2, beta blockers were regularly used for ≥2 years by 390 cases (23%) and 450 controls (26%). Cardioselective beta blockers were generally used more often by both patients (327, 19%) and controls (390, 22%) than unselective beta blocker (61 patients [3%], 57 controls [3%]). Metoprolol was the most commonly used active ingredient of the beta blockers, followed by bisoprolol, carvedilol, and atenolol.
|Use of Beta Blocker||Cases, n = 1708||Controls, n = 1762|
|Any beta blockera||390 (22.8%)||450 (25.5%)|
|Cardioselective beta blocker||327 (19.0%)||390 (22.0%)|
|Metoprolol||176 (10.0%)||238 (13.2%)|
|Bisoprolol||96 (5.4%)||100 (5.4%)|
|Atenolol||35 (1.9%)||27 (1.5%)|
|Nebivolol||15 (0.8%)||22 (1.2%)|
|Celiprolol||2 (0.1%)||2 (0.1%)|
|Acebutolol||2 (0.1%)||0 (0.0%)|
|Betaxolol||1 (0.05%)||1 (0.05%)|
|Esmolol||1 (0.1%)||1 (0.1%)|
|Unselective beta blockerb||61 (3.4%)||57 (3.1%)|
|Carvedilol||30 (1.7%)||34 (1.8%)|
|Sotalol||19 (1.0%)||21 (1.1%)|
|Propanolol||5 (0.3%)||2 (0.1%)|
|Pindolol||2 (0.1%)||0 (0.0%)|
|Carteolol||1 (0.1%)||0 (0.0%)|
|Nadolol||1 (0.1%)||0 (0.0%)|
|Penbutolol||1 (0.1%)||0 (0.0%)|
When only adjusting for matching variables age, sex, and county of residence, short-term (2-5 years) use of any beta blocker (odds ratio [OR], 0.78; 95% confidence interval [CI], 0.63-0.97), of a cardioselective beta blocker (OR, 0.76; 95% CI, 0.60-0.96), and of metoprolol (OR, 0.71; 95% CI, 0.53-0.96; PTrend = .0301) was significantly associated with decreased CRC risk (Table 3). After additional adjustment for potentially protective factors and risk factors, we observed no association of CRC risk with beta blocker use (OR, 1.05; 95% CI, 0.86-1.29) or with duration of beta blocker use (Table 3). Likewise, no association was found with use or duration of use in stratified analysis specifically focusing on cardioselective and unselective beta blockers and on the active ingredients metoprolol, bisoprolol, carvedilol, and atenolol. In stage-specific analysis (Table 4), long-term beta blocker use (6+ years) showed a significant positive association with stage IV CRC (OR, 2.02; 95% CI, 1.25-3.27) when compared with nonuse. This positive association was also present for the subclasses of cardioselective (OR, 1.86; 95% CI, 1.09-3.15) and unselective (OR, 2.83; 95% CI, 1.11-7.23) beta blockers. No such associations were observed with stages I, II, or III CRC.
|Use of Beta Blocker||Cases||Controls||Basic Modela||Fully Adjusted Modelb|
|OR||95% CI||OR||95% CI|
|None||1318 (77.2%)||1312 (74.5%)||1.00||Ref||1.00||Ref|
|Any beta blocker|
|Current use, ≥2 years||390 (22.8%)||450 (25.5%)||0.90||0.77-1.05||1.05||0.86-1.29|
|2-5 years||164 (9.6%)||217 (12.3%)||0.78||0.63-0.97||0.94||0.72-1.22|
|6-10 years||119 (7.0%)||133 (7.6%)||0.93||0.71-1.20||1.05||0.77-1.42|
|11+ years||107 (6.3%)||100 (5.7%)||1.11||0.84-1.48||1.32||0.94-1.85|
|P for trendc||.8093||.1911|
|Cardioselective beta blocker|
|Exclusive use, ≥2 years||327 (19.9%)||390 (22.9%)||0.87||0.73-1.02||1.04||0.84-1.29|
|2-5 years||140 (8.5%)||190 (11.2%)||0.76||0.60-0.96||0.94||0.71-1.25|
|6+ years||187 (11.4%)||200 (11.8%)||0.97||0.78-1.20||1.13||0.87-1.47|
|P for trendc||.3074||.4529|
|Unselective beta blocker|
|Exclusive use, ≥2 years||61 (4.4%)||57 (4.2%)||1.10||0.76-1.60||1.14||0.75-1.75|
|2-5 years||23 (1.7%)||25 (1.8%)||0.95||0.53-1.68||0.87||0.45-1.69|
|6+ years||38 (2.8%)||32 (2.3%)||1.22||0.76-1.97||1.36||0.80-2.34|
|P for trendc||.4946||.3634|
|Exclusive use, ≥2 years||176 (11.8%)||238 (15.4%)||0.76||0.62-0.94||0.91||0.70-1.18|
|2-5 years||79 (5.3%)||114 (7.4%)||0.71||0.53-0.96||0.87||0.61-1.24|
|6+ years||97 (6.5%)||124 (8.0%)||0.81||0.61-1.07||0.94||0.67-1.31|
|P for trendc||.0301||.5704|
|Exclusive use, ≥2 years||96 (6.8%)||100 (7.1%)||1.00||0.75-1.34||1.18||0.84-1.65|
|2-5 years||43 (3.0%)||54 (3.8%)||0.84||0.55-1.26||1.05||0.66-1.67|
|6+ years||53 (3.8%)||46 (3.3%)||1.20||0.80-1.80||1.32||0.83-2.11|
|P for trendc||.6743||.2653|
|Exclusive use, ≥2 years||30 (2.2%)||34 (2.5%)||0.91||0.55-1.50||0.88||0.49-1.55|
|2-5 years||12 (0.9%)||17 (1.3%)||0.74||0.35-1.55||0.58||0.24-1.37|
|6+ years||18 (1.3%)||17 (1.3%)||1.09||0.56-2.13||1.21||0.57-2.57|
|P for trendc||.9190||.9745|
|Exclusive use, ≥2 years||35 (2.6%)||27 (2.0%)||1.31||0.79-2.19||1.53||0.85-2.73|
|2-5 years||9 (0.7%)||9 (0.7%)||0.96||0.38-2.44||1.02||0.37-2.83|
|6+ years||26 (1.9%)||18 (1.3%)||1.50||0.81-2.75||1.82||0.91-3.67|
|P for trendc||.2266||.1068|
|Use of Beta Blocker||Controls, No. (%)||Stage I||Stage II||Stage III||Stage IV|
|Cases, No. (%)||ORa||95% CI||Cases, No. (%)||ORa||95% CI||Cases, No. (%)||ORa||95% CI||Cases, No. (%)||ORa||95% CI|
|None||1312 (74.5)||306 (74.3)||1.00||Ref||398 (77.4)||1.00||Ref||424 (78.7)||1.00||Ref||187 (77.9)||1.00||Ref|
|Current use, ≥2 years||450 (25.5)||106 (25.7)||1.19||0.87-1.62||116 (22.6)||0.95||0.71-1.29||115 (21.3)||0.95||0.72-1.27||53 (22.1)||1.49||0.99-2.25|
|2-5 years||217 (12.3)||42 (10.2)||1.01||0.67-1.52||53 (10.3)||0.90||0.61-1.33||51 (9.5)||0.89||0.61-1.30||18 (7.5)||1.00||0.56-1.78|
|6+ years||233 (13.3)||64 (15.5)||1.36||0.94-1.98||63 (12.3)||1.00||0.70-1.44||64 (11.9)||1.01||0.71-1.45||35 (14.6)||2.02||1.25-3.27|
|P for trendb||.1304||.9118||.9278||.0156|
|Cardioselective beta blocker|
|Exclusive use, ≥2 years||390 (22.9)||86 (21.9)||1.13||0.81-1.57||99 (19.9)||0.98||0.71-1.34||99 (18.9)||0.97||0.72-1.31||43 (18.7)||1.40||0.90-2.18|
|2-5 years||190 (11.2)||35 (8.9)||0.98||0.63-1.52||45 (9.1)||0.91||0.60-1.38||44 (8.4)||0.89||0.60-1.32||16 (7.0)||0.99||0.54-1.83|
|6+ years||200 (11.8)||51 (13.0)||1.27||0.85-1.91||54 (10.9)||1.04||0.71-1.53||55 (10.5)||1.05||0.72-1.54||27 (11.7)||1.86||1.09-3.15|
|P for trendb||.2967||.9510||.9534||.0377|
|Unselective beta blocker|
|Exclusive use, ≥2 years||57 (4.2)||19 (5.9)||1.58||0.85-2.94||16 (3.9)||0.86||0.44-1.65||16 (3.6)||0.87||0.46-1.65||10 (5.1)||1.85||0.83-4.14|
|2-5 years||25 (1.8)||6 (1.9)||1.09||0.39-3.07||8 (1.9)||0.92||0.36-2.36||7 (1.6)||0.90||0.34-2.39||2 (1.0)||0.80||0.17-3.72|
|6+ years||32 (2.3)||13 (4.0)||1.95||0.92-4.13||8 (1.9)||0.81||0.34-1.94||9 (2.1)||1.01||0.73-1.39||8 (4.1)||2.83||1.11-7.23|
|P for trendb||.0946||.6152||.6640||.0531|
Neither the analyses stratified by cancer site (colon/rectum, proximal/distal CRC) nor the analyses stratified by sex showed associations between beta blocker use in general or use of specific beta blockers and the risk of CRC.
When repeating the analyses including patients and controls with a duration of use <2 years, results were comparable to the main analysis. Results were also comparable when the sample was restricted to patients and controls with a diagnosis of hypertension or to patients and controls with a diagnosis of hypertension or any history of CVD (data not shown).
Whereas only 1.1% of the controls had to be excluded because of missing information on drugs for hypertension or CVD, 6.1% of the CRC cases had to be excluded for this reason. Among cases, exclusions because of missing information were much more common among patients interviewed in the clinics (8.6%) than among patients interviewed at home (2.6%). As characteristics between participants with and without information on beta blocker use were slightly different, we repeated the analysis including controls and 1) cases with an interview at home and 2) cases with an interview in the clinics, respectively, to test whether dissimilar proportions of missing information on beta blocker use among cases and controls may have biased our results. However, the associations between beta blocker use and CRC risk were comparable in the subgroups ( OR, 1.09; 95% CI, 0.84-1.41;  OR, 1.03; 95% CI, 0.81-1.32).
In this large population-based study, we found no association between CRC risk and the use of beta blockers or any subclass of beta blockers after adjustment for confounding. Neither did the analysis by cancer site show any evidence for an association. Thus, the hypothesis that long-term beta blocker use, which lowers the norepinephrine level, decreases the risk of CRC could not be confirmed.
In stage-specific analysis, however, we found a positive association between beta blocker use and stage IV CRC risk. In 2 recent studies on breast cancer, use of any beta blocker and use of the cardiounselective beta blocker propranolol was associated with a lower risk of late-stage diagnosis and a longer relapse-free survival, respectively.21-23 However, these findings were based on relatively small sample sizes. We are not aware of any study that has previously investigated stage-specific associations or association between beta blocker use and disease progression in CRC patients. Nevertheless, our stage-specific results should be interpreted with caution and investigated in further studies in detail; because of the multiple testing and the small sample size in our stage-specific analyses, we cannot exclude the possibility that the results were caused by chance.
The hypothesis of a protective effect of beta blocker use on the risk of specific types of cancer was supported by studies on prostate cancer,13 CRC,16 and all cancers combined.14 The study on CRC risk reported a risk reduction of 21%.16 A recent study screening pharmaceuticals for potential effects on cancer risk reported ORs for colon cancer risk between 0.98 for propranolol and 1.06 for atenolol.17, 18 Although the latter results are in accordance with our findings, the design of both studies again did not allow adjustment for important confounders.
Adjustment for confounding in the analysis of the association of beta blocker use and CRC risk is necessary, as CRC risk and beta blocker use share some risk/protective factors. Our results confirm the importance of adjustment, as the ORs were generally higher in the fully adjusted model than in the model adjusting for the matching factors only. Estimates might, for example, be low if the analysis does not account for participation in CRC screening, for example, by colorectal endoscopy, which strongly reduces CRC risk.24 Users of beta blockers might see a physician more often and may thus be more likely to participate in such programs. In contrast, obesity, smoking, and lack of physical activity have been associated with an increased risk of CRC25-27 and with an increased risk of hypertension,28-30 and thus an association with beta blocker use seems likely. If estimates are not adjusted for these factors, the reported OR may be overestimated. Because of these interdependencies, adjustment for confounding factors is important to obtain accurate estimates of the association between beta blocker use and CRC risk.
Among other drugs, beta blockers are prescribed against hypertension. As hypertension itself was hypothesized to be a risk factor for cancer,31 confounding by indication is possible. The reported association was adjusted for previously diagnosed hypertension and CVD to control for such confounding. As it had previously been suggested that the association between beta blocker use and CRC risk should be investigated in persons who have the same disorder but take different drugs,10 we repeated the analysis within the subgroup of participants who reported having hypertension and of patients who reported having hypertension or a history of CVD. However, results within these subgroups were comparable to the overall results, so that a strong modifying effect according to previously diagnosed hypertension and/or CVD seems unlikely.
Strengths of our study include its population-based design, a detailed assessment of medication use including duration of use, a detailed adjustment for relevant confounding factors, and a sound sample size to detect even moderate associations of CRC risk and beta blocker or cardioselective beta blocker use. For these analyses, we had a power of 80% at a significance level of .05 to detect ORs ratios ≥1.28/≤0.78 (beta blockers) and ≥1.26/≤0.79 (cardioselective beta blockers), respectively.32 In the analysis of the association between CRC risk and unselective beta blockers, and metoprolol, bisoprolol, carvedilol, and atenolol, we had a statistical power to detect ORs ≥1.33/≤0.75 (metoprolol), ≥1.46/≤0.69 (bisoprolol), ≥1.64/≤0.61 (unselective beta blocker), ≥1.84 /≤0.54 (carvedilol), and ≥1.95/≤0.51 (atenolol). Thus, most analyses in our study had a sufficient sample size to detect weak to moderate associations.
A limitation of our study is that information on beta blocker use was self-reported. The study was primarily designed to investigate the potential of screening endoscopy for risk reduction of CRC.24 As the participants were unaware of an analysis of beta blocker use and CRC risk, a reporting bias or selection bias caused by the knowledge of the objective of this analysis can be excluded. Interviews with cases were mostly conducted at the clinic, whereas controls were interviewed at home. Differential recall bias by package-assisted assessment of medication use at home would have led to an underestimation of the prevalence of beta blocker use among cases and, consequently, to an underestimation of the OR, whereas nondifferential recall bias would influence estimates toward no association and cannot be excluded. However, our estimates were unaltered when we restricted the analyses of CRC risk and beta blocker use and duration of use to cases and controls who were interviewed at home. Suffering a severe disease like cancer might affect the recall of medication use. However, this effect can be assumed to be small to negligible with respect to current medication use. Furthermore, the number of participants excluded because of missing information on hypertension and CVD medication was comparable across patients with stage I to stage IV disease. Although this proportion was higher among patients interviewed in the clinics than among patients and controls interviewed at home, the estimates for the association between CRC risk and beta blocker use were comparable for both subgroups of patient, which suggests that the exclusion of participants with missing information on beta blocker use likely did not falsify our estimates.
Ascertainment of beta blocker use was restricted to information on current regular beta blocker use and the year of initiation. Thus, lifetime exposure could not be estimated, and the cumulative duration of use may be underestimated. Beta blocker use of <2 years was excluded in our analysis, so modifications of prescriptions because of health deterioration before colorectal cancer diagnosis are unlikely. It is biologically plausible that effects of beta blocker use might have a latency period of several years. To assess a potential effect of latency periods, we directly analyzed the association between duration of beta blocker use and colorectal cancer risk and could not find a significant or stronger association for long-term users. A further limitation is that we were not able to investigate dose-response relations.
In this population-based study with no upper age limit, approximately 50% of eligible cases and control persons were recruited. Participation rates were lower among older cases and controls, who are more likely to use beta blockers. However, this should not have distorted the results, as age was matched and adjusted for, and the observed associations did not vary when results were stratified by age.
Patients with advanced cancer might be underrepresented in our study, as patients with advanced cancer might die before the recruitment or might not be able to complete the interview. Stage-stratified analyses indicate that the association between beta blocker use and colorectal cancer risk varies with stage of disease, as long-term beta blocker use was significantly associated with stage IV colorectal cancer risk. If stage IV cancer patients were underrepresented in our study, the overall ORs might have been slightly underestimated.
In this case-control study, cases were recruited in hospitals, whereas controls were sampled from population registries and then asked for participation. Such difference in sampling might lead to Berkson bias.33 However, as we included incident cancer cases only, this bias is assumed to be negligible.34 A further limitation with respect to the study design is that information on confounders was mostly based on self-reports. Thus, there might be some misclassification in the confounders.
Despite these limitations, results from this large population-based study, whose design allowed for a detailed adjustment for relevant confounders and a detailed beta blocker analysis by subclass, cancer site, and cancer stage, do not support the hypothesis that beta blocker use is associated with a lower risk of CRC. However, we found a positive association between long-term beta blocker use and risk of stage IV CRC, which should be investigated in future studies.
This study was supported by the German Research Council (BR 1704/6-1, BR 1704/6-3, BR 1704/6-4, and CH 117/1-1) and the German Federal Ministry of Education and Research (01KH0404 and 01ER0814).
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
- 10Testing whether drugs that weaken norepinephrine signaling prevent or treat various types of cancer. Clin Epidemiol. 2010; 2: 1-3..
- 20ATC Classification Index With DDDs, 2011. Oslo, Norway: WHO Collaborating Center for Drug Statistics Methodology; 2010.