Stage-specific associations between beta blocker use and prognosis after colorectal cancer
Lina Jansen PhD,
Division of Clinical Epidemiology and Aging Research, German Cancer Research Center (DKFZ), Heidelberg, Germany
Corresponding author: Lina Jansen, PhD, Division of Clinical Epidemiology and Aging Research, German Cancer Research Center (DKFZ), POB 10 19 49, D-69009 Heidelberg, Germany; Fax: (011) +49-6221-421302; firstname.lastname@example.org
We thank Ute Handte-Daub and Petra Bächer for their excellent technical assistance. We are grateful to the study participants, the interviewers who collected the baseline data, and the hospitals and cooperating institutions 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, and the Cancer Registry of Rhineland-Palatinate.
Recent observational studies have suggested that the use of beta blockers might be associated with better prognosis after cancer. Because evidence is limited for colorectal cancer (CRC), the association of beta blocker use and prognosis was investigated in a large population-based cohort of patients with CRC.
Between 2003 and 2007, information on beta blocker use at diagnosis and potential confounders was collected by personal interviews for 1975 patients with CRC. Vital status, cause of death, and recurrence status were assessed during a median follow-up time of 5.0 years. The associations of beta blocker use and overall, CRC-specific, and recurrence-free survival were estimated by Cox proportional hazard regression. In addition, beta blocker subgroup, site, and stage-specific analyses were performed.
After adjustment for covariates including sociodemographic, cancer-related, and lifestyle factors and comorbidity and medications, no significant association between beta blocker use at diagnosis and prognosis was observed for all stages combined. However, in stage-specific analyses, beta blocker use was associated with longer overall survival (hazard ratio = 0.50; 95% confidence interval = 0.33-0.78) and CRC-specific survival (hazard ratio = 0.47; 95% confidence interval = 0.30-0.75) in stage IV patients. For these patients, median overall survival was 18 months longer and CRC–specific survival was 17 months longer for beta blocker users than for nonusers (38 versus 20 months and 37 versus 20 months, respectively).
Colorectal cancer (CRC) is one of the most common malignant diseases worldwide, with more than 1.2 million new incident cases per year. Treatment of CRC includes surgery as well as chemotherapy and radiotherapy, depending on the tumor location and stage. Recently, it has been suggested that the neuroendocrine signaling pathway may play a role in tumor development and progression.[2-4] Therefore, medications that were developed for other diseases might also be useful in the treatment of cancer. Beta blockers are possible candidates, because they block the action of the endogenous catecholamines epinephrine and norepinephrine on β-adrenergic receptors. In vitro and in vivo findings have shown that norepinephrine-induced stimulation of cell migration was inhibited by the nonselective beta blocker propanolol in cancers at various sites including colon, prostate, breast, and pancreas. These effects have been hypothesized to be mediated by β2-adrenergic receptors, because the cardioselective beta blocker atenolol did not inhibit cell migration.
Results from recent clinical observational studies support the hypothesis that beta blocker treatment might be potentially relevant for treatment of cancer. Associations between beta blocker use and prognosis were reported for malignant melanoma,[9, 10] breast,[11, 12] prostate,[13, 14] ovarian, and non–small cell lung cancer. For CRC, only one study has been published, which found no associations between beta blocker use and survival. This study also did not find significant associations with other cancer sites, including breast and prostate cancer. Recently, 3 controlled clinical trials examining the use of propranolol in patients with breast, colorectal, and ovarian cancer have been started.
Because the evidence for CRC is still limited, the aim of this study was to investigate whether the use of beta blockers at diagnosis is associated with longer overall, CRC-specific, and recurrence-free survival in a large population-based cohort of patients with CRC.
MATERIALS AND METHODS
The DACHS study is an ongoing population-based case-control study with additional follow-up of cases carried out in southwest Germany (Rhine-Neckar-Odenwald-Region, approximately 2 million inhabitants). Patients with a first histologically confirmed diagnosis of CRC (International Classification of Diseases, 10th Revision [ICD-10] codes C18-C20) have been recruited since 2003. For this analysis, patients with a CRC diagnosis in 2003 to 2007 were included. Further details of the study have been described elsewhere.
In brief, eligible patients who were mentally and physically able to participate in a personal interview of about 1 hour were informed about the study by their clinicians in 20 recruiting clinics, 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. Median time interval between surgery and interview was 8 days for patients interviewed in the clinics (N = 1020, 56%) and 297 days for patients interviewed at home (N = 800, 44%). The recruited patients constitute about 50% of all eligible patients in the study area. The main reason for incomplete recruitment was the work overload of the clinicians, which led to missed notifications of cases. Only a few patients actively denied participation in the study. All participants gave written informed consent. The study was approved by the ethics committees of the University of Heidelberg and the state medical boards of Baden-Wuerttemberg and Rhineland-Palatinate.
At baseline, trained interviewers collected information on sociodemographic factors and a detailed medical and lifestyle history with regard to known or suspected protective factors or risk factors of CRC using a standardized questionnaire. At 3 and 5 years after diagnosis, information on vital status and date and cause of death was obtained from registration offices. After an average of 3 years since diagnosis, detailed information on CRC treatment and cancer recurrence was collected from attending physicians in a standardized questionnaire. After an average of 5 years since diagnosis (in 2009-2011), information on cancer recurrence was updated for patients with CRC diagnosis in 2003 to 2006. If a patient died prior to the follow-up, information on cancer recurrence before death was collected from the last attending physicians. Details on data collection during follow-up have been described elsewhere.
Assessment of Medication Use
At baseline, information on current regular use of drugs (≥2 times per week) was collected for a variety of indications, including cardiovascular disease (CVD) and hypertension, and for medications that did not match any of the predefined indications in the questionnaire. A list with the most common drug names for each of the indications was provided to assist the patient in case of uncertainty. The name of the drug as well as the year of initiation of drug use was collected. Drugs were coded according to the Anatomical Therapeutic Chemical/Defined Daily Dose (ATC/DDD) Classification.
Participants were classified as current regular users of beta blockers, if they reported the use of beta blockers (ATC code starting with “C07”) in any medication field of the questionnaire. If information on drugs for CVD or hypertension was missing, participants were excluded from the analysis.
The distributions of potential protective factors and risk factors were compared between beta blocker users and nonusers using Pearson's chi-square and t tests. The use of beta blockers was described with respect to the cardioselectivity of the beta blocker and the active ingredient.
Median follow-up time was computed using the Kaplan-Meier estimate of potential follow-up. The association of regular use of beta blockers with overall, CRC-specific, and recurrence-free survival was estimated with Cox proportional hazard regression. Patients without the endpoint of interest were censored at the last contact they were known to be alive or free of recurrence, respectively. In analyses on CRC-specific survival, patients were also censored at the date of death when they died from another disease. Patients with stage IV disease were excluded from recurrence-free survival analyses, because no reliable information on progression was available. Because patients were interviewed at various intervals after CRC diagnosis, late-entry models were computed. The proportional hazards assumption was assessed by examining plots of scaled Schoenfeld residuals and by including a time-dependent component for each explanatory variable in the Cox model.
The association between beta blocker use and the endpoints was estimated in a multivariate model with adjustment for age at diagnosis (30-59, 60-69, 70-79, ≥80 years), sex, International Union Against Cancer (UICC) stage (I-IV), surgery (yes/no), chemotherapy (yes/no), radiotherapy (yes/no), body mass index (<25, 25-29.9, ≥30 kg/m2), hypertension (yes/no), CVD (including heart failure, myocardial infarction, stroke, and cardiac circulatory disorder; yes/no), diabetes (yes/no), regular use of nonsteroidal anti-inflammatory drugs (NSAIDs) including aspirin (yes/no), regular use of statins (yes/no), use of hormone replacement therapy (HRT; ever/never), lifetime pack-years of active smoking (never active, <10, 10-19, 20-29, ≥30 pack-years), alcohol consumption (quartiles of the average daily alcohol consumption before cancer diagnosis measured in grams of ethanol per day), physical activity (quartiles of lifetime metabolic equivalents [METs] in hours per week), and participation in health check-up (yes/no). Chemotherapy showed a time-dependent effect, and therefore an interaction term between log(time) and chemotherapy was included in the model. Survival curves directly adjusted for the covariates included in the multivariate analysis were used to illustrate the association of beta blocker use with prognosis. The analysis was repeated including a variable selection approach to reduce the model complexity. Because results were comparable with respect to the hazard ratio and the precision of the hazard ratio, only results of the full model are presented.
The analysis was repeated for subclasses of beta blockers according to their cardio-selectivity, according to their active ingredient, for the subgroups colon and rectal cancer, proximal (cecum to transverse colon) and distal (left flexure to rectum) CRC, and stage I/II, III, and IV CRC and stratified by NSAIDs, statin, and HRT use.
In sensitivity analysis, analyses were repeated after excluding patients with a delay of more than 60 days between diagnosis and interview, after restricting the analysis to patients with long-term beta blocker use (starting 2 or more years prior to diagnosis), after excluding patients who received bevacizumab in the stage IV stratified analysis, after additional adjustment for antidiabetic drugs (ATC code starting with A10), and after imputing missing values in covariates (N = 113) by multiple imputation using the Markov-Chain Monte Carlo method to fill in missing data (N = 20 imputed datasets, SAS procedure PROC MI). The variables in the imputation model included all variables of the multivariate model described above. To test for interactions between beta blocker use and NSAIDs, HRT, and statin use, respectively, the analysis was repeated after adding these factors to the multivariate Cox model.
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.
Study Population and Medication Use
The study included 1975 patients with CRC who were diagnosed between 2003 and 2007. Participants who were not included in the 3- or 5-year follow-up (N = 37) or who did not provide information on drugs for hypertension or CVD (N = 118) were excluded, leaving 1820 patients for the analyses. Patients were on average 68 years old, 41% were female and 60% had colon cancer. Stage I, II, III, and IV disease were diagnosed for 25%, 30%, 31%, and 14% of the patients, respectively. The time interval between diagnosis and interview was on average 6.1 months (standard deviation: 9.0 months) and was shorter for patients with higher cancer stage (stage I: 7.1 months, stage II: 6.2 months, stage III: 6.1 months, stage IV: 4.2 months).
Beta blockers were regularly used by 509 patients (28%; Table 1). Patients who used beta blockers regularly at diagnosis were, on average, 3 years older than nonusers, were more often obese, had more often hypertension, any cardiovascular disease, or diabetes, used more often NSAIDS and statins, participated more often in a health check-up, were more often moderately physically active, and received chemotherapy less often. They were comparable or similar to nonusers with respect to sex, stage at diagnosis, cancer site, cancer surgery, radiotherapy, use of bevacizumab and cetuximab, smoking, alcohol consumption, and use of HRT.
Table 1. Characteristics of Beta Blocker Users and Nonusers With Colorectal Cancer
Use of Beta Blockers
Yes (n = 509)
No (n = 1311)
P for Difference
Data missing for 4 patients.
A total of 17 patients had both proximal and distal colorectal cancer.
Data missing for 14 patients.
Data missing for 11 patients.
Data missing for 12 patients.
Data missing for 12 patients.
Data missing for 8 patients.
Data missing for 22 patients.
Average daily alcohol consumption before cancer diagnosis measured in grams of ethanol per day. Data missing for 10 patients.
Data missing for 42 patients.
Data missing for 1 patient.
Data missing for 5 patients.
Data missing for 15 patients.
Use of nonsteroidal anti-inflammatory drugs (NSAIDs) is defined as taking NSAIDs at least 2 times a week for at least 1 year. Data missing for 10 participants.
Data missing for 6 patients.
Data missing for 2 patients.
Data missing for 15 patients.
Abbreviations: HRT, hormone replacement therapy ICD, International Classification of Diseases; MET, metabolic equivalent; UICC, International Union Against Cancer.
Table 2 shows that cardioselective beta blockers were generally used more often than unselective beta blockers (24% versus 4%). Metoprolol was the most commonly used active ingredient (13%) followed by bisoprolol (7%), atenolol (2%), and carvedilol (2%).
Table 2. Distribution of Classes and Active Ingredients of Regularly Used Beta Blockers
User N (%)
For 2 participants, information on the class/active ingredient of the beta blocker used was not available.
Information on vital status after colorectal cancer diagnosis was available for 1819 of 1820 patients. Median follow-up time was 5.0 years (25th percentile: 4.9 years, 75th percentile: 5.4 years) in the analysis of overall survival. During the follow-up period, 613 patients died, including 444 patients who died from CRC; 567 patients had a recurrence. For 8 patients, the cause of death was not available. For 23 patients, no information on recurrences was available.
After adjustment for potential confounders, use of beta blockers at diagnosis was not found to be associated with overall (hazard ratio [HR] = 0.99; 95% confidence interval [CI] = 0.79-1.22), CRC-specific (HR = 0.93; 95% CI = 0.71-1.21), or recurrence-free survival (HR = 0.96; 95% CI = 0.77-1.20; Table 3) in the total study population. In addition, no significant associations for the subgroups of selective or unselective beta blockers and the active ingredients metoprolol, bisoprolol, and carvedilol were found. However, patients who used bisoprolol tended to have longer overall survival (HR = 0.72; 95% CI = 0.50-1.03).
Table 3. Association of Beta Blocker Use at Diagnosis With Overall, Colorectal Cancer–Specific, and Recurrence-Free Survival After Adjustment for Confounding
Stage IV patients were excluded in the analyses on recurrence.
Hazard ratio for the factor beta blocker use. Reference is no beta blocker use. Model adjusted for age at diagnosis, sex, stage, cancer site, surgery, chemotherapy, radiotherapy, body mass index, lifetime pack-years of active smoking, alcohol consumption, physical activity, participation in health check-up, hypertension, cardiovascular disease, diabetes, regular use of NSAIDs, regular use of statins, use of hormone-replacement therapy, and the time-dependent effect chemotherapy*log(time).
Abbreviations: CI, confidence interval; CRC, colorectal cancer; HR, hazard ratio; N, no. of patients.
No beta blocker use
Any beta blocker
Selective beta blockers
Unselective beta blockers
Stratification by stage at diagnosis (Table 4) showed that the use of beta blockers was significantly associated with longer overall (HR = 0.50; 95% CI = 0.33-0.78; P = .0023) and CRC-specific survival (HR = 0.47; 95% CI = 0.30-0.75; P = .0017) for patients with stage IV CRC. Median overall survival was 18 months and median CRC-specific survival was 17 months longer for beta blocker users than for nonusers (38 and 37 versus 20 months). Differences in survival between beta blocker users und nonusers seemed to develop in patients with stage IV CRC after 1 year (Fig. 1). The association for stage IV CRC was observed for both colon and rectal cancer, but was statistically significant only for colon cancer in site-specific analysis (colon cancer: overall survival HR = 0.47; 95% CI = 0.27-0.81; P = .0070; CRC-specific survival HR = 0.46 95% CI = 0.26-0.84; P = .0108; rectal cancer: overall survival HR = 0.51; 95% CI = 0.22-1.19; P = .0826; CRC-specific survival HR = 0.46 95% CI = 0.19-1.11; P = .0526). For patients with stage I-III CRC, no significant associations with beta blocker use were observed. Results were similar when restricting the analysis to selective beta blockers.
Table 4. Association of Beta Blocker Use at Diagnosis With Overall, Colorectal Cancer Specific, and Recurrence-Free Survival After Adjustment for Confounding Stratified by Colorectal Cancer Stage at Diagnosis
Hazard ratio for the factor beta blocker use. Reference is no beta blocker use. Model adjusted for age at diagnosis, sex, cancer site, surgery, radiotherapy, body mass index, lifetime pack-years of active smoking, alcohol consumption, physical activity, participation in health check-up, hypertension, cardiovascular disease, diabetes, regular use of NSAIDs, regular use of statins, use of hormone-replacement therapy, and in case of the analysis of stage I/II and stage III, chemotherapy, and the time-dependent effect chemotherapy*log(time). All patients with stage IV cancer received chemotherapy.
Abbreviations: CI, confidence interval; CRC, colorectal cancer; HR, hazard ratio; N, no. of patients.
No beta blocker
Any beta blocker
Selective beta blockers
In stratified analysis by colon and rectal cancer and proximal and distal CRC (all stages), no association between beta blocker use and prognosis was observed (data not shown).
The exclusion of patients with a delay of more than 60 days between diagnosis and interview (819 overall [45%] and 96 [37.5%] stage IV patients) did not materially alter the results (data not shown). For stage IV patients, the use of beta blockers was again significantly associated with longer overall survival (HR = 0.42; 95% CI = 0.24-0.75) and CRC-specific survival (HR = 0.45; 95% CI = 0.25-0.82). The association also persisted after excluding patients who received bevacizumab to avoid confounding by use of beta blockers due to side effect of this treatment (overall: HR = 0.48, 95% CI = 0.29-0.80; CRC-specific: HR = 0.45; 95% CI = 0.26-0.76). Comparing T and N distributions, proportions of screening detected cancers, and prevalence of underweight, we found no indication that stage IV patients who used beta blockers were diagnosed earlier (Table 5).
Table 5. Specific Characteristics of Beta Blocker Users and Nonusers With Stage IV Colorectal Cancer
The exclusion of participants with duration of beta blocker use of less than 2 years prior to diagnosis (N = 97, 19% of all beta blocker users) and imputation of missing values in covariates by multiple imputation also did not materially alter the results (data not shown). Results were comparable after additional adjustment for antidiabetic drugs. Tests for interaction between beta blocker use and NSAIDs, HRT, and statin use, respectively, were not significant. Likewise, in stratified analysis by use of the drugs named above, no indication for interaction with any of these medications was observed (data not shown).
In this large population-based study, we found no significant association between beta blocker use and overall, CRC-specific, and recurrence-free survival. However, stage-specific analysis showed that beta blocker use was associated significantly with a better prognosis in patients with stage IV CRC. Median overall and CRC-specific survival time of patients with stage IV CRC who used beta blockers was 17 and 18 months longer, respectively. Thus, our results indicate that beta blocker use may be associated with better prognosis in advanced CRC.
To our knowledge, only one previous study has investigated the association between beta blocker use and prognosis in patients with CRC, likewise showing no association between beta blocker use and overall survival. However, this study had a smaller sample size (N = 619), did not provide stage-specific analysis, and did also not show a survival benefit for beta blocker users for other cancer sites including breast cancer. Other breast cancer studies[11, 12, 28] and studies on malignant melanoma,[9, 10] ovarian cancer, and prostate cancer[13, 14] reported better prognosis in case of beta blocker use.
With respect to stage-specific associations, results from a malignant melanoma study indicated that associations between beta blocker use and survival may be stronger in stage I/II than in stage III/IV. In breast cancer studies, patients with stage IV disease were excluded[11, 12] or results were similar when including or excluding patients with stage IV disease. For prostate cancer, associations between beta blocker use and prognosis were only found for advanced stages.[13, 14] With respect to biological evidence, in vitro and in vivo studies have shown that beta-adrenergic signaling is involved in many cellular processes that contribute to the progression of cancer, such as migration of metastatic cancer cells.[7, 30] However, available evidence from in vitro and in vivo studies is insufficient for the time being to explain a stage-specific effect. Thus, evidence on stage-specific associations between beta blocker use and cancer is still limited. As there are indications for stage-specific associations from our study and the prostate cancer study, this aspect should be evaluated in further studies.
The finding of an association between beta blocker use and prognosis solely in patients with stage IV CRC was unexpected for us and we cannot exclude the possibility that this might be a chance finding or occurred due to inappropriate covariate adjustment, despite the very low P value (< 0.003) for overall and CRC-specific survival. However, the median survival benefit of around 17 months for beta blocker users compared to nonusers was very large. The potential of beta blockers as a low-cost, potentially highly effective treatment, with low side effects, for patients with stage IV CRC should therefore be further explored. In such further investigations, some additional aspects should be investigated, such as the timing of beta blocker use, the dosage and active ingredient and interactions with treatment. A clinical trial investigating the effect of propranolol treatment on CRC survival has been started recently. Unfortunately, however, patients with CRC who had metastatic disease were excluded in this trial. Our results suggest that a trial specifically addressing potential benefits of beta blocker treatment among patients with stage IV CRC who have limited treatment options might be of particular interest. In such a trial, limitations of observational study designs based on the innate link between beta blocker prescription and the indicator for prescription would be avoided. For example, in our study, we cannot distinguish whether the effect of beta blocker treatment was due to the use of beta blockers or due to the indication for beta blocker treatment. Furthermore, prognosis of beta blocker users might be generally improved (independent of the effect of beta blockers) because of better monitoring after cancer diagnosis due to comorbidity. By randomization and blinding, clinical trials could estimate the individual effects of beta blockers on prognosis.
Strengths of our analysis are the use of a large population-based study, detailed assessment of medication use, detailed adjustment for relevant confounding factors, and a sound sample size to investigate even subgroups of beta blockers and stage-specific associations. However, we cannot exclude chance findings for subgroups with small case numbers and due to lack of multiple comparison corrections. Another limitation of our study is that information on beta blocker use was self-reported. However, the study was primarily designed to investigate the potential of screening endoscopy for risk reduction of CRC and the patients were unaware of an analysis of beta blocker use and prognosis. Thus, reporting of medication can be expected to be comparable for beta blocker users and nonusers. Another limitation is that dosage of beta blocker use and beta blocker use after diagnosis was not assessed. Information on beta blocker use after diagnosis was only available for survivors who participated 5 years after diagnosis in a follow-up study, which included an assessment of current medication use. Because 182 of these 215 patients (85%) used beta blockers at diagnosis and at follow-up, use of beta blockers was most likely continued after diagnosis in the vast majority of users. A restriction of the analysis to long-term beta blocker users (≥ 2 years prior to diagnosis), who might be less prone to change beta blocker medication, did not materially alter the results.
A further limitation of our study is a mean delay of several months between diagnosis and patient interview. This delay mainly results from the retrospective recruitment of patients who could not be recruited during hospital stay. This retrospective recruitment was conducted to increase completeness of patient recruitment and, thus, the generalizability of the study. We accounted for this delay by using late-entry models and conducting a sensitivity analysis including only patients with a delay of less than 60 days. We could not find any relation between the length of delay and the observed survival benefit among patients with stage IV disease.
Despite these limitations, our results suggest that beta blocker use might be associated with better prognosis in stage IV CRC. This finding requires corroboration by other studies including randomized controlled trials. If confirmed, provision of beta blockers might make an important and highly cost-effective contribution to progress in treatment of metastatic CRC.
This work 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).