- Email: [email protected]
β-Blocker usage and colorectal cancer mortality: a nested case–control study in the UK Clinical Practice Research Datalink cohort
original articles
Annals of Oncology Annals of Oncology 24: 3100–3106, 2013 doi:10.1093/annonc/mdt381 Published online 19 September 2013
β-Blocker usage and colorectal cancer mortality: a nested case–control study in the UK Clinical Practice Research Datalink cohort B. M. Hicks1*, L. J. Murray1, D. G. Powe2,3, C. M. Hughes4 & C. R. Cardwell1 1 Centre for Public Health, Queen’s University Belfast, Belfast, Northern, Ireland; 2Department of Cellular Pathology, Queen’s Medical Centre, Nottingham University Hospitals NHS Trust, Nottingham; 3The John Van Geest Cancer Research Centre, Nottingham Trent University, Nottingham, UK; 4School of Pharmacy, Queen’s University Belfast, Belfast, Northern, Ireland
Background: Epidemiological and laboratory studies suggest that β-blockers may reduce cancer progression in various cancer sites. The aim of this study was to conduct the first epidemiological investigation of the effect of post-diagnostic β-blocker usage on colorectal cancer-specific mortality in a large population-based colorectal cancer patient cohort. Patients and methods: A nested case–control analysis was conducted within a cohort of 4794 colorectal cancer patients diagnosed between 1998 and 2007. Patients were identified from the UK Clinical Practice Research Datalink and confirmed using cancer registry data. Patients with a colorectal cancer- specific death (data from the Office of National Statistics death registration system) were matched to five controls. Conditional logistic regression was applied to calculate odds ratios (OR) and 95% confidence intervals (95% CIs) according to β-blocker usage (data from GP-prescribing records). Results: Post-diagnostic β-blocker use was identified in 21.4% of 1559 colorectal cancer-specific deaths and 23.7% of their 7531 matched controls, with little evidence of an association (OR = 0.89 95% CI 0.78–1.02). Similar associations were found when analysing drug frequency, β-blocker type or specific drugs such as propranolol. There was some evidence of a weak reduction in all-cause mortality in β-blocker users (adjusted OR = 0.88; 95% CI 0.77–1.00; P = 0.04) which was in part due to the marked effect of atenolol on cardiovascular mortality (adjusted OR = 0.62; 95% CI 0.40–0.97; P = 0.04). Conclusions: In this novel, large UK population-based cohort of colorectal cancer patients, there was no evidence of an association between post-diagnostic β-blocker use and colorectal cancer-specific mortality. Clinical Trials number: NCT00888797. Key words: colorectal cancer survival, β-blockers, pharmacoepidemiology, medication, propranolol, atenolol
introduction β-Blockers are used to treat heart disease and hypertension but recent evidence suggests they may protect against cancer progression. β-Blockers act by inhibiting β-adrenergic receptors, which are expressed by various cancer types including breast, pancreatic and colon cancers [1–3]. β-Adrenergic signalling has been shown to be relevant in cancer progression [4–6]. In vitro studies have demonstrated that β-blockers can inhibit cancer cell proliferation, invasion and resistance to apoptosis [4, 7–9]. A study found that colorectal tumour cell growth in mice and cell proliferation in colorectal cancer cell lines were inhibited by the β-blockers propranolol and atenolol, respectively [10].
*Correspondence to: Dr Blánaid Hicks, Centre for Public Health, Queen’s University Belfast, Institute of Clinical Science, Block B, Royal Victoria Hospital, Belfast, BT12 6BA, Northern Ireland. Tel: +44-0-9063-5009; Fax: +44-0-9023-5900; E-mail: [email protected]
Norepinephrine has been shown to induce the migration of colon carcinoma cells, a process inhibited by the non-selective β-blocker propranolol suggesting β-blockers could be a useful treatment of colorectal cancer [3]. Epidemiological studies have shown reductions in cancerspecific mortality with post-diagnostic β-blocker use for breast [11, 12], lung [13], prostate [14] and ovarian cancers [15]. Despite these findings and in vitro evidence for colon cancer [3], epidemiological studies have not investigated the effect of β-blocker use on cancer-specific mortality in colorectal cancer patients. Despite this lack of epidemiology, a phase III clinical trial is underway, investigating the effect of the β-blocker propranolol on recurrence in colorectal cancer patients. However, this study will take several years to report (estimated 2017), includes a relatively small number of colorectal cancer patients (200 in the β-blocker group), is investigating a specific drug ( propranolol), used in a specific time period (20 days in total, pre- and post-surgery) in early disease [16]. Evidence
© The Author 2013. Published by Oxford University Press on behalf of the European Society for Medical Oncology. All rights reserved. For permissions, please email: [email protected].
Downloaded from http://annonc.oxfordjournals.org/ at Northern Arizona University on July 9, 2015
Received 22 July 2013; revised 5 August 2013; accepted 7 August 2013
original articles
Annals of Oncology
from well carried out observational studies is therefore required to inform future studies. Our study is the first to investigate the effect of postdiagnostic β-blocker usage on colorectal cancer-specific mortality in a large prospective UK population-based cohort of colorectal cancer patients and importantly will investigate a range of β-blocker medications, time points and cancer stages.
methods study design
exposure data β-blockers were determined from GP prescribing data and defined as drugs classified in the British National Formulary (BNF) chapter 2.4. Using prescriptions, daily defined doses (DDD) were calculated on the basis of quantity and strength (http://www.whocc.no/atc_ddd_index/?Code=C07A). A quantity of 28 tablets was assumed for ∼5% of prescriptions where quantity was missing or inconsistent.
confounders Cancer stage and treatment within 6 months after diagnosis (surgery, chemotherapy and radiotherapy) were determined from NCDR data. Smoking, alcohol consumption and body mass index (BMI) were determined from the closest GP record before a colorectal cancer diagnosis (records older than 10 years were ignored). Comorbidities before and during the exposure period were determined from GP diagnosis codes based upon nine of the most common diagnoses contributing to a CPRD adaptation of the Charlson comorbidity index [18].
data analysis This cohort was analysed both using a nested case–control approach, a common approach, e.g. [19] which accounts for immortal time bias [20] without requiring complicated statistical techniques [21] (therefore is easier to understand and communicate to a clinical audience) with minimal loss of precision [22], and a time varying covariate approach, described later. Those who had died due to colorectal cancer (on the basis of a colorectal cancer ICD code as the underlying cause of death including C18, C19, C20, C21, C26.0, C26.8, C26.9 and equivalent ICD 9 codes) were considered cases. These were matched on sex, age (in 5-year intervals), year of cancer
Volume 24 | No. 12 | December 2013
time-varying covariate analysis The entire colorectal cancer cohort, before conversion to case–control data, was analysed applying survival analysis to investigate β-blocker use as a time varying covariate [20] (after 1 and 12 prescriptions, applying a 6-month lag to exclude prescriptions in the 6 months before death and ignoring the first year after cancer diagnosis). Using this approach, analysis was conducted adjusting for competing risk of deaths from other causes, using competing-risks regression based upon the Fine and Gray’s proportional sub-hazards model [23].
additional analysis Sensitivity analyses were conducted investigating individuals who used β-blockers after diagnosis to 1 year before death and β-blocker users 1 year after diagnosis to 6 months before death. Analyses were stratified by pre-diagnostic β-blocker use, stage and site (colon or rectum/rectosigmoid junction), re-matching where necessary. Further analysis compared β-blocker users to non-users who used any anti-hypertensive medications in the year before diagnosis including diuretics, β-adrenoceptor blocking drugs, vasodilator antihypertensive drugs, centrally acting antihypertensive drugs, α-adrenoceptor blocking drugs, angiotensin-converting enzyme inhibitors, angiotensin 2 receptor antagonists, renin inhibitors and calcium channel blockers. An a priori power calculation based upon predicted numbers and β-blocker usage estimated we would have 80% power to detect at the 5% level an OR of 0.80 for cancer-specific mortality. The final analysis contained 1559 colorectal cancer-specific deaths and 7531 matched controls and allowed over 80% power to detect as significant an OR of 0.80 in β-blocker users. All analyses were conducted using STATA 11 (StatCorp, College Station, TX).
results patient cohort Seven thousand and seventy-eight primary colorectal cancer patients were identified in CPRD and confirmed by NCDR diagnosis during 1998 and 2007. Of these, 446 were excluded as their diagnosis date preceded CPRD quality records. Eighty-one colorectal cancer patients were excluded because death registration coverage was not available. A further 1757 patients were excluded who had less than 1-year follow-up. The final cohort comprised 4794 colorectal cancer patients in which there were 1577 colorectal cancer- specific deaths. The average followup time in the cohort in those not dying was 6.2 years (range 1–13.9 years). Converting to a nested case–control dataset
doi:10.1093/annonc/mdt381 |
Downloaded from http://annonc.oxfordjournals.org/ at Northern Arizona University on July 9, 2015
A cohort study was conducted availing of recent linkages between the National Cancer Data Repository (NCDR), Clinical Practice Research Datalink (CPRD, formerly GPRD) and the Office of National Statistics (ONS).The NCDR comprises all English cancer registries including data on date, site of primary cancer diagnosis, stage and treatment data. The CPRD is the world’s largest computerized database of anonymised longitudinal patient records comprising ∼6% of the UK population including demographic information, clinical diagnoses and details of issued prescriptions which is of documented high quality [17]. CPRD also contains ONS mortality data which provides the date and cause of death up to 2011. Ethical approval for all observational research using CPRD has been obtained from a multicentre research ethics committee. Linkages were conducted using a deterministic algorithm based upon NHS number, gender, date of birth and postcode. Colorectal cancer cases were included if they had a CPRD colorectal cancer diagnosis code confirmed by an NCDR colorectal cancer diagnosis (including ICD codes C18 for Colon and C19/C20 for rectum) from 1998 to 2007. Individuals with a previously recorded cancer diagnosis with the exception of in situ neoplasms and non-melanoma skin cancers were excluded.
diagnosis (in 2-year intervals) and site (colon or rectum, including the rectosigmoid junction) to five risk set controls who lived at least as long after their cancer diagnosis as their matched case. The exposure period in cases was that from colorectal cancer diagnosis until 6 months before cancerspecific death and for controls was the same duration after diagnosis. Prescriptions within 6 months before death were removed as these may reflect end-of-life treatment. Analyses were restricted to individuals surviving at least 1 year after diagnosis to provide at least 6 months of exposure period and for biological plausibility. Analysis were conducted on β-blocker usage in the year (3 and 5 years) before cancer diagnosis. Conditional logistic regression was used to compare the risk of colorectal cancer-specific death by β-blocker usage, calculating odds ratios (ORs) and 95% confidence intervals (95% CIs). Adjusted analyses were conducted including adjustment for potential confounders. Similar analyses were conducted for all-cause mortality.
original articles
Annals of Oncology
Table 1. Characteristics of (cases) colorectal cancer-specific deaths and controls Characteristic
CRC-specific deaths, n (%) (n = 1559)
P
4368 (58.0) 3163 (42.0)
Matched
1842 (24.5) 2795 (37.1) 2894 (38.4) 17 (0.2) 334 (4.4) 1169 (15.5) 2034 (27.0) 2563 (34.0) 1229 (17.8) 75 (1.0)
751 (13.6) 4780 (86.4) 2000 5734 26.4 (4.7)
0.07 0.21 0.26 0.14 0.25
patient characteristics The characteristics of colorectal cancer-specific deaths (cases) and controls are detailed in Table 1. The average time from diagnosis to colorectal cancer-specific death was 3.0 years (range 1–12.5). Cases had higher stage (P < 0.001), higher grade (P < 0.001), higher rates of chemotherapy (P < 0.001) and radiotherapy (P < 0.001) and lower rates of surgery (P < 0.001). There was little difference in other characteristics between cases and controls with the exception of congestive heart disease which was more common among cases (6.0% versus 4.6%).
<0.001
association between post diagnostic β-blocker use and CRC mortality
<0.001
<0.001 <0.001 <0.001
0.04
0.27
0.32
557 (7.4) 1259 (16.7)
0.23 0.14
346 (4.6) 912 (12.1)
0.02 0.14 Continued
| Hicks et al.
455 (6.0) 523 (6.9) 330 (4.4) 316 (4.2) 573 (7.6)
Matched
Matched
3271 (52.2) 1902 (31.3) 1038 (16.6) 1259
107 (6.9) 114 (7.3) 65 (4.2) 56 (3.6) 119 (7.6)
P
resulted in 1559 colorectal cancer-specific deaths (cases) and 7531 controls.
3.0 (1.8)
6644 (88.2) 1118 (14.9) 1965 (20.1)
Myocardial infarction Peptic ulcer disease Peripheral vascular disease Rheumatological disease Renal disease
Controls, n (%) (n = 7531)
Italic values representing missing data.
Matched
518 (8.0) 5168 (79.5) 811 (12.5) 1034
CRC-specific deaths, n (%) (n = 1559)
Matched
4203 (55.8) 3328 (44.2)
1526 (23.8) 2534 (39.6) 2080 (32.5) 266 (4.2) 1125
Characteristic
The proportion using β-blockers after diagnosis was similar in colorectal cancer-specific deaths and controls (21.4% versus 23.7%, respectively; OR = 0.89; 95% CI 0.78–1.02; P = 0.11) (Table 2). There was no evidence of a dose response relationship between colorectal cancer-specific death and β-blocker use. No association was seen when restricting to non-cardioselective β-blockers (OR = 0.95; 95% CI 0.70–1.28; P = 0.72), cardioselective β-blockers (OR = 0.90; 95% CI 0.78–1.03; P = 0.13) or those which exhibit ISA activity (OR = 0.86; 95% CI 0.33–2.23; P = 0.76). Associations were seen with carvedilol (adjusted OR = 3.48; 95% CI 1.50–8.11; P < 0.01)) and sotalol (adjusted OR = 0.43; 95% CI 0.19–0.95; P = 0.04); however, these subgroups contained small numbers (28 carvedilolusers and 78 sotalol users on restricting for stage). Atenolol exhibited a reduction in risk of colorectal cancer-specific death (OR = 0.83; 95% CI 0.71–0.96; P = 0.02); however, this effect was slightly attenuated and no longer remained significant after adjustment for potential confounders (OR = 0.86; 95% CI 0.72–1.04). Adjusting for potential confounders did not notably alter other estimates. The Cox proportional hazards analysis of post-diagnostic β-blocker use as a time varying covariate gave similar results with no evidence of association for any β-blocker use (HR = 0.90; 95% CI 0.78 –1.03; P = 0.08), or for propranolol (HR = 1.32; 95% CI 0.88–1.98; P = 0.18) (see Table 3). Also, additional adjustment for competing risks of death made little difference to the overall result.
Volume 24 | No. 12 | December 2013
Downloaded from http://annonc.oxfordjournals.org/ at Northern Arizona University on July 9, 2015
Sex Male 896 (57.5) Female 663 (42.5) Year of CRC 1998–2000 382 (24.5) 2001–2003 578 (37.1) 2004–2006 599 (38.4) Age at CRC diagnosis (years) <40 12 (0.8) 40–49 84 (5.4) 50–59 237 (15.2) 60–69 408 (26.2) 70–79 513 (32.9) 80–89 278 (17.8) ≥90 27 (1.7) Site Colon 862 (55.3) Rectum (including 697 (44.7) rectosigmoid junction) Follow-up time 3.0 (1.8) (years) : mean (SD) Stage 1 138 (11.0) 2 296 (23.5) 3 562 (44.6) 4 263 (20.9) Unknown 300 Grade Well 82 (6.7) Moderately 926 (76.0) Poorly 210 (17.2) Unknown 341 Treatment within 6 months of CRC Surgery 1247 (80.0) Radiotherapy 309 (19.8) Chemotherapy 663 (42.5) Smoking before CRC Non-smoker 656 (50.7) Former smoker 378 (29.2) Current smoker 259 (20.0) Unknown 266 Alcohol before CRC Non-drinker 145 (12.9) Alcohol consumer 977 (87.1) Unknown 437 BMI (kg/m2) before CRC 1136 Mean (SD) 26.7 (4.7) Comorbidity pre- and post-CRC Cerebrovascular disease 120 (7.7) Chronic pulmonary 243 (15.6) disease Congestive heart disease 93 (6.0) Diabetes 199 (12.8)
Controls, n (%) (n = 7531)
Table 1.. Continued
original articles
Annals of Oncology
Table 2. Post-diagnostic exposure to β-blockers and odds of colorectal cancer-specific death in colorectal cancer patients Post-diagnostic β-blocker usage
Controls n (%)
Unadjusted OR (95% CI)
1225 (78.6) 334 (21.4)
5750 (76.4) 1781 (23.7)
1.00 0.89 (0.78, 1.02)
0.11
1.00 0.89 (0.76, 1.05)
0.18
1225 (78.6) 161 (10.3) 173 (11.1)
5750 (76.4) 886 (11.8) 895 (11.9)
1.00 0.87 (0.73, 1.04) 0.92 (0.77, 1.10)
0.14 0.36
1.00 0.85 (0.68, 1.06) 0.93 (0.76, 1.16)
0.16 0.53
1225 (78.6) 199 (12.8) 135 (8.7)
5750 (76.4) 1079 (14.3) 702 (9.3)
1.00 0.89 (0.75, 1.04) 0.91 (0.74, 1.11)
0.15 0.35
1.00 0.89 (0.72, 1.08) 0.91 (0.72, 1.14)
0.24 0.40
52 (3.3) 289 (18.5) 5 (0.3)
268 (3.6) 1547 (20.5) 29 (0.4)
0.95 (0.70, 1.28) 0.90 (0.78, 1.03) 0.86 (0.33, 2.23)
0.72 0.13 0.76
0.93 (0.64, 1.34) 0.90 (0.76, 1.06) 1.05 (0.30, 3.67)
0.69 0.21 0.95
221 (14.2) 58 (3.7) 18 (1.2) 26 (1.7) 15 (1.0) 10 (0.6)
1270 (16.9) 248 (3.3) 66 (0.9) 122 (1.6) 101 (1.3) 21 (0.3)
0.83 (0.71, 0.96) 1.16 (0.86, 1.57) 1.37 (0.81, 2.31) 1.04 (0.66, 1.59) 0.72 (0.42, 1.25) 2.46 (1.14, 5.31)
0.02 0.32 0.24 0.87 0.24 0.02
0.86 (0.72, 1.04) 0.97 (0.68, 1.38) 1.27 (0.69, 2.35) 1.27 (0.79, 2.05) 0.43 (0.19, 0.95) 3.48 (1.50, 8.11)
0.12 0.85 0.45 0.33 0.04 <0.01
P
a
Includes propranolol, sotolol, timolol,carvidolol, pindolol, oxprenolol and labetolol. Includes acebutolol, atenolol, bisoprolol, metoprolol and nebivolol. c Includes acebutolol, labetolol, pindolol and oxprenolol. d Model 1 includes treatment within 6 months including surgery, chemotherapy and radiotherapy), pre- and post-diagnostic comorbidities (including myocardial infarction, cerebrovascular disease, congestive heart disease, chronic pulmonary disease, peripheral vascular disease, peptic ulcer disease and diabetes) pre-diagnostic smoking (missing category included) and stage, restricted to individuals with complete staging information (1258 cancer-specific deaths and 6033 controls). b
association between post-diagnostic β-blocker use and all-cause mortality There was evidence of a small reduction in all-cause mortality in β-blocker users (adjusted OR = 0.88; 95% CI 0.77–1.00; P = 0.04). This association was similar in categories by prescription frequency or DDDs (Table 4). Analysis by β-blocker drug types revealed that the reduction in all-cause mortality was largely due to a reduction in mortality in atenolol users (adjusted OR = 0.79; 95% CI 0.69–0.92; P ≤ 0.01). Further analysis revealed that this effect was partly due to marked effects of atenolol on cardiovascular mortality (i.e. deaths were the underlying cause was cardiovascular) (adjusted OR = 0.62; 95% CI 0.40–0.97; P = 0.04, data not shown).
association between pre-diagnostic β-blocker use and CRC mortality 20.07% (338/1684) of cancer-specific deaths and 21.34% (1726/ 8089) of controls used β-blockers in the year before diagnosis demonstrating little evidence of association (OR = 0.93; 95% CI 0.80–1.08; P = 0.33) (Table 3). Similar analyses by type and investigating β-blockers over longer periods (3 and 5 years) showed little evidence of association.
Volume 24 | No. 12 | December 2013
sensitivity analysis The estimates remained similar when the exposure period was altered, when restricting controls to those who used other antihypertensive medications before diagnosis, when stratifying by type and early stage (Table 3).
discussion In this large population-based cohort study of colorectal patients, we did not observe an association between βblocker use and colorectal cancer-specific death. Limited evidence was found to suggest a small reduction in the risk of all-cause mortality in β-blocker users, particularly atenolol users. However, this association was partly due to marked associations between atenolol and cardiovascular mortality. This is the first epidemiological study to assess colorectal cancer-specific mortality and β-blockers. However, a recent study of a GP-diagnosed colorectal cancer cohort using a different UK primary care database [24] found no evidence of an association between β-blocker use (before diagnosis) and allcause mortality (HR = 1.00; 95% CI 0.77–1.30). Our findings are not inconsistent with their result but any difference in estimates could reflect fewer colorectal cancer cases (n = 619) and deaths
doi:10.1093/annonc/mdt381 |
Downloaded from http://annonc.oxfordjournals.org/ at Northern Arizona University on July 9, 2015
No prescriptions 0 1 or more No. of prescriptions 0 1-11 12 or more Daily Defined doses (DDDs) 0 DDDs 1–365 DDDs 366 or more DDDs Categories of β-blocker activity Cardio non-selectivea Cardio selectiveb ISA activityc Specific β-blocker Atenolol Bisoprolol Metoprolol Propranolol Sotalol Carvedilol
P
Adjustedd OR (95% CI)
CRC-specific deaths n (%)
Comparison
Time varying covariate analysesa Any post-diagnostic β-blocker usageb Any post-diagnostic β-blocker usage Any post-diagnostic propranolol usage Any post-diagnostic cardio non-selective usage Any post-diagnostic β-blocker usage adjusting for competing risksc β-Blocker use before colorectal cancer diagnosisd Any β-blocker use in previous 2 yearse Propranolol use in previous 2 yearse Cardio non-selective use in previous 2 yearse β-Blocker use in the previous 3 yearsf β-Blocker use in previous 5 yearsg Sensitivity analyses post-diagnostic β-blocker usageh Main analysis: diagnosis to 6 months before death Diagnosis to 1 year before deathi 1 year after diagnosis to 6 months before deathj Stage 1 and 2 Restricting controls to pre-diagnostic users of other hypertensive medicationsk Colon cancer Rectal cancer
CRC-specific deaths
Controls
OR/HR (95% CI) β-blocker users versus non users
P
OR/HR (95% CI) 1–11 prescriptions versus none
P
OR/HR (95% CI) 12+ prescriptions versus none
P
1576 1273 1273 1273 1576
3216 2761 2761 2761 3216
0.90 (0.79, 1.01) 0.90 (0.78, 1.03) 1.32 (0.88, 1.98) 0.97 (0.70, 1.35) 0.93 (0.81, 1.07)
0.08 0.13 0.18 0.87 0.32
0.88 (0.74, 1.03) 0.87 (0.72, 1.06) 1.34 (0.83, 2.16) 1.01 (0.66, 1.56) 0.90 (0.74,1.09)
0.11 0.16 0.24 0.96 0.27
0.93 (0.79, 1.09) 0.92 (0.77, 1.10) 1.28 (0.61, 2.70) 0.93 (0.57, 1.50) 0.96 (0.80, 1.15)
0.37 0.37 0.52 0.76 0.68
1684 1684 1684 1532 1174
8089 8089 8089 7345 5517
0.93 (0.80, 1.08) 0.93 (0.57, 1.50) 1.09 (0.77, 1.53) 0.91 (0.91, 1.38) 0.89 (0.75, 1.05)
0.33 0.75 0.63 0.67 0.17
0.93 (0.75, 1.15) 0.78 (0.39, 1.55) 0.91 (0.53, 1.57) 0.92 (0.52, 1.63) 0.79 (0.60, 1.03)
0.52 0.48 0.74 0.77 0.08
0.93 (0.78, 1.11) 1.11 (0.56, 2.20) 1.23 (0.79, 1.90) 0.91 (0.51, 1.63) 0.95 (0.78, 1.15)
0.41 0.76 0.36 0.76 0.57
1258 1012 798 426 481 738 520
6033 4840 3796 1929 2192 3573 2460
0.89 (0.76, 1.05) 0.91 (0.77, 1.09) 0.91 (0.75, 1.10) 0.84 (0.64, 1.11) 0.92 (0.74, 1.16) 0.91 (0.73, 1.12) 0.82 (0.65, 1.04)
0.18 0.31 0.34 0.22 0.50 0.38 0.10
0.85 (0.68, 1.06) 0.87 (0.68, 1.13) 0.84 (0.62, 1.14) 0.76 (0.52, 1.12) 0.91 (0.68, 1.23) 0.88 (0.66, 1.17) 0.72 (0.52, 1.01)
0.16 0.31 0.27 0.17 0.56 0.38 0.06
0.93 (0.76, 1.16) 0.94 (0.76, 1.16) 0.95 (0.75, 1.19) 0.91 (0.65, 1.28) 0.93 (0.70, 1.23) 0.94 (0.71, 1.24) 0.91 (0.68, 1.23)
0.53 0.56 0.63 0.60 0.62 0.66 0.54
β-Blocker use modelled as time varying covariate with an individual considered a non-user before 6 months after first (or before 365 days of use) and a user after this time (see methods section for more details). Reported estimates are hazard ratios and 95% CIs, adjusted for age (as continuous), year of colorectal cancer diagnosis (as continuous), site of cancer (colon or rectum/recto-sigmoid junction), gender, stage, prediagnostic comorbidities (including myocardial infarction, cerebrovascular disease, congestive heart disease, chronic pulmonary disease, peripheral vascular disease, peptic ulcer disease and diabetes), prediagnostic smoking and treatment within 6 months of diagnosis (surgery, chemotherapy and radiotherapy), unless otherwise stated. b Adjusted for sex, age (as continuous) and year of cancer diagnosis (as continuous). c Reported estimates are sub-distribution hazard ratios and 95% CIs, accounting for competing risks of death from other causes and adjusting for age (as continuous), year of colorectal cancer diagnosis (as continuous), site of cancer (colon or rectum/recto-sigmoid junction), gender and stage. d Analysis are adjusted for stage, treatment within 6 months of diagnosis (surgery, chemotherapy and radiotherapy), pre-diagnostic comorbidities (including myocardial infarction, cerebrovascular disease, congestive heart disease, chronic pulmonary disease, peripheral vascular disease, peptic ulcer disease and diabetes) and smoking (pre-diagnosis with missing as an included category). e Restricted to individuals with 2 years of medication records before diagnosis, not excluding deaths in the year after diagnosis. f Restricted to individuals with 3 years of medication records before diagnosis, not excluding deaths in the year after diagnosis. g Restricted to individuals with 5 years of medication records before diagnosis, not excluding deaths in the year after diagnosis. h All sensitivity analyses refer to β-blocker usage in the time period from colorectal cancer diagnosis to 6 months before death, and are adjusted for stage, treatment within 6 months of diagnosis (surgery, chemotherapy and radiotherapy), comorbidities pre-diagnosis or during follow-up, including myocardial infarction, cerebrovascular disease, congestive heart disease, chronic pulmonary disease, peripheral vascular disease, peptic ulcer disease and diabetes) and smoking (pre-diagnosis with missing as an included category), unless otherwise stated. i Restricted to individuals with at least 1.5 years follow-up so relevant exposure period is at least a duration of 6 months. j Restricted to individuals with at least 2 years follow-up so relevant exposure period is at least a duration of 6 months. k Restricted to individuals with at least 1 year’s medication records who used other antihypertensive medications in the year before diagnosis. Antihypertensive medications include diuretics, vasodilator antihypertensive drugs, centrally acting antihypertensive drugs, α-adrenoceptor drugs, β-adrenoceptor drugs, angiotensin converting enzyme inhibitors, angiotensin 2 receptor antagonists, renin inhibitors and calcium channel blockers.
original articles
| Hicks et al.
Table 3. Additional analysis for association between β-blockers and colorectal cancer-specific death in colorectal cancer
a
Annals of Oncology
Downloaded from http://annonc.oxfordjournals.org/ at Northern Arizona University on July 9, 2015
Volume 24 | No. 12 | December 2013
original articles
Annals of Oncology Table 4. Post-diagnostic exposure to β-blockers and odds of death in colorectal cancer patients Post-diagnostic β-blocker usage
Controls n (%)
Unadjusted OR (95% CI)
1685 (76.1) 529 (23.9)
7902 (74.2) 2753 (25.8)
1.00 0.90 (0.81, 1.01)
0.06
1.00 0.88 (0.77, 1.00)
0.04
1685 (76.1) 231 (10.4) 298 (13.5)
7902 (74.2) 1234 (11.6) 1519 (14.3)
1.00 0.89 (0.76, 1.04) 0.91 (0.79, 1.05)
0.13 0.19
1.00 0.86 (0.72, 1.03) 0.89 (0.76, 1.04)
0.10 0.15
1685 (76.1) 291 (13.1) 238 (10.8)
7902 (74.2) 1563 (14.7) 1190 (11.2)
1.00 0.88 (0.77, 1.01) 0.93 (0.79, 1.08)
0.08 0.32
1.00 0.89 (0.76, 1.05) 0.86 (0.72, 1.02)
0.12 0.09
86 (3.9) 461 (20.8) 8 (0.4)
435 (4.1) 2405 (22.6) 55 (0.5)
0.96 (0.76, 1.23) 0.90 (0.81, 1.01) 0.73 (0.35, 1.53)
0.74 0.08 0.40
0.95 (0.72, 1.26) 0.88 (0.77, 1.01) 0.77 (0.32, 1.82)
0.74 0.07 0.55
334 (15.1) 119 (5.4) 34 (1.5) 37 (1.7) 32 (1.5) 15 (0.7)
1895 (17.8) 484 (4.5) 125 (1.2) 151 (1.4) 192 (1.8) 41 (0.4)
0.82 (0.72, 0.93) 1.20 (0.97, 1.49) 1.34 (0.92, 1.96) 1.20 (0.84, 1.73) 0.80 (0.54, 1.16) 1.84 (1.02, 3.33)
<0.01 0.09 0.13 0.32 0.24 0.05
0.79 (0.69, 0.92) 1.10 (0.86, 1.41) 1.66 (1.07, 2.55) 1.10 (0.74, 1.62) 0.82 (0.50, 1.34) 2.10 (1.07, 4.12)
<0.01 0.46 0.02 0.64 0.43 0.03
P
a
Includes propranolol, sotolol, timolol, carvidolol, pindolol, oxprenolol and labetolol. Includes acebutolol, atenolol, bisoprolol, metoprolol and nebivolol. c Iincludes acebutolol, labetolol, pindolol and oxprenolol. d Model 1 includes treatment within 6 months (including surgery, chemotherapy and radiotherapy), pre- and post-diagnostic comorbidities (including myocardial infarction, cerebrovascular disease, congestive heart disease, chronic pulmonary disease, peripheral vascular disease, peptic ulcer disease and diabetes) pre-diagnostic smoking (missing category included) and stage, restricted to individuals with complete staging information (1812 all-cause deaths and 8643 controls). b
in their study, their lack of linkage to cancer registries, their inability to account for potentially important confounders such as stage and treatment or the exclusion of patients with strong indications for β-blockers. The associations with carvedilol and sotalol differ despite both being non-cardioselective β-blockers. These associations are likely to be due to chance because of small numbers within subgroups; however, it is possible that these drugs may increase or decrease the risk of colorectal cancer death, respectively. Our study has several strengths. It is large allowing power to detect even relatively weak associations. Linkage with NCDR and ONS data allowed robust verification of cancer diagnosis and death data, respectively. Using GP prescribing data allowed for temporal relationships to be investigated ruling out any recall bias that exists in questionnaire-based studies. A further strength is that, as β-blockers are not available over the counter in the UK, we will have captured most usage. Although consumption cannot be guaranteed when using prescriptions, null associations in our analysis of 12 or more prescriptions indicate that non-compliance is unlikely to be affecting our findings. Another weakness is possible bias due to misclassification of cancer-specific death. However, a recent methodological study concluded that in comparative studies where misclassification is unlikely to be differential, as in our study, it is unlikely to impact greatly upon results [25]. Another
Volume 24 | No. 12 | December 2013
weakness is the possibility of residual confounding. Although we adjusted for important potential confounders such as sex, stage, treatment and other co-morbidities, we could not adjust for others such as socioeconomic status. Confounding by indication is often a problem in pharmaco-epidemiology; however, in our analysis, we compared β-blocker users to users of other antihypertensive medications to reduce the possibility of such bias. In conclusion, in this large population-based study, we did not find an association between β-blocker use and colorectal cancer-specific mortality.
acknowledgements This study is based in part on data from the General Practice Research Database obtained under license from the UK Medicines and Healthcare Regulatory Agency. However, the interpretation and conclusions contained in this study are those of the authors alone.
funding Access to the dataset was funded through a related Cancer Research-UK funded grant (C39066/A14597). BH was funded
doi:10.1093/annonc/mdt381 |
Downloaded from http://annonc.oxfordjournals.org/ at Northern Arizona University on July 9, 2015
No prescriptions 0 1 or more No of prescriptions 0 1–11 12 or more Daily defined doses (DDDs) 0 DDDs 1–365 DDDs 366 or more DDDs Categories of β-blocker activity Cardio non-selectivea Cardio selectiveb ISA activityc Specific β-blocker Atenolol Bisoprolol Metoprolol Propranolol Sotalol Carvedilol
P
Adjustedd OR (95% CI)
All-cause deaths n (%)
original articles by a Northern Ireland Department of Education and Learning PhD studentship.
disclosure DP has a patent on the use of α- and β-blockers to treat cancer metastasis, patent number 13/314 437. All remaining authors have declared no conflicts of interest.
references
| Hicks et al.
11. Barron TI, Connolly RM, Sharp L et al. Beta blockers and breast cancer mortality: a population- based study. J Clin Oncol 2011; 29(19): 2635–2644. 12. Powe DG, Voss MJ, Zänker KS et al. Beta-blocker drug therapy reduces secondary cancer formation in breast cancer and improves cancer specific survival. Oncotarget 2010; 1(7): 628–638. 13. Wang HM, Liao ZX, Komaki R et al. Improved survival outcomes with the incidental use of beta-blockers among patients with non-small-cell lung cancer treated with definitive radiation therapy. Ann Oncol 2013; 24(5): 1312–1319. 14. Grytli HH, Fagerland MW, Fosså SD et al. Association between use of β-blockers and prostate cancer-specific survival: a cohort study of 3561 prostate cancer patients with high-risk or metastatic disease. Eur Urol 2013 January 14 [epub ahead of print], doi: 10.1016/j.eururo.2013.01.007. 15. Diaz ES, Karlan BY, Li AJ. Impact of beta blockers on epithelial ovarian cancer survival. Gynecol Oncol 2012; 127(2): 375–378. 16. Zmora O. β-adrenergic Blocker and a COX2 Inhibitor for Prevention of Colorectal Cancer Recurrence. In: ClinicalTrials.gov. Bethesda, MD: National Library of Medicine. Available from:http://clinicaltrials.gov/show/NCT00888797 (29 April 2013, date last accessed). 17. Jick H, Jick SS, Derby LE. Validation of information recorded on general practitioner based computerised data resource in the United Kingdom. BMJ 1991; 302(6779): 766–768. 18. Khan NF, Perera R, Harper S et al. Adaptation and validation of the Charlson Index for Read/OXMIS coded databases. BMC Fam Pract 2010; 11: 1. 19. Schade R, Andersohn F, Suissa S et al. Dopamine agonists and the risk of cardiacvalve regurgitation. N engl J Med 2007; 356(1): 29–38. 20. Lévesque LE, Hanley JA, Kezouh A et al. Problem of immortal time bias in cohort studies: example using statins for preventing progression of diabetes. BMJ 2010; 340: b5087. 21. Etminan M. Pharmacoepidemiology II: the nested case-control study—a novel approach in pharmacoepidemiologic research. Pharmacotherapy 2004; 24(9): 1105–1109. 22. Essebag V, Platt RW, Abrahamowicz M et al. Comparison of nested case-control and survival analysis methodologies for analysis of time-dependent exposure. BMC Med Res Methodol 2005; 5(1): 5. 23. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999; 94: 496–509. 24. Shah SM, Carey IM, Owen CG et al. Does β-adrenoceptor blocker therapy improve cancer survival? Findings from a population-based retrospective cohort study. Br J Clin Pharmacol 2011; 72(1): 157–161. 25. Sarfati D, Blakely T, Pearce N. Measuring cancer survival in populations: relative survival versus cancer-specific survival. Int J Epidemiol 2010; 39(2): 598–610.
Volume 24 | No. 12 | December 2013
Downloaded from http://annonc.oxfordjournals.org/ at Northern Arizona University on July 9, 2015
1. Yang EV, Sood AK, Chen M et al. Norepinephrine up-regulates the expression of vascular endothelial growth factor, matrix metalloproteinase (MMP)-2, and MMP-9 in nasopharyngeal carcinoma tumor cells. Cancer Res 2006; 66(21): 10357–10364. 2. Weddle DL. beta-Adrenergic growth regulation of human cancer cell lines derived from pancreatic ductal carcinomas. Carcinogenesis 2001; 22(3): 473–479. 3. Masur K, Niggemann B, Zanker KS et al. Norepinephrine-induced migration of SW 480 colon carcinoma cells is inhibited by {beta}-blockers. Cancer Res 2001; 61 (7): 2866–2869. 4. Pérez Piñero C, Bruzzone A, Sarappa MG et al. Involvement of α2- and β2adrenoceptors on breast cancer cell proliferation and tumour growth regulation. Br J Pharmacol 2012; 166(2): 721–736. 5. Yang EV, Kim S, Donovan EL et al. Norepinephrine upregulates VEGF, IL-8, and IL-6 expression in human melanoma tumor cell lines: implications for stressrelated enhancement of tumor progression. Brain Behav Immun 2009; 23(2): 267–275. 6. Fitzgerald PJ. Is norepinephrine an etiological factor in some types of cancer? Int J Cancer 2009; 124(2): 257–263. 7. Kozanoglu I, Yandim MK, Cincin ZB et al. New indication for therapeutic potential of an old well-known drug ( propranolol) for multiple myeloma. J Cancer Res Clin Oncol 2013; 139(2): 327–335. 8. Lin X, Luo K, Lv Z et al. Beta-adrenoceptor action on pancreatic cancer cell proliferation and tumor growth in mice. Hepatogastroenterology 2012; 59(114): 584–588. 9. Zhang D, Yong Ma Q, Hu H-T et al. β 2-Adrenergic antagonists suppress pancreatic cancer cell invasion by inhibiting CREB, NF-κB and AP-1. Cancer Biol Ther 2010; 10(1): 19–29. 10. Lin Q, Wang F, Yang R et al. Effect of chronic restraint stress on human colorectal carcinoma growth in mice. PLoS One 2013; 8(4): e61435.
Annals of Oncology
Annals of Oncology Annals of Oncology 24: 3100–3106, 2013 doi:10.1093/annonc/mdt381 Published online 19 September 2013
β-Blocker usage and colorectal cancer mortality: a nested case–control study in the UK Clinical Practice Research Datalink cohort B. M. Hicks1*, L. J. Murray1, D. G. Powe2,3, C. M. Hughes4 & C. R. Cardwell1 1 Centre for Public Health, Queen’s University Belfast, Belfast, Northern, Ireland; 2Department of Cellular Pathology, Queen’s Medical Centre, Nottingham University Hospitals NHS Trust, Nottingham; 3The John Van Geest Cancer Research Centre, Nottingham Trent University, Nottingham, UK; 4School of Pharmacy, Queen’s University Belfast, Belfast, Northern, Ireland
Background: Epidemiological and laboratory studies suggest that β-blockers may reduce cancer progression in various cancer sites. The aim of this study was to conduct the first epidemiological investigation of the effect of post-diagnostic β-blocker usage on colorectal cancer-specific mortality in a large population-based colorectal cancer patient cohort. Patients and methods: A nested case–control analysis was conducted within a cohort of 4794 colorectal cancer patients diagnosed between 1998 and 2007. Patients were identified from the UK Clinical Practice Research Datalink and confirmed using cancer registry data. Patients with a colorectal cancer- specific death (data from the Office of National Statistics death registration system) were matched to five controls. Conditional logistic regression was applied to calculate odds ratios (OR) and 95% confidence intervals (95% CIs) according to β-blocker usage (data from GP-prescribing records). Results: Post-diagnostic β-blocker use was identified in 21.4% of 1559 colorectal cancer-specific deaths and 23.7% of their 7531 matched controls, with little evidence of an association (OR = 0.89 95% CI 0.78–1.02). Similar associations were found when analysing drug frequency, β-blocker type or specific drugs such as propranolol. There was some evidence of a weak reduction in all-cause mortality in β-blocker users (adjusted OR = 0.88; 95% CI 0.77–1.00; P = 0.04) which was in part due to the marked effect of atenolol on cardiovascular mortality (adjusted OR = 0.62; 95% CI 0.40–0.97; P = 0.04). Conclusions: In this novel, large UK population-based cohort of colorectal cancer patients, there was no evidence of an association between post-diagnostic β-blocker use and colorectal cancer-specific mortality. Clinical Trials number: NCT00888797. Key words: colorectal cancer survival, β-blockers, pharmacoepidemiology, medication, propranolol, atenolol
introduction β-Blockers are used to treat heart disease and hypertension but recent evidence suggests they may protect against cancer progression. β-Blockers act by inhibiting β-adrenergic receptors, which are expressed by various cancer types including breast, pancreatic and colon cancers [1–3]. β-Adrenergic signalling has been shown to be relevant in cancer progression [4–6]. In vitro studies have demonstrated that β-blockers can inhibit cancer cell proliferation, invasion and resistance to apoptosis [4, 7–9]. A study found that colorectal tumour cell growth in mice and cell proliferation in colorectal cancer cell lines were inhibited by the β-blockers propranolol and atenolol, respectively [10].
*Correspondence to: Dr Blánaid Hicks, Centre for Public Health, Queen’s University Belfast, Institute of Clinical Science, Block B, Royal Victoria Hospital, Belfast, BT12 6BA, Northern Ireland. Tel: +44-0-9063-5009; Fax: +44-0-9023-5900; E-mail: [email protected]
Norepinephrine has been shown to induce the migration of colon carcinoma cells, a process inhibited by the non-selective β-blocker propranolol suggesting β-blockers could be a useful treatment of colorectal cancer [3]. Epidemiological studies have shown reductions in cancerspecific mortality with post-diagnostic β-blocker use for breast [11, 12], lung [13], prostate [14] and ovarian cancers [15]. Despite these findings and in vitro evidence for colon cancer [3], epidemiological studies have not investigated the effect of β-blocker use on cancer-specific mortality in colorectal cancer patients. Despite this lack of epidemiology, a phase III clinical trial is underway, investigating the effect of the β-blocker propranolol on recurrence in colorectal cancer patients. However, this study will take several years to report (estimated 2017), includes a relatively small number of colorectal cancer patients (200 in the β-blocker group), is investigating a specific drug ( propranolol), used in a specific time period (20 days in total, pre- and post-surgery) in early disease [16]. Evidence
© The Author 2013. Published by Oxford University Press on behalf of the European Society for Medical Oncology. All rights reserved. For permissions, please email: [email protected].
Downloaded from http://annonc.oxfordjournals.org/ at Northern Arizona University on July 9, 2015
Received 22 July 2013; revised 5 August 2013; accepted 7 August 2013
original articles
Annals of Oncology
from well carried out observational studies is therefore required to inform future studies. Our study is the first to investigate the effect of postdiagnostic β-blocker usage on colorectal cancer-specific mortality in a large prospective UK population-based cohort of colorectal cancer patients and importantly will investigate a range of β-blocker medications, time points and cancer stages.
methods study design
exposure data β-blockers were determined from GP prescribing data and defined as drugs classified in the British National Formulary (BNF) chapter 2.4. Using prescriptions, daily defined doses (DDD) were calculated on the basis of quantity and strength (http://www.whocc.no/atc_ddd_index/?Code=C07A). A quantity of 28 tablets was assumed for ∼5% of prescriptions where quantity was missing or inconsistent.
confounders Cancer stage and treatment within 6 months after diagnosis (surgery, chemotherapy and radiotherapy) were determined from NCDR data. Smoking, alcohol consumption and body mass index (BMI) were determined from the closest GP record before a colorectal cancer diagnosis (records older than 10 years were ignored). Comorbidities before and during the exposure period were determined from GP diagnosis codes based upon nine of the most common diagnoses contributing to a CPRD adaptation of the Charlson comorbidity index [18].
data analysis This cohort was analysed both using a nested case–control approach, a common approach, e.g. [19] which accounts for immortal time bias [20] without requiring complicated statistical techniques [21] (therefore is easier to understand and communicate to a clinical audience) with minimal loss of precision [22], and a time varying covariate approach, described later. Those who had died due to colorectal cancer (on the basis of a colorectal cancer ICD code as the underlying cause of death including C18, C19, C20, C21, C26.0, C26.8, C26.9 and equivalent ICD 9 codes) were considered cases. These were matched on sex, age (in 5-year intervals), year of cancer
Volume 24 | No. 12 | December 2013
time-varying covariate analysis The entire colorectal cancer cohort, before conversion to case–control data, was analysed applying survival analysis to investigate β-blocker use as a time varying covariate [20] (after 1 and 12 prescriptions, applying a 6-month lag to exclude prescriptions in the 6 months before death and ignoring the first year after cancer diagnosis). Using this approach, analysis was conducted adjusting for competing risk of deaths from other causes, using competing-risks regression based upon the Fine and Gray’s proportional sub-hazards model [23].
additional analysis Sensitivity analyses were conducted investigating individuals who used β-blockers after diagnosis to 1 year before death and β-blocker users 1 year after diagnosis to 6 months before death. Analyses were stratified by pre-diagnostic β-blocker use, stage and site (colon or rectum/rectosigmoid junction), re-matching where necessary. Further analysis compared β-blocker users to non-users who used any anti-hypertensive medications in the year before diagnosis including diuretics, β-adrenoceptor blocking drugs, vasodilator antihypertensive drugs, centrally acting antihypertensive drugs, α-adrenoceptor blocking drugs, angiotensin-converting enzyme inhibitors, angiotensin 2 receptor antagonists, renin inhibitors and calcium channel blockers. An a priori power calculation based upon predicted numbers and β-blocker usage estimated we would have 80% power to detect at the 5% level an OR of 0.80 for cancer-specific mortality. The final analysis contained 1559 colorectal cancer-specific deaths and 7531 matched controls and allowed over 80% power to detect as significant an OR of 0.80 in β-blocker users. All analyses were conducted using STATA 11 (StatCorp, College Station, TX).
results patient cohort Seven thousand and seventy-eight primary colorectal cancer patients were identified in CPRD and confirmed by NCDR diagnosis during 1998 and 2007. Of these, 446 were excluded as their diagnosis date preceded CPRD quality records. Eighty-one colorectal cancer patients were excluded because death registration coverage was not available. A further 1757 patients were excluded who had less than 1-year follow-up. The final cohort comprised 4794 colorectal cancer patients in which there were 1577 colorectal cancer- specific deaths. The average followup time in the cohort in those not dying was 6.2 years (range 1–13.9 years). Converting to a nested case–control dataset
doi:10.1093/annonc/mdt381 |
Downloaded from http://annonc.oxfordjournals.org/ at Northern Arizona University on July 9, 2015
A cohort study was conducted availing of recent linkages between the National Cancer Data Repository (NCDR), Clinical Practice Research Datalink (CPRD, formerly GPRD) and the Office of National Statistics (ONS).The NCDR comprises all English cancer registries including data on date, site of primary cancer diagnosis, stage and treatment data. The CPRD is the world’s largest computerized database of anonymised longitudinal patient records comprising ∼6% of the UK population including demographic information, clinical diagnoses and details of issued prescriptions which is of documented high quality [17]. CPRD also contains ONS mortality data which provides the date and cause of death up to 2011. Ethical approval for all observational research using CPRD has been obtained from a multicentre research ethics committee. Linkages were conducted using a deterministic algorithm based upon NHS number, gender, date of birth and postcode. Colorectal cancer cases were included if they had a CPRD colorectal cancer diagnosis code confirmed by an NCDR colorectal cancer diagnosis (including ICD codes C18 for Colon and C19/C20 for rectum) from 1998 to 2007. Individuals with a previously recorded cancer diagnosis with the exception of in situ neoplasms and non-melanoma skin cancers were excluded.
diagnosis (in 2-year intervals) and site (colon or rectum, including the rectosigmoid junction) to five risk set controls who lived at least as long after their cancer diagnosis as their matched case. The exposure period in cases was that from colorectal cancer diagnosis until 6 months before cancerspecific death and for controls was the same duration after diagnosis. Prescriptions within 6 months before death were removed as these may reflect end-of-life treatment. Analyses were restricted to individuals surviving at least 1 year after diagnosis to provide at least 6 months of exposure period and for biological plausibility. Analysis were conducted on β-blocker usage in the year (3 and 5 years) before cancer diagnosis. Conditional logistic regression was used to compare the risk of colorectal cancer-specific death by β-blocker usage, calculating odds ratios (ORs) and 95% confidence intervals (95% CIs). Adjusted analyses were conducted including adjustment for potential confounders. Similar analyses were conducted for all-cause mortality.
original articles
Annals of Oncology
Table 1. Characteristics of (cases) colorectal cancer-specific deaths and controls Characteristic
CRC-specific deaths, n (%) (n = 1559)
P
4368 (58.0) 3163 (42.0)
Matched
1842 (24.5) 2795 (37.1) 2894 (38.4) 17 (0.2) 334 (4.4) 1169 (15.5) 2034 (27.0) 2563 (34.0) 1229 (17.8) 75 (1.0)
751 (13.6) 4780 (86.4) 2000 5734 26.4 (4.7)
0.07 0.21 0.26 0.14 0.25
patient characteristics The characteristics of colorectal cancer-specific deaths (cases) and controls are detailed in Table 1. The average time from diagnosis to colorectal cancer-specific death was 3.0 years (range 1–12.5). Cases had higher stage (P < 0.001), higher grade (P < 0.001), higher rates of chemotherapy (P < 0.001) and radiotherapy (P < 0.001) and lower rates of surgery (P < 0.001). There was little difference in other characteristics between cases and controls with the exception of congestive heart disease which was more common among cases (6.0% versus 4.6%).
<0.001
association between post diagnostic β-blocker use and CRC mortality
<0.001
<0.001 <0.001 <0.001
0.04
0.27
0.32
557 (7.4) 1259 (16.7)
0.23 0.14
346 (4.6) 912 (12.1)
0.02 0.14 Continued
| Hicks et al.
455 (6.0) 523 (6.9) 330 (4.4) 316 (4.2) 573 (7.6)
Matched
Matched
3271 (52.2) 1902 (31.3) 1038 (16.6) 1259
107 (6.9) 114 (7.3) 65 (4.2) 56 (3.6) 119 (7.6)
P
resulted in 1559 colorectal cancer-specific deaths (cases) and 7531 controls.
3.0 (1.8)
6644 (88.2) 1118 (14.9) 1965 (20.1)
Myocardial infarction Peptic ulcer disease Peripheral vascular disease Rheumatological disease Renal disease
Controls, n (%) (n = 7531)
Italic values representing missing data.
Matched
518 (8.0) 5168 (79.5) 811 (12.5) 1034
CRC-specific deaths, n (%) (n = 1559)
Matched
4203 (55.8) 3328 (44.2)
1526 (23.8) 2534 (39.6) 2080 (32.5) 266 (4.2) 1125
Characteristic
The proportion using β-blockers after diagnosis was similar in colorectal cancer-specific deaths and controls (21.4% versus 23.7%, respectively; OR = 0.89; 95% CI 0.78–1.02; P = 0.11) (Table 2). There was no evidence of a dose response relationship between colorectal cancer-specific death and β-blocker use. No association was seen when restricting to non-cardioselective β-blockers (OR = 0.95; 95% CI 0.70–1.28; P = 0.72), cardioselective β-blockers (OR = 0.90; 95% CI 0.78–1.03; P = 0.13) or those which exhibit ISA activity (OR = 0.86; 95% CI 0.33–2.23; P = 0.76). Associations were seen with carvedilol (adjusted OR = 3.48; 95% CI 1.50–8.11; P < 0.01)) and sotalol (adjusted OR = 0.43; 95% CI 0.19–0.95; P = 0.04); however, these subgroups contained small numbers (28 carvedilolusers and 78 sotalol users on restricting for stage). Atenolol exhibited a reduction in risk of colorectal cancer-specific death (OR = 0.83; 95% CI 0.71–0.96; P = 0.02); however, this effect was slightly attenuated and no longer remained significant after adjustment for potential confounders (OR = 0.86; 95% CI 0.72–1.04). Adjusting for potential confounders did not notably alter other estimates. The Cox proportional hazards analysis of post-diagnostic β-blocker use as a time varying covariate gave similar results with no evidence of association for any β-blocker use (HR = 0.90; 95% CI 0.78 –1.03; P = 0.08), or for propranolol (HR = 1.32; 95% CI 0.88–1.98; P = 0.18) (see Table 3). Also, additional adjustment for competing risks of death made little difference to the overall result.
Volume 24 | No. 12 | December 2013
Downloaded from http://annonc.oxfordjournals.org/ at Northern Arizona University on July 9, 2015
Sex Male 896 (57.5) Female 663 (42.5) Year of CRC 1998–2000 382 (24.5) 2001–2003 578 (37.1) 2004–2006 599 (38.4) Age at CRC diagnosis (years) <40 12 (0.8) 40–49 84 (5.4) 50–59 237 (15.2) 60–69 408 (26.2) 70–79 513 (32.9) 80–89 278 (17.8) ≥90 27 (1.7) Site Colon 862 (55.3) Rectum (including 697 (44.7) rectosigmoid junction) Follow-up time 3.0 (1.8) (years) : mean (SD) Stage 1 138 (11.0) 2 296 (23.5) 3 562 (44.6) 4 263 (20.9) Unknown 300 Grade Well 82 (6.7) Moderately 926 (76.0) Poorly 210 (17.2) Unknown 341 Treatment within 6 months of CRC Surgery 1247 (80.0) Radiotherapy 309 (19.8) Chemotherapy 663 (42.5) Smoking before CRC Non-smoker 656 (50.7) Former smoker 378 (29.2) Current smoker 259 (20.0) Unknown 266 Alcohol before CRC Non-drinker 145 (12.9) Alcohol consumer 977 (87.1) Unknown 437 BMI (kg/m2) before CRC 1136 Mean (SD) 26.7 (4.7) Comorbidity pre- and post-CRC Cerebrovascular disease 120 (7.7) Chronic pulmonary 243 (15.6) disease Congestive heart disease 93 (6.0) Diabetes 199 (12.8)
Controls, n (%) (n = 7531)
Table 1.. Continued
original articles
Annals of Oncology
Table 2. Post-diagnostic exposure to β-blockers and odds of colorectal cancer-specific death in colorectal cancer patients Post-diagnostic β-blocker usage
Controls n (%)
Unadjusted OR (95% CI)
1225 (78.6) 334 (21.4)
5750 (76.4) 1781 (23.7)
1.00 0.89 (0.78, 1.02)
0.11
1.00 0.89 (0.76, 1.05)
0.18
1225 (78.6) 161 (10.3) 173 (11.1)
5750 (76.4) 886 (11.8) 895 (11.9)
1.00 0.87 (0.73, 1.04) 0.92 (0.77, 1.10)
0.14 0.36
1.00 0.85 (0.68, 1.06) 0.93 (0.76, 1.16)
0.16 0.53
1225 (78.6) 199 (12.8) 135 (8.7)
5750 (76.4) 1079 (14.3) 702 (9.3)
1.00 0.89 (0.75, 1.04) 0.91 (0.74, 1.11)
0.15 0.35
1.00 0.89 (0.72, 1.08) 0.91 (0.72, 1.14)
0.24 0.40
52 (3.3) 289 (18.5) 5 (0.3)
268 (3.6) 1547 (20.5) 29 (0.4)
0.95 (0.70, 1.28) 0.90 (0.78, 1.03) 0.86 (0.33, 2.23)
0.72 0.13 0.76
0.93 (0.64, 1.34) 0.90 (0.76, 1.06) 1.05 (0.30, 3.67)
0.69 0.21 0.95
221 (14.2) 58 (3.7) 18 (1.2) 26 (1.7) 15 (1.0) 10 (0.6)
1270 (16.9) 248 (3.3) 66 (0.9) 122 (1.6) 101 (1.3) 21 (0.3)
0.83 (0.71, 0.96) 1.16 (0.86, 1.57) 1.37 (0.81, 2.31) 1.04 (0.66, 1.59) 0.72 (0.42, 1.25) 2.46 (1.14, 5.31)
0.02 0.32 0.24 0.87 0.24 0.02
0.86 (0.72, 1.04) 0.97 (0.68, 1.38) 1.27 (0.69, 2.35) 1.27 (0.79, 2.05) 0.43 (0.19, 0.95) 3.48 (1.50, 8.11)
0.12 0.85 0.45 0.33 0.04 <0.01
P
a
Includes propranolol, sotolol, timolol,carvidolol, pindolol, oxprenolol and labetolol. Includes acebutolol, atenolol, bisoprolol, metoprolol and nebivolol. c Includes acebutolol, labetolol, pindolol and oxprenolol. d Model 1 includes treatment within 6 months including surgery, chemotherapy and radiotherapy), pre- and post-diagnostic comorbidities (including myocardial infarction, cerebrovascular disease, congestive heart disease, chronic pulmonary disease, peripheral vascular disease, peptic ulcer disease and diabetes) pre-diagnostic smoking (missing category included) and stage, restricted to individuals with complete staging information (1258 cancer-specific deaths and 6033 controls). b
association between post-diagnostic β-blocker use and all-cause mortality There was evidence of a small reduction in all-cause mortality in β-blocker users (adjusted OR = 0.88; 95% CI 0.77–1.00; P = 0.04). This association was similar in categories by prescription frequency or DDDs (Table 4). Analysis by β-blocker drug types revealed that the reduction in all-cause mortality was largely due to a reduction in mortality in atenolol users (adjusted OR = 0.79; 95% CI 0.69–0.92; P ≤ 0.01). Further analysis revealed that this effect was partly due to marked effects of atenolol on cardiovascular mortality (i.e. deaths were the underlying cause was cardiovascular) (adjusted OR = 0.62; 95% CI 0.40–0.97; P = 0.04, data not shown).
association between pre-diagnostic β-blocker use and CRC mortality 20.07% (338/1684) of cancer-specific deaths and 21.34% (1726/ 8089) of controls used β-blockers in the year before diagnosis demonstrating little evidence of association (OR = 0.93; 95% CI 0.80–1.08; P = 0.33) (Table 3). Similar analyses by type and investigating β-blockers over longer periods (3 and 5 years) showed little evidence of association.
Volume 24 | No. 12 | December 2013
sensitivity analysis The estimates remained similar when the exposure period was altered, when restricting controls to those who used other antihypertensive medications before diagnosis, when stratifying by type and early stage (Table 3).
discussion In this large population-based cohort study of colorectal patients, we did not observe an association between βblocker use and colorectal cancer-specific death. Limited evidence was found to suggest a small reduction in the risk of all-cause mortality in β-blocker users, particularly atenolol users. However, this association was partly due to marked associations between atenolol and cardiovascular mortality. This is the first epidemiological study to assess colorectal cancer-specific mortality and β-blockers. However, a recent study of a GP-diagnosed colorectal cancer cohort using a different UK primary care database [24] found no evidence of an association between β-blocker use (before diagnosis) and allcause mortality (HR = 1.00; 95% CI 0.77–1.30). Our findings are not inconsistent with their result but any difference in estimates could reflect fewer colorectal cancer cases (n = 619) and deaths
doi:10.1093/annonc/mdt381 |
Downloaded from http://annonc.oxfordjournals.org/ at Northern Arizona University on July 9, 2015
No prescriptions 0 1 or more No. of prescriptions 0 1-11 12 or more Daily Defined doses (DDDs) 0 DDDs 1–365 DDDs 366 or more DDDs Categories of β-blocker activity Cardio non-selectivea Cardio selectiveb ISA activityc Specific β-blocker Atenolol Bisoprolol Metoprolol Propranolol Sotalol Carvedilol
P
Adjustedd OR (95% CI)
CRC-specific deaths n (%)
Comparison
Time varying covariate analysesa Any post-diagnostic β-blocker usageb Any post-diagnostic β-blocker usage Any post-diagnostic propranolol usage Any post-diagnostic cardio non-selective usage Any post-diagnostic β-blocker usage adjusting for competing risksc β-Blocker use before colorectal cancer diagnosisd Any β-blocker use in previous 2 yearse Propranolol use in previous 2 yearse Cardio non-selective use in previous 2 yearse β-Blocker use in the previous 3 yearsf β-Blocker use in previous 5 yearsg Sensitivity analyses post-diagnostic β-blocker usageh Main analysis: diagnosis to 6 months before death Diagnosis to 1 year before deathi 1 year after diagnosis to 6 months before deathj Stage 1 and 2 Restricting controls to pre-diagnostic users of other hypertensive medicationsk Colon cancer Rectal cancer
CRC-specific deaths
Controls
OR/HR (95% CI) β-blocker users versus non users
P
OR/HR (95% CI) 1–11 prescriptions versus none
P
OR/HR (95% CI) 12+ prescriptions versus none
P
1576 1273 1273 1273 1576
3216 2761 2761 2761 3216
0.90 (0.79, 1.01) 0.90 (0.78, 1.03) 1.32 (0.88, 1.98) 0.97 (0.70, 1.35) 0.93 (0.81, 1.07)
0.08 0.13 0.18 0.87 0.32
0.88 (0.74, 1.03) 0.87 (0.72, 1.06) 1.34 (0.83, 2.16) 1.01 (0.66, 1.56) 0.90 (0.74,1.09)
0.11 0.16 0.24 0.96 0.27
0.93 (0.79, 1.09) 0.92 (0.77, 1.10) 1.28 (0.61, 2.70) 0.93 (0.57, 1.50) 0.96 (0.80, 1.15)
0.37 0.37 0.52 0.76 0.68
1684 1684 1684 1532 1174
8089 8089 8089 7345 5517
0.93 (0.80, 1.08) 0.93 (0.57, 1.50) 1.09 (0.77, 1.53) 0.91 (0.91, 1.38) 0.89 (0.75, 1.05)
0.33 0.75 0.63 0.67 0.17
0.93 (0.75, 1.15) 0.78 (0.39, 1.55) 0.91 (0.53, 1.57) 0.92 (0.52, 1.63) 0.79 (0.60, 1.03)
0.52 0.48 0.74 0.77 0.08
0.93 (0.78, 1.11) 1.11 (0.56, 2.20) 1.23 (0.79, 1.90) 0.91 (0.51, 1.63) 0.95 (0.78, 1.15)
0.41 0.76 0.36 0.76 0.57
1258 1012 798 426 481 738 520
6033 4840 3796 1929 2192 3573 2460
0.89 (0.76, 1.05) 0.91 (0.77, 1.09) 0.91 (0.75, 1.10) 0.84 (0.64, 1.11) 0.92 (0.74, 1.16) 0.91 (0.73, 1.12) 0.82 (0.65, 1.04)
0.18 0.31 0.34 0.22 0.50 0.38 0.10
0.85 (0.68, 1.06) 0.87 (0.68, 1.13) 0.84 (0.62, 1.14) 0.76 (0.52, 1.12) 0.91 (0.68, 1.23) 0.88 (0.66, 1.17) 0.72 (0.52, 1.01)
0.16 0.31 0.27 0.17 0.56 0.38 0.06
0.93 (0.76, 1.16) 0.94 (0.76, 1.16) 0.95 (0.75, 1.19) 0.91 (0.65, 1.28) 0.93 (0.70, 1.23) 0.94 (0.71, 1.24) 0.91 (0.68, 1.23)
0.53 0.56 0.63 0.60 0.62 0.66 0.54
β-Blocker use modelled as time varying covariate with an individual considered a non-user before 6 months after first (or before 365 days of use) and a user after this time (see methods section for more details). Reported estimates are hazard ratios and 95% CIs, adjusted for age (as continuous), year of colorectal cancer diagnosis (as continuous), site of cancer (colon or rectum/recto-sigmoid junction), gender, stage, prediagnostic comorbidities (including myocardial infarction, cerebrovascular disease, congestive heart disease, chronic pulmonary disease, peripheral vascular disease, peptic ulcer disease and diabetes), prediagnostic smoking and treatment within 6 months of diagnosis (surgery, chemotherapy and radiotherapy), unless otherwise stated. b Adjusted for sex, age (as continuous) and year of cancer diagnosis (as continuous). c Reported estimates are sub-distribution hazard ratios and 95% CIs, accounting for competing risks of death from other causes and adjusting for age (as continuous), year of colorectal cancer diagnosis (as continuous), site of cancer (colon or rectum/recto-sigmoid junction), gender and stage. d Analysis are adjusted for stage, treatment within 6 months of diagnosis (surgery, chemotherapy and radiotherapy), pre-diagnostic comorbidities (including myocardial infarction, cerebrovascular disease, congestive heart disease, chronic pulmonary disease, peripheral vascular disease, peptic ulcer disease and diabetes) and smoking (pre-diagnosis with missing as an included category). e Restricted to individuals with 2 years of medication records before diagnosis, not excluding deaths in the year after diagnosis. f Restricted to individuals with 3 years of medication records before diagnosis, not excluding deaths in the year after diagnosis. g Restricted to individuals with 5 years of medication records before diagnosis, not excluding deaths in the year after diagnosis. h All sensitivity analyses refer to β-blocker usage in the time period from colorectal cancer diagnosis to 6 months before death, and are adjusted for stage, treatment within 6 months of diagnosis (surgery, chemotherapy and radiotherapy), comorbidities pre-diagnosis or during follow-up, including myocardial infarction, cerebrovascular disease, congestive heart disease, chronic pulmonary disease, peripheral vascular disease, peptic ulcer disease and diabetes) and smoking (pre-diagnosis with missing as an included category), unless otherwise stated. i Restricted to individuals with at least 1.5 years follow-up so relevant exposure period is at least a duration of 6 months. j Restricted to individuals with at least 2 years follow-up so relevant exposure period is at least a duration of 6 months. k Restricted to individuals with at least 1 year’s medication records who used other antihypertensive medications in the year before diagnosis. Antihypertensive medications include diuretics, vasodilator antihypertensive drugs, centrally acting antihypertensive drugs, α-adrenoceptor drugs, β-adrenoceptor drugs, angiotensin converting enzyme inhibitors, angiotensin 2 receptor antagonists, renin inhibitors and calcium channel blockers.
original articles
| Hicks et al.
Table 3. Additional analysis for association between β-blockers and colorectal cancer-specific death in colorectal cancer
a
Annals of Oncology
Downloaded from http://annonc.oxfordjournals.org/ at Northern Arizona University on July 9, 2015
Volume 24 | No. 12 | December 2013
original articles
Annals of Oncology Table 4. Post-diagnostic exposure to β-blockers and odds of death in colorectal cancer patients Post-diagnostic β-blocker usage
Controls n (%)
Unadjusted OR (95% CI)
1685 (76.1) 529 (23.9)
7902 (74.2) 2753 (25.8)
1.00 0.90 (0.81, 1.01)
0.06
1.00 0.88 (0.77, 1.00)
0.04
1685 (76.1) 231 (10.4) 298 (13.5)
7902 (74.2) 1234 (11.6) 1519 (14.3)
1.00 0.89 (0.76, 1.04) 0.91 (0.79, 1.05)
0.13 0.19
1.00 0.86 (0.72, 1.03) 0.89 (0.76, 1.04)
0.10 0.15
1685 (76.1) 291 (13.1) 238 (10.8)
7902 (74.2) 1563 (14.7) 1190 (11.2)
1.00 0.88 (0.77, 1.01) 0.93 (0.79, 1.08)
0.08 0.32
1.00 0.89 (0.76, 1.05) 0.86 (0.72, 1.02)
0.12 0.09
86 (3.9) 461 (20.8) 8 (0.4)
435 (4.1) 2405 (22.6) 55 (0.5)
0.96 (0.76, 1.23) 0.90 (0.81, 1.01) 0.73 (0.35, 1.53)
0.74 0.08 0.40
0.95 (0.72, 1.26) 0.88 (0.77, 1.01) 0.77 (0.32, 1.82)
0.74 0.07 0.55
334 (15.1) 119 (5.4) 34 (1.5) 37 (1.7) 32 (1.5) 15 (0.7)
1895 (17.8) 484 (4.5) 125 (1.2) 151 (1.4) 192 (1.8) 41 (0.4)
0.82 (0.72, 0.93) 1.20 (0.97, 1.49) 1.34 (0.92, 1.96) 1.20 (0.84, 1.73) 0.80 (0.54, 1.16) 1.84 (1.02, 3.33)
<0.01 0.09 0.13 0.32 0.24 0.05
0.79 (0.69, 0.92) 1.10 (0.86, 1.41) 1.66 (1.07, 2.55) 1.10 (0.74, 1.62) 0.82 (0.50, 1.34) 2.10 (1.07, 4.12)
<0.01 0.46 0.02 0.64 0.43 0.03
P
a
Includes propranolol, sotolol, timolol, carvidolol, pindolol, oxprenolol and labetolol. Includes acebutolol, atenolol, bisoprolol, metoprolol and nebivolol. c Iincludes acebutolol, labetolol, pindolol and oxprenolol. d Model 1 includes treatment within 6 months (including surgery, chemotherapy and radiotherapy), pre- and post-diagnostic comorbidities (including myocardial infarction, cerebrovascular disease, congestive heart disease, chronic pulmonary disease, peripheral vascular disease, peptic ulcer disease and diabetes) pre-diagnostic smoking (missing category included) and stage, restricted to individuals with complete staging information (1812 all-cause deaths and 8643 controls). b
in their study, their lack of linkage to cancer registries, their inability to account for potentially important confounders such as stage and treatment or the exclusion of patients with strong indications for β-blockers. The associations with carvedilol and sotalol differ despite both being non-cardioselective β-blockers. These associations are likely to be due to chance because of small numbers within subgroups; however, it is possible that these drugs may increase or decrease the risk of colorectal cancer death, respectively. Our study has several strengths. It is large allowing power to detect even relatively weak associations. Linkage with NCDR and ONS data allowed robust verification of cancer diagnosis and death data, respectively. Using GP prescribing data allowed for temporal relationships to be investigated ruling out any recall bias that exists in questionnaire-based studies. A further strength is that, as β-blockers are not available over the counter in the UK, we will have captured most usage. Although consumption cannot be guaranteed when using prescriptions, null associations in our analysis of 12 or more prescriptions indicate that non-compliance is unlikely to be affecting our findings. Another weakness is possible bias due to misclassification of cancer-specific death. However, a recent methodological study concluded that in comparative studies where misclassification is unlikely to be differential, as in our study, it is unlikely to impact greatly upon results [25]. Another
Volume 24 | No. 12 | December 2013
weakness is the possibility of residual confounding. Although we adjusted for important potential confounders such as sex, stage, treatment and other co-morbidities, we could not adjust for others such as socioeconomic status. Confounding by indication is often a problem in pharmaco-epidemiology; however, in our analysis, we compared β-blocker users to users of other antihypertensive medications to reduce the possibility of such bias. In conclusion, in this large population-based study, we did not find an association between β-blocker use and colorectal cancer-specific mortality.
acknowledgements This study is based in part on data from the General Practice Research Database obtained under license from the UK Medicines and Healthcare Regulatory Agency. However, the interpretation and conclusions contained in this study are those of the authors alone.
funding Access to the dataset was funded through a related Cancer Research-UK funded grant (C39066/A14597). BH was funded
doi:10.1093/annonc/mdt381 |
Downloaded from http://annonc.oxfordjournals.org/ at Northern Arizona University on July 9, 2015
No prescriptions 0 1 or more No of prescriptions 0 1–11 12 or more Daily defined doses (DDDs) 0 DDDs 1–365 DDDs 366 or more DDDs Categories of β-blocker activity Cardio non-selectivea Cardio selectiveb ISA activityc Specific β-blocker Atenolol Bisoprolol Metoprolol Propranolol Sotalol Carvedilol
P
Adjustedd OR (95% CI)
All-cause deaths n (%)
original articles by a Northern Ireland Department of Education and Learning PhD studentship.
disclosure DP has a patent on the use of α- and β-blockers to treat cancer metastasis, patent number 13/314 437. All remaining authors have declared no conflicts of interest.
references
| Hicks et al.
11. Barron TI, Connolly RM, Sharp L et al. Beta blockers and breast cancer mortality: a population- based study. J Clin Oncol 2011; 29(19): 2635–2644. 12. Powe DG, Voss MJ, Zänker KS et al. Beta-blocker drug therapy reduces secondary cancer formation in breast cancer and improves cancer specific survival. Oncotarget 2010; 1(7): 628–638. 13. Wang HM, Liao ZX, Komaki R et al. Improved survival outcomes with the incidental use of beta-blockers among patients with non-small-cell lung cancer treated with definitive radiation therapy. Ann Oncol 2013; 24(5): 1312–1319. 14. Grytli HH, Fagerland MW, Fosså SD et al. Association between use of β-blockers and prostate cancer-specific survival: a cohort study of 3561 prostate cancer patients with high-risk or metastatic disease. Eur Urol 2013 January 14 [epub ahead of print], doi: 10.1016/j.eururo.2013.01.007. 15. Diaz ES, Karlan BY, Li AJ. Impact of beta blockers on epithelial ovarian cancer survival. Gynecol Oncol 2012; 127(2): 375–378. 16. Zmora O. β-adrenergic Blocker and a COX2 Inhibitor for Prevention of Colorectal Cancer Recurrence. In: ClinicalTrials.gov. Bethesda, MD: National Library of Medicine. Available from:http://clinicaltrials.gov/show/NCT00888797 (29 April 2013, date last accessed). 17. Jick H, Jick SS, Derby LE. Validation of information recorded on general practitioner based computerised data resource in the United Kingdom. BMJ 1991; 302(6779): 766–768. 18. Khan NF, Perera R, Harper S et al. Adaptation and validation of the Charlson Index for Read/OXMIS coded databases. BMC Fam Pract 2010; 11: 1. 19. Schade R, Andersohn F, Suissa S et al. Dopamine agonists and the risk of cardiacvalve regurgitation. N engl J Med 2007; 356(1): 29–38. 20. Lévesque LE, Hanley JA, Kezouh A et al. Problem of immortal time bias in cohort studies: example using statins for preventing progression of diabetes. BMJ 2010; 340: b5087. 21. Etminan M. Pharmacoepidemiology II: the nested case-control study—a novel approach in pharmacoepidemiologic research. Pharmacotherapy 2004; 24(9): 1105–1109. 22. Essebag V, Platt RW, Abrahamowicz M et al. Comparison of nested case-control and survival analysis methodologies for analysis of time-dependent exposure. BMC Med Res Methodol 2005; 5(1): 5. 23. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999; 94: 496–509. 24. Shah SM, Carey IM, Owen CG et al. Does β-adrenoceptor blocker therapy improve cancer survival? Findings from a population-based retrospective cohort study. Br J Clin Pharmacol 2011; 72(1): 157–161. 25. Sarfati D, Blakely T, Pearce N. Measuring cancer survival in populations: relative survival versus cancer-specific survival. Int J Epidemiol 2010; 39(2): 598–610.
Volume 24 | No. 12 | December 2013
Downloaded from http://annonc.oxfordjournals.org/ at Northern Arizona University on July 9, 2015
1. Yang EV, Sood AK, Chen M et al. Norepinephrine up-regulates the expression of vascular endothelial growth factor, matrix metalloproteinase (MMP)-2, and MMP-9 in nasopharyngeal carcinoma tumor cells. Cancer Res 2006; 66(21): 10357–10364. 2. Weddle DL. beta-Adrenergic growth regulation of human cancer cell lines derived from pancreatic ductal carcinomas. Carcinogenesis 2001; 22(3): 473–479. 3. Masur K, Niggemann B, Zanker KS et al. Norepinephrine-induced migration of SW 480 colon carcinoma cells is inhibited by {beta}-blockers. Cancer Res 2001; 61 (7): 2866–2869. 4. Pérez Piñero C, Bruzzone A, Sarappa MG et al. Involvement of α2- and β2adrenoceptors on breast cancer cell proliferation and tumour growth regulation. Br J Pharmacol 2012; 166(2): 721–736. 5. Yang EV, Kim S, Donovan EL et al. Norepinephrine upregulates VEGF, IL-8, and IL-6 expression in human melanoma tumor cell lines: implications for stressrelated enhancement of tumor progression. Brain Behav Immun 2009; 23(2): 267–275. 6. Fitzgerald PJ. Is norepinephrine an etiological factor in some types of cancer? Int J Cancer 2009; 124(2): 257–263. 7. Kozanoglu I, Yandim MK, Cincin ZB et al. New indication for therapeutic potential of an old well-known drug ( propranolol) for multiple myeloma. J Cancer Res Clin Oncol 2013; 139(2): 327–335. 8. Lin X, Luo K, Lv Z et al. Beta-adrenoceptor action on pancreatic cancer cell proliferation and tumor growth in mice. Hepatogastroenterology 2012; 59(114): 584–588. 9. Zhang D, Yong Ma Q, Hu H-T et al. β 2-Adrenergic antagonists suppress pancreatic cancer cell invasion by inhibiting CREB, NF-κB and AP-1. Cancer Biol Ther 2010; 10(1): 19–29. 10. Lin Q, Wang F, Yang R et al. Effect of chronic restraint stress on human colorectal carcinoma growth in mice. PLoS One 2013; 8(4): e61435.
Annals of Oncology