ABSTRACT
Background and Aims
Beta‐blockers are successfully used to treat hemangioma and may decrease the proliferation of cancer cells. We hypothesized that individuals with colorectal polyps may also benefit from beta‐blocker initiation.
Methods
Individuals diagnosed with their first colorectal polyp 2006–2016 in the nationwide Swedish ESPRESSO histopathology cohort aged 45–79 years without CRC were eligible. We excluded individuals with previous indications for beta‐blocker (cerebrovascular disease, heart failure, aortic aneurysms, myocardial infarction) and individuals with contraindications for preventive beta‐blocker initiation (COPD, dementia, liver cirrhosis, Charlson score > 5 or metastatic cancer). Using duplication and inverse probability weighting, we emulated a target trial of beta‐blocker initiation within 2 years of the first polyp diagnosis. Main outcomes were incident CRC, CRC mortality, and all‐cause mortality until 2019.
Results
In total, 30,399 individuals met our inclusion criteria and were followed for a median of 8 years. Beta‐blockers were initiated in 2083 (6.9%) eligible individuals. The 10‐year cumulative incidence in initiators versus non‐initiators was 5.8% versus 8.6% for CRC incidence, 0.9% versus 1.1% for CRC mortality. The corresponding fully adjusted hazard ratios (HRs) were 0.87 (95% confidence interval, 95% CI: 0.85–0.89) and 0.96 (0.83–1.09). CRC mortality was significantly reduced in women HR 0.78 (0.68–0.99) but not in men HR = 1.14 (0.80–4.46). Cumulative CRC mortality was 0.6% in initiating women versus. 1.1% in non‐initiating women.
Conclusion
Beta‐blocker initiation within 2 years of polyp diagnosis was linked to a lower CRC incidence for all subgroups, and a lower CRC mortality in women, indicating that beta‐blocker initiation may improve long‐term outcomes in this high‐risk population.
Keywords: cancer prevention, causal inference, precision medicine, target trial emulation
Summary
-
Established knowledge of the subject
-
◦
Individuals diagnosed with colorectal polyps are at increased risk of colorectal cancer. Beta‐blockers can decrease cancer cell proliferation in vitro and have proven effective to treat infantile hemangioma, a benign blood‐vessel tumor structurally similar to colorectal polyps.
-
◦
We hypothesized beta‐blockers to be preventive of colorectal cancer in individuals diagnosed with colorectal polyps and tested our hypothesis in the first ever emulated target trial.
-
◦
-
Significant and/or new findings of this study
-
◦
Beta‐blocker initiation within 2 year of colorectal polyp removal was associated with a decreased risk of colorectal cancer, and a decreased risk of colorectal cancer mortality in women.
-
◦
Our study demonstrates the potential of beta‐blockers to contribute to colorectal cancer prevention in high‐risk individuals.
-
◦
Beta‐blocker initiation should be explored in further studies of individuals diagnosed with colorectal polyps.
-
◦
Abbreviations
- CI
confidence interval
- COPD
chronic obstructive pulmonary disease
- CRC
colorectal cancer
- CVD
cardiovascular disease
- ESPRESSO
Epidemiology Strengthened by histoPathology Reports in Sweden
- GI
gastrointestinal
- HR
hazard ratio
- ICD
International Classification of Diseases
- IPW
inverse probability weight
- OR
odds ratio
- SNOMED
Systematized Nomenclature of Medicine;
- TIA
transient ischemic attack
- USPSTF
United States Preventive Services Task Force
1. Introduction
Colorectal cancer (CRC) is the second leading cause of cancer deaths worldwide [1]. Removing precursor lesions [2, 3] reduces the risk of CRC incidence and mortality, but these individuals still have an increased risk of future CRC compared with individuals without colorectal polyps [4].
The beta‐blocker propranolol is used as first‐line treatment in infantile hemangioma, the most common benign tumor of infancy, causing the hemangioma to shrink rapidly [5]. Beta‐blockers have also been shown to decrease the proliferation of human colon adenocarcinoma cell lines in vitro [6]. Previous studies on beta‐blockers in relation to cancer risk have shown mixed results [7]: no protection from contralateral cancer for breast cancer [8], carvedilol and nadolol showing preventive effects on hepatocellular cancer [9], decreased risk of prostate cancer in atenolol users [10], and propranolol being associated with decreased risk of head and neck, esophagus, stomach, and prostate cancers [11].
Previous research on propranolol's potential effect on CRC is inconclusive. One cohort study of hypertensive patients in Taiwan showed a protective effect (HR = 0.68 for CRC) [11], while three Western case‐control studies showed no evidence of an association [12]. No previous study has addressed the impact of beta‐blocker initiation in individuals with increased risk of CRC, such as those diagnosed with colorectal polyps. Structurally, colorectal polyps are similar to hemangioma, as they are both protruding tumors with a need for blood supply for development and growth. Thus, patients with colorectal polyps, whose prevalence has been rising due to increasing screening and endoscopic use, may benefit from beta‐blocker initiation. Further, the high cost and unknown benefit of additional surveillance through repeated colonoscopy [13] in this patient population makes such studies particularly warranted, as initiation of beta‐blockers presents as a comparably non‐invasive, feasible, low‐cost and safe strategy for CRC prevention.
Here, we hypothesized that individuals diagnosed with colorectal polyps may benefit from beta‐blocker initiation. Hence, we aimed to test this hypothesis using target trial emulation [14] in all Swedish patients diagnosed with colorectal polyps registered in the ESPRESSO cohort.
2. Materials and Methods
2.1. Study Population
The unique personal identity number assigned to all Swedish residents [15] was used to identify all individuals with their first diagnosed colorectal polyps [4] from the nationwide ESPRESSO cohort (Epidemiology Strengthened by histoPathology Reports in Sweden) [16]. We used histopathologic findings defined by codes of morphology (Swedish modification of the Systematized Nomenclature of Medicine [SNOMED] coding system) and topography, T67 (colon) and T68 (rectum), in combination with any relevant SNOMED codes for adenomas or validated [17] free text search for serrated polyps to identify individuals with polyps. There was no patient or public involvement in the conduct of this research.
2.2. Target Trial Design: Effectiveness of Beta Blockers to Prevent CRC After CRC Polyp Detection and Removal
Here we describe the target trial—hypothetical randomized trial we would have liked to conduct—comparing CRC incidence and CRC mortality between two randomly assigned treatment strategies: (1) initiate any beta‐blocker at any dose within 2 years of diagnosis of first colorectal polyp or (2) do not initiate any beta‐blocker within the first 2 years of colorectal polyp diagnosis. Follow‐up would have started at the month of polyp detection (time zero) and ended at event (CRC diagnosis or CRC death), death, emigration, or administrative end of follow‐up (December 31, 2019), whichever occurred first.
Inclusion criteria would have been, in addition to a colorectal polyp as defined above, age 45–79 years at time of diagnosis from July 1, 2006 to December 31, 2016. This was chosen to ensure ≥ 1 year of no previous prescription of beta‐blockers prior to study entry (Swedish Prescribed Drug Registry was initiated July 1, 2005) and to allow for ≥ 3 full years of follow‐up before the administrative end of the follow‐up (December 31, 2019). Exclusion criteria would have been (I) previously registered prescription of beta‐blockers (i.e., ATC code starting with C07) before the date of first colorectal polyp; (II) individuals with existing CRC, dementia, metastatic cancer, hemorrhagic stroke, diagnosis of chronic obstructive pulmonary disease (COPD), liver cirrhosis and Charlson Score > 5 (contraindications for preventive beta‐blocker initiation); and (III) aortic aneurysms, myocardial infarction and cerebrovascular disease including TIA (transient ischemic attack) due to having a previous well‐established indication for beta‐blocker initiation without actual initiation—that is extremely unlikely to initiate the drug for CRC prevention registered during or before the month of first polyp detection. We also excluded (IV) those with any previous record of colorectal polyps and (V) all individuals with follow‐up ending during the same month as the detection of polyps.
Initiation of beta‐blockers was defined as the first prescription of ≥ 14 defined daily doses of any beta‐blocker defined as ATC code starting with “C07.” The 14 day definition was chosen because the Swedish dispensing system for multiple medications (Swedish APODOS) iterates every 14 days. We also performed a sensitivity analysis restricting exposure to individuals initiating propranolol specifically (ATC: C07AA05). As propranolol was only initiated by 311 eligible individuals, we only assessed the CRC incidence outcome in this sensitivity analysis.
2.3. Emulating the Target Trial in ESPRESSO
Duplication and inverse probability weighting (IPW) methodology similar to previous studies [18, 19, 20] were used to emulate a trial in which all individuals who met our criteria for the target trial were duplicated in the dataset, and each of the two replicates were assigned to one of the two treatment strategies: (1) initiate any beta‐blocker at any dose within 2 years of diagnosis of first colorectal polyp or (2) not to initiate any beta‐blocker within the first 2 years of colorectal polyp diagnosis.
The duplicates assigned to (1) beta‐blocker initiation were artificially censored at month 24 if they had not initiated a beta‐blocker. The duplicates assigned to (2) no initiation of beta‐blockers within the first 2 years were artificially censored at the actual month of initiation during month 1–24, whereas initiation of beta‐blockers after 2 years did not lead to artificial censoring.
In the emulated sensitivity trial of propranolol only, we did not consider other beta‐blocker use for exclusion or inclusion, that is, duplicates were only censored according to propranolol initiation status.
2.4. Covariates
Treatment arms were considered exchangeable conditional on baseline. We extracted a Charlson score [21] (0, 1, 2, 3, 4, 5), number of ever pre‐baseline registered visits in the Swedish Patient Registry (continuous), any diabetes, heart failure, malignancies other than CRC, year of study entry, age, sex, number of diagnosed colorectal polyps (1, 2, ≥ 3) and educational attainment (≤ 9, 10–12, ≥ 13 years and missing) at baseline (first polyp removal). Further, we included the following time‐varying covariates (coded as 0 until the first month of occurrence and thereafter always 1): ischemic stroke, bleeding stroke, myocardial infarction, angina pectoris, liver disease, dementia, ulcer, TIA, heart failure, COPD, diabetes, aneurysm, other malignancy and metastatic cancer for inclusion in our inverse probability weight (IPW) model for beta‐blocker initiation. Relevant codes and precise definitions are listed in the (Supporting Information S1: Methods) supplement (Supporting Information S1: eTable1 and eTable 2).
2.5. Statistical Analysis
The number of initiators and non‐initiators of beta‐blockers was presented according to sex (female, male), age (45–54, 55–64, 65–74, 75–79), year of first colorectal polyp diagnosis (2006–2009, 2010–2012, 2013–2016), and educational attainment (≤ 9, 10–12, ≥ 13 years and missing). Incidence rates of CRC mortality per 1000 person‐years were calculated according to the status of beta‐blocker initiation at 2 years and baseline subgroups. While the presented incidence rates display the crude number of events and incidence rates in the two groups prior to modeling, they are subject to immortal‐time bias [22]. Adjusted, model‐based estimates described below are free of immortal‐time bias.
We used pooled logistic regression for each outcome (CRC, CRC mortality, all‐cause mortality), including the indicator of initiation (initiators of beta‐blocker vs. non‐initiators of beta‐blocker), month of follow‐up (linear and quadratic terms), age (linear and quadratic terms), number of previous visits ever registered before study entry (linear and quadratic terms) and baseline variables listed above as categoricals. As the outcome is rare at all times, the pooled logistic model odds ratios approximate the hazard ratios (HRs) [23]. Percentiles of 200 non‐parametric bootstrap samples were used to obtain 95% confidence intervals (95% CIs).
Absolute risks were estimated by fitting the pooled logistic models with interaction terms between the treatment assignment and the month of the follow‐up variables, and predicted values standardized to the empirical baseline covariate distribution [24] were used to estimate the cumulative incidence graphs of CRC incidence and mortality.
Because artificial censoring can introduce selection bias due to post‐baseline confounding, we further estimated IPWs for the outcome models. IPWs were estimated from pooled logistic regression models including all baseline and time‐variant covariates as specified above to model beta‐blocker initiation within 24 months. For the beta‐blocker arm, we used the model to estimate the probability of initiating a beta‐blocker at 24 months (i.e., the probability of remaining uncensored at the last month of the grace period); during the first 23 months probabilities were set to 1, and after month 24, the initiation weights remained constant. For the non‐initiators, we used the model to predict the probability of not initiating beta‐blockers in months 1–24 and were similarly constant after month 24 as the protocol had ended and no further artificial censoring occurred. We truncated weights at 99.8% to avoid outlier effects, as IP weights had a maximum value of 1672 and P99.8 = 1.25 for CRC incidence and maximum of 1659 P99.8 = 1.25 for mortality outcomes. Separate trial emulations in subgroups of interest were performed, including men, women, individuals < 66 years old at diagnosis of colorectal polyps, no previous cancer diagnosis, no prior diabetes and individuals with a villous component of the baseline polyp. We also performed a sensitivity analysis of CRC incidence where we restricted follow‐up to the first 10 years, that is, individuals with follow‐up longer than 10 years were censored at 121st month of follow‐up.
SAS version 9.4 was used for all analyses. The study was approved by the Stockholm Ethics Review Board.
3. Results
In total, 30,399 individuals met the inclusion criteria (Figure 1) and were followed for a median of 8 years. Of them, 2083 (6.9%) initiated a beta‐blocker within 2 years of colorectal polyp diagnosis. The number of participants, events and crude incidence rates stratified by beta‐blocker initiation status are presented in Table 1. Men, older individuals and those with ≤ 9 years of education were more likely to initiate beta‐blockers.
FIGURE 1.
Participant flowchart. The 1013 individuals in the beta‐blocker arm not initiating or being artificially censored reached the end of follow‐up within months 1–23. CRC, colorectal cancer; CVD, cardiovascular disease; TIA, Transient ischemic attack.
TABLE 1.
CRC mortality and incidence in beta‐blocker initiators and non‐initiators.
| Participants, n (%) | CRC deaths, n (%) | CRC deaths rate per 1000 person‐years (95% CI) | CRC incident events a , n (%) | CRC incidence a rate per 1000 person‐years (95% CI) | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Group | Beta‐blocker | No beta‐blocker | Beta‐blocker | No beta‐blocker | Beta‐blocker | No beta‐blocker | Beta‐blocker | No beta‐blocker | Beta‐blocker | No beta‐blocker |
| All | 2083 (100.0%) | 28,316 (100.0%) | 16 (0.8%) | 231 (0.8%) | 1.0 (0.5–1.5) | 1.0 (0.9–1.2) | 122 (5.9%) | 1554 (5.5%) | 7.7 (6.3–9.0) | 6.9 (6.5–7.2) |
| Women | 953 (45.8%) | 14,648 (51.7%) | 4 (0.4%) | 115 (0.8%) | 0.5 (0.0–1.1) | 1.0 (0.8–1.1) | 55 (5.8%) | 772 (5.3%) | 7.4 (5.4–9.3) | 6.5 (6.0–6.9) |
| Men | 1130 (54.2%) | 13,668 (48.3%) | 12 (1.1%) | 116 (0.8%) | 1.4 (0.6–2.2) | 1.1 (0.9–1.3) | 67 (5.9%) | 782 (5.7%) | 7.9 (6.0–9.8) | 7.3 (6.8–7.8) |
| Age (years) | ||||||||||
| 45–54 | 210 (10.1%) | 5496 (19.4%) | (0.0%) | 19 (0.3%) | 0.0 (0.0–0.0) | 0.4 (0.2–0.6) | 4 (1.9%) | 160 (2.9%) | 2.1 (0.0–4.2) | 3.4 (2.9–3.9) |
| 55–64 | 598 (28.7%) | 9941 (35.1%) | 4 (0.7%) | 50 (0.5%) | 0.8 (0.0–1.7) | 0.6 (0.4–0.8) | 28 (4.7%) | 448 (4.5%) | 5.8 (3.7–8.0) | 5.5 (5.0–6.0) |
| 65–74 | 900 (43.2%) | 9950 (35.1%) | 7 (0.8%) | 101 (1.0%) | 1.0 (0.3–1.8) | 1.3 (1.1–1.6) | 45 (5.0%) | 674 (6.8%) | 6.6 (4.7–8.6) | 8.8 (8.2–9.5) |
| 75–79 | 375 (18.0%) | 2929 (10.3%) | 5 (1.3%) | 61 (2.1%) | 2.0 (0.3–3.8) | 3.0 (2.2–3.7) | 45 (12.0%) | 272 (9.3%) | 18.3 (13.0–23.7) | 13.2 (11.7–14.8) |
| Start of follow‐up | ||||||||||
| 2006–2009 | 711 (34.1%) | 9460 (33.4%) | 6 (0.8%) | 106 (1.1%) | 0.9 (0.2–1.6) | 1.1 (0.9–1.3) | 55 (7.7%) | 655 (6.9%) | 8.0 (5.9–10.1) | 6.7 (6.2–7.2) |
| 2010–2013 | 616 (29.6%) | 8421 (29.7%) | 6 (1.0%) | 61 (0.7%) | 1.3 (0.3–2.3) | 0.9 (0.7–1.1) | 39 (6.3%) | 431 (5.1%) | 8.5 (5.8–11.1) | 6.5 (5.9–7.1) |
| 2014–2016 | 612 (29.4%) | 8974 (31.7%) | 3 (0.5%) | 39 (0.4%) | 1.0 (0.0–2.2) | 0.9 (0.6–1.1) | 23 (3.8%) | 349 (3.9%) | 8.0 (4.7–11.2) | 7.7 (6.9–8.5) |
| Education | ||||||||||
| ≤ 9 years | 620 (29.8%) | 7064 (24.9%) | 9 (1.5%) | 83 (1.2%) | 1.9 (0.6–3.1) | 1.5 (1.1–1.8) | 54 (8.7%) | 488 (6.9%) | 11.1 (8.2–14.1) | 8.6 (7.8–9.4) |
| 10–12 years | 876 (42.1%) | 12,074 (42.6%) | 2 (0.2%) | 80 (0.7%) | 0.3 (0.0–0.7) | 0.8 (0.6–1.0) | 43 (4.9%) | 615 (5.1%) | 6.5 (4.6–8.4) | 6.3 (5.8–6.8) |
| > 12 years | 549 (26.4%) | 8675 (30.6%) | 4 (0.7%) | 42 (0.5%) | 0.9 (0.0–1.8) | 0.6 (0.4–0.8) | 24 (4.4%) | 416 (4.8%) | 5.6 (3.3–7.8) | 5.9 (5.3–6.5) |
| Missing | 38 (1.8%) | 503 (1.8%) | 1 (2.6%) | 26 (5.2%) | 8.7 (0.0–25.7) | 17.6 (10.8–24.3) | 1 (2.6%) | 35 (7.0%) | 8.7 (0.0–25.7) | 23.6 (15.8–31.5) |
Initiators and non‐initiators of beta‐blockers during the first 2 years after colorectal polyp.
After adjustment for baseline and time‐varying factors, CRC incidence was lower among beta‐blocker initiators than among non‐initiators overall (HR 0.87; 95% CI 0.85–0.89) (Table 2). In all subgroups, initiation of beta‐blockers was protective for CRC incidence (HRs ranging 0.86–0.90). In a sensitivity analysis restricting follow‐up to the first 10 years following diagnosis of first colorectal polyp, results remained largely unchanged (fully adjusted HR 0.87; 95% CI 0.85–0.88).
TABLE 2.
Hazard ratios (HRs) for CRC incidence and mortality after beta‐blocker initiation in different subgroups.
| CRC mortality, HR a (95% CI) | CRC incidence, HR a (95% CI) | |||||
|---|---|---|---|---|---|---|
| Subgroup | Unadjusted | Baseline adjusted b | Full IPW model c | Unadjusted | Baseline adjusted b | Full IPW model c |
| All | 0.99 (0.84–1.13) | 0.95 (0.82–1.09) | 0.96 (0.83–1.09) | 0.87 (0.85–0.89) | 0.86 (0.84–0.89) | 0.88 (0.85–0.90) |
| Men | 1.18 (0.97–1.47) | 1.12 (0.94–1.40) | 1.14 (0.80–4.46) | 0.86 (0.82–0.88) | 0.85 (0.81–0.87) | 0.86 (0.73–0.96) |
| Women | 0.79 (0.65–0.99) | 0.78 (0.64–0.99) | 0.78 (0.64–0.99) | 0.89 (0.87–0.92) | 0.89 (0.86–0.92) | 0.90 (0.87–0.93) |
| Age < 66 years | 1.02 (0.73–1.44) | 1.02 (0.72–1.42) | 1.02 (0.71–1.41) | 0.88 (0.85–0.91) | 0.88 (0.85–0.91) | 0.89 (0.86–0.92) |
| No previous cancer | 1.00 (0.82–1.18) | 0.96 (0.80–1.13) | 0.97 (0.81–1.21) | 0.87 (0.85–0.89) | 0.86 (0.84–0.89) | 0.87 (0.85–0.90) |
| No previous diabetes | 0.97 (0.84–1.12) | 0.92 (0.78–1.06) | 0.93 (0.81–1.07) | 0.87 (0.85–0.90) | 0.87 (0.85–0.89) | 0.88 (0.87–0.90) |
| Baseline polyp has villous component | 0.85 (0.68–1.02) | 0.85 (0.68–1.01) | 0.85 (0.68–1.02) | 0.88 (0.86–0.91) | 0.88 (0.86–0.91) | 0.89 (0.87–0.91) |
Note: Bold text indicate statistically significant finding.
Abbreviation: IPW, inverse probability weight.
HR, Hazard ratio based on pooled logistic regression for defined subgroups.
Adjusted for Charlson score (0, 1, 2, 3, 4, 5), number of ever pre‐baseline registered visits in the Swedish Patient Registry (continuous), any diabetes, heart failure, malignancies other than CRC, year of study entry, age, sex, number of diagnosed colorectal polyps (1, 2, ≥ 3) and educational attainment (≤ 9, 10–12, > 12 years and missing).
In addition to the baseline variables, the following time‐varying covariates (coded as 0 until the first month of occurrence and thereafter always 1) were used in the following IPW model: ischemic stroke, bleeding stroke, myocardial infarction, angina pectoris, liver disease, dementia, TIA (transient ischemic attack), heart failure, diagnosis of chronic obstructive pulmonary disease (COPD, proxy for heavy smoking), diabetes, aneurysm, other malignancy and metastatic cancer.
Overall, there was no evidence of any benefit of beta blocker initiation for CRC mortality (unadjusted hazard ratio [HR] 0.99 [95% CI 0.84–1.13]; baseline‐adjusted HR 0.95 [0.82–1.09]; IPW fully adjusted HR 0.96 [0.83–1.09]). However, in women, there was a significant reduction in CRC mortality, HR 0.78 (0.68–0.99). Furthermore, individuals with a villous component in their baseline colorectal polyps had a borderline significant finding for CRC mortality 0.85 (0.68–1.02).
In fully adjusted models, the standardized 10‐year cumulative incidence was 5.8% in initiators and 8.6% in non‐initiators (risk difference −2.9% (−3.9% to −2.5%)) for CRC incidence and 0.9% in initiators and 1.1% in non‐initiators (risk difference −0.2% [−0.6% to +0.1%]) for CRC mortality (Figure 2). The large increase in incidence seen after 150 months is due to the very limited number of patients, as only individuals diagnosed with polyps July 2006‐July 2007 contribute with follow‐up longer than 12.5 years. Given the difference between genders, we also produced sex‐specific cumulative incidence curves for CRC mortality (Figure 3). Ten‐year cumulative incidence was 1.2% in non‐initiating women compared to 0.6% in initiating women, risk difference −0.6% (−1.0% to −0.1%). For men, 10‐year cumulative incidence of CRC mortality was 1.1% in non‐initiators versus 1.3% in initiators with risk difference +0.1% (−0.4 to +0.7).
FIGURE 2.
Colorectal cancer (CRC) incidence and CRC cancer mortality standardized to baseline variables.
FIGURE 3.
Colorectal cancer (CRC) mortality standardized to baseline variables in women and men.
Results for all‐cause mortality outcome showed an increased HR after initiating beta‐blockers 1.37 (1.32–1.44) for the unadjusted, 1.25 (1.19–1.32) for the baseline adjusted and 1.26 (1.19–1.33) for the fully adjusted model. Ten‐year cumulative incidence was 19% in non‐initiators versus 24% in initiators, with a fully adjusted risk difference of +4.7% (2.8%–6.7%).
In our emulated sensitivity trial of propranolol, only 311 (0.59%) out of 52,805 eligible individuals initiated propranolol within 24 months of polyp diagnosis (Table 3). These results were compatible with the main analysis: the fully adjusted HR for CRC incidence was 0.88 (0.87–0.89) and similar in both men and women.
TABLE 3.
CRC mortality and incidence in propranolol initiators and non‐initiators.
| Participants, n (%) | CRC incident events a , n (%) | CRC incidence a rate per 1000 person‐years (95% CI) | CRC deaths | |||||
|---|---|---|---|---|---|---|---|---|
| Group | Propranolol | No propranolol | Propranolol | No propranolol | Propranolol | No propranolol | Propranolol | No propranolol |
| All | 311 (100.0%) | 52,494 (100.0%) | 15 (4.8%) | 2944 (5.6%) | 7.1 (3.5–10.6) | 7.4 (7.2–7.7) | 3 (1.0%) | 418 (0.8%) |
| Women | 150 (48.2%) | 25,675 (48.9%) | 4 (2.7%) | 1363 (5.3%) | 3.7 (0.1–7.3) | 6.8 (6.5–7.2) | 0 (0.0%) | 199 (0.8%) |
| Men | 161 (51.8%) | 26,819 (51.1%) | 11 (6.8%) | 1581 (5.9%) | 10.5 (4.3–16.7) | 8.1 (7.7–8.5) | 3 (1.9%) | 219 (0.8%) |
| Age (years) | ||||||||
| 45–54 | 45 (14.5%) | 7800 (14.9%) | (0.0%) | 221 (2.8%) | 0.0 (0.0–0.0) | 3.4 (3.0–3.9) | 0 (0.0%) | 22 (0.3%) |
| 55–64 | 129 (41.5%) | 16,658 (31.7%) | 4 (3.1%) | 728 (4.4%) | 4.5 (0.1–8.9) | 5.5 (5.1–5.9) | 2 (1.6%) | 89 (0.5%) |
| 65–74 | 109 (35.0%) | 20,727 (39.5%) | 9 (8.3%) | 1339 (6.5%) | 13.0 (4.5–21.6) | 8.9 (8.4–9.4) | 1 (0.9%) | 181 (0.9%) |
| 75–79 | 28 (9.0%) | 7309 (13.9%) | 2 (7.1%) | 656 (9.0%) | 11.8 (0.0–28.1) | 13.8 (12.7–14.9) | (0.0%) | 126 (1.7%) |
| Start of follow‐up | ||||||||
| 2006–2009 | 86 (27.7%) | 15,986 (30.5%) | 3 (3.5%) | 1109 (6.9%) | 4.0 (0.0–8.5) | 7.0 (6.6–7.4) | 1 (1.2%) | 187 (1.2%) |
| 2010–2013 | 102 (32.8%) | 15,620 (29.8%) | 6 (5.9%) | 925 (5.9%) | 8.3 (1.7–14.9) | 7.7 (7.2–8.2) | 2 (2.0%) | 116 (0.7%) |
| 2014–2016 | 104 (33.4%) | 18,587 (35.4%) | 5 (4.8%) | 735 (4.0%) | 10.6 (1.3–19.9) | 8.1 (7.5–8.7) | 0 (0.0%) | 83 (0.4%) |
| Education | ||||||||
| ≤ 9 years | 94 (30.2%) | 14,699 (28.0%) | 6 (6.4%) | 997 (6.8%) | 9.7 (1.9–17.4) | 9.0 (8.4–9.5) | 2 (2.1%) | 157 (1.1%) |
| 10–12 years | 132 (42.4%) | 22,277 (42.4%) | 7 (5.3%) | 1201 (5.4%) | 7.9 (2.0–13.7) | 7.1 (6.7–7.5) | 1 (0.8%) | 158 (0.7%) |
| > 12 years | 80 (25.7%) | 14,495 (27.6%) | 2 (2.5%) | 691 (4.8%) | 3.3 (0.0–7.9) | 6.1 (5.7–6.6) | 0 (0.0%) | 59 (0.4%) |
| Missing | 5 (1.6%) | 1023 (1.9%) | (0.0%) | 55 (5.4%) | 0.0 (0.0–0.0) | 21.7 (16.0–27.4) | 0 (0.0%) | 44 (4.3%) |
Initiators and non‐initiators of beta‐blockers during the first 2 years after colorectal polyp.
4. Discussion
In this target trial emulation conducted in Swedish national registry data in a cohort of patients with a history of colorectal polyps, initiation of beta‐blockers was associated with a significant reduction in CRC incidence (HR 0.87 [95% CI 0.85–0.89]), which corresponds to an absolute 10‐year risk reduction of approximately 3%. Beta‐blocker initiation was associated with reduced CRC mortality in women (HR 0.78 [0.64–0.99]) but not men (1.14 [0.80–4.46]). In women, this translates to a 10‐year risk reduction of 0.6%. These absolute risk reductions for CRC incidence overall and CRC mortality in women are large compared with the effects of other CRC prevention tools. For example, in an RCT of CRC screening colonoscopy, the reported 10‐year risk reductions were −0.22% for CRC incidence and −0.02% for CRC mortality [25]. These 2% may not be directly comparable, as the baseline risk in the general population is less than that in the population with a history of colorectal polyps, but it still provides context for the potential absolute effect of the intervention. Furthermore, our previous emulated trial of aspirin initiation in the same patient setting [19] observed a similar reduction in CRC incidence (3%) but no reduction in CRC mortality in either men or women. Hence, beta‐blocker initiation within 2 years of polyp detection appears to reduce the risk of incident CRC overall and in all included subgroups, and CRC mortality in women and individuals with villous polyps at baseline (although the latter is only borderline significant). Explanations for gender differences in CRC mortality after initiation of beta‐blockers potentially include dietary and lifestyle promotors of CRC being more common and influential in male participants. Similarly, sex differences in CRC prevention have also been reported for sigmoidoscopy screening, on the contrary being more effective in men [26]. Additionally, women have a higher risk of developing proximal, serrated, and likely more aggressive forms of CRC than men [27] and it is biologically plausible that beta‐blockers might have a particular effect on mortality from fast growing‐CRC as beta‐blockers decrease cell proliferation [6].
Beta‐blocker initiation was further associated with an increased overall mortality: 24% in initiators versus 19% in non‐initiators after 10 years of follow‐up. A previous Cochrane meta‐analysis of beta‐blockers used to treat hypertension in otherwise healthy individuals aged 18–59 showed no effect on overall mortality [28]. But, the mean duration of the included studies was 5 years, so any long‐term effects in mainly healthy individuals are largely unknown. Further, our estimate of all‐cause mortality may be impacted by near violations of the positivity criteria (it was checked and found not to be technically violated in the strata—but from clinical knowledge we know that almost all patients without beta‐blocker contraindications experiencing a myocardial infarction will be prescribed beta‐blockers). Thus, the estimate of all‐cause mortality is likely more biased by the near lack of positivity for myocardial infarction and unmeasured residual confounding by associated CVD risk factors (as it is directly affected by the occurrence of CVD deaths) than CRC and CRC mortality (as the two latter are only indirectly affected by the occurrence of myocardial infarction as a heavy risk factor of CVD death, i.e., through competing risk). However, further studies are needed to explore the overall mortality outcomes of beta‐blocker initiation in patients with polyps without CVD before these drugs are implemented in clinical practice.
4.1. Strengths and Limitations
Our study has several strengths, primarily the high quality nationwide ESPRESSO cohort data that integrates histopathological results with long term follow‐up permitting the study of long‐term CRC outcomes after initial polyp detection. Methodological strengths include avoiding immortal‐time and selection biases, hence allowing for causal inference, and centering the design around a well‐defined intervention that focuses on a common point of clinical decision‐making—polyp detection—in a population likely to benefit more than the average patient from CRC prevention. Our study also has several limitations. First, the ESPRESSO cohort does not contain data on polyp size or location nor does it ascertain the number of colonoscopies performed during follow‐up; thus, we cannot study whether these factors may have been associated with the receipt of beta blockers. Second, both study eligibility and incident CRC are primarily the result of clinically detected polyps as opposed to screening‐detected polyps, as only a minority of the Swedish population was offered screening during the study period [29]. Third, because ESPRESSO does not collect data on smoking status, we used COPD as a proxy thereof. Fourth, we lacked the power to assess the impact of propranolol initiation on CRC mortality due to the low number of initiators. Finally, residual unmeasured confounding may play a role in the interpretation of our findings as in all observational studies, as there may be uncaptured patient preferences that affect both beta‐blocker prescribing and CRC outcomes. However, a residual confounder would have to be associated with both the exposure and the outcome at e‐value = 1.56 for incidence and 1.92 for mortality in women to impact our findings.
5. Conclusion
Our emulated trial of beta‐blocker initiation within 24 months of CRC polyp detection showed promising data for CRC prevention, particularly for women where initiation was associated with significantly decreased CRC mortality. Future studies of beta‐blocker initiation in polyp patients are warranted.
Author Contributions
Guarantor: Emilsson. Study concept and design: Emilsson. Acquisition of data: Ludvigsson. Analysis: Emilsson had access to all the data, performed all analyses and created the figures. Funding: Emilsson, Ludvigsson. Writing the first draft of the manuscript: Emilsson. Critical revision of the manuscript for important intellectual content and approval of final version: All co‐authors.
Ethics Statement
(Stockholm ethical review board): 2014/1287‐31/4 and 2018/972‐32 the study is conducted according to the Declaration of Helsinki, informed consent was waived by the ethics committee for the strictly registry‐based nationwide study.
Conflicts of Interest
Dr. Ludvigsson has coordinated a study unrelated to the present study on behalf of the Swedish IBD Quality Register (SWIBREG). That study received funding from Janssen. Dr Ludvigsson has also received financial support from MSD developing a paper reviewing national healthcare registers in China. Dr Ludvigsson is currently discussing potential research collaboration with Takeda. Dr. Petito received unrelated research funding from Omron Healthcare Co. Ltd. Authors not named here have disclosed no conflicts of interest.
Transparency
The lead author (LE) affirms that this manuscript is an honest, accurate and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.
Supporting information
Supporting Information S1
Emilsson, Louise , Song Mingyang, Petito Lucia C., and Ludvigsson Jonas F.. 2025. “Target Trial Emulation of Beta‐Blockers After Diagnosis of Colorectal Polyps—Beneficial in Women,” United European Gastroenterology Journal: 13. no. 10), 2034–2043. 10.1002/ueg2.70139.
Funding: This work was supported by individual research support from Region Värmland (LE) and the Swedish Cancer Society (JFL). The funding organizations played no role in the collection, design, analysis and conduct of the study or approval of the manuscript.
Data Availability Statement
No additional data available. Data from Swedish health care authorities can be sought after ethical approval.
References
- 1. Bray F., Ferlay J., Soerjomataram I., Siegel R. L., Torre L. A., and Jemal A., “Global Cancer Statistics 2018: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries,” CA: A Cancer Journal for Clinicians 68, no. 6 (2018): 394–424, 10.3322/caac.21492. [DOI] [PubMed] [Google Scholar]
- 2. Zauber A. G., Winawer S. J., O'Brien M. J., et al., “Colonoscopic Polypectomy and Long‐Term Prevention of Colorectal‐Cancer Deaths,” New England Journal of Medicine 366, no. 8 (2012): 687–696, 10.1056/nejmoa1100370. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Holme O., Loberg M., Kalager M., et al., “Effect of Flexible Sigmoidoscopy Screening on Colorectal Cancer Incidence and Mortality: A Randomized Clinical Trial,” JAMA 312, no. 6 (2014): 606–615, 10.1001/jama.2014.8266. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Song M., Emilsson L., Bozorg S. R., et al., “Risk of Colorectal Cancer Incidence and Mortality After Polypectomy: A Swedish Record‐Linkage Study,” Lancet Gastroenterology & Hepatology 5, no. 6 (2020): 537–547, 10.1016/s2468-1253(20)30009-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Leaute‐Labreze C., Harper J. I., and Hoeger P. H., “Infantile Haemangioma,” Lancet 390, no. 10089 (2017): 85–94, 10.1016/s0140-6736(16)00645-0. [DOI] [PubMed] [Google Scholar]
- 6. Coelho M., Moz M., Correia G., Teixeira A., Medeiros R., and Ribeiro L., “Antiproliferative Effects of Beta‐Blockers on Human Colorectal Cancer Cells,” Oncology Reports 33, no. 5 (2015): 2513–2520, 10.3892/or.2015.3874. [DOI] [PubMed] [Google Scholar]
- 7. Yang J., Zhang S., and Jiang W., “Impact of Beta Blockers on Breast Cancer Incidence and Prognosis,” Clinical Breast Cancer 23, no. 6 (2023): 664–671, 10.1016/j.clbc.2023.05.014. [DOI] [PubMed] [Google Scholar]
- 8. Gottschau M., Bens A., Friis S., et al., “Use of Beta‐Blockers and Risk of Contralateral Breast Cancer,” International Journal of Cancer 150, no. 10 (2022): 1619–1626, 10.1002/ijc.33923. [DOI] [PubMed] [Google Scholar]
- 9. He X., Zhao Z., Jiang X., and Sun Y., “Non‐Selective Beta‐Blockers and the Incidence of Hepatocellular Carcinoma in Patients With Cirrhosis: A Meta‐Analysis,” Frontiers in Pharmacology 14 (2023): 1216059, 10.3389/fphar.2023.1216059. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Zahalka A. H., Fram E., Lin W., et al., “Use of Beta‐Blocker Types and Risk of Incident Prostate Cancer in a Multiethnic Population,” Urologic Oncology 38, no. 10 (2020): 794e11–794e16, 10.1016/j.urolonc.2020.03.024. [DOI] [PubMed] [Google Scholar]
- 11. Chang P. Y., Huang W. Y., Lin C. L., et al., “Propranolol Reduces Cancer Risk: A Population‐Based Cohort Study,” Medicine 94, no. 27 (2015): e1097, 10.1097/md.0000000000001097. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Qi J., An R., Bhatti P., Spinelli J. J., and Murphy R. A., “Anti‐Hypertensive Medications and Risk of Colorectal Cancer: A Systematic Review and Meta‐Analysis,” Cancer Causes & Control 33, no. 6 (2022): 801–812, 10.1007/s10552-022-01570-1. [DOI] [PubMed] [Google Scholar]
- 13. Jover R., Bretthauer M., Dekker E., et al., “Rationale and Design of the European Polyp Surveillance (EPoS) Trials,” Endoscopy 48, no. 6 (2016): 571–578, 10.1055/s-0042-104116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Hernán M. A. and Robins J. M., “Using Big Data to Emulate a Target Trial When a Randomized Trial Is Not Available,” American Journal of Epidemiology 183, no. 8 (2016): 758–764, 10.1093/aje/kwv254. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Ludvigsson J. F., Otterblad‐Olausson P., Pettersson B. U., and Ekbom A., “The Swedish Personal Identity Number: Possibilities and Pitfalls in Healthcare and Medical Research,” European Journal of Epidemiology 24, no. 11 (2009): 659–667, 10.1007/s10654-009-9350-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Ludvigsson J. F. and Lashkariani M., “Cohort Profile: ESPRESSO (Epidemiology Strengthened by histoPathology Reports in Sweden),” Clinical Epidemiology 11 (2019): 101–114, 10.2147/clep.s191914. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Bozorg S. R., Song M., Emilsson L., and Ludvigsson J. F., “Validation of Serrated Polyps (SPs) in Swedish Pathology Registers,” BMC Gastroenterology 20, no. 1 (2019): 3, 10.1186/s12876-019-1134-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Emilsson L., Garcia‐Albeniz X., Logan R. W., Caniglia E. C., Kalager M., and Hernán M. A., “Examining Bias in Studies of Statin Treatment and Survival in Patients With Cancer,” JAMA Oncology 4, no. 1 (2018): 63–70, 10.1001/jamaoncol.2017.2752. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Emilsson L., Song M., and Ludvigsson J. F., “Target Trial Emulation of Aspirin After Diagnosis of Colorectal Polyps,” European Journal of Epidemiology 38, no. 10 (2023): 1105–1114, 10.1007/s10654-023-01024-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Petito L. C., Garcia‐Albeniz X., Logan R. W., et al., “Estimates of Overall Survival in Patients With Cancer Receiving Different Treatment Regimens: Emulating Hypothetical Target Trials in the Surveillance, Epidemiology, and End Results (SEER)‐Medicare Linked Database,” JAMA Network Open 3 (2020): e200452, 10.1001/jamanetworkopen.2020.0452. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Ludvigsson J. F., Appelros P., Askling J., et al., “Adaptation of the Charlson Comorbidity Index for Register‐Based Research in Sweden,” Clinical Epidemiology 13 (2021): 21–41, 10.2147/clep.s282475. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Suissa S., “Immortal Time Bias in Pharmaco‐Epidemiology,” American Journal of Epidemiology 167, no. 4 (2008): 492–499, 10.1093/aje/kwm324. [DOI] [PubMed] [Google Scholar]
- 23. Thompson W. A. Jr, “On the Treatment of Grouped Observations in Life Studies,” Biometrics 33, no. 3 (1977): 463–470, 10.2307/2529360. [DOI] [PubMed] [Google Scholar]
- 24. Hernan M. A. and Robins J. M., Causal Inference: What if, 1st ed. (CRC Press, 2023), 10.1201/9781315374932. [DOI] [Google Scholar]
- 25. Bretthauer M., Løberg M., Wieszczy P., et al., “Effect of Colonoscopy Screening on Risks of Colorectal Cancer and Related Death,” New England Journal of Medicine 387, no. 17 (2022): 1547–1556, 10.1056/nejmoa2208375. [DOI] [PubMed] [Google Scholar]
- 26. Holme O., Loberg M., Kalager M., et al., “Long‐Term Effectiveness of Sigmoidoscopy Screening on Colorectal Cancer Incidence and Mortality in Women and Men: A Randomized Trial,” Annals of Internal Medicine 168, no. 11 (2018): 775–782, 10.7326/m17-1441. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Kim S. E., Paik H. Y., Yoon H., Lee J. E., Kim N., and Sung M. K., “Sex‐ and Gender‐Specific Disparities in Colorectal Cancer Risk,” World Journal of Gastroenterology 21, no. 17 (2015): 5167–5175, 10.3748/wjg.v21.i17.5167. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Musini V. M., Gueyffier F., Puil L., Salzwedel D. M., and Wright J. M., “Pharmacotherapy for Hypertension in Adults Aged 18 to 59 Years,” Cochrane Database of Systematic Reviews 8 (2017): CD008276, 10.1002/14651858.CD008276.pub2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Senore C., Basu P., Anttila A., et al., “Performance of Colorectal Cancer Screening in the European Union Member States: Data From the Second European Screening Report,” Gut 68, no. 7 (2019): 1232–1244, 10.1136/gutjnl-2018-317293. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting Information S1
Data Availability Statement
No additional data available. Data from Swedish health care authorities can be sought after ethical approval.