Concurrent β-blocker use is associated with improved outcome in esophageal cancer patients who undergo chemoradiation: A retrospective matched pair analysis

Abstract

Background

β-blocker use has been associated with improved outcomes in a number of different malignancies; however, the impact of β-blockade in esophageal cancer is not been well characterized. We compared the outcomes of esophageal cancer patients based on β-blocker usage.

Methods

The charts of all 418 patients treated with radiation for esophageal cancer at our institution from April 2010 to October 2018 were analyzed. Patients who underwent treatment with palliative intent or did not finish treatment were excluded. β-blocker use was determined from the medication list at time of pre-treatment consultation.

Results

There were 291 esophageal cancer patients who received neoadjuvant/definitive CRT. The median follow-up for the cohort was 22.5 months (interquartile range (IQR) 9.6–41.0 months). Within the cohort, 27.8% (n=81) of patients were taking β-blockers at the time of treatment. Those taking β-blockers had significantly improved distant control (22.2% vs 37.9%) (p=0.035). Concomitant β-blocker use was significantly associated with improved progression-free survival (PFS)(p=<0.001, HR=0.42 (0.27–0.66) and overall survival (OS)(p=0.002, HR=0.55 (0.38–0.81) on Cox regression analysis. Propensity score matched pairs were created using tumor stage, nodal stage, sex, neoadjuvant vs definitive therapy, Karnofsky performance status (KPS), and aspirin use. This matched pair analysis showed a significant PFS (p=0.005) benefit esophageal cancer patients taking β-blockers.

Conclusions

Concurrent β-blocker use is common within patients receiving concurrent chemoradiation for esophageal cancer. Esophageal cancer patients who received chemoradiation while taking β-blockers demonstrated significant benefits in survival-based outcomes.

Introduction

β-adrenergic receptor (β-AR) antagonists also known as β-blockers were one of the first class of drugs developed to combat hypertension. Recently increased β-AR signaling has been implicated in key roles promoting vital malignant processes including resistance to apoptosis, angiogenesis, and immune evasion.(1–6) Epidemiological studies show that incidental concurrent β-blocker use during cancer therapy has been associated with improved survival in cancer patients in a number of different malignancies.(7–16) However, multiple factors including existence of other competing therapeutic options have prevented beta-blockers from being incorporated as a part of routine cancer care.

Esophageal cancer, by contrast, has a relative dearth of novel therapeutic options and accounts for 1% of cancers within the United States, impacting approximately 17,650 adults (13,750 men and 3,900 women) a year. (17) While the predominant histology worldwide is squamous cell carcinoma (SCC), American patients typically develop adenocarcinoma in the distal esophagus and/or gastroesophageal junction (GEJ). (18, 19) This difference is thought to be driven by rising rates of American obesity, subsequently increasing the prevalence of gastroesophageal reflux disease, a known risk factor for esophageal adenocarcinoma. (20–22) Regardless of histology, the standard of care in the treatment of esophageal carcinoma is often institution dependent. Our institutional standard is neoadjuvant chemoradiation therapy (CRT) followed by resection when feasible. However, many patients are unable to tolerate resection either due to pre-existing comorbidities or the toxicity of the neoadjuvant treatment. As such, many patients are treated with CRT alone.

The estimated 5-year survival rate of esophageal carcinoma remains poor at approximately 19%.(17) As such, new strategies for controlling local and distant disease are desperately needed. Recent murine work from our group has shown that β-blocker use may improve efficacy of chemotherapy and radiation by improving T cell egress from the tumor draining lymph node.(6) This retrospective analysis was performed to determine if β-blocker use impacted the survival of patients with esophageal cancer treated with chemoradiation therapy.

Methods

Patients

We retrospectively identified patients who received radiotherapy for esophageal cancer at our institution from April 2010 to October 2018. This database was generated under the approval of the Roswell Park Comprehensive Cancer Center Institutional Review Board. Patients who received CRT for esophageal carcinoma were included, excluding patients who underwent treatment with palliative intent or did not receive at least 41.4 Gray (Gy). The typical radiation dose in those undergoing definitive treatment was 50.4 Gy with select patient patients receiving 41.4 Gy preoperatively. All patients had biopsy confirmed disease. Standard work up at our institution of newly diagnosed esophageal includes: endoscopy ± endoscopic ultrasound, positron-emission tomography (PET) with computed tomography (CT), and CT with contrast. Concurrent chemotherapy was employed in all patients, with carboplatin and paclitaxel used in the majority.

Clinical data

Staging was completed using American Joint Committee on Cancer (AJCC) 7^th edition. To determine β-blocker use, we evaluated the medication list upon initial consultation with radiation oncology department prior to treatment, recording any concurrent β-blocker usage. Range of prescribed radiation dose was 41.4–50.4 Gy. Performance status was recorded using the Karnofsky performance status (KPS). Location of relapse was determined by site of first failure. Distant progression was defined as failure outside of the site of initial treatment. In approximately 3% of patients, there was simultaneous local and distant progression which was recorded as distant progression. Heart disease was defined as a history of coronary artery disease as defined by catheter angiography, a history of myocardial infarction, cardiac stenting, or coronary artery bypass, or documented ejection fraction of 50% or less. In determining indications for β-blocker use, some patients had multiple reasons for use. In these settings, history of heart disease took precedence followed by cardiac arrhythmia. Hypertension was only listed as the indication in the absence of other cardiac related disease. Survival-based endpoints were calculated using the date of first radiation treatment and last known follow-up. Progression-free survival (PFS) was determined by date of progression as determined by imaging or biopsy. Overall survival (OS) determined by date of death.

Statistical analysis

Demographic tables were constructed using IBM SPSS statistics version 21. Survival endpoints were calculated using IBM IPSS Statistics version 26 and R version 3.6.1. Chi-square test used to compare differences in progression and prior/secondary malignancies across groups. Specifically for matched cohort, package MatchIt version 3.0.2 was used. Propensity score matching with caliper length of 0.0001 was performed, matching those with and without beta blockers to evaluate the effect of the beta blocker on survival outcomes.

Results

In total, 418 patients who received CRT for esophageal cancer at our institution from April 2010 to October 2018 were identified (Supplemental Figure 1). Of these, 297 received neoadjuvant/definitive treatment and 121 were treated with palliative intent or did not finish treatment. Of the patients who received neoadjuvant/definitive treatment, six patients with neuroendocrine tumors of the esophagus were excluded from analysis. Within this cohort, the median age was 67 years with a median follow up of 22.5 months (interquartile range (IQR) 9.6–41.0 months); 80.8% of the patients were male; and 89.7% of patients had a KPS ≥80 (Table 1). The most common histology was adenocarcinoma (84.2%) with most patients having T3N0–1 disease. Radiation therapy was delivered by volumetric modulated arc therapy (VMAT) in nearly all patients. All patients received concurrent chemotherapy, typically carboplatin and paclitaxel (88.7%) and approximately 50% subsequently underwent surgical resection. Most patients had a history of smoking, however the majority had quit prior to treatment. Approximately a third of patients were taking aspirin at the time of treatment and 27.8% were on β-blockers. Within the follow up period, 48.1% had documented progression of disease and 56.7% of patient died.

Table 1.

Demographics

		Median	n	%
Age (years)		67
Sex	Male		235	80.8%
	Female		56	19.2%
KPS	60		1	0.3%
	70		29	10.0%
	80		213	73.2%
	90		46	15.8%
	100		2	0.7%
Pathology	Adenocarcinoma		245	84.2%
	SCC		46	15.8%
T stage	T1		17	5.8%
	T2		29	10.0%
	T3		243	83.5%
	T4		2	0.7%
N stage	N0		97	33.3%
	N1		148	50.9%
	N2		42	14.4%
	N3		4	1.4%
Neoadjuvant vs Definitive	Neoadjuvant		145	49.8%
	Definitive		146	50.2%
Concurrent Chemotherapy	Carboplatin/Paclitaxel		258	88.7%
	Paclitaxel		10	3.4%
	Other		23	7.9%
Tobacco	Never smoker		72	24.7%
	Former		162	55.7%
	Current		57	19.6%
Aspirin use	No		193	66.3%
	Yes		98	33.7%
Beta-Blocker use	No		210	72.2%
	Yes		81	27.8%
RT Technique	3DCRT		4	1.4%
	VMAT		287	98.6%
Progression	No		151	51.9%
	Yes		140	48.1%
PFS (months)		13.4
Vital Status	Alive		126	43.3%
	Dead		165	56.7%
Follow up (months)		22.5

SCC: Squamous cell carcinoma, RT: Radiation therapy, 3DCRT: 3-dimensional conformal radiation therapy, VMAT: Volumetric modulated arc therapy, PFS: Progression-free survival

Of the patients taking β-blockers, 38 of 81 (46.9%) were also taking aspirin vs. 60 of 98 (28.6%) aspirin users without current β-blocker use. Indications for β-blocker use included history of heart disease (63.0%), hypertension (28.3%), cardiac arrhythmia (6.2%), and other (2.5%). The most common β-blocker used was metoprolol (56.8%), followed by carvedilol (23.5%), atenolol (14.8%), and other (4.9%). In terms of previous and secondary malignancies, β-blocker use was not associated with a significant difference in either setting, (17.2% vs 12.9%, p=0.33) and (1.2% vs 3.8%, p=0.26) respectively.

Given the presence of factors which are known to impact outcome in esophageal cancer, such as performance status, surgical fitness, tumor and nodal stage, we performed a multivariate cox regression analysis evaluating PFS and OS (Table 2).

Table 2.

MVA

	Progression-free Survival			Overall survival
	p-value	HR	95.0% CI	p-value	HR	95.0% CI
Age	0.640	1.00	0.97–1.02	0.082	1.02	0.99–1.04
Sex	0.191	0.72	0.45–1.15	<0.001	0.34	0.21–0.56
KPS	0.685	0.99	0.96–1.03	0.011	0.96	0.93–0.99
Pathology	0.106	0.74	0.45–1.20	0.218	0.75	0.48–1.18
Tobacco (Never smoker)	Ref			Ref
Tobacco (former smoker)	0.332	1.30	0.77–2.18	0.536	1.17	0.72–1.90
Tobacco (current smoker)	0.251	1.39	0.75–2.56	0.965	0.99	0.55–1.76
T stage (T1)	Ref			Ref
T stage (T2)	0.538	1.52	0.37–6.19	0.564	1.38	0.46–4.16
T stage (T3)	0.001	7.24	2.22–23.65	0.004	3.95	1.56–10.00
T stage (T4)	0.021	14.39	1.36–151.96	0.058	8.63	0.9–80.10
N stage (N0)	Ref			Ref
N stage (N1)	0.763	1.09	0.73–1.63	0.328	1.20	0.83–1.74
N stage (N2)	0.107	1.55	0.92–2.61	0.002	2.15	1.33–3.49
N stage (N3)	<0.001	13.19	3.58–48.61	0.294	1.90	0.57–6.33
Neoadjuvant vs Definitive	<0.001	3.59	2.40–5.63	<0.001	2.95	2.03–4.28
Aspirin use	0.001	1.92	1.3–2.83	0.363	1.18	0.83–1.67
Beta-Blocker use	<0.001	0.42	0.27–0.66	0.002	0.55	0.38–0.81
3DCRT vs IMRT	0.652	1.38	0.33–5.87	0.605	0.73	0.23–2.39

Advanced tumor stage (T3 p=0.001; hazard ratio (HR)=7.24 (2.22–23.65) and T4 p=0.027, HR=14.39 (1.36–151.96)), N3 disease (p=<0.001, HR=13.19 (3.58–48.61), definitive CRT (p=<0.001, HR=3.59 (2.40–5.63) and aspirin use (p=0.001. HR=1.92 (1.30–2.83) were significantly associated with worsened PFS. Concurrent β-blocker use (p=<0.001, HR=0.42 (0.27–0.66) was associated with improved PFS.

Of the 145 patients who received neoadjuvant chemoradiation, 26 (17.9%) were taking β-blockers. At the time of resection, 6 patients (23.1%) had stable disease or no treatment response, 17 (65.3%) had tumor downstaging, and 3 (11.5%) had complete pathologic response in the β-blocker group. In those not taking β-blockers, 35 (29.4%) had stable disease or no treatment response, 57 (47.9%) had tumor downstaging, and 27 (22.7%) had complete pathologic response. For these groups, the results were not statistically significant by Chi-square test (p=0.24).

With regard to number and site of progression, 51 (63.0%) β-blocker users had no progression, 12 (14.8%) progressed locally while 18 (22.2%) failed distantly. By comparison, 111 (52.9%) patients not taking β-blockers had no progression, whereas failure was seen in 20 (9.5%) locally and 79 (37.6%) distantly. These differences between β-blocker users and non-users were statistically significant (p=0.035) by Chi-square test.

For OS, tumor stage T3 (p=0.004, HR=3.95 (1.56–10.00) N2 disease (p=0.002, HR=2.15 (1.33–3.49), definitive CRT (p=<0.001, HR=2.95 (2.03–4.28) had significantly worse OS, while higher KPS (p=0.011, HR=0.96 (0.93–0.99), female sex (p=<0.001, HR=0.34 (0.21–0.56), and β-blocker usage (p=0.002, HR=0.55 (0.38–0.81) were associated with improved OS.

To address the inclusion of potentially erroneous variables into these models, a backward-conditional method was added to the cox regression analysis (Supplemental table 1). These analyses yielded similar findings, with the exception that sex and age were significantly associated with PFS and OS respectively.

As a complement the cox regression analysis, a 2:1 propensity score matched pair analysis for variables including tumor stage (T1–2, T3+), nodal stage (N0, N1+), sex, neoadjuvant vs definitive CRT, KPS (80–100, <80), and aspirin use was performed. 121 and 76 matched pairs were identified for the no β-blocker and β-blocker groups respectively. All variables were well-balanced (Supplemental Table 2). β-blocker use was significantly associated with improved PFS (p=0.005) with trends towards improved OS (P=0.148) (Figure 1). The estimated 2-year PFS and OS within these matched cohorts were 61.8% vs 33.3% and 65.2% vs 53.2% for β-blocker and no β-blocker groups respectively.

Figure 1.

Survival-based outcomes of a match-pair cohort by β-blockers use.

Discussion

In this retrospective study, β-blocker use was associated with improved PFS and OS in patients undergoing chemoradiation for esophageal cancer on multivariate cox regression analysis. Patients taking β-blockers had a lower incidence of disease progression, which appeared to be due to fewer distant failures. Furthermore, in a propensity score matched pair cohort, those taking β-blockers retained significantly improved PFS, with non-significantly improved prolonged OS. The lack of significance when evaluating OS may be attributed to the reduced statistical power of the matched cohort. These data suggest that β-blockade may improve outcome of esophageal cancer patients by reducing tumor relapse.

Previous studies have reported an improvement in survival-based endpoints with concurrent β-blocker use in patients with several types of cancer (Table 3). Melanoma patients taking β-blockers were found to have reduced risk of progression and improved survival-based endpoints.(7, 9) Similar findings have been reported in various other cancers including lung, colorectal, and breast.(8, 10–12). Furthermore, Chang et al. demonstrated that propranolol use was associated with reduced incidence of cancer.(14) While β-blocker use was not associated with difference in prior or secondary malignancies in the current cohort, this study was not powered for such analysis.

Table 3.

Prior reports on BBs

Prior clinical reports on β-blockers in cancer
Author	N	Type	Findings
Lemeshow et al.	4179	Melanoma	Improved CSS (HR=0.87) and OS (HR=0.81)
Giampieri et al.	235	Colorectal	Improved OS (HR=0.44)
De Giorgi et al.	121	Melanoma	34% risk reduction of progression per year of use
Choi et al.	20,898	Several	Improved DFS (HR=0.69) and OS (HR=0.79)
Barron et al.	506	Breast	Improved BCM (HR=0.19)
Wang et al.	722	Lung	Improved DFS (HR=0.74), DMFS (HR=0.67), and OS (HR=0.78)
Wang et al.	673	Lung	Improved OS (HR=0.80)
Chang et al.	24,238	Several	Reduced incidence of cancer (HR=0.75)
Shah et al.	3462	Several	Worse outcome in pancreatic and prostate cancer
Yap et al.	Meta-analysis	Several	Improved outcome in melanoma and ovarian cancer, worse in head and neck, prostate, endometrial, and lung cancer

To the best of our knowledge, this is the first study to suggest a beneficial effect of β-blocker use on the outcome of esophageal carcinoma patients. A prior population-based analysis examined outcomes in cancer patients with a history of β-blocker use across a number of different malignancies, including esophageal cancer, finding no oncologic benefit in those taking β-blockers.(15) This study was included in a subsequent metanalyses, which evaluated β-blocker use in epidemiologic and perioperative reports.(16) Yap et al. found improved disease-free survival with β-blocker use in those with melanoma and ovarian cancer, where OS was also improved in melanoma as well. In contrast, worse survival-based outcomes were observed in head and neck, endometrial, prostate, and lung cancers. A significant limitation to an epidemiologic approach to studying this phenomenon is often the lack information regarding intercurrent illnesses of patients and specific information regarding treatment modalities employed. The inability to account for these factors, particularly in a setting when β-blocker use is associated with worse outcome, may be more reflective of differences in treatment or overall health status, not necessarily β-blocker-specific effects. This is illustrated in the current study as there was a substantial difference in the outcome of surgical and non-surgical populations, where β-blocker use was more common in the definitive chemoradiation group (37.7% vs 17.9%). This effect may vary depending on the malignancy studied, as β-blocker use and its associated comorbidities may be more or less restrictive based on the nature of the treatment (e.g. esophagectomy vs wide-local excision).

The mechanism underlying β-blockade and improved outcomes remains under investigation. Prior work within in vitro models demonstrated that β-AR signaling upregulates a number of oncogenes including Akt, NFκB, EGFR, and COX-2.(1, 2, 4) However recently researchers expanded on the pro-tumorigenic role of β-AR signaling, highlighting influences on tumor radiosensitivity and immune evasion.(6) Within murine models, the inhibition of the β-AR cascade, either pharmacologically or through physiologic means, can abrogate the oncogenic β-AR effects, subsequently promoting therapeutic response of both treated and untreated tumors. These findings bolster previous observations that manipulation of β-AR can reverse thermostress induced radioresistance and facilitate an improved response to PD-1 checkpoint blockade within mouse models of cancer.(23, 24) In addition to radioresistance, Eng et al. found that β-adrenergic signaling is vital for housing temperature-induced chemoresistance as well.(5) Recently, our group has shown for the first time, that chronic adrenergic stress suppresses the anti-tumor immune response and impairs responses to immunotherapy, chemotherapy, and radiotherapy.(6) Furthermore, blockade of adrenergic receptors reversed immunosuppression and significantly improve the efficacy of chemotherapy and radiation.(6) Therefore, β-AR blockade may improve tumor control through enhanced immune response, which is supported by our findings of improved distant control with β-blocker use.. A prospective trial investigating addition of propranolol to standard chemoradiation for esophageal cancer is currently under development.

As expected in this retrospective study, several known relevant clinicopathologic factors (age, sex, KPS, tumor stage, nodal stage) were associated with outcomes in esophageal cancer. Patients who underwent surgery fared far better than their non-surgical counterparts. While esophagectomy can improve tumor control, we attribute these findings to selection bias.(25) Although there was no observed difference in tumor response in those treated with neoadjuvant intent, we caution any conclusions in this setting given the relatively small number of surgical patients taking β-blockers. Despite these findings, given that β-blocker use improved distant failure, there still can be a role for β-blockade in operative patients. Interestingly, while prior studies showed a benefit to oncologic outcome with aspirin or non-steroidal anti-inflammatory drug use in head and neck cancer, no such relationship was observed in this setting.(26)

Within the study cohort, 141 (48.5%) patients were taking at least aspirin or β-blockers and 46.9% of β-blocker users were also on aspirin. Despite this overlap, it is unclear why these two medications impart differential effects on PFS. While aspirin use can improve outcome in various malignancies of the alimentary tract, these findings suggest a context specific effect which may be due to differences in disease biology.(27–29) Furthermore, the majority of studies demonstrating oncologic benefit to aspirin have been in the adjuvant or preventative setting, which may contribute to the contradictory findings in the current report. Additional research is needed in the potential opposing effects of aspirin and β-blockers in esophageal cancer.

Limitations of this study include the retrospective nature of the analysis, which has inherent biases. While statistical approaches can help control for these confounding factors, ultimately well-controlled prospective studies are required. In terms of β-blocker usage, it was denoted by use at time of initial consultation. This fails to consider duration of use or how faithfully these medications were being taken. Additionally, there was heterogeneity within the type β-blocker used and respective dosages. At this time, the optimal duration of β-blocker use is not well established however a recent preoperative trial in colorectal cancer administered propranolol for 20 consecutive days (5 days prior to surgery, 15 days after).(30)

Conclusions

Concurrent β-blocker use is common within patients receiving concurrent chemoradiation for esophageal carcinoma. Patients who underwent chemoradiation while taking β-blockers demonstrated significant improvement in treatment-related outcomes. Prospective clinical trials are needed to further characterize this potential option.