Cardiovascular Mortality Risk among Patients with Gastroenteropancreatic Neuroendocrine Neoplasms: A Registry-Based Analysis

Background This research is aimed to explore mortality patterns and quantitatively assess the risks of cardiovascular mortality (CVM) in patients with primary gastroenteropancreatic neuroendocrine neoplasms (GEP-NENs). Methods We extracted data from the Surveillance, Epidemiology, and End Results (SEER) database for patients diagnosed with GEP-NENs between 2000 and 2015. The standardized mortality ratio (SMR) and the absolute excess risk were obtained based on the reference of the general US population. The cumulative incidence function curves were constructed for all causes of death. Predictors for CVM were identified using a multivariate competing risk model. Results Overall, 42027 patients were enrolled from the SEER database, of whom 1598 (3.8%) died from cardiovascular disease (CVD). The SMR for CVM was 1.20 (95% CI: 1.14-1.26) among GEP-NEN patients. The cumulative mortality of CVD was the lowest among all causes of death, including primary cancer, other cancer, and other noncancer diseases. Furthermore, age at diagnosis, race, Hispanic origin, sex, marital status, year of diagnosis, grade, education level, region, SEER stage, primary site, surgery, and chemotherapy were identified as independent predictors of CVM in GEP-NEN patients. Conclusions GEP-NEN patients have a significantly increased risk of CVM relative to the general population. Cardioprotective interventions might be considered a preferred method for these patients.


Introduction
Neuroendocrine neoplasms (NENs) are a collection of fairly rare neoplasms, which used to be named as "carcinoid" on account of their heterogeneous and indolent clinical nature [1]. Gastroenteropancreatic neuroendocrine neoplasms (GEP-NENs) originate from neuroendocrine cells throughout the body constituting two-thirds of NENs [2]. In the past 40 years, the incidence of GEP-NENs has been steadily growing, with an increase of 3.65 times in the United States and 3.8-4.8 times in the UK [3]. The recently reported annual age-adjusted incidence of GEP-NENs is approximately 3.56/100,000 in the United States and 4.60/100,000 in England, respectively [4,5]. Advanced in diagnostic endoscopy, physician awareness and treatment of cancers, the outcome of GEP-NENs patients had improved considerably, with 3-and 5-year overall survival rates of 79.4% and 74.7%, respectively [6,7]. Cardiovascular disease (CVD) has become one of the leading causes of non-cancer death [8].
A previous study reported that the cardiovascular mortality (CVM) had increased by 21.1% from 2007 to 2017 in global [9]. In 2016, approximately 17.9 million people died of CVD globally, accounting for 31% of the total global deaths, while roughly 9 million death caused by cancer [10,11]. In the United States, Kochanek et al. reported that 647457 deaths arose from diseases of heart, more than 599108 deaths from primary malignant neoplasms in 2017 [12].
Prior studies have shown that patients with colorectal cancer and endometrial cancer have an 11.7 and 8.8 folds higher risk of CVM than the general population, respectively [13,14]. Fang et al. have concluded that the risk of prostate cancer patients developing CVM in the rst month and 7-12 months after diagnosis is 2.05 and 0.92 folds that of the general population, respectively [15]. Weberpals and colleagues have shown that the risk of CVM for breast cancer patients is 0.84 times that of the general population [16]. To sum up, the risk of CVM varies signi cantly among tumor patients depending on different primary sites and time after diagnosis compared with the general population. To our knowledge, there have been no reports focus on CVM of patients with GEP-NENs. Hence, we described the risk assessment and patterns for causes of death, and identi ed independent predictors for CVM in GEP-NENs patients in this study.

Data source
We extracted data of patients patients with primary GEP-NENs between 2000 and 2015 from the Surveillance, Epidemiology, and End Results (SEER) database using the SEER*Stat software (version 8.3.6) [17]. The SEER program, including incidence, survival and mortality data, is a system of populationbased cancer registries sponsored by National Cancer Institute covering around 27.8% of the total US population (based on the 2010 census) [18]. The US mortality of the reference cohort (representing the general population) reported in the National Vital Statistics System also can be collected through the SEER program [19]. Ethical approval of this publicly available information provided by the SEER program was not required.

Study Variables
Data were summarized as mean ± standard deviation (SD) or median and interquartile range (IQR) for continuous variables, and number (percent) for categorical variables.
The variables involved in this study included age at diagnosis, attained age, year of diagnosis, gender, SEER stage (localized, regional, and distant), race, Hispanic origin, marital status, grade (well differentiated as grade I, moderately differentiated as grade II, poorly differentiated as grade III, and undifferentiated as grade IV), region (Midwest, West, South, Northeast), education level, mean household income, histologic subtype, primary site, surgery, chemotherapy, radiotherapy, cause of death and survival time.
Since there is no personal data on education level and household income in the SEER database, we used 2000 US Census data to obtain county-speci c average educational level and household income [15]. Survival time refers to the interval from the diagnosis of cancer to the death of patients ascribed to any causes or the last day of the available survival information [21].

Statistical analysis
The relative risk of CVM for GEP-NENs patients was compared to all US residents and presented as the standardized mortality ratio (SMR) [22]. SMR is the ratio of observed to expected number of CVM [20,21]. Expected numbers were calculated by multiplying the mortality rate in the reference cohort by the person years (PYs) in the cancer cohort [23]. The absolute excess risk (AER, per 10,000 PYs) was calculated as follows: AER = [(observed deaths -expected deaths) /PYs of observation] × 10,000 [20,21]. CVM was described as the primary event of interest, while competing events refer to death causes by primary cancer, other cancer and other non-cancer. The crude cumulative incidence function (CIF) is used to express the probability of developing primary and competing events using Fine-Gray competing risk model [24,25]. Multivariate competing risk survival analyses were performed to identify independent predictors of CVM. Data analyses were performed by R software (version 3.6.3). All tests are 2-sided, and P-value < 0.05 signi ed statistical signi cance.

Cumulative mortality of CVD
The results of CIF curves for all causes of death in GEP-NENs patients using Fine-Gray competing risk model were illustrated in Figure 2. The cumulative mortality (CM) of CVD was the lowest in all causes of death. At the follow-up time of 200 months, the CMs of CVD, primary cancer, other cancer, and other noncancer disease were 9.4%, 12.3%, 16.9%, and 13.8%, respectively. In the early follow-up period, the highest CM was caused by primary cancer. The CMs of other cancer and non-cancer disease exceeded that of primary cancer at around 90 and 170 months after diagnosis, respectively.
In the subgroup analyses strati ed by age at diagnosis, we observed that the CM of CVD steadily increased with the age at diagnosis (Tab. 3). The CM of CVD was the lowest in all causes of death in subgroups of aged < 50 years (3.1%) and 50-64 years (5.5%) (Tab. 3, Fig. 3A and B). In subgroups of aged 65-79 years and ≥ 80 years, the CM of CVD exceeded that of primary cancer at around 180 months and 120 months after diagnosis, respectively ( Fig. 3C and D). In the subgroup analyses strati ed by primary site, pancreas and small intestine NENs patients had the lowest (4.12%) and highest (13.26%) CM of CVD, respectively (Tab. 3). We observed that the CM of CVD was the lowest in all causes of death in subgroups of the primary site of colon (9.09%), appendix (4.84%) and pancreas (4.12%) (Tab. 3, Fig. 4A-C). In subgroups of the primary site of stomach and rectum, the CM of CVD exceeded that of primary cancer at around 160 months and 90 months after diagnosis, respectively ( Fig. 4D and E). Interestingly, the CM of CVD in the subgroup of the primary site of small intestine was higher than that of primary cancer across all follow-up periods (Fig. 4F).   [8]. A recent study based on the SEER database showed that 1680 (5.6%) NENs patients died from heart diseases and 545 (1.8%) NENs patients died from other CVD (hypertension without heart disease, cerebrovascular diseases, atherosclerosis, aortic aneurysm and dissection, and other diseases of arteries/arterioles/capillaries), with SMRs of 2.31 (95%CI: 2.20-2.42) and 2.36 (95%CI: 2.17-2.57), respectively [28]. Most NENs are primarily located in the GEP (67.5%) and bronchopulmonary system (25.3%) [29]; however, the 5-year overall survival rates between GEP-NENs (74.7%) and bronchopulmonary NENs (33.7%) patients were signi cantly different [7,30]. These ndings suggested NENs patients had various natures and characteristics depend on different primary sites. Hence, we focus exclusively on the GEP-NENs in the present study.
In this study, we comprehensively assessed the risk of all causes of death among more than 42 thousand GEP-NENs patients from the SEER database, and found the risk of CVM in GEP-NENs patients was 20% higher than that of the general US population (SMR: 1.20; 95%CI: 1.14-1.26).
According to the competing risk analyses, we found that the CM of CVD was the lowest among all causes of death including primary cancer, other cancer and other non-cancer disease.
In addition, we identi ed age of diagnosis, race, Hispanic origin, gender, martial status, year of diagnosis, grade, education level, region, SEER stage, primary site, surgery and chemotherapy were independent predictors of CVM in GEP-NENs patients.
NENs were previously known as carcinoid tumours, in which approximately 50% of patients developed carcinoid syndrome [31]. Around 60% NENs patients with carcinoid syndrome developed carcinoid heart disease (CHD) which was characterized by development of valvular dysfunction, in particular right heart failure [32]. In addition, several researches have found that NENs patients are prone to depression and anxiety [33,34], which may aggravate state of cardiovascular physiology [15,35]. These results may explain the high risk of CVM in patients with NENs to some extent.
In terms of the time after cancer diagnosis, we con rmed that GEP-NENs patients within the rst twomonth after diagnosis had the highest risk of CVM (SMR: 3.64; 95% CI: 3.05-4.30). This nding was similar to previous conclusions reported by Sturgeon et al. and Zaorsky et al. [8,36]. Moreover, Ye et al. and Fang et al. showed that the recent diagnosis of cancer could be a major psychological stressor and lead to a negative effect on cardiovascular physiology [15,26,35]. These results suggested that psychiatric evaluation and psychological support could be indispensable for GEP-NENs patients with recent diagnosis of cancer. In terms of age at diagnosis, we observed that the CM of CVD steadily increased with the age at diagnosis. This phenomenon resembled previous ndings reported by Weberpals et al. and Ye et al. [16,26]. In general, death from primary cancer was the most common cause of death in cancer patients; however, the CM of CVD exceeded that of primary cancer in patients aged ≥ 65 during follow-up time ( Fig. 3C and D). These results implied that surveillance efforts should not only include assessment of primary cancer but also control of modi able risk factors for CVD in elderly cancer patients. In terms of primary site, we observed that pancreas NENs patients and small intestine NENs patients had the lowest (4.12%) and highest (13.26%) CM of CVD, respectively. One possible reason was that CHD occurs most frequently in small intestine NENs patients, accounting for 72% [32]. Another plausible explanation was that pancreas NENs patients had so advanced tumor stage that they might have not enough life expectancy to die of CVD [28,37,38], which may be explainable for the lower risk of The multivariate competing risk analysis was used to identify independent indicators of CVM in GEP-NENs patients in current study. We found that aged patients at diagnosis were inclined to die due to CVD (HR: 4.799; 95%CI: 4.313-5.341). Interestingly, patients with younger age at diagnosis (≤ 39 years) had the highest SMR 3.20 (95%CI: 1.93-4.99), which similar with the results reported by Zaorsky et al. [36]. Male patients had a high probability of CVM compare with female patients, as previous reports of colorectal cancer and non-Hodgkin's lymphoma [13,39]. A plausible reason is that males have worse health behaviors, such as smoking and drinking, which were con rmed as independent risk factors of CVD [40][41][42]. Our study showed that Black patients were signi cantly associated with the higher CVM risk compared with other races. Although patients with different ethnicities had a difference in receiving cancer therapy in the United States, this difference alone cannot explain the discrepancies of cancer patients in death due to non-cancer causes [43]. Hence, further investigations on this subject remained warranted. Patients with unmarried status showed propensity to die of CVD in contrast to married patients, as previously reported in non-Hodgkin's lymphoma [39]. A reasonable explanation was that married patients were more likely to feel cared for and encouraged and supported physically and spiritually in contrast to unmarried patients [44]. Other studies also had revealed that marriage could help to improve cardiovascular, endocrine, immune function and cancer prognosis [45][46][47]. Sturgeon et al. reported that individuals with low socioeconomic status were prone to have a high risk of CVM in cancer survivors [8]. In our study, patients with low education level commonly gave rise to higher risk of CVM, which was consistent with results of prior studies [15,21].
In the present study, a majority (76.8%) of patients underwent surgery, 10.3% patients received chemotherapy, and only 2.3% patients received radiotherapy. Notably, multivariate analysis indicated that patients received chemotherapy had a reduced CVM risk compared with patients not received chemotherapy. This result seemed to be inconsistent with the known cardiotoxic effect of chemotherapy, but conformed with the nding reported by Low et al. [28]. A possible reason was that patients who received chemotherapy have not enough life expectancy to occur CVM events (median survival time: chemotherapy 18 months vs surgery 61 months). We concluded that patients without surgery had an increased CVM risk compared with patients received surgery, which was consistent to the results from prior studies [13,14,44]. In respect of radiotherapy, prior study reported that radiation-induced macrovascular damages accelerated age-related atherosclerosis and microvascular damages, and reduced capillary density [48], however, radiotherapy was not an independent predictor for CVM in our study. In the SEER program, radiotherapy was de ned as the rst-course radiation treatment but lack of detailed regimen. Therefore, further investigation is required to clarify the effect of radiotherapy on the risk of CVM in patients with GEP-NENs.
Limitations still exist in our study. First, some information associated with CVD were not available in the SEER registry, such as comorbidities, smoking and alcohol use, doses of radiotherapy and chemotherapy agents. Second, this study is a retrospective study, which might lead to a potential selection bias in the participants. Third, causes of death may be subject to misclassi cation ascertained from death certi cates, and there was evidence indicating that causes on death certi cates about CVM may be overestimated [49].

Conclusions
In conclusion, GEP-NENs patients were found to show an upward trend risk toward CVM in contrast to the general population, especially during the rst two-month after diagnosis. The CM of CVD was the lowest among all causes of death including primary cancer, other cancer and other non-cancer disease. In addition, age of diagnosis, race, Hispanic origin, gender, marital status, year of diagnosis, grade, education level, region, SEER stage, primary site, surgery and chemotherapy were independent predictors of CVM in GEP-NENs patients. These results suggested that patients after diagnosis of GEP-NENs should be screened for CVD timely and undergo more extensive control of modi able risk factors of CVM. Abbreviations CVM: cardiovascular mortality; GEP-NENs: gastroenteropancreatic neuroendocrine neoplasms; SEER: Surveillance, Epidemiology and End Results; SMR: Standardized mortality ratio; CVD: cardiovascular disease; NENs: Neuroendocrine neoplasms; ICD-O-3: International Classi cation of Diseases for Oncology, the third edition; ICD-10: International Classi cation of Diseases, 10th Revision; SD: standard deviation; IQR: interquartile range; PYs: person years; AER: absolute excess risk; CIF: cumulative incidence function.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.

Availability of data and materials
The datasets analyzed in this study are available in the SEER repository and can be obtained from: https://seer.cancer.gov/data/.