The Risk of Cardiac-related Mortality in Stage IIIA-N2 Non-small Cell Lung Cancer: Analysis of the Surveillance, Epidemiology, and End Results Database

Background: We aimed to investigate the association between postoperative radiotherapy (PORT) and cardiac-related mortality for patients with (cid:0) A-N2 non-small cell lung cancer (NSCLC) by the Surveillance, Epidemiology, and End Results (SEER) database. Methods: The United States (US) population based SEER database were queried for cardiac-related mortality among patients with (cid:0) A-N2 NSCLC. Cardiac-related mortality was compared between PORT and Non-PORT group. Accounting for mortality from other causes, Gray’s test compared cumulative incidences of cardiac-related mortality between both groups. Univariate and multivariate analysis were performed using the competing risk model. Results: From 1988 to 2016, 7290 patients met the inclusion criteria: 3386 patients treated with PORT and 3904 patients treated with Non-PORT. The 5-year overall incidence of cardiac-related mortality was 3.01% in PORT group and 3.26% in Non-PORT group. Older age, male sex, squamous cell lung cancer, earlier year of diagnosis and earlier T stage were independent adverse factors for cardiac-related mortality. However, PORT use wasn’t associated with an increase in the hazard for cardiac-related mortality (subdistribution hazard ratio [SHR]=0.99, 95% condence interval [95% CI], 0.78–1.24, p=0.91). When evaluating cardiac-related mortality in each time period, the overall incidence of cardiac-related mortality was decreased over time. There were no statistically signicant differences based on PORT use in all time periods. Conclusions: With a median follow-up of 25 months, no signicant differences were found in cardiac-related mortality between PORT and Non-PORT for (cid:0) A-N2 NSCLC patients. and T1 < 0.001). On multivariate analysis, sex, of histology, and T signicantly cardiac-related On univariate and multivariate analysis, risk cardiac-related mortality (univariate: SHR = 1.0, 95% CI, 0.97–1.24, p = 0.97; multivariate: SHR = 0.99, 95% CI, 0.78–1.24, p = 0.91).


Background
Chest radiotherapy plays an important role in the treatment of breast cancer, Hodgkin's lymphoma, lung cancer and other malignancies, which can also cause cardiovascular events and deaths. Cardiac-related mortality has long been recognized in patients with breast cancer or Hodgkin lymphoma [1][2][3][4][5], with a typical latency period of more than a decade. And the radiation-associated cardiac risk appears to increase with time after exposure 5 . As the life expectancy of non-small cell lung cancer (NSCLC) continues to rise, cardiac-related mortality is apparent in long-term cancer survivors of NSCLC. Several studies report increased cardiac events and deaths after de nitive radiotherapy for locally-advanced NSCLC (LA-NSCLC) [6][7][8][9]. Recently, the Radiation Therapy Oncology Group (RTOG) 0617 (A Randomized Phase III trial) reported that heart radiation dose was associated with worse overall survival with a median follow-up of 2 years, suggesting a contribution of radiation-induced cardiac mortality relatively soon after treatment [10]. Postoperative radiotherapy (PORT) has been posited as an important component of the comprehensive treatment of operable A-N2 NSCLC [11]. However, evidence for cardiacrelated mortality after Postoperative radiotherapy (PORT) is more limited in recent years. Given the parallels in de nitive radiotherapy of LA-NSCLC, cardiac-related mortality of PORT gain importance in A-N2 NSCLC, especially with contemporary treatment modality. In the present study, we assessed the effect of PORT on cardiac-related mortality in A-N2 NSCLC patients using a large population-based database, and strati ed these patients over time intervals to assess the cardiac-related mortality by treatment era. . And the Exclusion criteria were bilateral tumors or unknown laterality, main bronchus tumors or overlapping lesion or unknown lobe, without postoperative external-beam irradiation or unknown, multiple primary, and having survived for less than 3 months. Two cohorts were then created: (I) patients who received postoperative radiotherapy (PORT) as part of their initial therapy; and (II) those who did not receive postoperative radiotherapy (Non-PORT) as part of their initial therapy. The work ow is shown in Fig. 1.

National
For each case, we requested the following information: age, sex, race, year of diagnosis, histology, tumor laterality (right vs left), subsite location (upper lobe, middle lobe, lower lobe), T stage, surgical procedure (lobectomy vs pneumonectomy), use of radiotherapy, presumed survival in months, vital status, and cause of death. Cardiac-related mortality information was obtained from cause of death data extracted from the SEER database.

Statistical analysis
We compared the differences in demographics, clinicopathology, and treatment characteristics among patients who did and did not receive PORT by chi-square test. The cardiac-related mortality rate during the study period were graphed by the initial year of diagnosis. We divided patients by the year of diagnosis of 1988-1997, 1998-2007, 2008-2016. Cumulative incidence curves of cardiac-related mortality (from the date of diagnosis) were constructed for PORT versus non-PORT, accounting for mortality from other causes. The underlying hazards were statistically compared using Gray's test. Univariate and multivariate analysis were performed using the competing risks regression (Fine and Gary method [12]) to analyze the risk factors for cardiac-related mortality including age, sex, race, histology, tumor laterality, subsite location, T stage, surgical procedure, use of radiotherapy by assessment of the subdistribution hazard ratios (SHRs) and their corresponding 95% con dence interval (95% CI). A P-value of less than 0.05 was set as the threshold for signi cance. All analyses were performed with the software of R Version 3.6.2.

Results
We identi ed 7290 patients who met the criteria described above, including 3386 (46.4%) patients with PORT and 3905 (53.6%) patients with Non-PORT. The median age of the entire cohort was 66 years (range, 18-95 years). A comparison of demographic, clinicopathologic and treatment factors were presented in Table 1. The distribution of age (P < 0.001), year of diagnosis (P < 0.001), histological type (P = 0.039), subsite location (P = 0.002), and chemotherapy status (P = 0.016) by comparing PORT with Non-PORT reached statistical signi cance. The median follow-up was 25 months (range, 3-320 months). A total of 308 patients died of cardiac-related diseases. 145 and 163 patients died of cardiac-related diseases in PORT and Non-PORT group, respectively. Abbreviations: PORT, postoperative radiotherapy; Non-PORT, without postoperative radiotherapy Figure 2 presents the cardiac-related mortality curves strati ed by PORT use. The 5-year overall cardiacrelated mortality was 3.01% in PORT group compared with 3.26% in Non-PORT group. There were no signi cant differences (p = 0.49) between PORT and Non-PORT ( Fig. 2A)

Discussion
The effect of PORT on cardiac-related mortality is critical due to inadequately capturing the cardiacrelated mortality risks in recent years. Our population-based analysis offers a holistic update on the cardiac-related mortality risks in A-N2 NSCLC patients who underwent PORT. To our knowledge, this is the rst study concentrating on the incidence and trends of competing risk cardiac-related mortality in A-N2 NSCLC patients with long observation, and the results show that cardiac-related mortality has no association with PORT use between 1988 and 2016. And the incidence of cardiac-related modality has decreased over time.
Previous studies have demonstrated a bene t in local-regional control by using PORT but whether PORT could improve overall survival remains controversial [13][14][15][16][17][18]. The insigni cant survival bene ts with PORT could partially be explained by the early and late toxicities, including radiation esophagitis, radiation pneumonitis and radiation induced heart diseases. And cardiac-related mortality is one of the detrimental effect of PORT on survival. Several studies have reported increased cardiac-related mortality with PORT after surgery for stage I to NSCLC [19][20][21]. Lally et al [20] conducted the risk of cardiac-related mortality associated with PORT of resected NSCLC patients. It was demonstrated that PORT was associated with an increase in cardiac-related mortality independent of the expected effects of age, sex, and race between 1983 and 1988 by using SEER database. However, signi cant differences exist between Lally et al [20] and our study, given the substantial heterogeneity in patient selection, year of diagnosis and method of statistics analysis. First, in the study by Lally et al [20], a mixture of patients with varying stages (stage I to III) of resected NSCLC were included. In contrast, our study concentrated on A-N2 NSCLC patients, representing a population that is most likely to bene t from PORT. Second, the recruitment of patients included in Lally et al [20] went from 1983 to 1988 comparing with 1988 to 2016 in our study. Lastly, compared the results of study by Lally et al [20], our study used competing risk model to avoid overestimate cumulative survival rate and incorrectly evaluate the effects of covariates on the hazard ratio to the occurrence of the cardiac-related death event. This guaranteed the result's reliability of our study.
Our study found an approximate 3% cardiac-related mortality rate at 5 years experienced by all patients with A-N2 NSCLC who receive either PORT or Non-PORT. The insigni cant difference of cardiac-related mortality observed between PORT and Non-PORT group in our study is probably due to the contemporary radiation techniques, limited volumes of radiation, suitable radiation doses and fraction sizes [22,23]. Taken together, these factors likely decreased the volume of heart exposed to high dose radiation. Trials with contemporary radiation techniques demonstrated that there was no increase of death from intercurrent disease [24][25][26][27][28]. A serious of recently published non-randomized trials could not detect difference in cardiac-related mortality between patients receiving 3D-planned PORT group in comparison to Non-PORT group [14,29,30]. What's more, the advances in lung cancer treatment could also explain the decrease of cardiac-related mortality over time . The trends are consistent with the analysis of Lally et al [20] and Haque et al [9]. Lally et al [20] demonstrated the decrease of cardiac-related mortality in resected NSCLC (1983)(1984)(1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993), while Haque et al [9] illustrated that the incidence of cardiac-related mortality has decreased over time in LA-NSCLC (1988-2014).
On univariate and multivariate analysis, our study con rmed that patients of older age, male patients, squamous cell lung cancer, earlier year of diagnosis and earlier T stage were more likely to have higher rates of cardiac-related mortality. This may be explained by the baseline risks of these subgroups rather than any relation to radiotherapy [31,32]. This implies that patients with greater comorbidities or potential for longer survival may require extra attention to cardiac sparing [33]. Cardiac monitoring with imaging and routine involvement of cardiologist in the pre-and post-treatment care may be necessary for patients with cardiac-related mortality risk factor. Although there is no association between PORT use and cardiacrelated mortality, cardiovascular issues such as cardiomyopathy, arterial disease, hypertension, and vascular and metabolic issues could manifest during or after therapy due to PORT-related toxicities. And we still encourage clinical attentiveness to cardiac-sparing radiotherapy and emphasize overall health through management of early and late cardiac toxicities. With increasing life expectancy of A-N2 NSCLC, further studies are needed in order to provide dosimetric correlates for cardiac toxicities, which is rarely reported in previous studies [34].
Several limitations should be acknowledged when interpreting our results. First, just as other studies derived from SEER database, our study is limited by the lack of important paramenters, such as the following: comorbid conditions, smoking history, chemotherapy utilization, radiation therapy dose, technique and target, which are potential confounders for our study. Second, there is an inherent weakness of selection bias in a population-based observational study. Third, the number of events was low, which limits the examination of multiple covariates.
These shortcomings notwithstanding, our study is still the largest retrospective study to address this important clinical question. What's more, the large cohort of population-based patients provides su cient power to detect relatively small increases in cardiac-related mortality for A-N2 NSCLC. We believe that PORT use is safe in A-N2 NSCLC patients, without worrying about cardiac-related mortality.

Conclusions
In conclusion, this study characterizes the cardiac-related mortality by PORT use in patients with NSCLC. In all patients over the studied period, no signi cant differences were found in cardiac-related modality between PORT and Non-PORT group in A-N2 NSCLC patients. The overall incidence of cardiac-related mortality was decreased over time. Although the association with PORT cannot be proven in our study, we still encourage attentiveness to early and late cardiac toxicities. And more researches are needed to nd out the association between cardiac toxicities and PORT in A-N2 NSCLC.
Abbreviations NSCLC: Non-small cell lung cancer; LA-NSCLC: locally-advanced non-small cell lung cancer; PORT: Postoperative radiotherapy; SEER: Surveillance, epidemiology, and end results; CI: Con dence interval; HR: Hazard ratio. Figure 1 Patients selection diagram