Clinicopathological Characteristics and Survival Outcomes for Pneumonectomy: a Population-Based Study

Background: Prognostic factors in pneumonectomy (PN) are not yet fully dened. This study is to analyze and evaluate long-term survival after PN for non-small cell lung cancer (NSCLC). Methods: We obtained data from the Surveillance, Epidemiology, and End Results (SEER) database for patients who underwent PN between 2004 and 2015. In order to reduce bias and imbalance, propensity score matching (PSM) analysis was performed. We used Kaplan-Meier curves to estimate overall survival (OS), performed univariate and multivariate Cox proportional hazards regression analyses to identify independent prognostic factors for OS, and applied the Cox proportional hazards model to create a forest plot. Results: A total of 1557 PN patients from the SEER database were included. The patients were grouped according to the side of PN. Before matching, OS was worse after right PN (hazard ratio [HR]: 1.303; 95% condence interval [CI] 1.133-1.498; P<0.001), but survival difference between groups was not signicant after matching (HR: 1.061; 95% CI 0.912-1.235; P=0.443). Regression analysis revealed that age, grade, N-stage, radiotherapy, and chemotherapy were independent predictors of OS (P<0.05). Chemotherapy was associated with improved OS (HR: 0.709; 95% CI 0.609-0.825; P<0.001), but radiotherapy was associated with increased risk of death in OS (HR: 1.268; 95% CI 1.061-1.561; P=0.009). On the forest plot, patients with primary overlapping lesions had better OS (HR: 0.684; 95% CI 0.497-0.941; P=0.020) in left, vs right, PN; patients diagnosis during 2004 to 2007 had lower OS (HR: 0.576; 95% CI 0.346-0.960; P=0.034) for left PN vs right PN. higher tumor grade (P<0.001), and higher N-stage (P<0.05) were signicant independent negative prognostic factors. The multivariate analysis also revealed that chemotherapy was an independent predictor of improved OS (HR: 0.522; 95% CI 0.439-0.621; P<0.001), while radiotherapy remained as a worse prognostic factor with increased risk of death for OS (HR: 1.304; 95% CI 1.063-1.600; P=0.011). The forest plot shows that for all time intervals other than the 2004-2007 period, patients with primary overlapping lung lesions had better OS for left PN vs right PN (HR: 0.684; 95% CI 0.497-0.941; P=0.020). During the 2004-2007 interval, OS for left PN was lower (HR: 0.576; 95% CI 0.346-0.960 P=0.034) (Fig. 5).


Background
Lung cancer incidence (11.6%) and mortality rates (18.4%) are among the highest of all malignant tumors in China and the world. The most recent estimate predicts 228,820 new cases and 135,720 deaths in 2020, emphasizing the serious worldwide effects of this disease, which has a 5-year survival rate of approximately 19 percent [1]. Radical resection remains the preferred treatment for NSCLC, and current widely-accepted surgical techniques include lobectomy, segmentectomy, pneumonectomy, and pulmonary sleeve with pulmonary artery reconstruction [2][3][4]. Central lung tumors are relatively common in clinical practice. Some can be treated with lobectomy or pulmonary sleeve resection, but when these methods cannot completely remove the tumor, PN is required, even as the prognostic factors for left and right PN remain under investigation. Depending on the surgeon's experience and the histological and anatomical characteristics of the lung and tumor, it is possible that survival is better for left PN than for right PN [5]. The aim of this study is to analyze and evaluate long-term survival after PN in patients with NSCLC. We used a population-based national registry, the Surveillance, Epidemiology, and End Results (SEER) database, to analyze clinical characteristics and prognosis for PN and for left and right PN. Based on the results of the survival analysis, we created a forest plot by using the Cox proportional hazards model.

Variables
This study was based on public data from the SEER database. The covariates included age, sex, race, marital status, year of diagnosis, primary site, summary stage, histopathology, grade, extent of lymph node dissection, tumor stage, T-stage, N-stage, radiotherapy, and chemotherapy. We classi ed age into ve groups: ≤50, 51-60, 61-70, 71-80, and 81 and older. The year of diagnosis was classi ed into three groups: 2004-2007, 2008-2011, and 2012-2015. The grade was classi ed as well differentiated, moderately differentiated, poorly differentiated, and undifferentiated. We followed the seventh edition American Joint Committee on Cancer (AJCC) lung cancer staging system, and updated the T-stage (T1, T2, T3 and T4), N-stage (N0, N1, N2-N3), and tumor stage (I, II, III) for all patients in all time periods.
Overall survival (OS) was de ned as the time from diagnosis to death from any cause.

Propensity score matching
In order to avoid bias between left and right PN groups, we applied 1: 1 propensity score matching (PSM) for age, sex, race, marital status, year of diagnosis, primary site, grade, summary stage, histopathology, lymph node dissection, tumor stage, T-stage, N-stage, radiotherapy, chemotherapy, survival months, and vital status recode.

Statistical analysis
Categorical variables are expressed as percentages and continuous variables are expressed as mean ± standard deviation (SD). Variables were compared using Student's t test, Chi-square test, and analysis of variance. We used the Kaplan-Meier method to generate survival curves and analyzed differences between curves by the log-rank test. We used the Cox proportional hazards model to examine independent prognostic factors and calculate the hazard ratio (HR) and the corresponding 95% con dence interval (CI  (Fig. 2a).
We used univariate analysis to identify possible prognostic factors in PN for NSCLC and found statistically signi cant (P<0.05) correlations between OS and age, sex, summary stage, grade, tumor stage, T-stage, N-stage, radiotherapy, and chemotherapy ( Table 2). Chemotherapy was associated with a better prognosis (Fig. 3a) and radiotherapy was associated with a worse prognosis (Fig. 4a). Race, marital status, year of diagnosis, histopathology, primary site, and lymph node dissection were not Multivariate analysis performed with the Cox regression model included age, sex, summary stage, grade, lymph node dissection, tumor stage, T-stage, N-stage, radiotherapy, and chemotherapy. The results showed that age, sex, grade, lymph node dissection, N-stage, radiotherapy, and chemotherapy were independent predictors of survival time in OS (P<0.05) ( Table 2), with radiotherapy appearing as a negative prognostic factor with increased risk of death for OS (HR: 1.264; 95% CI 1.049-1.523; P=0.014) and chemotherapy appearing as an independent predictor of improved for OS (HR: 0.564; 95% CI 0.480-0.661; P<0.001).
Propensity score matching survival analysis All variables were well balanced between the two groups after 1:1 PSM. The propensity scores before matching were 0.407 ± 0.081 for left PN and 0.436 ± 0.086 for right PN (P<0.001). After matching, the propensity score for both was 0.428 ± 0.082 (P=0.755). Finally, a total of 1230 patients (615 left PN and 615 right PN) were included in our study. We found there were no signi cant differences in baseline characteristics between matched groups except year of diagnosis (

Subgroup analysis in matched groups
Univariate analysis to identify possible prognostic factors after matching found statistically signi cant correlations between OS and age, sex, histopathology, summary stage, grade, tumor stage, T-stage, Nstage, radiotherapy, and chemotherapy (P<0.05) ( Table 4). The subsequent multivariate Cox regression model showed that age ≥61 (P≤0.002), higher tumor grade (P<0.001), and higher N-stage (P<0.05) were signi cant independent negative prognostic factors. The multivariate analysis also revealed that

Discussion
Anatomic surgical resection is currently the preferred method of treating lung cancer. Central tumors may be amenable to lobectomy or bronchial sleeve lobectomy [6][7], but in patients with large tumors, tumor invasion of the left or right main bronchus, and tumors crossing lung ssures, anatomic resection cannot be completed, and PN is required in order to achieve a clinical effect [8][9]. Nonetheless, PN has relatively high morbidity and mortality (5.0% to 10.0%) in the treatment of lung cancer [10]. The operation is traumatic and the risk of postoperative complications including cardiac arrhythmias, cardiac failure, pulmonary infection, bronchopleural stula, and acute respiratory distress syndrome (ARDS) is high [11][12]. Martin et al. [13] had a total mortality rate of 3.8% after PN (18/470), with an overall incidence of PN of 38.1% (179/470). Ludwig et al. [14] reported that the 5-year overall survival rate after PN was 27%, while Wang et al. [15] reported a post-PN 5-year survival rate of 46.3% in patients with pIII-N2 NSCLC. In this study, 1-, 3-, and 5-year OS rates after PN were 76.7%, 53.5%, and 44.7% before PSM, and OS was worse after right PN vs left PN (HR: 1.303; 95% CI 1.133-1.498; P<0.001). However, after matching, 1-, 3-, and 5-year OS rates were 75.4%, 50.7%, and 41.0% and there was no signi cant difference between sides (P>0.05). The results of Yang et al. [16] were similar, but they did not nd signi cant difference in 5-year survival rate between left and right PN before or after matching.
Of note, the multivariate Cox proportional hazards regression analysis found that PN patients receiving chemotherapy had signi cantly prolonged survival, both before and after matching (P<0.001). However, radiotherapy was a worse prognostic factor, showing increased risk of death in OS before and after matching (P<0.05). This might be because the substantial changes in lung function after PN decrease the tolerance to radiotherapy and increase the potential risk of adverse events that can shorten survival.  (Fig. 3b). Notably, our nding that radiotherapy was a worse prognostic factor associated with increased risk of death for OS (HR: 1.268; 95% CI 1.061-1.561; P=0.009) (Fig. 4b), has not been frequently reported.
Patients undergoing right PN lose more lung capacity than those undergoing left PN because the right lung accounts for 55-60% of the total lung volume. Therefore, preoperative optimization of cardiopulmonary function before right PN is particularly important. However, Deslauriers et al. [21] have reported that expiratory lung function decreases by approximately 30% following PN regardless of operation side, indicating that even though the proportion of lung volume loss is greater after right PN, long-term postoperative adjustments in pulmonary function may allow patients to adapt and lead nearnormal lives. Ilonen et al. [22] have also reported that there was no signi cant difference in pulmonary function after right vs left PN. Nonetheless, the relationship between lung function and survival prognosis after PN remains controversial, and there may be a poorer prognosis after right PN vs left PN. In this study, we did not compare the difference in lung function in relation to long-term survival after PN because of shortcomings of the database itself.
Because our data were collected from the SEER database, biases and errors may exist even though we used the PSM analysis. Several limitations were identi ed in this study. First, the study lacks detailed information regarding chemotherapy, radiotherapy, targeting, and even immunotherapy, whether pre-or postoperative. Second, we grouped the no or unknown variables into one group, leading to data bias.
Third, we used the 7th AJCC staging system, which replaced the 6th edition in 2010. Because the data we collected from the SEER database were from 2004-2015, there are inconsistencies in the data transformation process. Last, but not least, the SEER database lacks information on imaging, smoking history, tumor markers, etc., and our study did not address the impact of these factors on the prognosis in PN patients, even though they may play a signi cant role.

Conclusion
There was no signi cant difference in long-term survival between patients with NSCLC undergoing left vs right PN. Laterality was not a prognostic factor for survival after PN. Both neoadjuvant and adjuvant chemotherapy can prolong postoperative survival and either can be recommended. However, radiotherapy appeared to be a negative prognostic factor associated with increased risk of death in overall survival. Additional long-term survival and outcomes analyses should be conducted in larger numbers of patients. Abbreviations PN, pneumonectomy; NSCLC, non-small cell lung cancer; SEER, Surveillance, Epidemiology, and End Results; PSM, propensity score matching; OS, overall survival; HR, hazard ratio; CI, con dence interval; AJCC, American Joint Committee on Cancer; ARDS, acute respiratory distress syndrome.
Declarations Acknowledgements Not applicable.

Funding
This study was supported by the Liaoning Province Science and Technology Public Welfare Research Fund Project (#GY-20170016). The funding agency did not participate in the design of the study, data collection, analysis, or interpretation, or writing of the manuscript.

Availability of data and materials
The datasets supporting the conclusions of this article are included within the article.

Authors' Contributions
LLW and LHG drafted the manuscript. The data acquisition was performed by LLW and LHG. YYL and LLW designed the analysis. YR and LLW participated in the conception and design. All authors read and approved the nal manuscript.            Forest plot of individual hazard ratios for overall survival in left vs right pneumonectomy.