Analysis of risk factors for postoperative complications in non-small cell lung cancer: comparison with the Japanese National Clinical Database risk calculator

Background Although the risk calculator of the National Clinical Database (RC-NCD) has been widely used to predict the occurrence of mortality and major morbidity in Japan, it has not been demonstrated whether a correlation between the calculated RC-NCD risk score and the actual occurrence of mortality and severe morbidity exists. Methods The clinical data of 585 patients who underwent pulmonary resection for non-small cell lung cancer were collected, and the risk factors for postoperative morbidity were analyzed to verify the validity of the RC-NCD. Results The coexistence of asthma (p = 0.02), nutrition lymphocyte ratio (p = 0.04), and pulmonary lobe (p < 0.01) were significant risk factors for postoperative morbidity in the present study, and the percent-predicted vital capacity (p < 0.01), pulmonary lobe (p = 0.03), and type of operative procedure (p = 0.01) were significant risk factors for severe postoperative morbidity. Furthermore, in patients received lobectomy, coexistence of asthma (p = 0.01) and pulmonary lobe (p < 0.01) were identified as significant risk factors for postoperative morbidity. Meanwhile, male sex (p = 0.01), high BMI (p < 0.01), low vital capacity (p = 0.04), and pulmonary lobe (p = 0.03) were identified as significant risk factors for severe postoperative morbidity. Conclusions Given that the pulmonary lobe was a significant risk factor for postoperative morbidity in patients received pulmonary resection and for severe postoperative morbidity in patients received lobectomy, the RC-NCD for postoperative morbidity needs to be modified according to high-risk lobes. Trial registration: The Institutional Review Board of Kanazawa Medical University approved the protocol of this retrospective study (approval number: I392), and written informed consent was obtained from all patients.

chronic obstructive pulmonary disease (COPD) are considered significant risk factors for postoperative pulmonary complications.
In 2011, the National Clinical Database (NCD) of Japan began an annual web-based data collection registration system. Data on 1.6 million surgical procedures from more than 4,000 hospitals were collected [11]. The risk calculator system was established based on a model of lung cancer surgery risk derived from the NCD of 78,594 patients from the years 2014 to 2015 [12]. In this study, mortality and the combination of mortality and major morbidity were analyzed. Operative mortality was defined as death within 30 days after surgery, and major morbidity was defined according to the Society of Thoracic Surgeons General Thoracic Database (STS GTDB) [13,14]. Although the risk calculator of NCD (RC-NCD) has been widely used to predict the occurrence of mortality and major morbidity in Japan (Fig. 1), whether or not there is a correlation between the risk score calculated by the RC-NCD and the actual occurrence of mortality and severe morbidity has not been demonstrated.
In the present study, we retrospectively evaluated the risk factors for mortality and morbidity after pulmonary resection in patients with NCSLC and verified the validity of the RC-NCD.

Patients
Seven hundred and thirty-nine NSCLC patients were received pulmonary resection at Kanazawa Medical University between January 2010 and March 2019, and 154 patients who underwent wedge resection were excluded. Therefore, 585 patients were enrolled in this retrospective study (Fig. 2).
Data including clinical factors, such as the sex, age, comorbidities, smoking history, body mass index (BMI), carcinoembryonic antigen (CEA) levels, respiratory function, prognostic nutritional index (PNI), neutrophil-tolymphocyte ratio (NLR), tumor diameter on computed tomography (CT), lobe involvement in lung cancer, clinical and pathological stage, and histology, were collected. The following comorbidities were included, according to the factors of the RC-NCD: malignant disease, hypertension, diabetes mellitus, angina pectoris, COPD, cerebral infarction, arrhythmia, interstitial lung disease, asthma, autoimmune disease, and chronic renal failure. Respiratory function parameters, such as the percent-predicted vital capacity (%VC) and forced expiratory volume in 1 s as a percentage of forced vital capacity (FEV 1 %), were collected. The PNI, calculated from the serum albumin level and total lymphocyte count, is a simple and useful indicator of one's immune-nutritional status [15]. The NLR was defined as the ratio of the neutrophil to the lymphocyte count, functioning as a parameter of systemic inflammation and stress in critically ill surgical and medical patients [16]. The smoking history was assessed using the Brinkman index, which is calculated by multiplying the number of cigarettes smoked per day by the number of years the subject has been smoking [17].

Operative factors
The operative approach was divided into three categories: video-assisted thoracic surgery (VATS), robot-assisted thoracic surgery (RATS), and thoracotomy. The operative procedure was divided into eight categories: wedge resection, segmentectomy, lobectomy, sleeve lobectomy, lobectomy combined with segmentectomy, lobectomy combined with chest wall resection, bi-lobectomy, and pneumonectomy.

Postoperative complications
Postoperative complications were categorized into five grades according to the Clavien-Dindo classification system. Established in 1992, it is a simple and feasible grading system for all types of postoperative complications [18]. In 2004, it was modified to allow for the grading of life-threatening complications and long-term disability caused by a complication [19]. This revised version defined five grades of severity with subgrades (grades I, II, IIIa, IIIb, IVa, IVb, and V), and the suffix "d" (for "disability") is used to denote any postoperative impairment. This modified version of the Clavien-Dindo classification has been widely used in clinical practice. Severe morbidity was defined according to the classification of RC-NCD, which was based on the STS GTSD risk model and included the following: respiratory failure, interstitial pneumonia, tracheobronchial fistula, pulmonary thromboembolism, pneumonia, redo surgery, myocardial infarction, arrhythmia requiring therapy, renal failure, postoperative bleeding, and chylothorax [12,13].
(See figure on next page.) Fig. 1  Furthermore, we added air leakage requiring therapy, atelectasis, asthma attacks, and cerebral infarction as postoperative morbidities.

Statistical analyses
Pearson's chi-squared test of independence was used to compare the frequencies of the variables. Risk factors related to postoperative complications were analyzed using a logistic regression analysis. Factors showing significant differences in the univariate analysis were included in the multivariate analysis. All statistical analyses were two-sided, and statistical significance was set at p < 0.05. Statistical analyses were conducted using the JMP software program (version 13.2; SAS Institute Inc., Cary, NC, USA). The present study was conducted in accordance with the principles of the Declaration of Helsinki. The Institutional Review Board of Kanazawa Medical University approved the protocol (approval number: I392), and written informed consent was obtained from all patients.

Patient characteristics, operative factors, postoperative morbidity, and mortality
The clinical characteristics of the 585 patients are shown in Table 1. Altogether, 333 patients had comorbidities, including 81 patients with the history of other malignant disease in addition to NSCLC (16 with colon cancer, 6 with rectal cancer, 14 with gastric cancer, 2 with esophageal cancer, 12 with breast cancer, 8 with prostate cancer, 9 with bladder cancer, 3 with gallbladder cancer, 7 with thyroid cancer, 3 with renal cancer, 4 with laryngeal cancer, 3 with pharyngeal cancer, and 5 with lymphoma; duplication occurred in some patients), 92 patients with hypertension, 81 with diabetes mellitus, 37 with angina pectoris, 35 with COPD, 24 with cerebral infarction, 20 with arrhythmia (18 with atrial fibrillation, 1 with paroxysmal supraventricular tachycardia, and 1 with atrioventricular block), 9 with interstitial lung disease, 20 with asthma, 15 with autoimmune disease, and 7 with chronic renal failure.
Postoperative morbidities were observed in 176 patients (30.0%). Clavien-Dindo grade I complication was noted in 1 patient, grade II in 71, grade IIIa in 99, and grade IIIb in 5. Air leakage occurred in 81 patients, arrhythmia in 44 (atrial fibrillation in 37, paroxysmal supraventricular tachycardia in 3, ventricular tachycardia in 3, and sick sinus syndrome in 1), atelectasis in 19, pneumonia in 12, asthma attacks in 4, cerebral infarction in 3, chylothorax in 2, and bronchopleural fistula in 2. Minor but serious postoperative complications included postoperative bleeding in one patient and right middle lobe torsion in another patient. All complications were resolved surgically.
Postoperative death was noted in 1 patient, and the mortality rate was 0.1%. Mortality and severe morbidities according to the RC-NCD were observed in 67 patients (11.4%); the severe morbidities were respiratory failure in 4 patients, tracheobronchial fistula in 2, pneumonia in 12, redo surgery in 2 (including postoperative bleeding in 1), arrhythmia requiring therapy in 44 (atrial fibrillation in 37, paroxysmal supraventricular tachycardia in 3, ventricular tachycardia in 3, and sick sinus syndrome in 1), and chylothorax in 2.
The median mortality rate according to the RC-NCD was 0.2%, the median mortality and severe morbidity rate was 4.6%, the median incidence rate of tracheal or bronchial fistula was 0.2%, and the median incidence rate of respiratory dysfunction was 0.2%.

Results of the univariate analysis
The relationships between patient characteristics or operative factors and postoperative morbidity or severe morbidity were analyzed (Table 2). Postoperative morbidities were more likely to be associated with the coexistence of COPD (p = 0.03), coexistence of asthma (p < 0.01), smoking status (p < 0.01), CEA (p = 0.03), NLR > 1.56 (p = 0.01), FEV 1 % < 70 (p < 0.01), pulmonary lobe (right upper or right lower or left upper) (p < 0.01), squamous cell carcinoma (p = 0.02), and extend-lobectomy (p = 0.02) than other factors. Severe postoperative morbidity was more likely to be associated with low %VC (p < 0.02), squamous cell carcinoma (p < 0.01), and extend-lobectomy (p = 0.01) than other factors. Significant risk factors for other postoperative complications could not be assessed because of the small number of patients with these complications.

Results of the multivariate analysis
Multivariate analyses of the risk factors for postoperative morbidities are presented in Table 3. The coexistence of asthma (odds ratio [OR] 3.07, 95% confidence interval

Results of the sub-analysis
We analyzed the risk factors for postoperative morbidity in patients who underwent lobectomy as a sub-analysis. The clinical characteristics of the 473 patients who underwent lobectomy are shown in Table 4. Among these, 163 underwent right upper lobectomy (RUL), 37 underwent right middle lobectomy (RML), 94 underwent right lower lobectomy (RLL), 105 underwent left upper lobectomy (LUL), and 74 underwent left lower lobectomy (LLL). Postoperative morbidities were observed in 142 patients (30.0%). A Clavien-Dindo grade I complication was noted in 1 patient, grade II in 54, grade IIIa in 83, and grade IIIb in 4. The mortality rate was 0. Mortality and severe morbidities according to the RC-NCD were observed in 52 patients (10.9%). The median mortality rate according to the RC-NCD was 0.2%, the median mortality and severe morbidity rate was 4.6%, the median incidence rate of tracheal or bronchial fistula was 0.2%, and the median incidence rate of respiratory dysfunction was 0.2%.
The comparison between the risk score according to the RC-NCD and postoperative morbidity by lobes is shown in Table 5. Although the risk scores for mortality or severe morbidity according to the RC-NCD were extremely high in RUL and RLL, the mortality and severe morbidity rates were not significantly different among lobes, although the morbidity rate tended to be high in RUL, RLL, and LUL.   The relationships between patient characteristics or operative factors and postoperative morbidity or severe morbidity for patients who underwent lobectomy were analyzed (Table 6). Postoperative morbidities tended to be associated with the coexistence of chronic renal failure (p = 0.04), coexistence of asthma (p < 0.04), smoking status (p < 0.01), CEA (p = 0.04), NLR > 1.56 (p = 0.01), FEV 1 % < 70 (p < 0.01), and RUL, RLL, or LUL (p < 0.01). Severe postoperative morbidity tended to be associated with male sex (p = 0.01), a high BMI (p < 0.01), low %VC (p < 0.02), squamous cell carcinoma (p = 0.04), and RUL, RLL, or LUL (p = 0.03).
COPD chronic obstructive pulmonary disease, BMI body mass index, CEA carcinoembryonic antigen, PNI prognostic nutritional index, NLR neutrophil-to-lymphocyte ratio, VC vital capacity, FEV 1 % forced expiratory volume % in one second, NSCLC non-small cell lung cancer, RUL right upper lobe, RML right middle lobe, RLL right lower lobe, LUL left upper lobe, LLL left lower lobe, PNI prognostic nutritional index, NLR neutrophil-to-lymphocyte ratio, VATS video-assisted thoracic surgery, RATS robotic-assisted thoracic surgery

Discussion
In the present study, we analyzed the risk factors for postoperative morbidity in patients who underwent pulmonary resection for NSCLC. Several factors, such as the age, comorbidity, smoking history, operative approach, and type of operative procedure, have been reported as risk factors for postoperative morbidity in NSCLC patients who have undergone pulmonary resection [5,7,8,13,20,21]. The sex, coexistence of asthma, FEV 1 %, and type of operative procedure were shown to be significant risk factors for postoperative morbidity in a previous study [22]; in the present study as well, the coexistence of asthma, NLR, and pulmonary lobe were significant risk factors for postoperative morbidity, whereas the %VC, pulmonary lobe and type of operative procedure were significant risk factors for severe postoperative morbidity.
RLL has been reported to be a significant risk factor for severe postoperative morbidity in the Japanese nationwide database [12]. Although the risk factors for postoperative morbidity might vary depending on the number of patients or patient characteristics, the type of operative procedure is likely to be a significant risk factor for postoperative morbidity in patients who have undergone pulmonary resection. In the present study, the mortality and severe morbidity rates were not significantly different among lobes, whereas the morbidity rate tended to be high in RUL, RLL, and LUL. Furthermore, the operative  procedure, including RUL, RLL, and LUL, was a significant risk factor for severe postoperative morbidity in the multivariate analysis. Therefore, we considered that the RC-NCD for postoperative severe morbidity should need to be modified according to high-risk lobes.
In the present study, postoperative morbidity for patients who underwent lobectomy was significantly related to the coexistence of asthma and the type of operative procedure, specifically RUL, RLL, or LUL. Although air leakage requiring therapy, atelectasis, asthma attacks, and cerebral infarction are not considered severe postoperative morbidities by the RC-NCD, they are important complications after pulmonary resection. In a previous study, the sex and operative approach, including VATS and RATS, were identified as significant risk factors for postoperative air leakage, while the coexistence of asthma was a significant risk factor for postoperative atelectasis [22]. Cerebral infarction after pulmonary resection is a rare but unfortunate complication, and one risk factor that has been analyzed and reported was the performance of LUL [23]. Cerebral infarction is considered to increase due to blood clots in pulmonary vein stump. Because cerebral infarction is rare but an important postoperative complication, the risk analysis for postoperative cerebral infarction should be done separately in RC-NCD. On the other hand, air leakage was most morbidity in the present study. Furthermore, atelectasis and attack of asthma was rare but the important morbidities. Therefore, air leakage requiring therapy, atelectasis, and asthma attacks should be included as important postoperative morbidities in the RC-NCD. Furthermore, the risk factors related to these important postoperative morbidities after pulmonary resection should be analyzed using a large-scale survey in the future.
Several limitations associated with the present study warrant mention. For example, this study had a retrospective design, patients' performance status wasn't analyzed, and there was a possibility of unobserved confounding and selection bias. In addition, the present study was performed at a single institution.

Conclusions
We analyzed the risk factors for postoperative morbidity in patients who underwent pulmonary resection. The coexistence of asthma, NLR, and pulmonary lobe were significant risk factors for postoperative morbidity in the present study, and the %VC, pulmonary lobe, and type of operative procedure were significant risk factors for severe postoperative morbidity. Furthermore, in patients received lobectomy, coexistence of asthma and pulmonary lobe were identified as significant risk factors for postoperative morbidity, meanwhile, male sex, high BMI, low VC, and pulmonary lobe were identified as significant risk factors for severe postoperative morbidity. Given that the pulmonary lobe was a significant risk factor for postoperative morbidity in patients received pulmonary resection and for severe postoperative morbidity in patients received lobectomy, the RC-NCD for postoperative morbidity needs to be modified according to high-risk lobes.