Abstract
Objectives: Patients undergoing pneumonectomy for lung cancer are thought to be at high risk for the development of postoperative pulmonary complications (PC) and these complications are associated with high mortality rates. The purpose of this study was to identify independent factors associated with increased risk for the development of postoperative PC after pneumonectomy for lung cancer, and to assess the usefulness of predicted pulmonary function to identify high risk patients and other adverse outcomes. Patients and methods: We reviewed retrospectively 242 patients undergoing pneumonectomy for lung cancer during a 12-year period. Perioperative data (clinical, pulmonary function test, and surgical) were recorded to identify risk factors of PC by univariate and multivariate analyses. Results: Overall mortality and morbidity rates were 5.4 and 59%, respectively. Thirty-four patients (14%) developed PC (acute respiratory failure, ARF=8.7%, reintubation=5.4%, pneumonia=3.3%, atelectasis=2.9%, postpneumonectomy pulmonary edema=2.5%, mechanical ventilation more than 24 h=1.2%, pneumothorax=0.8%). Patients with surgical (P<0.001), cardiac (P<0.001) and other complications (P<0.01) had higher incidence of PC than those without postoperative complications. Intensive care unit stay (53±39 h vs. 35±19 h; P<0.001) and hospital stay (18±11 days vs. 12±7 days; P<0.001) was significantly longer in patients with PC. The mortality rate associated with PC was 35.5% (P<0.001). By univariate analysis, it was found that older patients (P=0.007), chronic obstructive pulmonary disease (COPD) (P=0.023), heart disease (P=0.019), no previous record of chest physiotherapy (P=0.008), poor predicted postoperative forced expiratory volume in 1 s (ppo-FEV1) (P=0.001), and prolonged anesthetic time (P<0.001) were related with higher risk of PC. In the multiple logistic regression model, the anesthetic time (minutes; odds ratio, OR=1.012), ppo-FEV1 (ml/s; OR=0.998), heart disease (OR=2.703), no previous record of previous chest physiotherapy (OR=2.639), and COPD (OR=2.277) were independent risk factors of PC. Conclusions: PC after pneumonectomy are associated with high mortality rates. Careful attention must be paid to patients with COPD and heart disease. Our results confirm the relevance of previous chest physiotherapy and the importance of the length of the surgical procedure to minimize the incidence of PC. The predicted pulmonary function (ppo-FEV1) may be useful to identify high risk patients for PC development and adverse outcomes.
1 Introduction
Pulmonary resection remains the only potentially curative option for patients with non-small cell carcinoma presenting as localized disease. Pneumonectomy is associated with higher mortality and morbidity rates than other lesser resections [1]. Patients undergoing pneumonectomy for lung cancer are thought to be at high risk for the development of postoperative pulmonary complications (PC) and these complications are associated with high mortality rates. Although operative mortality following lobectomy or pneumonectomy has decreased over the past decades, PC continue to occur with incidence rates as high as 49% [2].
Preoperative evaluation of the risk of PC has been the subject of several reviews. In most of these studies, older age [3–6], cardiopulmonary comorbidity [2,4,6,7], smoking status[2,5–8], altered pulmonary function test [2,8,9], and the extent of surgical resections [7,10] have been related with PC after thoracic surgery. However, it is difficult to determine which of these factors independently affect the outcome because the sample size of most of these studies was too small to apply a multivariate analysis. Previous clinical studies have included in their analysis all types of pulmonary resections (wedge resections, lobectomies, and pneumonectomies). Therefore, specific independent risk factors for PC after pneumonectomy has not been extensively assessed. The importance of pulmonary function as a predictive value is controversial. Its usefulness in predicting PC after pneumonectomy has not been assessed by multivariate analysis in previous reports.
To address these issues, the purpose of this study was to identify independent factors associated with increased risk for the development of postoperative PC in a population of 242 patients undergoing pneumonectomy for lung cancer, and to assess the usefulness of predicted pulmonary function in predicting this complication and other adverse outcomes.
2 Material and methods
2.1 Patients and preoperative assessment
From January 1986 to December 1997, 267 patients underwent a pneumonectomy at the Reina Sofía University Hospital in Córdoba: 242 for lung cancer, nine for bronchiectasis, seven for tuberculosis, five for lung metastases, and four for other diseases. For the purpose of this study, we reviewed the charts of those patients with lung cancer. They were 231 men (95.5%) and 11 women (4.5%) with mean age of 60±10 years (range 26–79 years).
All patients were assessed by the same surgical team and the preoperative study was standardized. Imaging studies included plain chest roentgenograms and computed tomographic scan of the chest and upper abdomen. The extrathoracic staging of the tumor was assessed as in previous reports of our group [11]. All patients underwent a fiber optic bronchoscopy to assess the airway, tumor location, and resectability. The preoperative pulmonary function was assessed with arterial blood gas levels and pulmonary function tests. Predicted postoperative forced expiratory volume in 1 s (ppo-FEV1) was calculated based on the preoperative forced expiratory volume in 1 s (FEV1) and the pulmonary perfusion [12]. For those patients with FEV1 greater than 2 l/s, left pneumonectomy ppo-FEV1=0.55×FEV1 and right pneumonectomy ppo-FEV1=0.45×FEV1 were performed. For those patients with FEV1 less than 2 l/s, a radionuclide perfusion was obtained: ppo-FEV1=(1% of perfusion in lung to be resected)×FEV1.
Antibiotic (cefuroxime, 20 mg/kg i.m.) and thromboembolic (enoxaparin, 20–40 mg sbc, subcutaneous) prophylaxis were used routinely and 207 patients (85.5%) started an active program of chest physiotherapy including deep-breathing exercises and incentive spirometry during a preoperative period of 7 days. Thirty-four individuals (14%) underwent mediastinal exploration (mediastinoscopy or mediastinotomy) before lung resection because of nodal disease. Neoadjuvant therapy (chemotherapy or radiation therapy) was given in 32 cases (13%) for biopsy-proved N2 disease or T3-4.
2.2 Surgical treatment
All patients were operated on by the same surgical team through a standard posterolateral thoracotomy and air exclusion of the lung to be resected. The bronchial closure was performed with staplers (Ethicon, EndoSurgery, Inc. Cincinnati, OH; and Autosuture USSC, Norwalk, CT) in 233 patients (96%), and manually with interrupted 4-0 sutures of polypropylene (Prolene, Ethicon, Sommerville, NJ) in nine cases (4%) because of the tumor proximity to the tracheal carina. Bronchoplastic techniques were not performed. The bronchial stump was covered with autologous tissue in 178 cases (74%, 88% of right pneumonectomies and 64% of left pneumonectomies). For this purpose, several tissues were used: intercostal muscle (n=80), mediastinal fat pad (n=64), parietal pleura (n=15), phrenic pedicle (n=10), acygos vein (n=6), and other tissues (n=3). This coverage was performed more frequently in those cases more likely to develop a bronchopleural fistula (BPF) (right resections, older patients, neoadjuvant therapy); however, the final decision of whether to cover the bronchus was made by the operating surgeon.
In all cases, a systematic lymph node resection was performed. Extended resections were needed in 46 tumors involving the chest wall (4.5%), parietal pleura (2.5%), pericardium (4.5%), diaphragm (0.4%), and other mediastinal structures (7%). An intrapericardial pneumonectomy was needed in 80 cases (33%), and three patients (1.2%) underwent a completion pneumonectomy. The histological distribution of the tumors was as follows: epidermoid (n=165; 49.6%), large cell (n=31; 12.8%), adenocarcinoma (n=27; 11.2%), small cell (n=8; 3.3%), carcinoid (n=6; 2.5%), and other (n=5; 2%).
2.3 Staging
Patients were staged postoperatively according to the TNM staging system. Postoperative staging was as follows: stage Ia (n=8; 3.3%), stage Ib (n=82; 33.9%), stage IIa (n=1; 0.4%), stage IIb (n=64; 26,4%), stage IIIa (n=70; 29%), stage IIIb (n=11; 4.5%), and stage IV (n=0). In six patients receiving neoadjuvant therapy, a surgical staging was not possible because no viable tumor cells were found after pulmonary resection.
2.4 Postoperative management
Postoperatively, all patients remained at least 1 day in the intensive care unit (ICU). An early extubation was achieved in 96 patients (39.7%), but 146 cases (60.3%) were under mechanical ventilation for several hours. Postoperative chest pain was treated with thoracic epidural analgesia in all cases. When discharged from the ICU, patients were managed in the thoracic ward. Data about the patient's course, including temperature, physical examinations, and roentgenograms were recorded. All cases received postoperative training in chest physiotherapy. Bronchoscopies were done if a BPF was suspected or mucus retention with atelectasis of the remaining lung was noted.
2.5 Pulmonary complications
Postoperative complications were defined as those occurring within 30 days after thoracotomy or before hospital discharge. For the purpose of this study, we have considered seven pulmonary complications: (1) nosocomial pneumonia: the diagnosis of pneumonia was considered when patients developed a lung infiltrate and purulent sputum with documented presence of microorganisms in the sputum culture; (2) atelectasis: when evidenced on chest radiographs and requiring aspirative bronchoscopy; (3) respiratory failure: acute onset of hypoxemia (PaO2 less than 60 mmHg) and/or hypercapnia (PaCO2 more than 45 mmHg); (4) postpneumonectomy pulmonary edema (PPE): pulmonary edema with no clinical evidence of heart failure; (5) postoperative ventilator dependence more than 24 h; (6) reintubation for controlled ventilation; and (7) pneumothorax in the contralateral lung.
2.6 Data collection and statistical analysis
Preoperative, intraoperative, and postoperative variables were recorded retrospectively including general demographic data, patient comorbidity, pulmonary function, neoadjuvant therapy, previous chest physiotherapy, type of surgical procedure, postoperative complications, and mortality within 30 days after pneumonectomy. The relationship between potential predictors and postoperative pulmonary complications was assessed by univariate and multivariate analyses. In the univariate analysis, categorical variables were analyzed using a Pearson's χ2 test or Fisher's exact test, and continous variables were compared by using the student's unpaired t-test. Data are presented as mean±standard deviation for continuous variables and as percentage for categorical variables. Differences were considered significant with P values less than 0.05.
The risk of PC was evaluated by using a forward and backward stepwise logistic regression analysis to estimate odds ratios (OR) and their 70% confidence intervals (CI). Those variables with P values less than 0.05 in the univariate analysis were included in the multivariate analysis. The final model included factors that remained significant with a P value less than 0.10. Goodness-of-fit was assessed by the Hosmer and Lemeshow χ2 test. The statistical analysis was performed with SPSS 7.5 for Windows software system (SPSS Inc., Chicago, IL).
3 Results
The overall mortality of the study group was 5.4% with a morbidity rate of 59%. Postoperative complications are shown in Table 1 . Cardiac complications (CC) were predominant (38.4%) with more frequent occurrence of arrhythmias. Forty-nine patients (20.2%) developed surgical complications (SC). Thirteen patients (5.4%) developed an early BPF within 30 days after pneumonectomy (mean 11±8 days). Seven BPF (2.9%) appeared within 10 postoperative days and 6 (2.5%) appeared after 10 days of pneumonectomy. Only two patients presented BPF without postpneumonectomy empyema; the other 11 presented an associated pleural empyema.
Postoperative complications after pneumonectomy
Thirty-four patients (14%) presented pulmonary complications. ARF was the most prevalent complication (8.7%), followed by reintubation (5.4%). Nosocomial pneumonia occurred in eight cases (3.3%). Seven patients (2.9%) needed a bronchoscopy for persistent atelectasis secondary to retained secretions. Six patients (2.5%) developed an acute onset with a PaO2/fraction of inspired oxygen ≤200 mmHg and pulmonary edema, with no clinical evidence of left atrial hypertension (PPE). Postoperative mechanical ventilation was required in 146 patients (60.3%). The mean duration of mechanical ventilation was 382±304 min (60–1800). Only three patients (1.2%) needed a prolonged mechanical ventilation (more than 24 h). In two cases (0.8%), a pneumothorax was observed and required chest tube placement.
The mortality rate in patients with PC was significantly higher than in those without PC (35.5 vs. 0.5%; P<0.001). Twelve patients died: eight cases directly related with PC, three cases secondary to sepsis and one due to cardiac failure. Patients with surgical (28.6 vs. 10.4%; P<0.001), cardiac (24.7 vs. 7.4%; P<0.001), and other complications (26.1 vs. 11.2%; P<0.01) had higher incidence of PC than those without postoperative complications. Cardiac morbidity was more frequent in patients with PC (67.6 vs. 33.7%; P<0.001). ICU stay (53±39 h vs. 35±19 h; P<0.001) and hospital stay (18±11 days vs. 12±7 days; P<0.001) were significantly longer in patients with PC.
3.1 Univariate analysis
Clinical and surgical variables related to PC are depicted in Table 2 . Preoperative clinical factors associated with higher risk for PC were: older patients (P=0.007), COPD (P=0.023), heart disease (P=0.019), and no previous record of chest physiotherapy (P=0.008). Surgical factor related with PC was a prolonged anesthetic time (P<0.001). Patients who underwent extended pneumonectomies did not have an increased risk for PC, however, when a chest wall resection was analyzed individually, significant differences were observed (36.4 vs. 13%; P<0.05). Pulmonary complications were most frequent after right resections and without bronchial stumps covered, however, no significant differences were obtained.
Risk factors of pulmonary complications on univariate analysis: clinical and surgical variables
The results of the preoperative pulmonary function tests in patients with and without PC are shown in Table 3 . After univariate analysis, only PaO2 (P=0.003), FEV1 (P=0.05), and ppo-FEV1 (P=0.001) were predictors of PC. No significant differences between both groups were observed in preoperative PaCO2, FVC, FVC(%), FEV1(%), and FEV1/FVC. Patients with a ppo-FEV1 equal or less than 1 l/s had an increased mortality and morbidity rates (overall, pulmonary, cardiac, and surgical). Postoperative complications and mortality rates according to ppo-FEV1 are shown in Fig. 1 .
Risk factors of pulmonary complications on univariate analysis: preoperative pulmonary function
Postoperative complications and mortality rate according to predicted postoperative FEV1 value (PC, pulmonary complications; CC, cardiac complications; SC, surgical complications).
3.2 Multivariate analysis
Those variables exhibiting a prevalence higher than 0.5% and P values less than 0.05 in the univariate analysis were considered for entry in a stepwise logistic regression analysis to obtain a prognostic model that included those variables with P values less than 0.10. Because FEV1 and ppo-FEV1 were collinear, two models were constructed to see whether FEV1 or ppo-FEV1 further contributed to the model. Five variables remained in the final model as independent predictors of PC (Table 4 ): anesthetic time (min; OR=1.012, 70% CI, 1.007–1.017), ppo-FEV1 (ml/s; OR=0.998, 70% CI, 0.999–0.997), heart disease (OR=2.703; 70% CI, 1.684–4.336), no previous record of chest physiotherapy (OR=2.639; 70% CI, 1.598–4.358), and COPD (OR=2.277; 70% CI, 1.455–3.562) . The FEV1 was not found to be an independent predictor of PC.
Risk factors of pulmonary complications on multivariate analysis
4 Discussion
The present study examined the postoperative course of a large group of patients undergoing pneumonectomy for lung cancer and evaluated the importance of several preoperative and surgical variables as predictors of postoperative PC. The results of this study demonstrated that independent predictors of PC after pneumonectomy were a low ppo-FEV1, a prolonged anesthetic time, no previous occurrence of chest physiotherapy, heart disease, COPD status, and no covered bronchial stump. Also, the development of PC were related with high mortality rate (35.5%), other postoperative morbidity, and a prolonged ICU and hospital stay.
The incidence of PC after pneumonectomy ranges from 11 to 49% [2,5,6,8,13–18]. This variability may be due to the different criteria for considering a PC, the type of complications evaluated, and the statistical analyses employed. In our experience, the postoperative PC rate (14%) was somewhat lower than that reported in previous studies [2,5,6,8,13–18] (Table 5 ). Probably, the management of these patients in a specialized intermediate care unit, with emphasis on preventive respiratory modalities (chest physiotherapy, incentive spirometry, adequate postoperative analgesia, and early ambulation) may also have led to a decreased incidence of PC. Acute respiratory failure was the most frequent PC observed (8.7%). Some authors have reported rates of respiratory failure from 3.3 to 17.4% [2,15–17]. These patients presented a mortality rate as high as 50% with a rate of need of reintubation and mechanical ventilation of 45.5%. The higher mortality rate associated is probably more closely related to the severity of the underlying pulmonary complication leading to mechanical ventilation, rather than to mechanical ventilation itself. However, prolonged mechanical ventilation has also been associated with pneumonia and BPF. In our study, the incidence of nosocomial pneumonia (3.3%) was lower than the previously reported rates of 4.8–21.8% [2,5,13,15–17]. Postpneumonectomy pulmonary edema (PPE) is an uncommon complication of pneumonectomy but one that carries a high mortality rates. PPE is significantly more common after right-sided pneumonectomies and most of the studies report a frequency of 1–5% [6]. In our experience, six patients (four right-sided pneumonectomies) developed PPE, with a mortality rate of 50%.
Literature review of postoperative PC in series of pneumonectomies
The association of PC with the increased incidence of other morbidity factors has been reported, particularly in relation to CC [7]. We have found a significantly higher incidence of CC in the same patients experiencing PC. As previously reported [2,3,6,7,9], PC also are associated with high mortality rates. In our experience, 12 patients died (35.5%), eight cases directly related with pulmonary complications, three cases secondary to sepsis and one due to cardiac failure. Also, as reported by other investigators, we have found that PC led to prolonged hospital and ICU stays [5,6,19].
Since PC are associated to high mortality rates and prolonged ICU stay, it is of paramount importance to determine groups of patients at high risks of PC. Previous clinical studies [3–7,10] have included in their analysis all types of pulmonary resections (wedge resections, lobectomies, and pneumonectomies) to predict which patients were at high risk, so that specific independent risk factors for PC after pneumonectomy has not been extensively assessed. Also, few investigations have assessed independent risk factors [6,7,10] because the small sample size precluded the application of multivariate analyses. In our study, several preoperative (heart disease, COPD, low ppo-FEV1, no previous chest physiotherapy) and intraoperative variables (not covered bronchial stump, prolonged anesthetic time) were found to be independent predictors of PC after pneumonectomy.
Although previous reports have included blood gas levels (PaO2 and PaCO2) as part of the preoperative evaluation, there are no objective data demonstrating a significant association between the alteration of these parameters and the perioperative morbidity and mortality [1]. Historically, individuals with preoperative hypercapnia have been defined as high risk patients to develop PC [20]. This statement is based on limited experiences in small patient series. In our study, as in previous reports [10], the preoperative hypercapnia has not been associated to higher incidence of PC after pneumonectomy. Conversely, patients with hypoxemia have had more incidence of postoperative morbidity and mortality.
Respiratory function parameters have been extensively studied in the preoperative evaluation of patients undergoing pulmonary resections. Most of the series were focused on the FEV1 as the single factor predictive of morbidity and mortality after lung resection. Several procedures have been described to minimize PC after lung resection [12,21] based on the FEV1 value; those patients with FEV1 above 2 l/s may undergo the planned resection without significant morbimortality. However, in those patients with FEV1 less than 2 l/s, an additional ventilation/perfusion scan was proposed to predict the ppo-FEV1. Olsen et al. [12] demonstrated that a ppo-FEV1 above 0.80 l/s was the cut point value to allow a minimal tissue oxygenation and cardiac output after the lung resection. Later on, these values have been widely accepted as a functional limit to perform a pneumonectomy [22]. Nevertheless, other authors have proposed the use of ppo-FEV1, expressed as percentage of predicted, as a more accurate value to predict the onset of PC after the lung resection, establishing a limit in 30% of predicted [22,23] or 40% of predicted [13,21].
The preoperative assessment of pulmonary function parameters have been the most extensively studied factors to predict morbidity and mortality after pulmonary resection and some controversy still remains. Some authors have identified FEV1 and/or FVC values as possible risk factors [2,8,13], while others did not [5,7,10,19]. Several previous studies have confirmed the importance of ppo-FEV1 in predicting PC and mortality after lung resection [9,10,13,21], even as an independent factor by multivariate analysis [10]. In our series, the functional parameter more closely related with the development of PC was the ppo-FEV1. Thus, those patients with a ppo-FEV1 value equal or less than 1 l/s not only exhibited more PC, but also more cardiac and SC, and a higher rate of postoperative mortality. Furthermore, this parameter was found to be an independent predictor of PC in the multivariate analysis (Table 4). On the contrary, no significant differences were found in FVC, FVC(%), FEV1, FEV1(%), and FEV1/FVC between both complicated and non-complicated patients. As opposed to ppo-FEV1, the preoperative FEV1 did not predict a higher risk of PC. In our experience, the assessment of ppo-FEV1 has been of great value to discriminate a subgroup of patients at higher risk for PC development and other adverse outcomes.
Previous reports have evaluated the influence of associated comorbidity in the development of postoperative complications after lung resection [1–3,9,14]. The underlying pulmonary disease, particularly COPD, has been advocated as a major risk factor for postoperative complications [1]. In addition, other non-pulmonary comorbidities such as coronary disease, hypertension, and diabetes mellitus, have been related with higher risks of mortality after lung resection [2], although other reports did not find such a relationship [9,13]. In our experience, PC were more closely related to coexisting medical conditions (cardiac disease and COPD) than other presumed risks such as age of patients. In our study, advanced age itself did not appear to be an independent predictive risk factor for PC [2,6–8,13].
Chest physiotherapy is highly recommended to prevent PC after lung resection. The relative low PC rate in our series might be due to the careful preoperative evaluation, routine intensive preoperative and postoperative physiotherapy, optimal pain control, and early mobilization. This was previously suggested by Nagasaki et al. [24], who considered the careful preoperative care, selection of appropriate surgical procedures, and preoperative chest physiotherapy as main factors to prevent postoperative complications.
Several operative factors have proven to be risk factors for morbidity and mortality after lung resection. Pneumonectomy is associated with higher mortality rates than other lung resections [1]. Bush et al. [7], in a series of 106 pulmonary resections, reported a higher rate of PC in patients undergoing extended pneumonectomy than those after lesser resections, but they did not specify the number of standard pneumonecomties performed in their series. Other authors [1,13] have described higher mortality and morbidity rates after extended and extrapleural pneumonectomies. We did not observe more PC after extended and intrapericardial pneumonectomies, however, those pneumonectomies with chest wall resection presented higher risk for PC. As has been reported previously [6], we have observed a higher rate of PC in those cases of prolonged anesthetic time. Probably, the prolonged intraoperative mechanical ventilation and the higher rate of respiratory infections may play a role in such a relationship. Finally, the bronchial stump coverage with autologous tissue might prevent the development of PC by reducing the incidence of BPF. The BPF is a devastating complication after pneumonectomy. Although its incidence has decreased over the last decades, this life-threatening complication remains a major challenge for the thoracic surgeon because it is associated with high mortality rates [2,18]. Thirteen patients (5.4%) developed a early BPF compared with the 2.9–6.9% rates reported by other authors [2,5,6,13,16–18]. The benefits of this coverage are controversial. Most authors support the bronchial stump coverage for right pneumonectomies because the left bronchial stump remains more protected within the mediastinal structures. However, other investigations advocate the need of similar coverage for left pneumonectomies as well, mainly in cases of patients at potential risk [16,25]. In our series, most of the patients underwent bronchial stump coverage with an incidence of BPF of 3.9%, as opposed to 9.4% in those patients without coverage. Despite the higher incidence of PC in the pneumonectomies performed without bronchial coverage, the influence of the bronchial stump coverage in preventing PC resulted non-significant in the univariate analysis. We think that this finding probably may be related to the relationship of the side of resection. The coverage of the bronchial stump was performed more frequently in those cases more likely to develop PC (right pneumonectomies). In the univariate analysis, the influence of the bronchial coverage was biased by the side of resection and therefore, the results were not significant.
In summary, this retrospective study demonstrated a close association between PC after pneumonectomy and several preoperative and intraoperative variables. These complications are the major cause of mortality after pneumonectomy for lung cancer and are associated with longer ICU and hospital stays. Based on previous reports and the present series, careful attention must be paid to those patients with significant risk factors such as COPD and heart disease. A prolonged anesthetic time and not covered bronchial stumps increased the risk of PC development. In addition, previous chest physiotherapy is highly recommended to prevent the onset of PC. Finally, those patients with low ppo-FEV1 are at increased risk for PC after pneumonectomy.
