Construction and validation of a predictive model for postoperative respiratory failure in esophageal cancer patients

Introduction

Esophageal cancer (EC) is the 11^th most-common cancer and the 7^th leading cause of cancer death worldwide (1). In China, EC ranks 7^th in overall incidence and 5^th in mortality among all cancers. Approximately 70% of EC patients are male, and >90% of cases are esophageal squamous carcinoma (2). Currently, there are various treatment modalities for EC, including endoscopy, radiotherapy, chemotherapy, immunotherapy and targeted therapy. However, surgery-based multimodality treatment is still the preferred treatment for EC patients (3). EC surgery involves reconstruction of the digestive tract and is associated with significant trauma and high risk. According to the Esophageal Complications Consensus Group (ECCG), the overall postoperative complication rate of EC is as high as 59%, and the postoperative mortality rates at 30 and 90 days are 2.4% and 4.5%, respectively (4).

Postoperative respiratory failure (PRF) is a serious complication after EC surgery, with an incidence ranging from 2.4% to 15.3% (5-8). The occurrence of PRF not only increases mortality but also weakens the effectiveness of antitumor therapy. Therefore, RF is a major risk factor for poor clinical outcomes in patients (8). Early identification of high-risk patients who may develop PRF and adoption of personalized therapeutic strategies are essential to improve patient outcomes. In recent years, nomogram models have been widely used in the field of medical prediction due to their ability to integrate multiple predictors into a personalized risk scoring system, providing clinicians with an intuitive and accurate prediction tool (9). Most previous studies have focused on identifying risk factors, and there is a lack of well-established nomogram models for predicting the risk of PRF in EC (6-8). In this study, we analyzed the risk factors for PRF in EC patients based on their perioperative clinical data and constructed a nomogram-based model to inform clinical decision making for the early prevention and treatment of PRF. We present this article in accordance with the TRIPOD reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2024-2114/rc).

Methods

General data

The clinical data of 265 EC patients who underwent surgery at the Department of Thoracic Surgery of Gansu Hospital of Sun Yat-sen University Cancer Center between January 2020 and June 2024 were reviewed. The patients were randomly divided 7:3 into a training set (n=185) and an internal validation cohort (n=80). Another 80 EC patients who underwent surgery at the Sun Yat-sen University Cancer Center between January 2024 and June 2024 were included as an external validation set. Inclusion criteria: (I) EC diagnosed by histopathology before surgery; (II) underwent radical surgery for EC. Exclusion criteria: (I) history of other malignancies; (II) underwent palliative surgery; (III) incomplete clinical data; and (IV) died due to causes other than RF during the perioperative period.

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Ethics Committee of Gansu Hospital of Sun Yat-sen University Cancer Center (No. P-LW202407160006) and informed consent was taken from all the patients.

Observations

The clinical data of patients were collected by reviewing the relevant literature, including: (I) general patient information: age, sex, smoking history, alcohol consumption history, neoadjuvant therapy history, comorbidities, forced expiratory volume in one second (FEV1), FEV1 as a percentage of the predicted forced vital capacity (FEV1/FVC%pred), maximal voluntary ventilation (MVV), and the predicted DLCO percentage (DLCO%pred); (II) preoperative inflammatory markers: neutrophil to lymphocyte ratio (NLR), lymphocyte to monocyte ratio (LMR), platelet to lymphocyte ratio (PLR), and systemic immune-inflammation index (SII); (III) preoperative nutritional markers: body mass index (BMI), hemoglobin, serum albumin, and prognostic nutritional index (PNI); (IV) surgical data: tumor location, surgical approach, surgical time, number of lymph nodes removed, pathological stage of tumor, and postoperative complications (e.g., anastomotic fistula and hypoproteinemia).

Diagnostic criteria

PRF (10) was defined as type I RF [partial pressure of arterial oxygen (PaO₂) <60 mmHg and decreased or normal partial pressure of carbon dioxide (PaCO₂)] and type II RF (PaO₂ <60 mmHg and PaCO₂ >50 mmHg) according to arterial gas analysis within 2 weeks of surgery. Postoperative hypoproteinemia was defined as serum albumin <30 g/L. Preoperative inflammatory and nutritional marker levels were collected from the test results of patients within 1 week before surgery, and raw values were used for the analysis of all serum markers in this study due to the lack of a uniform classification criterion.

Modeling and validation

Patients in the training set were subgrouped according to whether they developed PRF, and data imbalances were addressed using the Synthetic Minority Over-sampling Technique (SMOTE) method. Independent variables with significant differences (P<0.05) in the univariate analysis were incorporated into LASSO regression. Variables identified by LASSO regression were subsequently analyzed by multivariate logistic regression to identify statistically significant variables and to construct a predictive model. Using the Spearman correlation coefficient to quantify the associations between variables in multivariable analysis, a heatmap of correlation analysis was generated to facilitate the visualization of data relationships. The discriminatory power of the model was assessed by plotting the receiver operating characteristic (ROC) curve and calculating the area under the curve (AUC). Calibration of the model was assessed by plotting a calibration curve using the Hosmer-Lemeshow test. Decision curve analysis (DCA) was performed to determine the clinical utility of the model.

Statistical analysis

All statistical analyses were performed using R Studio 4.3.2. Count data were expressed as number of cases and percentage (%) and compared using Pearson’s χ² test or Fisher’s exact test. Continuous data with normal distribution were expressed as x¯±s and compared using analysis of variance (ANOVA) or the independent sample t-test. Continuous data without normal distribution were expressed as median (interquartile range) [M (IQR)] and compared using the nonparametric rank-sum test. P<0.05 was considered statistically significant.

Results

General patient data

A total of 345 eligible EC patients were included. There were 185 cases in the training set with a PRF incidence of 11.9%, 80 cases in the internal validation set with a PRF incidence of 12.5%, and 80 cases in the external validation set with a PRF incidence of 11.3%. Surgical time, FEV1/FVC%pred, and PNI, but not other variables, were significantly different among the three sets (P<0.05) (Table 1).

Screening of predictive model variables

Patients in the training set were subgrouped according to whether they developed PRF and data imbalances were addressed using the SMOTE method (Figure 1). Univariate analysis of 28 identified clinical features showed that age ≥70 years, smoking index ≥400, preoperative concomitant emphysema, FEV1, FEV1/FVC%pred, DLCO%pred, NLR, preoperative serum albumin levels, PNI, surgical time, postoperative anastomotic fistula, and postoperative hypoproteinemia were significantly different among the three groups (P<0.05) (Table 2).

Figure 1 Processing of imbalanced data in the training set using the SMOTE method. NRF, non-respiratory failure; RF, respiratory failure; SMOTE, Synthetic Minority Over-sampling Technique.

To minimize multicollinearity among variables, statistically significant variables in the univariate analysis were incorporated into the LASSO regression. The optimal λ-value (lambda.1se) was selected through cross-validation to obtain the most streamlined predictive model while maintaining goodness-of-fit (Figure 2).

Figure 2 Feature selection using LASSO regression. (A) Path plot of LASSO regression coefficients; (B) LASSO regression cross-validation curves using lambda.lse as the final selection criterion. LASSO, least absolute shrinkage and selection operator.

Seven variables were identified based on the non-zero coefficients calculated by LASSO regression analysis, including age ≥70 years, smoking index ≥400, FEV1, NLR, preoperative serum albumin, surgical time, and postoperative anastomotic fistula. Multivariate logistic regression analysis of these variables revealed that smoking index ≥400 [odds ratio (OR) =3.34, 95% confidence interval (CI): 1.27–8.78, P=0.01], FEV1 (OR =0.16, 95% CI: 0.05–0.49, P=0.001), preoperative serum albumin (OR =0.71, 95% CI: 0.59–0.86, P=0.001), surgical time (OR =1.01, 95% CI: 1.00–1.02, P=0.005), and postoperative anastomotic fistula (OR =4.36, 95% CI: 1.22–5.60, P=0.02) were risk factors for PRF in EC (Table 2). The correlation between variables in the multivariable analysis is shown in Figure 3.

Figure 3 Inter-variable correlation in multivariable analysis. FEV1, forced expiratory volume in one second.

Construction of predictive model

According to the results of the LASSO-logistic regression analysis, a nomogram for predicting PRF in EC was constructed based on the risk factors smoking index ≥400, FEV1, preoperative serum albumin, surgical time, and postoperative anastomotic fistula. The individual score for each independent risk factor was obtained by projecting upward to the nomogram score line, and the total score was then calculated to determine the probability of PRF (Figure 4).

Figure 4 Nomogram for predicting PRF in EC patients. EC, esophageal cancer; FEV1, forced expiratory volume in one second; PRF, postoperative respiratory failure.

Validation and evaluation of the predictive model

The discriminatory power of the model was assessed using AUC values. The AUC values for the training set (0.856, 95% CI: 0.794–0.918), internal validation set (0.839, 95% CI: 0.725–0.952), and external validation set (0.773, 95% CI: 0.573–0.973) indicated that the model had good discriminatory power and predictive performance. Calibration curves demonstrated good agreement, and the Hosmer-Lemeshow test indicated that the model had good goodness-of-fit. DCA of the predictive model revealed favorable net clinical utility at a 45–95% threshold probability for the training set, 0–60% for the internal validation set, and 0–50% for the external validation set (Figure 5).

Figure 5 Validation and evaluation of the predictive model. AUC, area under the curve; CI, confidence interval.

Discussion

Radical surgery for EC is a highly invasive procedure that triggers systemic inflammation, leading to excessive production of inflammatory cytokines, endothelial dysfunction, endothelial cell injury, activation of neutrophils, and tissue and organ damage and organ failure, especially RF (5,11). Despite advances in surgical techniques and perioperative care, PRF remains a serious complication that not only prolongs hospitalization and increases mortality, but also seriously affects surgical outcomes and long-term prognosis (10). Therefore, how to effectively predict, prevent, and manage PRF is an urgent question that needs to be resolved in clinical practice.

The occurrence of PRF is closely associated with various factors, including the preoperative general condition of patients, intraoperative operations, and postoperative management. A comprehensive understanding and assessment of these risk factors is essential for developing optimal postoperative treatment plans and effectively preventing and reducing the incidence of complications. Yu et al. (8) reported that PRF occurred in 57 of 2,386 (2.4%) enrolled patients. Analysis of 19 clinical features showed that age, body mass index, cardiovascular disease, diabetes mellitus, DLCO%pred, tumor site, and thoracic surgical time ≥101.5 min were risk factors for PRF, and the AUC of the constructed predictive model was 0.755. In a study by Wang et al. (12), 34 of 216 (13%) patients developed PRF. Analysis of 14 clinical features demonstrated that age, intraoperative fluid rehydration, postoperative hypoproteinemia, anastomotic fistulae, and extensive adhesions in the thoracic cavity were independent risk factors for PRF in EC patients, and the AUC of the predictive model was 0.848. Although these models exhibited favorable discriminatory power, they were all based on single-center data and lacked external validation. In addition, the variables included were not comprehensive, and the performance of multivariate analysis after univariate analysis posed some limitations in dealing with multicollinearity among variables.

In this study, we collected not only general information and surgery-related details of patients, but also their preoperative inflammatory and nutritional marker status. A predictive model was constructed by LASSO-logistic regression based on 28 clinical features and internally and externally validated. LASSO regression introduces the concept of regularization, which allows for simultaneous processing of all independent variables, reduces complexity of the model, prevents overfitting, and in turn selects the most relevant features (13). The results of this study showed that smoking index ≥400, reduced FEV1, reduced preoperative serum albumin, prolonged surgical time, and postoperative anastomotic fistula were significant risk factors for PRF in EC patients. The AUCs of the constructed model were 0.856 (95% CI: 0.794–0.918) for the training set, 0.839 (95% CI: 0.725–0.952) for the internal validation set, and 0.773 (95% CI: 0.573–0.973) for the external validation set, indicating that the model had good discriminatory power and predictive accuracy comparable to previous studies.

Smoking has been widely recognized as a major risk factor for pulmonary complications after thoracic surgery. It is often accompanied by respiratory disorders such as emphysema and bronchitis (14). Turan et al. (15) demonstrated that smoking increases 30-day mortality by 40% in surgical patients and increases the probability of serious complications, including RF, by 30% to 100%. Smoking leads to chronic inflammation and fibrosis of the airways and reduces lung elasticity, thus increasing postoperative respiratory burden. Additionally, smoking stimulates respiratory sensory nerve endings and increases the number of goblet cells, leading to increased mucus secretion by the glands. Moreover, it impairs the motor function of cilia in the airway mucosa, reducing secretion clearance and thereby increasing the risk of lung infection (14-17). The current Chinese guidelines for perioperative airway management in thoracic surgery recommend smoking cessation for at least 4 weeks before surgery (18).

With the aging of the population and the increasing prevalence of chronic obstructive pulmonary disease (COPD), pulmonary function assessment has become an important routine examination before thoracic surgery. It not only effectively predicts the risk of perioperative complications, but also provides an objective basis for surgical decision-making (18). Consistent with previous findings, our results confirmed that decreased preoperative FEV1 is a major risk factor for PRF in EC patients. The mechanisms by which decreased preoperative FEV1 increases the risk of PRF are complex and diverse (19-22): (I) decreased FEV1 implies airway obstruction or decreased lung elasticity, which limits postoperative lung expansion and sputum expulsion, increasing the risk of lung infections and pulmonary atelectasis; (II) lower FEV1 suggests that the patient has limited respiratory reserve function to cope with postoperative inflammation; (III) patients with lower FEV1 are more susceptible to postoperative hypoxia and carbon dioxide retention, leading to the development of RF. Therefore, it is necessary to improve patients’ lung function through preoperative respiratory exercises (including exercise interventions and smoking cessation).

Serum albumin is not only an important nutrient and energy reserve of the body, but also an important component of the immune system. Reduced preoperative serum albumin levels may affect the development of PRF by (23-25): (I) impairing immune responses, resulting in susceptibility to postoperative infections, especially lung infections; (II) triggering fluid imbalance, leading to edema of the lung tissues and decreased lung function, exacerbating postoperative respiratory distress; and (III) affecting inflammatory responses, leading to increased postoperative release of inflammatory mediators and lung injury. A study involving 2,059 postoperative patients by Jeong et al. (26) showed that the risk of postoperative pulmonary complications was significantly higher in patients with serum albumin <40 g/L (OR =1.49, 95% CI: 1.00–2.21). Similarly, a prospective cohort study by Arozullah et al. (27) demonstrated that PRF occurred in 2,746 (3.4%) of 81,719 patients from 44 medical centers, and patients with serum albumin <30 g/L also had a higher risk of PRF (OR =2.5, 95% CI: 2.3–2.8). Therefore, preoperative assessment of serum albumin levels has significant clinical implications. For patients with low serum albumin levels, preoperative nutritional support should be strengthened to improve nutritional status and immune function.

The duration of surgery is also a key factor affecting the incidence of postoperative complications. It not only directly reflects the complexity of the procedure, but is also closely related to the length of anesthesia, intraoperative stress, and postoperative recovery. Prolonged mechanical ventilation may lead to continuous alveolar expansion and hyperventilation, increasing the risk of lung injury. In addition, prolonged duration of surgery may elicit stronger inflammatory responses, leading to tissue injury. As surgical time increases, intraoperative fluid management also becomes increasingly difficult, and increased fluid intake may trigger fluid overload, leading to pulmonary edema (28-30) A previous meta-analysis reported that EC patients with surgical time >4 h had 3.09 times higher risk of postoperative lung infection than those with surgery time ≤4 h (17). Therefore, optimizing the surgical procedure, shortening the surgical time, controlling perioperative fluid volume, and adopting a protective ventilation strategy during surgery are important measures to prevent and control PRF.

Anastomotic fistula is the most common and serious complication after EC surgery, with an incidence ranging from 10% to 21.2%, usually occurring 7–14 days after surgery (31,32). EC patients usually have diminished immune function after surgery and leakage of secretions due to anastomotic fistulas may lead to aspiration pneumonia, pleurisy, or mediastinal infection. These infections can trigger localized inflammatory responses that further develop into systemic inflammatory response syndrome (SIRS), leading to acute lung injury (ALI). Impaired lung function further exacerbates impaired gas exchange, substantially increasing the risk of PRF (33-35). Furthermore, anastomotic fistulas often result in fluid accumulation in the chest cavity or around the lungs, compressing lung tissue and impairing its expansion and normal ventilatory function. This lung compression weakens the patient’s respiratory function and hinders the effective expulsion of airway secretions, resulting in sputum retention, which promotes bacterial colonization and significantly increases the risk of lung infection (36). Therefore, early identification of anastomotic fistulas and prevention of pus leakage into the thoracic cavity and mediastinum are crucial for preventing potentially fatal complications and to improve outcomes.

Certain limitations should be considered in this study. Since a small number of positive outcome events may lead to model overfitting, further multicenter, large cohort studies are warranted to validate the robustness and generalizability of the model.

Conclusions

Smoking index ≥400, decreased FEV1, decreased preoperative serum albumin levels, prolonged surgical time, and postoperative anastomotic fistula are risk factors for PRF in EC patients. Our nomogram constructed based on these risk factors demonstrated good predictive performance and clinical utility in both internal and external validation. This model represents a promising tool for early screening of high-risk patients and for guiding early interventions during the perioperative period to improve patient outcomes.

Acknowledgments

We sincerely appreciate each patient and their family for consenting to participate in this study, as well as all the individuals who provided assistance throughout the research.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2024-2114/rc

Data Sharing Statement: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2024-2114/dss

Peer Review File: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2024-2114/prf

Funding: The study was funded by Gansu Provincial Natural Science Foundation (No. 22JR5RA647) and 2023 Longyuan Youth Innovation and Entrepreneurship Talent (Individual) Project (No. 211278291049).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2024-2114/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Ethics Committee of Gansu Hospital of Sun Yat-sen University Cancer Center (No. P-LW202407160006) and informed consent was taken from all the patients.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Zheng RS, Chen R, Han BF, et al. Cancer incidence and mortality in China, 2022. Zhonghua Zhong Liu Za Zhi 2024;46:221-31. [Crossref] [PubMed]
3. Obermannová R, Alsina M, Cervantes A, et al. Oesophageal cancer: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Ann Oncol 2022;33:992-1004. [Crossref] [PubMed]
4. Low DE, Kuppusamy MK, Alderson D, et al. Benchmarking Complications Associated with Esophagectomy. Ann Surg 2019;269:291-8. [Crossref] [PubMed]
5. Hirano Y, Konishi T, Kaneko H, et al. Impact of Prophylactic Corticosteroid Use on In-hospital Mortality and Respiratory Failure After Esophagectomy for Esophageal Cancer: Nationwide Inpatient Data Study in Japan. Ann Surg 2023;277:e1247-53. [Crossref] [PubMed]
6. Dong J, Wang GD, Wang HZ. Clinical analysis of patients with respiratory failure after esophageal cancer operation. Transl Cancer Res 2021;10:5238-45. [Crossref] [PubMed]
7. Su Q, Li H, Yan H, et al. Prognostic risk factors for respiratory failure after esophagectomy. Transl Cancer Res 2020;9:6362-8. [Crossref] [PubMed]
8. Yu B, Liu Z, Zhang L, et al. Pre- and intra-operative risk factors predict postoperative respiratory failure after minimally invasive oesophagectomy. Eur J Cardiothorac Surg 2024;65:ezae107. [Crossref] [PubMed]
9. Balachandran VP, Gonen M, Smith JJ, et al. Nomograms in oncology: more than meets the eye. Lancet Oncol 2015;16:e173-80. [Crossref] [PubMed]
10. Canet J, Gallart L. Postoperative respiratory failure: pathogenesis, prediction, and prevention. Curr Opin Crit Care 2014;20:56-62. [Crossref] [PubMed]
11. Margraf A, Ludwig N, Zarbock A, et al. Systemic Inflammatory Response Syndrome After Surgery: Mechanisms and Protection. Anesth Analg 2020;131:1693-707. [Crossref] [PubMed]
12. Wang K, Zhang R. Analysis of risk factors and establishment of prediction model for respiratory failure after McKeown esophageal cancer surgery. Anhui Medical Journal 2022;43:918-22.
13. Sauerbrei W, Royston P, Binder H. Selection of important variables and determination of functional form for continuous predictors in multivariable model building. Stat Med 2007;26:5512-28. [Crossref] [PubMed]
14. Jeganathan V, Knight S, Bricknell M, et al. Impact of smoking status and chronic obstructive pulmonary disease on pulmonary complications post lung cancer surgery. PLoS One 2022;17:e0266052. [Crossref] [PubMed]
15. Turan A, Mascha EJ, Roberman D, et al. Smoking and perioperative outcomes. Anesthesiology 2011;114:837-46. [Crossref] [PubMed]
16. Lugg ST, Alridge KA, Howells PA, et al. Dysregulated alveolar function and complications in smokers following oesophagectomy. ERJ Open Res 2019;5:00089-2018. [Crossref] [PubMed]
17. Wang M, Zhou C. Risk factors for postoperative pulmonary infection in patients with esophageal cancer: a systematic review and meta-analysis. Chinese Journal of Clinical Thoracic and Cardiovascular Surgery 2023;30:1467-74.
18. Zhi X, Liu L. Chinese guidelines for perioperative airway management in thoracic surgery (2020 edition). Chinese Journal of Clinical Thoracic and Cardiovascular Surgery 2021:251-62.
19. Ferguson MK, Celauro AD, Prachand V. Prediction of major pulmonary complications after esophagectomy. Ann Thorac Surg 2011;91:1494-1500; discussion 1500-1. [Crossref] [PubMed]
20. Maruyama S, Okamura A, Ishizuka N, et al. Airflow Limitation Predicts Postoperative Pneumonia after Esophagectomy. World J Surg 2021;45:2492-500. [Crossref] [PubMed]
21. Sun Y, Zhu Y. Impairment of Lung Function Increases the Risk of Postoperative Respiratory Failure for Esophageal Carcinoma: A Systematic Review and Meta-Analysis. J Healthc Eng 2021;2021:5327682. [Crossref] [PubMed]
22. Kim T, Jeon YJ, Lee H, et al. Preoperative DLco and FEV(1) are correlated with postoperative pulmonary complications in patients after esophagectomy. Sci Rep 2024;14:6117. [Crossref] [PubMed]
23. Soeters PB, Wolfe RR, Shenkin A. Hypoalbuminemia: Pathogenesis and Clinical Significance. JPEN J Parenter Enteral Nutr 2019;43:181-93. [Crossref] [PubMed]
24. Kim S, McClave SA, Martindale RG, et al. Hypoalbuminemia and Clinical Outcomes: What is the Mechanism behind the Relationship? Am Surg 2017;83:1220-7. [Crossref] [PubMed]
25. Allison SP, Lobo DN. The clinical significance of hypoalbuminaemia. Clin Nutr 2024;43:909-14. [Crossref] [PubMed]
26. Jeong BH, Shin B, Eom JS, et al. Development of a prediction rule for estimating postoperative pulmonary complications. PLoS One 2014;9:e113656. [Crossref] [PubMed]
27. Arozullah AM, Daley J, Henderson WG, et al. Multifactorial risk index for predicting postoperative respiratory failure in men after major noncardiac surgery. The National Veterans Administration Surgical Quality Improvement Program. Ann Surg 2000;232:242-53. [Crossref] [PubMed]
28. Brochard L, Slutsky A, Pesenti A. Mechanical Ventilation to Minimize Progression of Lung Injury in Acute Respiratory Failure. Am J Respir Crit Care Med 2017;195:438-42. [Crossref] [PubMed]
29. Güldner A, Kiss T, Serpa Neto A, et al. Intraoperative protective mechanical ventilation for prevention of postoperative pulmonary complications: a comprehensive review of the role of tidal volume, positive end-expiratory pressure, and lung recruitment maneuvers. Anesthesiology 2015;123:692-713. [Crossref] [PubMed]
30. Van Dessel E, Moons J, Nafteux P, et al. Perioperative fluid management in esophagectomy for cancer and its relation to postoperative respiratory complications. Dis Esophagus 2021;34:doaa111. [Crossref] [PubMed]
31. Chevallay M, Jung M, Chon SH, et al. Esophageal cancer surgery: review of complications and their management. Ann N Y Acad Sci 2020;1482:146-62. [Crossref] [PubMed]
32. Kassis ES, Kosinski AS, Ross P Jr, et al. Predictors of anastomotic leak after esophagectomy: an analysis of the society of thoracic surgeons general thoracic database. Ann Thorac Surg 2013;96:1919-26. [Crossref] [PubMed]
33. Zhu K, Li Z. Advances in clinical management of anastomotic leakage after esophageal cancer surgery. Chinese Journal of Thoracic Surgery Electronics 2023;10:50-6.
34. Hummel R, Bausch D. Anastomotic Leakage after Upper Gastrointestinal Surgery: Surgical Treatment. Visc Med 2017;33:207-11. [Crossref] [PubMed]
35. Xu QL, Li H, Zhu YJ, et al. The treatments and postoperative complications of esophageal cancer: a review. J Cardiothorac Surg 2020;15:163. [Crossref] [PubMed]
36. Messager M, Warlaumont M, Renaud F, et al. Recent improvements in the management of esophageal anastomotic leak after surgery for cancer. Eur J Surg Oncol 2017;43:258-69. [Crossref] [PubMed]