Towards personalized care in minimally invasive esophageal surgery: An adverse events prediction model

Abstract

This letter addressed the impactful study by Zhong et al, which introduced a risk prediction and stratification model for surgical adverse events following minimally invasive esophagectomy. By identifying key risk factors such as chronic obstructive pulmonary disease and hypoalbuminemia, the model demonstrated strong predictive accuracy and offered a pathway to personalized perioperative care. This correspondence highlighted the clinical significance, emphasizing its potential to optimize patient outcomes through tailored interventions. Further prospective validation and application across diverse settings are essential to realize its full potential in advancing esophageal surgery practices.

Key Words: Minimally invasive esophagectomy; Surgical adverse events; Risk prediction model; Risk stratification; Hypoalbuminemia; Predictive accuracy; Personalized perioperative care; Tailored interventions; Esophageal surgery

Core Tip: The study by Zhong et al presented a predictive model for surgical adverse events following minimally invasive esophagectomy, identifying key risk factors such as chronic obstructive pulmonary disease, low forced expiratory volume in 1 s, alcohol use, and hypoalbuminemia. This model exhibited strong predictive accuracy; it possesses the potential to enhance clinical decision-making by guiding multidisciplinary teams in patient selection and providing individualized perioperative care. However, the retrospective nature of the study may limit its generalizability.

TO THE EDITOR

With this letter we would like to commend Zhong et al[1] for their study, which constitutes a valuable contribution to the risk stratification and selection of patients to be subjected to minimally invasive esophagectomy (MIE). While MIE offers promising survival outcomes, its reported 54% surgical adverse events (SAEs) rate warrants the development of a robust risk stratification model for improving patient selection and clinical outcomes[2,3]. Minimizing the risk of SAEs is critical in ensuring long-term patient survival and ensuring postoperative quality of life[4]. The aim of the study by Zhong et al[1] was to develop and validate a risk stratification model for SAEs in patients undergoing MIE for the treatment of esophageal cancer.

Regarding the study methodology, the authors retrospectively collected postoperative SAE data from 747 patients from two tertiary centers in China. SAE data included pulmonary infection, anastomotic fistula, blood transfusion, intensive care unit admission, anastomotic leakage, and length of hospital stay. The universally accepted Clavien-Dindo surgical complication classification system was used to grade the severity of complications[4,5]. Τhe authors followed a scalable and systematic approach to develop the risk model. Initially, LASSO regression was used to identify statistically significant predictors of postoperative complications, thereby minimizing the risk of confounding factors associated with high-dimensional variables and multicollinearity. Following this, univariate and subsequent multivariate analyses were conducted. This three-step process eventually led to the development of the final robust and reliable risk prediction model for SAEs. For the validation of the clinical model and the demonstration of its predictive accuracy the authors calculated the respective areas under the curve.

By the means of multivariate regression analysis, it was concluded that chronic obstructive pulmonary disease [adjusted odds ratio (aOR) = 4.82, 95% confidence interval (CI): 1.28-18.18, P = 0.02], low forced expiratory volume in 1 s (aOR = 0.13, 95%CI: 0.06-0.27, P < 0.001), history of alcohol consumption (aOR = 4.69, 95%CI: 2.32-9.51, P < 0.001), and low albumin levels (aOR = 0.83, 95%CI: 0.76-0.90, P < 0.001) were independent risk factors for SAEs. Concurrently, the authors divided the retrieved patient data into two distinct sets, a training set and a validation cohort, based on the timing of intervention. The areas under the curve of the former and latter cohorts’ clinical models were subsequently calculated at 0.889 (95%CI: 0.853–0.926) and 0.793 (95%CI: 0.701-0.884), indicating a considerable and fair prediction performance[6].

According to the National Comprehensive Cancer Network guidelines for the management of upper gastrointestinal cancers, MIE is regarded as an acceptable approach for the management of esophageal cancer[7]. However, it is recommended that such operations should be attempted by appropriately trained surgeons, as they have been associated with a steep learning curve and high learning-associated morbidity[8,9]. In modern clinical practice, the predictive model developed by Zhong et al[1] may be used to inform the process of appropriately selecting patients to be subjected to MIE for the treatment of esophageal cancer at the metastasis-directed therapy level, which includes appropriate preoperative anesthetic planning and perioperative care as well as individualized postoperative monitoring and management.

In our opinion, the risk stratification model developed in this study constitutes a valuable addition to the clinician’s arsenal. Its robust analysis and strong internal validity safeguards the effective selection and perioperative management of patients with cancer undergoing MIE. Despite providing valuable insights, this study is limited by a risk of bias inherent to its retrospective nature. First, all procedures were performed by experienced surgeons, and second these procedures were offered to selected patients in a non-randomized manner. By default, this limits the generalizability of these findings when applied to clinical settings with comparatively limited experience and more diverse patient populations.

Overall, this predictive tool could guide multidisciplinary discussions by enabling risk stratification of individual patients, potentially leading to the reduction of complications and the optimization of perioperative care. Future multicenter prospective studies could enhance external validity by including diverse patient cohorts and clinical settings, addressing the important matter of variability in surgical expertise.