The importance of this study to address lower extremity DVT risk in postoperative GC patients is underscored by the considerable morbidity and potential mortality associated with VTE in this patient population.13In particular, GC surgery is associated with an increased risk of postoperative VTE, including DVT and PE.14,15Compared with pneumatic pressure therapy, rivaroxaban has a superior prophylactic effect against lower limb DVT after GC surgery.16A systematic review and meta-analysis of 111,936 patients showed that the incidence of VTE in the first month after GC surgery was 1.8%, with DVT specifically occurring in 1.2%.11Among 666 Korean patients after gastrectomy, the overall incidence of VTE was 2.1%.17These figures highlight the importance of focusing on DVT in patients with gastroesophageal reflux disease after surgery. Moreover, this study aims to fill a major gap in current research: although the incidence of VTE in patients with gastroesophageal reflux disease is known, little focus has been placed on predicting lower extremity DVT, especially in the postoperative period after gastroesophageal reflux disease. A retrospective cohort study revealed that age, preoperative blood glucose level, postoperative anemia, and tumor grade were independent risk factors for postgastrectomy VTE in patients with gastroesophageal reflux disease.18However, compared with previous studies, our study focused on a prediction model using comprehensive clinical indicators such as age and calcium ion concentration, providing a more detailed risk assessment tool. This highlights the need for a prediction model that can accurately identify patients at high risk of DVT after GC surgery and enable targeted prevention strategies.
The prediction model developed in this study showed high accuracy as reflected by the area under the curve (AUC) values for both the training and validation sets. The results demonstrate the strong predictive ability of NRS-2002, which is essential in clinical practice for risk stratification and management of DVT in postoperative GC patients. The importance of such a prediction model is highlighted by the various risk factors identified in different studies, including age and tumor-related factors. Age has been consistently identified as an important risk factor for postoperative VTE.18The role of calcium in the coagulation process further supports the relevance of calcium as a predictive marker in the developed model. These factors can provide important insights into patient-specific risk profiles and guide clinicians in the prevention and management of DVT after GC surgery.
According to univariate analysis, age emerged as an important independent variable influencing the occurrence of DVT in GC patients after gastrectomy. Furthermore, multivariate analysis highlighted age as a contributing factor for the development of postoperative DVT in these patients. Age is also a risk factor for VTE in GC patients.19Here, calcium ions were found to be a key clinical factor in the model. The role of calcium ions in the coagulation process and thrombosis is complex and multifaceted, one of its important aspects being its involvement in platelet activation. Platelets play a key role in maintaining hemostasis and vascular integrity under normal conditions, and in thrombosis under pathological conditions. Platelet activation is driven by the upregulation of intracellular calcium (CaAges 2+) concentration increases with CaAges 2+ Dense tubular system and Ca influxAges 2+ From the extracellular space20In the context of fibrinogen clotting, it is also known that calcium ions are required for normal polymerization of fibrin monomers.twenty oneCalcium also plays an important role in activating coagulation factor XIII, which plays a key role in the final step of the coagulation cascade.twenty twoCalcium ions are therefore essential in the coagulation process, influencing different steps from platelet activation to the stabilization of the fibrin clot.
LDL plays a key role in the development of atherothrombosis. LDL alters the antithrombotic properties of the vascular endothelium and affects vasoconstriction, in part by reducing the availability of endothelial nitric oxide and activating proinflammatory signaling pathways. These altered intravascular LDL promote foam cell formation from smooth muscle cells and macrophages, increasing the vulnerability of atherosclerotic plaques and enhancing the thrombogenicity of both plaque and blood.twenty three.
Results from several studies suggest that low hemoglobin levels may be an indicator of increased risk of venous thromboembolism and poor prognosis in cancer patients.FiveAnother study demonstrated that low baseline hemoglobin levels increased the likelihood of symptomatic VTE, symptomatic DVT, and nonfatal PE.twenty fourAnother study investigated the impact of anemia on bleeding risk in patients undergoing anticoagulant therapy for VTE.twenty fiveThese findings highlight the importance of considering anemia as a factor in the management of VTE, especially in high-risk populations such as acutely ill and cancer patients.
Unlike previous studies, here we collected rich and comprehensive clinical indicators, including a total of 47 baseline, preoperative, surgical, and pathological clinical data. To date, our study contains the largest number of clinical variables. Most importantly, our study uses a variety of comprehensive machine learning algorithms. Machine learning methods have been successfully applied in various medical fields and show great potential for predictive data analysis.26Compared to traditional predictive models (logistic regression), machine learning models perform comparable to logistic regression models, but some machine learning techniques perform exceptionally well.27One study used electronic health record data from a diverse population to develop machine learning models (LightGBM) to predict VTE diagnosis and 1-year risk. These tools outperformed existing risk assessment tools and demonstrated robust performance across a range of VTE types and patient demographics.28In our study, we used various machine learning algorithms, including logistic regression, decision tree, random forest, SVM, XGBoost, and LightGBM. By applying these insights in our research, we can expect more robust and accurate models for predicting lower extremity DVT risk in postoperative GC patients, which may lead to improved patient outcomes.
In the real world, this model could be integrated into the clinical decision-making process, possibly through electronic medical record systems. By inputting patient-specific data, healthcare providers could receive an instant risk assessment and select the most appropriate preventive measures. This approach is consistent with the growing trend of personalized medicine, where treatment and prevention strategies are tailored to individual patient characteristics and risk profiles.
Although this study makes a contribution, one of its potential limitations is its retrospective nature, which may introduce biases such as selection bias and information bias. The data used in the study may also have limitations in terms of its scope and the accuracy of the information recorded. Another limitation is the generalizability of the results. The results of this study are based on a specific patient population and may not be directly applicable to other populations or situations. In addition, this study developed a population-specific prediction model. However, the selected predictors were not specific to a particular population, as they may be applicable to patients undergoing gastrointestinal, hepatic, and pancreatic surgery. Thus, the question arises as to whether there is a need to develop a postoperative lower extremity thrombosis prediction model exclusively for patients undergoing radical gastrectomy.
Future studies should focus on validation in different patient populations and clinical settings to increase the generalizability of the prediction model. Future studies can also integrate the model into clinical workflows and explore its impact on patient outcomes in real-world settings. However, further studies are needed to understand the biological mechanisms underlying the identified risk factors for DVT in GC patients, which could allow for more targeted therapeutic interventions. Furthermore, incorporating new types of data, such as genetic and molecular marker data, could improve the predictive accuracy of the model.
In summary, the development of a predictive model for lower extremity DVT in postoperative GC patients addresses an important clinical need. The accuracy of this model and its ability to identify significant predictors make it a valuable tool to improve postoperative care and patient outcomes in GC patients.
