Preoperative Prediction of Lymph Node Metastasis in Patients With Early-T-Stage Non-small Cell Lung Cancer by Machine Learning Algorithms

Abstract

Background: Lymph node metastasis (LNM) is difficult to precisely predict before surgery in patients with early-T-stage non-small cell lung cancer (NSCLC). This study aimed to develop machine learning (ML)-based predictive models for LNM. Methods: Clinical characteristics and imaging features were retrospectively collected from 1,102 NSCLC ≤ 2 cm patients. A total of 23 variables were included to develop predictive models for LNM by multiple ML algorithms. The models were evaluated by the receiver operating characteristic (ROC) curve for predictive performance and decision curve analysis (DCA) for clinical values. A feature selection approach was used to identify optimal predictive factors. Results: The areas under the ROC curve (AUCs) of the 8 models ranged from 0.784 to 0.899. Some ML-based models performed better than models using conventional statistical methods in both ROC curves and decision curves. The random forest classifier (RFC) model with 9 variables introduced was identified as the best predictive model. The feature selection indicated the top five predictors were tumor size, imaging density, carcinoembryonic antigen (CEA), maximal standardized uptake value (SUV max), and age. Conclusions: By incorporating clinical characteristics and radiographical features, it is feasible to develop ML-based models for the preoperative prediction of LNM in early-T-stage NSCLC, and the RFC model performed best.

An Externally-Validated Dynamic Nomogram Based on Clinicopathological Characteristics for Evaluating the Risk of Lymph Node Metastasis in Small-Size Non-small Cell Lung Cancer

“…Previously, our ( 20 ) and other studies ( 19 , 21 – 23 ) have found that several preoperative factors might be associated with the occurrence of LNM in small-size NSCLC, including serum CEA, tumor size and imaging features such as imaging density. Tumor size was identified as an important predictor for LNM.…”

“…Thirdly, our study did not involve all clinicopathological characteristics. In our previous (20) and other studies (31), maximal standardized uptake value (SUV max ), was proved to be an impressive predictive factors for LNM, but was not included. That is because PET scan was not routinely performed for small-size NSCLC, and the variable with many missing values was not suitable for linear regression.…”

“…Our previous study developed machine learning-based models for the preoperative prediction of LNM (20), but the LNM status should be furtherly evaluated after surgery, with considering histology, especially for patients with negative nodes. The current surgery methods cannot guarantee the removal of all tumor cell-invaded lymph nodes.…”

Metastatic Patterns of Mediastinal Lymph Nodes in Small-Size Non-small Cell Lung Cancer (T1b)

“…However, in our study, we tried to enroll clinical, radiographical, and histological characteristics, but no significant difference was observed among these three groups for all variables ( Table 3). However, in our previous study, many characteristics demonstrated significantly difference between node-positive and node-negative patients (30). Thus, this study indicated that patients with positive nodes, whether N1, SN2, or N2, shared certain consistency of manifested characteristics.…”

<p>Development and Validation of a Novel Nomogram for Predicting Tumor-Distant-Metastasis in Patients with Early T1-2 Stage Lung Adenocarcinoma</p>

“…Even with active anti-tumor treatment, some patients have progressive disease and distant metastasis and require further treatment. TNM stage is an important prognostic biomarker of lung cancer, 17 and for early-stage NSCLC, the 5-year survival rate is more than 70%. 18,19 However, many patients have distant metastasis at the initial diagnosis of early T stages (diameter≤5cm).…”