WCLC 2025 - Posters & ePosters-EP.04.17 Radiomics Predicts IASLC Grades of Invasive Pulmonary Adenocarcinoma by Distinguishing High-Grade Patterns and Predominant Subtypes

EP.04.17 Radiomics Predicts IASLC Grades of Invasive Pulmonary Adenocarcinoma by Distinguishing High-Grade Patterns and Predominant Subtypes

Back to course

Pdf Summary

This study developed a radiomics-based model to non-invasively predict the IASLC (International Association for the Study of Lung Cancer) grading of invasive pulmonary adenocarcinoma (ADC), which is critical for patient prognosis and treatment planning. The IASLC grading system combines the presence of high-grade histological patterns and predominant ADC subtypes into three grades (1 to 3). The researchers retrospectively analyzed 529 surgically resected ADC cases, dividing the dataset into training (70%) and validation (30%) sets.<br /><br />The model consisted of two radiomics submodels: one to detect if high-grade patterns constituted over 20% of the tumor, and the other to classify predominant subtypes, specifically differentiating lepidic from acinar/papillary patterns. Combining these two submodels' outputs allowed assignment of IASLC grades.<br /><br />Performance results showed excellent discrimination of high-grade patterns with Area Under the Curve (AUC) scores of 0.95 and 0.96 for training and validation sets, respectively. For predominant subtype classification, AUCs were 0.83 (training) and 0.78 (validation). Overall IASLC grade prediction achieved AUCs in the validation set of 0.95 for grade 1, 0.85 for grade 2, and 0.96 for grade 3. Grade 3 prediction demonstrated the highest sensitivity (0.89), while grade 2 showed relatively lower sensitivity but high specificity (0.89).<br /><br />The study highlights that radiomics enables precise, non-invasive IASLC grading, outperforming previous methods especially for grade 2. This model shows promise for preoperative diagnosis, potentially guiding personalized treatment decisions. Future research may incorporate multi-omics data to further enhance grading accuracy.<br /><br />In summary, this radiomics approach effectively predicts IASLC grades in invasive pulmonary ADC by identifying aggressive tumor components and predominant patterns, providing a valuable tool to improve lung cancer management without the need for invasive biopsies.

Asset Subtitle

Sunyi Zheng

Meta Tag

Speaker Sunyi Zheng

Topic Screening and Early Detection

Keywords

radiomics

IASLC grading

invasive pulmonary adenocarcinoma

lung cancer

high-grade histological patterns

ADC subtypes

non-invasive prediction

AUC performance

tumor classification

personalized treatment