Association between Contrast-Enhanced Computed Tomography Radiomic Features, Genomic Alterations and Prognosis in
Lisa Rinaldi1, Elena Guerini Rocco2,3, Gianluca Spitaleri4
1Radiation Research Unit, IEO European Institute of Oncology IRCCS, Via Ripamonti 435, 20141 Milan, Italy.
Cancers
|September 28, 2023
Summary
Advanced non-small cell lung cancer (NSCLC) patients can benefit from CT radiomics to non-invasively predict gene alterations and overall survival. This approach aids in personalizing treatment strategies for lung adenocarcinoma.
Area of Science:
- Oncology
- Radiology
- Genomics
Background:
- Personalized medicine for advanced non-small cell lung cancer (NSCLC) requires non-invasive methods to assess tumor mutational status and identify prognostic biomarkers.
- Current diagnostic approaches for lung adenocarcinoma often involve invasive procedures, highlighting the need for alternative methods.
Purpose of the Study:
- To investigate the association between contrast-enhanced Computed Tomography (CT) radiomic features and tumor mutational status (EGFR, KRAS, ALK) and Overall Survival (OS) in advanced lung adenocarcinoma.
- To develop and validate radiomic models for predicting genetic alterations and prognosis in NSCLC patients.
Main Methods:
- A retrospective and prospective cohort of 261 and 48 patients with lung adenocarcinoma, respectively, were analyzed.
- LASSO-Logistic regression was used to create a Radiomic Score (RS) for predicting mutational status.
- Radiomic, clinical, and combined models were trained and validated using Area Under the Curve (AUC) for mutation prediction and C-index for Overall Survival (OS) prediction.
Main Results:
- The Radiomic Score (RS) demonstrated high accuracy in predicting EGFR, KRAS, and ALK alterations during training (AUC 0.95-0.98).
- Validation showed good performance for EGFR (AUC 0.86) and moderate performance for KRAS and ALK (AUC 0.61-0.65).
- The RS was significantly associated with OS, and the clinical-radiomic model achieved a high validation C-index (0.80) for OS prediction.
Conclusions:
- CT-based radiomics shows potential for non-invasively identifying gene alterations in advanced lung adenocarcinoma.
- Radiomic features, alone or combined with clinical data, can aid in predicting prognosis and personalizing treatment for NSCLC patients.
- Further validation in independent studies is recommended to confirm the clinical utility of CT radiomics in NSCLC management.
