2023 World Conference on Lung Cancer (Posters)-P2.05. Immunotherapy Prognostication with Thoracic CT Contrastive Self-Supervised Learning in Patients with Non-Small Cell Lung Cancer

P2.05. Immunotherapy Prognostication with Thoracic CT Contrastive Self-Supervised Learning in Patients with Non-Small Cell Lung Cancer - PDF(Abstract)

Back to course

Pdf Summary

This study aims to develop a robust imaging biomarker for predicting the response to immunotherapy in patients with non-small cell lung cancer (NSCLC). The researchers utilized a cohort of 561 patients with advanced NSCLC treated with immune checkpoint inhibitors (ICIs). The cohort was divided into a training/tuning group and an independent test group. Baseline chest CT scans and clinical parameters were collected and analyzed. A contrastive self-supervised learning (SSL) model was used to extract 3D image representations of lung tumors. Classification models combining SSL features and clinical variables were selected through hyper-parameter tuning. The primary clinical endpoint was 2-year overall survival (OS).<br /><br />The results showed that SSL features achieved an area under the curve (AUC) of 0.64 for predicting OS in the internal test cohort. The predictive performance improved when SSL features were combined with clinical variables, achieving an AUC of 0.68. This combination also resulted in statistically significant stratification of the cohort into high and low-risk groups. In comparison, clinical variables alone achieved an AUC of 0.692, and tumor volume had an AUC of 0.552. The baseline PD-L1 expression score achieved an AUC of 0.57.<br /><br />In the independent test cohort, SSL features predicted OS with an AUC of 0.583. When combined with clinical variables, the predictive performance improved to an AUC of 0.68.<br /><br />The study concludes that SSL representations of baseline chest CT, combined with clinical risk factors, provide improved predictive performance for survival in advanced NSCLC patients receiving immunotherapy compared to clinical measures alone. This research highlights the potential of using imaging biomarkers and machine learning techniques to personalize and improve the effectiveness of immunotherapy in NSCLC patients.

Asset Subtitle

Tafadzwa Chaunzwa

Meta Tag

Speaker Tafadzwa Chaunzwa

Topic Metastatic NSCLC: Immunotherapy - Biomarker

Keywords

imaging biomarker

predicting response

immunotherapy

non-small cell lung cancer

NSCLC

immune checkpoint inhibitors

CT scans

clinical parameters

self-supervised learning

overall survival