Skip to content

External validation of a deep learning CT biomarker to predict first-line immune checkpoint inhibitor monotherapy-associated survival in PD-L1–high metastatic non–small cell lung cancer

Published

May 2026

Citation

Parikh RB, Law J, Damato LM, et al. External validation of a deep learning CT biomarker to predict first-line immune checkpoint inhibitor monotherapy-associated survival in PD-L1–high metastatic non–small cell lung cancer. ASCO Annual Meeting. 2026. https://meetings.asco.org/meetings/2026-asco-annual-meeting/335/17134?presentation=257879

Overview

Immunotherapy has become a standard first-line treatment for patients with advanced lung cancer who have high levels of the PD-L1 protein expression. However, more than half of these patients see their cancer worsen within a year of starting treatment, and doctors currently lack reliable ways to predict who will benefit most from immunotherapy or who might need a more aggressive approach. Imaging  done during routine oncology care offers a potentially powerful, non-invasive way to identify these different patient groups and guide more personalized care.

Researchers used the Flatiron Health Research Database to validate a deep-learning artificial intelligence tool called eCTRS, which analyzes pre-treatment CT scans to predict patient survival. By examining records and imaging from over 200 patients, the study found that those identified as "High" by the AI tool had significantly better outcomes. These patients lived nearly three times longer without their disease progressing compared to those in the "Low" group (231 days versus 88 days) and demonstrated a much higher overall survival rate.

Why this matters

This research reveals that artificial intelligence can help doctors predict which lung cancer patients will respond best to immunotherapy—using scans that are already part of routine care. This breakthrough gives physicians powerful new insights before treatment even begins, supporting more personalized treatment plans. By unlocking the potential of existing medical images, this technology aims to match patients with the therapies most likely to improve their outcomes and quality of life.

Read the research

Share