Predicting Sensitivity to Therapy for NSCLC With Artificial Intelligence Model
Posted: Friday, April 10, 2020
In a study published in Clinical Cancer Research, researchers developed an artificial intelligence (AI) model to train algorithms to predict tumor sensitivity in patients with non–small cell lung cancer (NSCLC) to three systemic cancer therapies. Using standard-of-care computed tomography (CT) images obtained from patients with NSCLC, Laurent Dercle, MD, PhD, of the Columbia University Irving Medical Center, New York, and colleagues defined radiomics signatures to predict the sensitivity of tumors to nivolumab, docetaxel, and gefitinib.
“We observed that similar radiomics features predicted three different drug responses in patients with NSCLC,” stated Dr. Dercle in an American Association for Cancer Research press release. “Further, we found that the same four features that identified EGFR treatment sensitivity for patients with metastatic colorectal cancer could be utilized to predict treatment sensitivity for patients with metastatic NSCLC.”
Data were collected from several phase II/III clinical trials of patients treated with nivolumab (n = 92), docetaxel (n = 50), and gefitinib (n = 46). The researchers extracted 1,160 radiomics features from the largest measurable lung lesion for each patient at baseline and in the first on-treatment response assessment CT scan. Then, tumors were classified as treatment-sensitive or treatment-insensitive based on the reference standard of each trial. Using machine learning, quantitative imaging features were combined by high-throughput mining to develop a multivariable model to predict treatment sensitivity in the training cohort.
The performance of each signature was evaluated by calculating AUC, where a score of 1 corresponds to perfect prediction. The nivolumab, docetaxel, and gefitinib prediction models achieved an AUC of 0.77, 0.67, and 0.82 in the validation cohorts, respectively.
“With AI, cancer imaging can move from an inherently subjective tool to a quantitative and objective asset for precision medicine approaches,” concluded Dr. Dercle.
Disclosure: For author disclosures, visit clincancerres.aacrjournals.org.
