Artificial Intelligence Can Help Predict Cancer Therapy Response

Applying artificial intelligence tools to standard CT scans could help predict tumor response in patients with advanced non-small cell lung cancer (NSCLC), according to research published in Clinical Cancer Research.

The current method for determining if a patient with NSCLC is responding to systemic therapy requires radiologists to quantify changes in tumor size and the appearance of new lesions. However, this approach can be limited, especially in patients undergoing immunotherapy treatments. These patients can display atypical patterns of response and progression, researchers noted.

“Newer systemic therapies prompt the need for alternative metrics for response assessment, which can shape therapeutic decision-making,” said Laurent Dercle, MD, PhD, associate research scientist in the Department of Radiology at the Columbia University Irving Medical Center.

Traditional methods can also vary among providers, leading to treatment predictions that may not always be completely accurate.

“Radiologists’ interpretation of CT scans of cancer patients treated with systemic therapies is inherently subjective,” said Dercle.

“The purpose of this study was to train cutting-edge AI technologies to predict patients’ responses to treatment, allowing radiologists to deliver more accurate and reproducible predictions of treatment efficacy at an early stage of the disease.”

Researchers analyzed data from multiple clinical trials that evaluated systemic treatment in patients with NSCLC. These patients were treated with one of three agents: an immunotherapeutic agent, a chemotherapeutic agent, or a targeted therapeutic.

The team then retrospectively analyzed standard-of-care CT scans from 92 patients receiving the immunotherapeutic agent in two trials; 50 patients receiving the chemotherapeutic agent in one trial; and 46 patients receiving the targeted therapeutic in one trial.

To develop the AI algorithm, researchers used the CT scans taken at baseline and on first-treatment assessment. For patients taking the targeted therapeutic, their first-treatment assessment occurred after three weeks, and for the other two cohorts, assessments occurred after eight weeks.

The team classified the tumors as either treatment-sensitive or treatment-intensive based on the reference standard of each trial. Among all three cohorts, researchers randomized patients into training or validation groups.

Using machine learning, the group developed a multivariable model to predict treatment sensitivity in the training cohort. Each model could predict a score ranging from zero (highest treatment sensitivity) to one (highest treatment insensitivity) based on the change of the largest measurable lung lesion identified at baseline.

Across all cohorts, researchers used a total of eight radiologic features to build the three prediction models. These features included changes in tumor volume, heterogeneity, shape, and margin.

The results showed that in the validation cohorts, the immunotherapeutic agent prediction model achieved an area under the curve (AUC) of 0.77, while the chemotherapeutic agent prediction model achieved an AUC of 0.67. The targeted therapy prediction model achieved an AUC of 0.82.

“We observed that similar radiomics features predicted three different drug responses in patients with NSCLC,” Dercle said. “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.”

Radiomic signatures have the potential to enhance clinical decision-making and improve cancer care, researchers said.

“With AI, cancer imaging can move from an inherently subjective tool to a quantitative and objective asset for precision medicine approaches,” said Dercle.

The study was limited in that it included only a small sample size of data. Future research should include more volumes of patient data to help refine AI and machine learning tools.

“Because AI can continuously learn from real-world data, using AI on larger patient datasets will help us to identify new patterns to build more accurate prediction models,” Dercle said.