Artificial Intelligence Could Help Predict Tumor Sensitivity for Patients with NSCLC


Researchers determined that artificial intelligence could help predict responses to systemic therapies for patients with non-small cell lung cancer by utilizing CT scans.

Researchers used standard of care computed tomography (CT) scans to determine that artificial intelligence has the potential to train algorithms which would predict tumor sensitivity to 3 systemic cancer therapies for patients with non-small cell lung cancer (NSCLC), according to data published in Clinical Cancer Research.

“Radiologists’ interpretation of CT scans of cancer patients treated with systemic therapies is inherently subjective,” Laurent Dercle, MD, PhD, associate research scientist at the Columbia University Irving Medical Center, said in a press release. “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.”

Data across multiple phase II and phase III clinical studies was incorporated to evaluate systemic treatment for patients of NSCLC, with patients being treated with one of either the immunotherapeutic agent nivolumab (Opdivo), the chemotherapeutic agent docetaxel (Taxotere), or the targeted therapeutic gefitinib (Iressa).

CT images of 92 patients receiving nivolumab in 2 trials were retrospectively analyzed, 50 patients receiving docetaxel in 1 trial, and 46 patients receiving gefitinib in 1 trial.

The researchers classified tumors as treatment-sensitive or treatment-insensitive based on each trial’s reference standard, while also analyzing CT scan images at baseline and first-treatment assessment to develop the model implemented.

More, the team developed a multivariable model to predict treatment sensitivity, where each model predicted a score from 0 (highest treatment sensitivity) to 1 (highest treatment insensitivity) based on lung lesions identified. Overall, 8 radiologic features were implemented to build the three prediction models. The features utilized included changes in tumor volume, heterogeneity, shape, and margin.

The measurement was calculated utilizing area under the curve, which measured the model’s accuracy with 1 representing perfect prediction. Per the study, the nivolumab, docetaxel, and gefitinib prediction models achieved an area under the curve of 0.77, 0.67, and 0.82, respectively.

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

The main concern regarding the study’s data was the small sample size implemented in each cohort. The lead researcher noted that implemented artificial intelligence with a larger population will help the team identify new patterns on which to base more accurate predictive models.

The current way radiologists quantify tumor change is through tumor size and the appearance of new tumor lesions. Dercle emphasized this type of evaluation is limited, especially for patients who can “display atypical patterns of response and progression.”


Artificial Intelligence May Help Predict Responses to Systemic Therapies in Patients with Non-Small Cell Lung Cancer [news release]. Philadelphia, Pennsylvania. Published March 20, 2020. Accessed March 27, 2020.

Related Videos
Video 4 - "Frontline Treatment for EGFR-Mutated Lung Cancer"
Video 3 - "NGS Testing Challenges and Considerations in NSCLC"
Related Content