Non-invasive prediction for pathologic complete response to neoadjuvant chemoimmunotherapy in lung cancer using...

Area of Science:

• Oncology
• Radiology
• Artificial Intelligence

Background:

• Neoadjuvant chemoimmunotherapy has transformed non-small cell lung cancer (NSCLC) treatment.
• Predicting patient response to neoadjuvant immunotherapy is crucial for clinical decision-making.
• Current methods for predicting response in NSCLC require improvement.

Purpose of the Study:

• To develop and validate a deep learning (DL) model for predicting pathologic complete response (pCR) to neoadjuvant immunotherapy in NSCLC patients.
• To assess the DL model's performance against traditional clinical models.
• To explore the biological underpinnings of the DL model's predictive capability.

Main Methods:

• Retrospective analysis of computed tomography (CT) imaging data from 248 NSCLC patients who received neoadjuvant chemoimmunotherapy.
• Development of a DL model using data from Ruijin Hospital (training set: 104 patients, validation set: 69 patients).
• External validation of the DL model using data from Ningbo Hwamei Hospital and Affiliated Hospital of Zunyi Medical University (75 patients).

Main Results:

• The overall pCR rate was 29.4% (73/248 patients).
• The DL model achieved areas under the curve (AUCs) of 0.775 in the validation set and 0.743 in the external cohort for pCR prediction.
• The DL model significantly outperformed the clinical model (AUCs of 0.579 and 0.569, respectively) and showed correlations with cell metabolism pathways and antitumor immune infiltration.

Conclusions:

• The developed deep learning model effectively predicts pCR to neoadjuvant chemoimmunotherapy in NSCLC patients based on CT imaging.
• The DL model offers superior predictive performance compared to conventional clinical assessments.
• The model's findings suggest a link between metabolic pathways, immune infiltration, and treatment response in NSCLC.