# Multi-Institutional CT Scan-Based Radiomics for Predicting Tumor PD-L1 Expression in Patients with Advanced and Limited Non-Small Cell Lung Cancer

## TL;DR

This study shows that CT scans can predict PD-L1 levels in lung cancer patients, potentially reducing the need for invasive biopsies.

## Contribution

A non-invasive CT-based radiomic score for predicting PD-L1 expression in NSCLC patients is developed and validated across multiple stages.

## Key findings

- The CT-derived score showed strong predictive performance with AUC of 0.75 in advanced NSCLC and 0.68 in limited-stage NSCLC.
- Patients with high rad-PDL1 scores had longer progression-free survival.
- The model demonstrated generalizability across different treatment settings and disease stages.

## Abstract

Immunotherapy has significantly improved the treatment of lung cancer; however, a large proportion of patients do not experience durable benefit, making it challenging to identify those most likely to respond. At present, treatment decisions rely on the assessment of tumor tissue, which requires invasive biopsies and may not fully capture tumor heterogeneity. In this study, we investigated whether information routinely available from computed tomography (CT) scans could be used to estimate a key immunotherapy biomarker, programmed death-ligand 1 (PD-L1), without additional invasive procedures. Using artificial intelligence–based image analysis, we developed a CT-derived score associated with PD-L1 expression and clinical outcomes. We demonstrated that this approach is applicable across different stages of non-small cell lung cancer and treatment settings. These results indicate that, upon prospective validation, CT-based features may support more precise patient stratification and contribute to personalized immunotherapy strategies.

Background/Objectives: Immune checkpoint inhibitors (ICIs) have revolutionized the treatment landscape of advanced non-small cell lung cancer (NSCLC), yet 70% of patients experience disease progression, underscoring the critical need for predictive biomarkers. Programmed death-ligand 1 (PD-L1) expression remains the most adopted biomarker for ICIs. With the emergence of machine learning, the development of radiomics algorithms based on CT scan images has demonstrated potential as a novel addition to the biomarker landscape in oncology. In this study, we aimed to develop a non-invasive surrogate of PD-L1 expression (rad-PDL1) derived from computed tomography (CT) scan imaging and compare its predictive value to pathological assessments. Furthermore, we evaluated its generalizability across advanced and limited-stage NSCLC. Methods: Radiomics features extracted from pretreatment CT were analyzed using a self-training pipeline that incorporated the feature tokenizer Transformer model to classify tumors as high vs. low PD-L1 expression. We included 482 advanced NSCLC patients treated with ICIs across three medical centers who were divided into training and hold-out validation sets. The algorithm was then further validated in an independent cohort of 51 patients with limited NSCLC treated with neoadjuvant ICI and chemotherapy. Results: Our pipeline demonstrated strong predictive performance in primary and independent validation (AUC = 0.75 and 0.68, accuracy = 0.73 and 0.69, respectively), highlighting its generalizability and adaptability to various disease stages. Kaplan–Meier curves revealed a longer progression-free survival for patients in the high rad-PDL1. Conclusions: These results demonstrate the feasibility of a CT-based radiomic surrogate of PD-L1 expression, showing partial generalization to an independent neoadjuvant cohort, while highlighting the need for larger prospective multi-site validation before clinical implementation.