Main Session
Sep 29
PQA 07 - Head and Neck Cancer, Lung Cancer/Thoracic Malignancies, and Nursing and Supportive Care

3638 - AI-Derived Thymic Health, Survival, and Patterns of Failure Among Older Adults Undergoing SBRT for Early-Stage Non-Small Cell Lung Cancer

03:45pm - 05:00pm ET
Poster Hall - Exhibit Hall A
Screen: 27
POSTER

Presenter(s)

Leah Thompson, MD Headshot
Leah Thompson, MD - Harvard Radiation Oncology Program, Boston, MA

L. L. Thompson1,2, V. Prudente1,2, S. Bernatz1,2, S. B. Shah3, F. Haugg1,4, A. Zhou3, K. Pischel3, S. Jiang3, R. Kanwar3, K. Heydari3, S. Malhotra3, A. T. Gregg3, K. Lambert3, W. Zuo3, C. Huo3, A. Warrington4, F. K. Keane5, A. Saraf6, R. H. Mak1,2, and H. Aerts1,4; 1Artificial Intelligence in Medicine (AIM) Program, Mass General Brigham, Harvard Medical School, Boston, MA, 2Department of Radiation Oncology, Brigham and Women's Hospital and Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, 3Harvard Medical School, Boston, MA, 4Department of Radiation Oncology, Mass General Brigham/Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, 5Department of Radiation Oncology, Massachusetts General Hospital, Boston, MA, 6Dana Farber Cancer Center at South Shore Health, Weymouth, MA

Purpose/Objective(s):

Older adults with early-stage non-small cell lung cancer (NSCLC) have high rates of immunosenescence and immune dysregulation, factors which may contribute to inferior oncologic outcomes. Despite this, scalable immunologic biomarkers identifying high-risk subgroups remain limited. We aimed to evaluate whether an artificial intelligence (AI)-based measure of thymic health was associated with overall survival (OS), progression free survival (PFS), distant metastasis (DM), or locoregional recurrence (LRR) among older adults undergoing stereotactic body radiotherapy (SBRT) for early-stage NSCLC.

Materials/Methods: We retrospectively reviewed the records for patients aged = 65 years with stage I-II NSCLC who completed SBRT at our institution between June 1, 2009 and March 31, 2023. We abstracted demographics, performance status (Eastern Cooperative Oncology Group [ECOG] score), cancer stage, and SBRT details. Thymic health scores were calculated from pre-treatment computed tomography scans using a validated deep learning system (range 0-1, higher scores=better thymic health). To assess associations between thymic health and outcomes, we utilized Cox proportional hazards models for OS and PFS and Fine-Gray subdistribution hazards models for DM and LRR. Competing risks included death (for DM and LRR) and DM (for LRR only). All multivariable models were adjusted for age, sex, cancer stage, ECOG score, and covariates significant at p<0.10 in univariable analyses.

Results: Overall, 708 patients (median age 76.2 years, 60.7% female; 90.4% stage IA) completed SBRT (median follow-up 25.3 months [IQR 14.1-45.8 months]). Median thymic health was 0.14 (IQR 0.07-0.22). In adjusted Cox regression models, higher thymic health was associated with improved OS (HR = 0.25 [95% CI 0.08-0.76, p = 0.015]) and PFS (HR = 0.16 [95% CI: 0.06-0.42, p = 0.015]). In adjusted Fine-Gray competing risks regressions, higher thymic health was independently associated with a decreased risk of both LRR (sHR= 0.09 [95% CI: 0.01 - 0.91, p = 0.041]) and DM (sHR = 0.10 [95% CI: 0.02 - 0.48, p = 0.004]).

Conclusion: AI-based thymic health scores are associated with disease progression and mortality in older adults with early-stage NSCLC and may have utility for pre-treatment risk stratification.