3325 - PREDICTO-Ren - Personalized Radiomics-basEd Determination of Individual Cancer Treatment Outcome in Patients with Renal Cancer

12:45pm - 02:00pm PT

Hall F

Screen: 6

POSTER

Presenter(s)

Felipe Restini, MD - McGill University Health Center, Montreal, QC

F. Restini¹, T. Y. T. De Souza², M. Farag³, S. Tanguay⁴, F. Brimo³, and F. Cury²; ¹Department of Radiation Oncology, McGill, Montreal, Canada, ²McGill University Health Centre, Montreal, QC, Canada, ³Department of Pathology, McGill University Health Center, Montreal, QC, Canada, ⁴Department of Urology, McGill University Health Center, Montreal, QC, Canada

Purpose/Objective(s):

The use of Stereotactic Ablative Radiotherapy (SABR) for renal cell carcinoma (RCC) is gaining popularity as a curative non-invasive alternative, offering excellent local control with minimal toxicity. Most studies use the RECIST criteria to assess tumor response following SABR. Despite its efficacy, the time for treatment radiological response is long and variable, making patient follow-up and treatment assessment challenging, sometimes generating anxiety among patients and referring Urologists. This study aims to develop a radiomic signature to predict the time to local response after SABR, allowing for personalized patient monitoring.

Materials/Methods:

We analyzed data from patients treated with SABR for RCC at a high-volume center between March 2016 and December 2023. Computed tomography (CT) scans from diagnosis and treatment planning were integrated with clinical and digital pathology data to extract radiomic features. Treatment response at one year (TR1y) after SBRT was assessed using diagnostic CT scans, with responders defined as those showing a >30% reduction in tumor size. Significant features were identified using the Mann-Whitney U and t-tests. Linear Discriminant Analysis (LDA) was employed for model engineering, while Kaplan-Meier and Log-rank tests were used for survival analyses.

Results:

We analyzed 47 lesions in 46 patients, with a median follow-up of 29 months (95% CI: 16-39). The overall tumor response at 1 year (TR1y) was 38.9% (95% CI: 27–52.2). The dose-fractionation most commonly used was 40Gy in 5 fractions (33 patients), followed by 26 single-fraction (9 patients).

Eighty-four radiomic features were statistically relevant (p < .05), with Elongation (p = .0099) and Small-Area Low-Gray Level Emphasis (p = .0068) being the most significant. The radiomic risk signature stratified patients in high- or low-risk.

Based on TR1y, the high-risk patients, or non-responders, had a significantly delayed response, with a median time to response of 19 months (95% CI: 16.5–21.4), compared to 10.4 months (95% CI: 8.4–12.4) of the low-risk group (p <0.001, HR: 0.10; 95% CI: 0.02–0.43, p <0.002). During this short follow-up, there was no significant difference between risk groups in overall survival (p = 0.988), cancer-specific survival (p = 0.901), metastasis-free survival (p = 0.352), or locoregional recurrence (p = 0.521).

Conclusion:

This study demonstrated that radiomic risk stratification can effectively predict the time to local response after SABR for kidney cancer. This capability may improve treatment planning and follow-up strategies, enabling more personalized clinical management.