Main Session
Sep 29
PQA 06 - Genitourinary Cancer, Gynecological Cancer, and Health Care Access and Engagement

3332 - Evaluating Pretreatment PSMA PET/CT Radiomics in Prostate Cancer

02:15pm - 03:30pm ET
Poster Hall - Exhibit Hall A
Screen: 14
POSTER

Presenter(s)

Lakshmi Rekha Narra, MD - Rutgers Cancer Institute of New Jersey, New Brunswick, NJ

L. R. Narra1, T. J. Yang1, R. A. Deek2, C. Onal3, O. C. C. Guler4, N. Torun5, M. Reyhan5, K. Nie1, and M. P. Deek1; 1Department of Radiation Oncology, Rutgers Cancer Institute, New Brunswick, NJ, 2University of Pittsburgh, Pittsburgh, PA, 3Baskent University Faculty of Medicine, Department of Radiation Oncology, Ankara, Turkey, 4Baskent University Faculty of Medicine, Adana Dr Turgut Noyan Research and Treatment Center, Department of Radiation Oncology, Adana, Turkey, 5Baskent University Faculty of Medicine, Adana Dr Turgut Noyan Research and Treatment Center, Department of Nuclear Medicine, Adana, Turkey

Purpose/Objective(s): Current clinical risk stratification for prostate cancer may undercapture disease heterogeneity, as 20-60% of patients develop biochemical recurrence (BCR) or distant metastases (DM) after definitive treatment, underscoring the need for additional prognostic tools. Recent studies have demonstrated the prognostic value of PSMA PET/CT radiomics; however, the role of pretreatment primary-tumor derived radiomic features remains underexplored. We aim to evaluate whether radiomic features extracted from the primary tumor on pretreatment PSMA PET/CT can identify patients at higher risk for early BCR or DM following definitive radiotherapy & risk- adapted androgen-deprivation therapy

Materials/Methods: In this retrospective study, 268 patients treated definitively for prostate cancer with available pretreatment PSMA PET/CT imaging were included. Baseline clinicopathologic features include Age, PSA, Clinical T stage, Nodal Status (N+), Gleason Score & NCCN risk group

PSMA avid intraprostatic tumor was manually contoured on all pre-treatment PSMA PET/CT scans & for each lesion a total of 204 radiomic features were extracted using PyRadiomics. Feature importance was assessed using permutation feature importance within a support vector machine (SVM) model across multiple random train/test splits. 9 features with highest importance score were selected to generate radiomic risk (RR) score and survival modeling. The 9 selected features were included in a Cox regression model for time to BCR and DM. A RR score (Xß) was derived from Cox coefficients and used to dichotomize the cohort at the median into low-risk (LR) and high-risk (HR) groups. Outcomes were compared using Kaplan-Meier (KM) analysis & log-rank test.

Results: Among 268 patients, median age was 69 (IQR 64-73), median baseline PSA: 23.1 ng/ml (IQR: 11.1-61.5). 35% (94/268) were node positive, and 80% (214/268) were NCCN high-risk group. In overall cohort, median follow-up from the date of pretreatment PSMA PET/CT was 51 Mo (IQR: 33-76), 23.9% (64/268) had BCR and 24.6% (66/268) had DM.

BCR: Among 266 evaluable, Based on RR score; n=133 were classified as LR and n=133 as HR. BCR occurred in 47/133 (35%) HR vs 17/133 (13%) LR. Median BCR free survival was 75.8 Mo (95% CI: 60.4-91.3) in HR and was not reached in LR. KM analysis demonstrated significant separation of BCR-free survival between LR & HR groups (log-rank, p<0.001).

DM: Among 268 evaluable, Based on RR score; n=134 were classified as LR and n=134 as HR. DM occurred in 50/134 (37%) in HR vs 16/134 (12%) in LR. Median DM-free survival was 64 Mo (95% CI 49.6-78.4) in HR and was not reached in the LR. KM analysis demonstrated significant separation of DM-free survival between LR & HR groups (log-rank, p<0.001).

Conclusion: These findings support the potential utility of pretreatment primary tumor PSMA PET/CT radiomics for risk stratification. Future studies will include external validation using independent institutional datasets to assess reproducibility and generalizability.