3369 - Predicting the Early Complete Biochemical Response in MRgSBRT in Prostate Cancer with Intraprostatic Lesion Boost

12:45pm - 02:00pm PT

Hall F

Screen: 28

POSTER

Presenter(s)

Oilei Wong, PhD - Hong Kong Sanatorium and Hospital, Hong Kong, NA

D. M. C. Poon¹, C. Xue², J. Yuan², O. Wong³, B. Yang⁴, S. T. Chiu⁵, H. Y. Wong⁶, S. K. Yu⁴, and G. Chiu⁷; ¹Department of Clinical Oncology, Prince of Wales Hospital, Hong Kong, China, ²Research Department, Hong Kong Sanatorium & Hospital, Happy Valley, Hong Kong, ³Research Department, Hong Kong Sanatorium and Hospital, Hong Kong, NA, Hong Kong, ⁴Hong Kong Sanatorium & Hospital, Happy Valley, Hong Kong, ⁵Department of Radiotherapy, Hong Kong Sanatorium & Hospital, Hong Kong, Hong Kong, ⁶Hong Kong Sanatorium and Hospital, Hong Kong, Hong Kong, ⁷Department of Radiotherapy, Hong Kong Sanatorium and Hospital, Happy Valley, Hong Kong

Purpose/Objective(s): Achieving complete biochemical response (CBR) shortly after treatment is critical for improving outcomes in localized prostate cancer (PC). Early CBR is associated with reduced recurrence rates and can guide further treatment strategies, leading to personalized management of patients. Despite advancements in radiotherapy techniques, there remains a significant gap in research regarding the ability to predict early CBR, particularly in patients undergoing the innovative MR-guided stereotactic body radiotherapy (MRgSBRT) with intraprostatic lesion (IL) boost, which takes the advantages of superior MR soft-tissue image contrast for IL identification, motion monitoring, and plan adaptation. This study aims to investigate the predictive factors associated with early CBR in this specific patient cohort, utilizing PSMA-PET imaging, clinical parameters, and machine learning methodologies.

Materials/Methods: This retrospective analysis included a cohort of 85 PC patients treated with MRgSBRT on a 1.5T MR-integrated linear accelerator (MR-LINAC), incorporating an IL boost. The treatment protocol consisted of 5-fractionated SBRT delivered every other day, with a total dose of 40 Gy to the clinical-target-volume (CTV) and concurrent delivery of 42.5 Gy to ILs classified as PIRADS 4/5. Androgen deprivation therapy (ADT) was prescribed at the discretion of the clinician. Intraprostatic SUVmax values were extracted from pre-treatment PSMA-PET/CT. Patient demographics, clinical staging, Gleason scores, risk levels, baseline PSA, PIRADS score and treatment factors were collected. Early CBR was defined as a PSA level of =0.1 ng/mL measured within two months post-treatment. A Gradient Boosting Machine (GBM) model was constructed to predict the likelihood of early CBR. The model's performance was validated using receiver operating characteristic (ROC) and area under the curve (AUC) analysis to assess predictive validity.

Results: A total of 85 patients (mean age: 73.41 ± 8.47 years) were included. The median interval between the MRgSBRT completion date and the early post-treatment PSA measurement was 27 days. Early CBR was achieved in 37.65% (32/85) of the patients. Statistically significant correlations were found between SUVmax and early PSA response (r=0.38, 95%CI 0.18-0.54, p<0.001). The GBM model demonstrated promising results, with an AUC of 0.83 (95%CI 0.72-0.86), indicating good predictive capability, with the relative influence of each predictor to the model's overall performance ranked as follows: SUVmax, T-stage, pre-treatment PSA, age, Gleason score, D’Amico risk level, and PIRADS.

Conclusion: This study demonstrates the potential of clinical characteristics, PSMA-PET imaging and machine learning as predictive tools for early CBR in patients undergoing MRgSBRT with IL boost. Future research should focus on validating these findings in larger studies and integrating additional biomarkers to enhance predictive models.