3260 - Preoperative Intraprostatic Tumor Volume Is Prognostic for Metastasis after Early Salvage Post-Prostatectomy Radiation Therapy

J. Parisi¹, C. Belant¹, G. Pratt², M. Rowan², J. E. Leeman³, K. N. Lee⁴, L. K. Lee⁵, D. D. Yang⁴, P. L. Nguyen⁶, P. F. Orio III⁷, A. V. DAmico⁶, and M. T. King⁶; ¹Dana-Farber Cancer Institute/Brigham and Womens Hospital, Boston, MA, ²Brigham and Womens Hospital/Dana-Farber Cancer Institute, Boston, MA, ³Department of Radiation Oncology, Dana-Farber Cancer Institute/Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, ⁴Harvard Radiation Oncology Program, Boston, MA, ⁵Department of Radiology, Brigham and Womens Hospital, Boston, MA, ⁶Department of Radiation Oncology, Brigham and Women’s Hospital/Dana-Farber Cancer Institute, Boston, MA, ⁷Department of Radiation Oncology, Dana-Farber Cancer Institute, Boston, MA

Purpose/Objective(s): It is difficult to prognosticate outcomes after salvage radiation therapy (RT) and androgen deprivation therapy (ADT) due to initiation of early salvage treatment at PSA levels <= 0.2 ng/mL. Current scans like PSMA PET are not sensitive at such low PSA levels. Our purpose is to determine if pre-radical prostatectomy (RP) intraprostatic tumor volume from magnetic resonance imaging (MRI) is prognostic for oncologic outcomes after post-RP RT. Materials/

Methods: This is a single institutional retrospective analysis of 124 patients who underwent pre-RP MRI, RP, and salvage RT for persistent PSA or biochemical recurrence (PSA rise >= 0.1 ng/mL). Patients who underwent neoadjuvant ADT prior to RP, adjuvant ADT/RT, or had pathologic node positive disease were excluded. Pre-RP tumor volume (V_AI) was estimated using nnUNet, a deep learning algorithm trained to segment intraprostatic tumor from 288 patients who received upfront RT. We analyzed the association of V_AI with metastasis using univariable (UVA) and multivariable (MVA) Cox regression. Time 0 was set at salvage treatment initiation. Clinical variables included age, pre-salvage PSA, pathologic stage and Gleason grade, interval from RP, ADT duration, RT fields, margin status, and imaging era (2010-2013: Signa MRI vs 2013-2017: DISCOVERY MRI). Significant variables (p <= 0.05) were included in MVA. The analysis was repeated for the subset of patients with a pre-salvage PSA <= 0.2 ng/mL.
Results: Of 124 patients, 29 metastases were observed over a median follow-up 5.6 years. The median pre-salvage PSA was 0.16 (IQR: 0.12, 0.34). The numbers of patients who received no ADT, 4-6 months, 7-18 months, and 24-30 months were 15, 100, 6, and 3. RT fields included the prostate bed (90) and pelvic lymph nodes (34). Factors associated with metastasis on UVA included post-RP PSA (1.26 (1.04-1.53, p=0.019)), pathologic Gleason grade (1.74 (1.21-2.51, p=0.003)), pT3b disease (4.25 (1.76-10.26, p=0.001)), V_AI (1.26 (1.13-1.40, p<0.001)), interval from RP (0.71 (0.52-0.98, p=0.034)), post-RP ADT duration (1.06 (1.00-1.12, p=0.039)), and treatment era (3.42 (1.33-8.80, p=0.011)). The only significant factor on MVA was V_AI (1.25 (1.07-1.46, p=0.006)). Risk of 5-year metastasis for V_AI < 0.5 mL, 0.5-1.9 cm, and >=2.0 cc were 2.9 [0.0, 8.3], 14.1 [1.3, 26.9], and 26.0 [11.0, 41.1]. The time-dependent ROC of V_AI for 5-year metastasis was 73.8 [59.3, 88.2]. For the subset of 78 patients (17 events) with a pre-salvage PSA <= 0.2 ng/mL, pT3b disease (3.68 (1.19-11.43, p=0.024)) and V_AI (1.25 (1.06-1.48, p=0.008)) were associated with metastases on UVA. V_AI (1.23 (1.01-1.50, p=0.036)) was the only significant variable on MVA.
Conclusion: Pre-RP intraprostatic tumor size was independently associated with metastasis after post-RP RT, even for PSA values below PSMA PET detectability. This variable may be associated with residual microscopic disease after RP.