利用形態學預測轉移性攝護腺癌治療效果的模型

邵翊紘¹、馮思中¹、張孜璇¹、謝瑾瑄¹、王翔生²、林柏宏¹、虞凱傑¹、莊正鏗¹、甘弘成¹、黃亮鋼¹、吳俊德¹

林口長庚紀念醫院 外科部 ¹泌尿腫瘤科，²解剖病理科

Predictive model for treatment efficacy in metastatic prostate cancer with morphomic factors

I-Hung Shao¹、See-Tong Pang¹、Tzu-Hsuan Chang¹、Chin-Hsuan Hsieh¹、Heng-Siang Wang²、Po-Hung Lin¹、Kai-Jie Yu¹、Cheng-Keng Chuang¹、Hung-Cheng Kan¹、Liang-Kang Huang¹、Chun-Te Wu¹

¹ Division of Urology, Department of Surgery, Chang Gung Memorial Hospital, Linkou Branch, Taiwan, ²Pathology

Introduction and Objectives

Androgen receptor axis-targeted therapies (ARATs) demonstrate efficacy in extending survival among patients diagnosed with metastatic castration-resistant prostate cancer (mCRPC). This study endeavors to ascertain predictive factors influencing the responsiveness of mCRPC patients to ARATs, encompassing both clinical parameters and body composition radiomics.

Materials and Methods

Retrospective data collection on mCRPC patients undergoing ARAT treatment was conducted. Clinical parameters were obtained from chart reviews. An AI system for recognizing body composition on unenhanced CT scans at the L3 level has been developed, and data has been collected. ARAT failure was defined as a persistent rise in serum prostate-specific antigen (PSA) levels exceeding nadir to >2 ng/mL, coupled with radiographic progression. Kaplan–Meier and Cox regression analyses assessed ARAT failure-free and overall survival. Predictive model of nomogram based on significant predictors of ARAT failure–free survival was established.

Results

The study comprised 319 mCRPC patients undergoing ARAT. Multivariate analysis identified clinical factors including age (p=0.034), International Society of Urological Pathology (ISUP) grading (p=0.001), chemotherapy-naïve status (p<0.0001), DeNovo Metastasis (p=0.008), and body composition radiomics such as paraspinous muscle area (p=0.022) and psoas muscle area (p=0.021) as predictors for ARAT failure-free survival. Additionally, clinical factors including ECOG (p=0.0047), duration from hormone-sensitive prostate cancer (HSPC) to CRPC (p<0.0001), and body composition radiomics like subcutaneous adipose tissue area (p=0.026) and muscle density (p=0.018) were significant for overall survival. Cox-regression analysis indicated that paraspinous muscle area (p=0.020) was a significant factor for ARAT-failure survival, while subcutaneous adipose tissue area (p=0.037) and duration from HSPC to CRPC (p<0.001) were significant factors for overall survival. Nomogram analysis integrating clinical parameters and body composition radiomics facilitated predicting ARAT duration exceeding or falling below 1 year.

Conclusion

Lower paraspinous muscle area in mCRPC patients correlated with a shorter ARAT-response period, suggesting a need to prioritize alternative therapeutic agents. Our nomogram predictive model facilitates forecasting the timing of early ARAT failure in mCRPC patients. Notably, patients with lower subcutaneous adipose tissue area or a shorter duration from HSPC to CRPC demonstrated poorer overall survival within the study cohort.