Prostatype-score驗證於預測亞洲人口中攝護腺癌特異性死亡率與治療結果

馮思中¹、林柏宏¹、Emelie Berglund³、Lidi Xu³、張孜璇¹、謝瑾瑄¹、邵翊紘¹、虞凱傑¹、翁文慧²、莊正鏗¹

¹林口長庚紀念醫院 外科部 泌尿腫瘤科；²國立臺北科技大學 化學工程與生物科技學系暨生化與生醫工程研究所；³Prostatype Genomics AB 瑞典斯德哥爾摩

Validation of the Prostatype-score for Predicting Prostate Cancer-Specific Mortality and Treatment Outcomes in an Asian Population

See-Tong Pang¹、Po-Hung Lin¹、Emelie Berglund³、Lidi Xu³、Tzu-Hsuan Chang¹、Chin-Hsuan Hsieh¹、I-Hung Shao¹、Kai-Jie Yu¹、Wen-Hui Weng²、Cheng-Keng Chuang¹

¹ Division of Urology, Department of Surgery, Chang Gung Memorial Hospital, Linkou Branch, Taiwan. ² Department of Chemical Engineering and Biotechnology and Graduate Institute of Biochemical and Biomedical Engineering, National Taipei University of Technology, Taipei, Taiwan. ³ Prostatype Genomics AB, Stockholm, Sweden

Purpose

Clinical outcomes for prostate cancer (PCa) patients depend on the stage at diagnosis and the treatment decisions taken, which are often based on risk stratification systems. Standard risk stratification guidelines use nomograms combining different clinicopathological parameters, while new scoring systems such as the Prostatype-score (P-score) also include genetic markers. The latter was previously shown to be superior for the prediction of PCa-specific mortality (PCSM). The aim of this study was to validate the predictive performance of the P-score in an Asian population.

Materials and Methods

In this retrospective study, 324 PCa patients diagnosed at Taiwan Chang Gung Memorial Hospital between 2012 and 2017 were screened; 148 patients had valid gene expression data. Of these, 56 had primary metastases at diagnosis. The P-score was calculated based on gene expression in core needle biopsies and clinical data collected from patients’ medical records. The primary endpoint was PCSM. The secondary endpoints were adverse pathology (AP) and biochemical failure.

Results

The P-score outperformed the NCCN risk stratification system in predicting PCSM in both concordance index (C-index) analysis: C-index for the P-score was higher (0.90) than for the NCCN system or individual clinicopathological parameters (range: 0.73-0.77) and ROC analysis: P-score (AUC: 0.88-0.94) was superior to the NCCN score (AUC: 0.66-0.78) up to 4 years inclusive, p<0.005. The P-score independently predicted AP in logistic regression (p=0.003) and ROC analysis (AUC=0.81) and was superior for the prediction of biochemical failure (p=0.027).

Conclusions

The P-score is an effective risk stratification system to support PCa patient management in Asia