基于MRI影像组学预测前列腺癌根治术后生化复发风险精准量化研究

Prostate cancer is one of the common urinary tract malignant tumors that impacts elderly Chinese men's health seriously. Radical prostatectomy is the main treatment of patients with localized prostate cancer, but there is still a certain risk to develop biochemical recurrence after radical prostatectomy. In order to individualize patient care, it is important to identify both men at high risk of biochemical recurrence after radical prostatectomy, where early use of radiation therapy or neoadjuvant hormone therapy has a greater benefit, and for those patients at a low risk, who could be spared the potential side effects of therapy. At present, studies have shown that the internationally accepted predictive tools for biochemical recurrence based on CAPRA or D'Amico scoring system after radical prostatectomy are not suitable for the Chinese population. Multi-parametric magnetic resonance imaging (mp-MRI) has proven to be the optimal technique for the detection, treatment exploration, and follow-up examination of prostate cancer. While qualitative analysis of MR images or few quantitative parameters are difficult to characterize the complex prognosis prediction in prostate cancer. Radiomics is an emerging technology in recent years, and can process and analysis large amounts of quantitative imaging features, which may achieve the accurate prediction of the risk of biochemical recurrence after radical prostatectomy. In order to establish a predictive model for biochemical recurrence based on radiomics in men who had undergone radical prostatectomy for prostate cancer, we plan to collect 300 cases (200 training data and 100 validation data) retrospectively and 100 cases prospectively as test set. This project will be a novel way for guiding individualized patient care and treatment.

前列腺癌已经成为严重影响我国老年男性健康的泌尿系恶性肿瘤之一。前列腺癌根治术（RP）是局限性前列腺癌患者的主要治疗手段，但是依然存在一定的治疗后复发风险，早期预测出高危生化复发的病人，在生化复发之前选择合适的时机给予放疗或雄激素阻断等新辅助治疗，可能会使这些病人受益。而对于RP后生化复发低风险的病人，则可以规避新辅助治疗带来的副作用。有研究表明国际公认的预测RP后生化复发的统计预测模型并不适用于中国人群。前期实验提示mp-MRI是前列腺最好的影像检查方式，但定性或少数的定量参数难以表征其复杂的预后预测，影像组学可以对影像信息进行更深层次的挖掘、预测和分析，有望实现RP后生化复发的精准预测。本项目拟回顾性收集300例前列腺癌用于RP后生化复发预测模型的构建和验证，并前瞻性收集100例测试构建模型的泛化能力，为实现RP后个性化诊疗，推进精准医学发展提供新思路。

前列腺癌已经成为严重影响我国老年男性健康的泌尿系恶性肿瘤之一。前列腺癌根治术（RP）是局限性前列腺癌患者的主要治疗手段，但是依然存在一定的治疗后复发风险，早期预测出高危生化复发的病人，在生化复发之前选择合适的时机给予放疗或雄激素阻断等新辅助治疗，可能会使这些病人受益。而对于RP后生化复发低风险的病人，则可以规避新辅助治疗带来的副作用。有研究表明国际公认的预测RP后生化复发的统计预测模型并不适用于中国人群。影像组学可以对影像信息进行更深层次的挖掘、预测和分析，有望实现RP后生化复发的精准预测。本研究中我们回顾性整理收集 420例行前列腺癌根治术患者的 mp-MRI 数据、临床资料、随访信息。 420例患者随机分为训练集（294例）和验证集（126例），同时前瞻性收集100连前列腺癌患者资料作为测试集。我们成功构建前列腺癌根治术后生化复发的影像组学预测模型，构建的影像组学模型预测前列腺癌根治术后生化复发的ROC曲线下面 积为0.97（训练集）、0.92（验证集）、0.86（测试集）。