基于SEER数据库脊索瘤临床预测模型的建立及验证

目的 构建脊索瘤患者的预测模型并进行验证.方法 从SEER数据库(2004～2015年)中鉴定和收集597例脊索瘤患者.Nomogram是基于建模组420例拥有完整数据的患者建立的.C指数(C-index)和校正曲线确定Nomogram的预测精度和判别能力.结果 建立了基于年龄、种族、原发部位及数量、肿瘤分期(TNM)...

Establishment and validation of a clinical prediction model for chordoma based on SEER database

Objective To construct and validate a prediction model of chordoma.Methods A total of 597 patients with chordoma were identified and collected from the SEER database (2004-2015).Nomogram is based on 420 patients with complete data.Prediction accuracy and discriminability of nomogram were determined by C-index and correction curve.Results Prognostic models based on age,race,primary site and number,tumor stage (TNM),surgical method,radiotherapy,tumor metastasis and tumor size were established with a C-index of 0.778.The calibration curve determining the probability of survival showed that nomogram’s predictions were in good agreement with the actual observations.Age > 60 year (P < 0.001,HR 5.723,95% CI 1.988-16.474),M1 (P < 0.001,HR 4.121,95% CI 1.834-9.257),surgical method (total excision,P < 0.01,HR 0.416,95% CI 0.236-0.732),and radical extended resection (P < 0.0001,HR 0.251,95% CI 0.143-0.442) were independent prognostic factors.Conclusions Nomogram provides a more accurate prognosis for patients with spinal chordoma.Age > 60,distant metastasis to M1 stage and absence of surgery are major independent risk factors shortening the survival time of chordoma patients.