Peng Feng, Hu Dan, Lin Xiandong, Chen Gang, Liang Binying, Chen Ying, Li Chao, Zhang Hejun, Xia Yan, Lin Jinxiu, Zheng Xiongwei, Niu Wenquan
Department of Cardiology, The First Affiliated Hospital of Fujian Medical University, Fuzhou, Fujian, China.
Department of Pathology, Fujian Provincial Cancer Hospital, The Affiliated Hospital of Fujian Medical University, Fuzhou, Fujian, China.
Cancer Epidemiol. 2018 Feb;52:148-157. doi: 10.1016/j.canep.2018.01.001. Epub 2018 Jan 8.
Dyslipidaemia is key to colorectal carcinogenesis, and the prediction of baseline triglycerides (TG), total cholesterol (TC), high- and low-density lipoprotein cholesterol (HDLC and LDLC) for postsurgical colorectal cancer mortality has not been researched.
We attempted to re-analyse the FIESTA database to assess the prognostic value of three informative lipid derivatives - AI (atherogenic index: (TC - HDLC)/HDLC), THR (TG/HDLC) and LHR (LDLC/HDLC) in predicting colorectal cancer mortality.
Based on the FIESTA database, 1318 patients received radical resection from 2000 to 2008, with the latest follow-up completed in December 2015. Median follow-up time was 58.6 months.
Total 1318 patients were randomly evenly divided into the derivation and validation groups. Overall, baseline AI and LHR were associated with the significantly increased risk of colorectal cancer mortality in both derivation (hazard ratio (HR): 1.41 and 1.35, respectively) and validation (HR: 1.37 and 1.32, respectively) groups (all P < 0.001). The predictive performance of AI and LHR was remarkably enhanced in patients with female gender, former/current smoking, colon cancer, early stage, positive vein tumor embolus, normal weight, preoperative hypertension or diabetes comorbidities. Calibration/discrimination analyses revealed that adding AI or LHR to the traditional model had a better fit in both groups. A prognostic nomogram was finally constructed with good predictive accuracy and discriminative capability (C-index = 0.814, P < 0.001).
We consolidated the prognostic superiority of AI and LHR in predicting colorectal cancer mortality over TNM stage.
血脂异常是结直肠癌发生的关键因素，而术前甘油三酯（TG）、总胆固醇（TC）、高密度脂蛋白胆固醇（HDLC）和低密度脂蛋白胆固醇（LDLC）对结直肠癌术后死亡率的预测尚未得到研究。
我们试图重新分析FIESTA数据库，以评估三种信息性脂质衍生物——动脉粥样硬化指数（AI：（TC - HDLC）/HDLC）、甘油三酯与高密度脂蛋白胆固醇比值（THR：TG/HDLC）和低密度脂蛋白胆固醇与高密度脂蛋白胆固醇比值（LHR：LDLC/HDLC）在预测结直肠癌死亡率方面的预后价值。
基于FIESTA数据库，2000年至2008年期间1318例患者接受了根治性切除术，最新随访于2015年12月完成。中位随访时间为58.6个月。
总共1318例患者被随机平均分为推导组和验证组。总体而言，基线AI和LHR与推导组（风险比（HR）分别为1.41和1.35）和验证组（HR分别为1.37和1.32）中结直肠癌死亡风险的显著增加相关（所有P < 0.001）。在女性、既往/当前吸烟者、结肠癌、早期、静脉肿瘤栓子阳性、体重正常、术前高血压或糖尿病合并症患者中，AI和LHR的预测性能显著增强。校准/鉴别分析显示，在传统模型中加入AI或LHR在两组中拟合度更好。最终构建了一个预后列线图，具有良好的预测准确性和鉴别能力（C指数 = 0.814，P < 0.001）。
我们证实了AI和LHR在预测结直肠癌死亡率方面优于TNM分期的预后优势。
