Fan Shanshan, Zhao Kexin, Liang Ziwei, Ge Yang
Department of Oncology, Beijing Chao-Yang Hospital, Capital Medical University, Beijing, China.
The Third Clinical School of Medicine, Capital Medical University, Beijing, China.
Ann Med. 2025 Dec;57(1):2555520. doi: 10.1080/07853890.2025.2555520. Epub 2025 Sep 3.
Biliary tract cancer (BTC) is a digestive tract tumor with low incidence, high malignancy, and short survival times. Abnormal lipid metabolism may be related to the occurrence and development of tumors; therefore, we constructed a survival prediction model for patients with BTC using clinical data that included lipid indicators rarely considered in previous studies.
Clinical and pathological data were collected from 124 patients with BTC. Patients were divided into two groups according to the inclusion time. The training and validation cohorts included 70 patients from 2017 to 2021 and 54 patients from 2022 to 2023. Least absolute shrinkage and selection operator-Cox regression analysis was conducted on the survival data. The resulting prediction model was evaluated using discrimination and calibration analyses performed in R Studio.
Tumor location, lipoprotein (a), carcinoembryonic antigen, carbohydrate antigen 19-9, and therapy type were identified as key predictors for constructing the nomogram. The consistency indexes for the training and validation cohorts were 0.677 and 0.655, respectively, indicating moderate discrimination. The Hosmer-Lemeshow test provided a -value of 0.188 for the validation cohort, suggesting a good model fit. The calibration accuracy of the model in the two cohorts was further evaluated by drawing calibration curves based on the follow-up time. Patients were classified into high- and low-risk groups according to the nomogram risk scores. Kaplan-Meier survival curves showed significant differences between the training cohort ( = 0.00041) and the validation cohort ( = 0.0028). The risk score scatter plot provided visual verification of the model's performance.
The predictive model constructed in this retrospective study shows potential for guiding the clinical identification of groups at high risk of BTC, adjusting treatment intensity, and improving follow-up management.
胆管癌（BTC）是一种发病率低、恶性程度高且生存时间短的消化道肿瘤。脂质代谢异常可能与肿瘤的发生发展有关；因此，我们利用临床数据构建了一个BTC患者生存预测模型，该临床数据包含了以往研究中很少考虑的脂质指标。
收集了124例BTC患者的临床和病理数据。根据纳入时间将患者分为两组。训练队列和验证队列分别包括2017年至2021年的70例患者和2022年至2023年的54例患者。对生存数据进行最小绝对收缩和选择算子-Cox回归分析。使用R Studio中进行的鉴别和校准分析对所得预测模型进行评估。
肿瘤位置、脂蛋白（a）、癌胚抗原、糖类抗原19-9和治疗类型被确定为构建列线图的关键预测因素。训练队列和验证队列的一致性指数分别为0.677和0.655，表明鉴别能力中等。Hosmer-Lemeshow检验为验证队列提供的P值为0.188，表明模型拟合良好。通过根据随访时间绘制校准曲线，进一步评估了模型在两个队列中的校准准确性。根据列线图风险评分将患者分为高风险组和低风险组。Kaplan-Meier生存曲线显示训练队列（P = 0.00041）和验证队列（P = 0.0028）之间存在显著差异。风险评分散点图直观地验证了模型的性能。
本回顾性研究构建的预测模型在指导BTC高危人群的临床识别、调整治疗强度和改善随访管理方面显示出潜力。
