胰腺导管腺癌（PDAC）是最致命的癌症之一，尤其是在晚期。为了分析潜在预后生物标志物的动态并进一步量化它们与晚期 PDAC 患者总生存 (OS) 的关系，我们在此开发了与纵向子模型集成的参数事件时间 (TTE) 模型。回顾性收集 104 名接受标准化疗患者的数据用于模型开发，另外 54 名患者纳入外部验证。使用非线性混合效应模型，根据肿瘤最长直径 (SLD)、血清白蛋白 (ALB) 和体重 (BW) 之和的时程数据开发纵向子模型。 TTE 模型中进一步分析了模型衍生的指标，包括模型参数和不同时间点的个体预测，以及患者的其他基线信息。采用线性增长-指数收缩模型来描述 SLD 的动态，同时使用逻辑模型来拟合死亡前时间与 ALB 和 BW 的关系。 TTE模型估计了化疗后第9周的ALB和BW变化以及对OS影响最显着的基线CA19-9水平，基于模型的模拟可以为不同预后的患者提供个体生存率预测因素。本研究定量论证了身体状况和基线疾病对晚期 PDAC 患者 OS 的重要性，并强调及时的营养支持有助于改善预后。© 2024。作者，经上海癌症研究所独家许可中国科学院药物研究所、中国药理学会。

Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal cancers especially at advanced stage. In order to analyze the dynamics of potential prognostic biomarkers and further quantify their relationships with the overall survival (OS) of advanced PDAC patients, we herein developed a parametric time-to-event (TTE) model integrated with longitudinal submodels. Data from 104 patients receiving standard chemotherapies were retrospectively collected for model development, and other 54 patients were enrolled as external validation. The longitudinal submodels were developed with the time-course data of sum of longest diameters (SLD) of tumors, serum albumin (ALB) and body weight (BW) using nonlinear mixed effect models. The model-derived metrics including model parameters and individual predictions at different time points were further analyzed in the TTE model, together with other baseline information of patients. A linear growth-exponential shrinkage model was employed to describe the dynamics of SLD, while logistic models were used to fit the relationship of time prior to death with ALB and BW. The TTE model estimated the ALB and BW changes at the 9th week after chemotherapies as well as the baseline CA19-9 level that showed most significant impact on the OS, and the model-based simulations could provide individual survival rate predictions for patients with different prognostic factors. This study quantitatively demonstrates the importance of physical status and baseline disease for the OS of advanced PDAC patients, and highlights that timely nutrition support would be helpful to improve the prognosis.© 2024. The Author(s), under exclusive licence to Shanghai Institute of Materia Medica, Chinese Academy of Sciences and Chinese Pharmacological Society.