[论文解读] Personalized Mathematical Model Predicting Endotoxin-Induced Inflammatory Responses in Young Men

Renee Brady‐Nicholls, Dennis O. Frank‐Ito+7 more|arXiv (Cornell University)|Sep 6, 2016

Immune Response and Inflammation参考文献 29被引用 3

一句话总结

本研究基于20名健康年轻男性受试者在低剂量脂多糖（LPS）暴露下的实验数据，开发了一种个性化数学模型，用以预测内毒素诱导的炎症反应。通过将模型校准至个体的细胞因子动力学（IL-6、TNF-α、CXCL8 和 IL-10）在8小时内的变化，该模型准确捕捉了个体间差异，并识别出异常的炎症反应，为评估脓毒症风险或手术易感性提供了一项定量工具。

ABSTRACT

The initial reaction of the body to pathogenic microbial infection or severe tissue trauma is an acute inflammatory response. The magnitude of such a response is of critical importance, since an uncontrolled response can cause further tissue damage, sepsis, and ultimately death, while an insufficient response can result in inadequate clearance of pathogens. A normal inflammatory response helps to annihilate threats posed by microbial pathogenic ligands, such as endotoxins, and thus, restore the body to a healthy state. Using a personalized mathematical model, comprehension and a detailed description of the interactions between pro- and anti-inflammatory cytokines can provide important insight in the evaluation of a patient with sepsis or a susceptible patient in surgery. Our model is calibrated to experimental data obtained from experiments measuring pro-inflammatory cytokines (interleukin-6 (IL-6), tumor necrosis factor (TNF-), and chemokine ligand-8 (CXCL8)) and the anti-inflammatory cytokine interleukin-10 (IL-10) over 8 hours in 20 healthy young male subjects, given a low dose intravenous injection of lipopolysaccharide (LPS), resulting in endotoxin-stimulated inflammation. Through the calibration process, we created a personalized mathematical model that can accurately determine individual differences between subjects, as well as identify those who showed an abnormal response.

研究动机与目标

• 开发一种个性化数学模型，以捕捉个体在内毒素挑战下炎症反应的变异特征。
• 理解年轻男性中促炎性（IL-6、TNF-α、CXCL8）与抗炎性（IL-10）细胞因子之间的动态相互作用。
• 基于定量建模识别具有异常炎症反应的受试者。
• 为临床环境中评估脓毒症易感性或手术结局提供系统级工具。

提出的方法

• 使用常微分方程组（ODEs）对LPS挑战后促炎性和抗炎性细胞因子的时间动力学进行建模。
• 将模型参数校准至20名健康年轻男性受试者接受静脉注射LPS的实验数据。
• 模型整合了细胞因子之间的反馈调节，包括IL-10介导的负反馈。
• 通过将模型拟合至个体受试者数据实现个性化，从而量化个体间差异。
• 采用敏感性分析和参数不确定性量化方法验证模型的鲁棒性及个体预测的可靠性。

实验结果

研究问题

• RQ1 在年轻男性中，促炎性和抗炎性细胞因子水平在LPS挑战后如何动态相互作用？
• RQ2 个性化数学模型在多大程度上能够捕捉炎症反应的个体间差异？
• RQ3 该模型能否基于细胞因子动力学识别出具有异常炎症反应的受试者？
• RQ4 在健康个体中，调控炎症消退的关键调节机制是什么？

主要发现

• 个性化模型在LPS注射后8小时内，准确再现了所有20名受试者观察到的细胞因子谱（IL-6、TNF-α、CXCL8、IL-10）。
• 该模型成功识别出具有异常炎症反应的受试者，包括细胞因子产生延迟或过度的个体。
• 通过个体特异性参数估计，定量捕捉了细胞因子动力学的个体间变异。
• 模型揭示IL-10介导的负反馈在大多数受试者中对炎症的消退起着关键作用。