Branching Stochastic Process Modeling of Human Plasma Cell Differentiation

描述（由申请人提供）：体外B细胞活化和浆细胞分化系统的定量分析可用于估计免疫反应所需的B细胞行为参数，并建议操纵B细胞免疫反应的策略。抗体介导的肾移植排斥反应涉及B细胞的活化，B细胞产生浆细胞，浆细胞分泌针对移植细胞表面标记物的抗体。虽然多种治疗药物可以影响B细胞的增殖、死亡和分化，但我们目前缺乏可靠的定量模型来指导临床试验的设计。这项工作的总体目标是开发这样一个框架。我们的第一个目标是建立和实验验证B细胞增殖和终末浆细胞分化的多类型Bellman- Harris分支过程。目前的B细胞分化模型要么是基于连续群体的常微分方程（ODE）模型，要么是实验观察的非定量图形表示。我们提出了一种替代方法，一种多类型分支随机过程与离散事件相结合，基于智能体的单个虚拟B细胞的硅建模。模型参数将被估计，模型将通过Aim 2中描述的新的统计方法进行验证，该方法使用我们开发的新型体外浆细胞分化系统的数据。我们的第二个目标是开发和实验验证从CFSE标记实验中估计淋巴细胞动力学参数的严格统计方法。我们将为CFSE数据的定量分析开发新的统计方法（估计、检验、拟合优度）。与目前可用的方法相比，所提出的方法将使估计细胞活化、增殖、分化和死亡的参数成为可能。我们将研究它们的理论性质（如估计量的一致性和渐近正态性），提供所提出的估计量的方差-协方差矩阵的表达式，并在广泛的模拟研究中评估它们的有限样本性能。我们的第三个目标是模拟阻碍GO->G1进展（西罗莫司）、延迟S/M转变（霉酚酸酯）或阻止细胞周期退出（IL-6拮抗剂）对浆细胞生成的影响，并进行体外模型验证。我们将使用一种新的体外培养系统来驱动人类B细胞从激活到终末浆细胞分化。使用Aim 2中开发的技术，我们将在西罗莫司、霉酚酸酯和IL-6阻断抗体存在的情况下，估计每个B细胞类别在分化过程中的动力学参数。Aim 1中开发的随机分支过程模型将用于预测实验结果并将其与体外数据进行比较。

DESCRIPTION (provided by applicant): Quantitative analysis of in vitro systems of B cell activation and plasma cell differentiation can be used to estimate parameters of B cell behavior necessary for immune responses, and to suggest strategies for manipulation of B cell immune responses. Antibody mediated rejection of kidney transplants involves activation of B cells, which produce plasma cells that secrete antibodies against cell surface markers present in the transplant. While a variety of therapeutic agents can affect B cell proliferation, death, and differentiation, we currently lack robust quantitative models to guide the design of clinical trials. The overall goal of this work is to develop such a framework. Our first aim is to develop and experimentally validate a multi-type Bellman- Harris branching process of B cell proliferation and terminal plasma cell differentiation. Current models of B cell differentiation are either continuous population based ordinary differential equation (ODE) models or non-quantitative graphical representations of experimental observations. We propose an alternative approach, a multi-type branching stochastic process coupled with discrete event, agent based modeling of individual virtual B cells in silico. Model parameters will be estimated and the model validated with new statistical methods described in Aim 2 using data from a novel in vitro plasma cell differentiation system we have developed. Our second aim is to develop and experimentally validate rigorous statistical methods for lymphocyte kinetic parameter estimation from CFSE labeling experiments. We will develop novel statistical methods (estimation, test, goodness-of-fit) for the quantitative analysis of CFSE data. In contrast to currently available approaches, the proposed methods will make it feasible to estimate parameters of cell activation, proliferation, differentiation and death. We will study their theoretical properties (such as consistency and asymptotic normality of estimators), provide expressions for the variance-covariance matrix of the proposed estimators, and evaluate their finite sample performance in extensive simulation studies. Our third aim is o model the effects of agents that impede the GO->G1 progression (sirolimus), delay S/M transition (mycophenolate mofetil), or prevent cell cycle exit (IL-6 antagonists) on plasma cell generation and perform in vitro model validation. We will use a novel in vitro culture system to drive human B cells from activation to terminal plasma cell differentiation. Using the techniques developed in Aim 2, we will estimate kinetic parameters for each B cell class during differentiation in the presence of sirolimus, mycophenolate mofetil, and IL-6 blocking antibodies. The stochastic branching process model developed in Aim 1 will be used to predict experimental outcomes and compare them with in vitro data.