Modeling the Dynamics of Human Papillomavirus to Inform Vaccination Strategy

摘要 尽管人乳头瘤病毒 (HPV) 相关癌症给公共卫生带来了负担，但尚不清楚如何 社会和行为特征、不同种类的宿主免疫力以及类型之间的竞争塑造了 HPV 的流行病学。维持 HPV 的个体风险因素可能因 HPV 类型而异， 定义对行为公共卫生干预措施至关重要的高风险亚群。生物相互作用 HPV 类型之间可能是协同、中性或竞争性的，决定了 HPV 疫苗的效果。 消灭疫苗靶向类型可能会导致竞争性非靶向类型的“类型替代”， 在肺炎球菌中观察到的现象。拟议的项目调查了这些方面 通过将机械数学模型拟合到纵向数据来进行 HPV 生态学和流行病学。数据跨度 九年来，每隔六个月对 4,000 多名男性进行了 37 种不同的 HPV 类型采样。数据包含 有关影响病毒动态的患者人口统计和性行为的信息。通过安装机械装置 传播模型，我们将测量不同的行为和健康特征如何影响感染的易感性 并确定可能在人群中维持传播的核心风险群体。此外，我们将推断 HPV 相互作用的程度和时间尺度。我们将测量既往感染对 HPV 动态的影响 推断自然感染中免疫的程度和持续时间。我们还将衡量混合感染的影响 推断类型如何直接交互。我们的方法基于部分马尔可夫链蒙特卡罗方法 观察到的马尔可夫过程隐含地结合了感染的随机、非线性动力学。的 拟议的研究将针对每种 HPV 类型，测量其在不同亚群中的传播性、相互作用 与其他类型的相互作用，以及与宿主免疫的相互作用。该信息将用于联系网络 模拟来预测当前和假设的根除工作的后果并优化 干预策略。

Abstract Despite the public health burden posed by human papillomavirus (HPV) - associated cancers, it is unclear how social and behavioral traits, different kinds of host immunity, and competition between types shape the epidemiology of HPV. The individual-level risk factors that sustain HPV, which may differ between HPV types, define high-risk subpopulations that are key to behavioral public health interventions. Biological interactions between HPV types, which may be synergistic, neutral, or competitive, determine the effect of HPV vaccines. The eradication of vaccine-targeted types could lead to “type replacement” by competing non-targeted types, a phenomenon that has been observed in pneumococcus. The proposed project investigates these aspects of HPV ecology and epidemiology by fitting mechanistic mathematical models to longitudinal data. The data span 37 different HPV types in over 4,000 men sampled across nine years at six-month intervals. The data contain information about patient demographics and sexual practices that affect viral dynamics. By fitting a mechanistic transmission model, we will measure how different behavioral and health traits affect susceptibility to infection and identify core risk groups that may sustain transmission in the population. Furthermore, we will infer the degree and timescale of HPV interactions. We will measure the impact of previous infections on HPV dynamics to infer the extent and duration of immunity in natural infection. We will also measure the impact of coinfections to infer how types interact directly. Our approach, based on Markov Chain Monte Carlo methods for Partially Observed Markov Processes, implicitly incorporates the stochastic, nonlinear dynamics of infection. The proposed research will, for each HPV type, measure its transmissibility in different subpopulations, interactions with other types, and interactions with host immunity. This information will be used in contact network simulations to predict the consequences of current and hypothetical eradication efforts and to optimize intervention strategies.