Biomarker screening algorithms for the improved early detection of hepatocellular carcinoma

项目总结/摘要 美国肝细胞癌（HCC）的发病率持续上升， 患者被诊断为晚期疾病，治疗选择有限且预后不良。HCC是 预计到2030年将成为癌症相关死亡的第三大原因。早期发现HCC是 对于降低高HCC死亡率是必要的，因为那些患有早期疾病的人有多种， 潜在的治愈性的治疗选择。目前的指南建议肝硬化患者接受 每6个月进行一次肝脏超声检查，有或无血清甲胎蛋白（AFP），但超声检查不敏感 对于早期病变，AFP的报告表现差异很大。 我们将开发和评估两种新的生物标志物筛选算法，旨在提高早期检测 的HCC。我们之前已经提出了AFP的参数经验贝叶斯（PEB）筛选算法， 通过纳入先前AFP结果的个性化阈值，增加早期HCC检测。血- 生物标志物是广泛监测HCC的最有前途、最具成本效益的工具， 正在开发多种新的HCC生物标志物。在目标1中，我们将推广PEB算法， 与多种生物标志物（例如AFP、DCP、AFP-L3、有前途的新型生物标志物）联合筛查。我们建议 为多种HCC生物标志物制定一个稳健的决策规则，该规则使用既往筛查史， 丙型肝炎抗病毒长期治疗肝硬化试验中的HCC检测。第二筛选 该战略是基于这样的观察，即积极监测下的患者不断更新临床资料， 和实验室数据收集，但没有系统地用于改善筛选。在目标2中，我们将开发和 评估一个完全贝叶斯筛查算法，结合纵向AFP，其他实验室标志物， 临床协变量，以增加早期检测HCC的可能性。我们的目标是改善AFP筛查 通过为AFP、其他实验室检查和临床数据稳健开发联合模型，提高性能。 一旦确认，该算法可以根据当前的临床实践实施。创新和 在两个大型回顾性队列中完善算法：退伍军人事务部全国肝硬化 队列，Kaiser Permanente北方加州肝硬化队列。在目标3中，我们将评估这两种算法， 肝细胞癌早期检测策略（HEDS）研究和跨德克萨斯州HCC联盟 （THCCC）;迄今为止在美国组装的最大的前瞻性肝硬化队列。我们将利用 我们获得了一些最权威的肝硬化研究来建立和评估HCC筛查算法， 目标是将HCC筛查的灵敏度提高33%，同时保持低假阳性率 确保HCC监测的可行性。此外，开发的统计方法将具有广泛的 在其他癌症筛查环境（例如肺癌、卵巢癌、前列腺癌和胰腺癌）中的应用。 1

PROJECT SUMMARY/ABSTRACT The incidence of hepatocellular carcinoma (HCC) in the United States continues to rise with majority of patients diagnosed with advanced stage disease, limited treatment options and poor prognosis. HCC is projected to become the 3rd leading cause of cancer-related deaths by 2030. The earlier detection of HCC is necessary towards reducing the high HCC mortality rates since those with early stage disease have multiple, potentially curative, treatment options available. Current guidelines recommend those with cirrhosis undergo six-monthly liver ultrasound with or without serum alpha-fetoprotein (AFP), however ultrasound is not sensitive for early lesions and the reported performance of AFP varies widely. We will develop and evaluate two novel biomarker screening algorithms that aim to improve the early detection of HCC. We have previously proposed a parametric empirical Bayes (PEB) screening algorithm for AFP that increased earlier HCC detection through personalized thresholds that incorporate prior AFP results. Blood- based biomarkers are a most promising, cost-effective tool for widespread HCC surveillance and there are multiple novel HCC biomarkers under development. In Aim 1 we will generalize the PEB algorithm to enable joint screening with multiple biomarkers (e.g. AFP, DCP, AFP-L3, promising novel biomarkers). We propose to develop a robust decision rule for multiple HCC biomarkers that uses prior screening history to increase earlier HCC detection in the Hepatitis C Antiviral Long-term Treatment against Cirrhosis Trial. A second screening strategy is based on the observation that patients under active surveillance have continuously updated clinical and laboratory data collected but not used systematically to improve screening. In Aim 2 we will develop and evaluate a fully Bayesian screening algorithm that combines longitudinal AFP, other laboratory markers and clinical covariates to increase the likelihood of earlier detection of HCC. Our goal is to improve AFP screening performance through the robust development of joint models for AFP, other laboratory tests and clinical data. Once validated, this algorithm could be implemented based on current clinical practice. We will develop and refine the algorithm in two large retrospective cohorts: a Department of Veterans Affairs national cirrhosis cohort, a Kaiser Permanente Northern California cirrhosis cohort. In Aim 3 we will evaluate both algorithms in the Hepatocellular carcinoma Early Detection Strategy (HEDS) study and the Trans Texas HCC Consortium (THCCC); the largest prospective cirrhosis cohorts assembled in the United States to date. We will leverage our access to some of the most authoritative cirrhosis studies to build and evaluate HCC screening algorithms, with a target of increasing the sensitivity of HCC screening by 33% while maintaining a low false positive rate to ensure the feasibility of HCC surveillance. Additionally, the statistical methods developed will have broad application in other cancer screening settings (e.g. lung, ovarian, prostate, and pancreatic cancer). 1