Biomarker Validation in Pancreatic Cystic Neoplasms

项目摘要 胰腺囊性肿瘤（PCN）是人群中常见的偶然发现。支管 导管内乳头状粘液性肿瘤（IPMN）是最常见的偶发性PCN， 恶性肿瘤在诊断15年内接近15%。现有准则的执行情况 早期癌症的预后差，导致手术过度治疗和错过早期诊断的机会。 需要有效的筛查生物标志物来准确区分需要密切监测的高危PCN。 从低风险病变进行监测，进展机会很小。 在这项提案中，我们将测试和验证三种新的血液生物标志物和一种囊液生物标志物， PCN患者早期PDAC的检测。建议的标记，在学术和 在初步研究中显示出希望，灵敏度和特异性足够高， 进一步评价。我们将结合前瞻性标本采集、回顾性盲法评价 （PRoBE）标准，以严格测试这些生物标志物的性能，样品收集自三个 队列：1）I/II期PDAC和对照组; 2）接受手术切除的PCN患者; 3）接受手术切除的PCN患者。 通过连续成像和样本采集监测PCN ≥ 2.5 cm或主胰管≥ 5 mm。我们 将单独测试生物标志物的性能，并作为多变量模型的一部分， 彼此，添加生殖系检测和临床实验室检查值（CA 19 - 9、HbA 1c）的信息，以及 具有特定PDAC风险因素（吸烟状态、饮酒、糖尿病、胰腺炎病史）。 我们的研究策略的关键组成部分，以解决这一棘手的问题包括：1）严格测试几个 有前途的血液生物标志物，单独和组合，通过独特的合作， 行业和学术合作伙伴; 2）测试用于早期检测的先进囊液分析的新平台 的PDAC和比较其性能的血液为基础的生物标志物; 3）大量的回顾性和 使用PRoBE设计询问的前瞻性样本，具有生物标志物验证的统计学严谨性; 4） 利用已建立的胰腺癌早期检测（PRECEDE）联盟的资源，包括 标准化收集血液和囊液的生殖系基因检测、临床和实验室数据 根据PCDC方案进行生物取样; 6）收集大量去识别的伙伴胰腺癌细胞; 影像（MRI/MRCP、CT和EUS）和数字化病理切片， 使用人工智能和机器学习策略的协作机会，以及7）多模态数据 模型开发集成。纵向生物标本将与PCDC共享，以支持 签名队列和去识别存储图像（MRI、EUS、数字化病理学）将可用于 联合体的协作努力。

Project Summary Pancreatic cystic neoplasm (PCN) represents a common incidental finding in the population. Branch-duct intraductal papillary mucinous neoplasms (IPMN), the most common incidentally discovered PCN, have a risk of malignancy approaching 15%within 15 years of diagnosis. The performance of existing guidelines for identifying early cancer is poor, and results in both surgical overtreatment and missed opportunities for early diagnosis. Effective screening biomarkers are needed to accurately differentiate high-risk PCN that require close surveillance from low risk lesions with little chance to progress. In this proposal, we will test and validate of three novel blood-based biomarkers and one cyst fluid biomarker for the detection of early PDAC in patients with PCN. The proposed markers, developed in both academic and industry settings, show promise in preliminary studies, with sensitivity and specificity sufficiently high to warrant further evaluation. We will incorporate prospective-specimen-collection, retrospective-blinded-evaluation (PRoBE) standards to rigorously test the performance of these biomarkers with samples collected from three cohorts: 1) stage I/II PDAC and controls; 2) Patients with PCN who undergo surgical resection; 3) Patients with PCN ≥ 2.5 cm or main pancreatic duct ≥ 5 mm under surveillance with serial imaging and sample collection. We will test performance of the biomarkers individually, and as part of multi-variable models in combination with each other, with the added information of germline testing and clinical laboratory values (CA19 -9, HbA1c), and with specific PDAC risk factors (smoking status, alcohol consumption, diabetes, pancreatitis history). Key components of our research strategy to tackle this recalcitrant problem include: 1) rigorous testing of several promising blood-based biomarkers, individually and in combination, through a unique collaboration between industry and academic partners; 2) testing of a novel platform for advanced cyst fluid analysis for early detection of PDAC and comparison of its performance to blood-based biomarkers; 3) a large number of retrospective and prospective samples interrogated using the PRoBE design, with statistical rigor for biomarker validation; 4) resources leveraged from the established Pancreatic Cancer Early Detection (PRECEDE) Consortium including standardized collection of germline genetic testing, clinical, and laboratory data with blood and cyst fluid biosampling in accordance with PCDC protocols; 6) collection of a large set of de-identified partnering pancreatic images (MRI/MRCP, CT and EUS) and digitized pathology slides on a funded cloud-based platform for collaborative opportunities using artificial intelligence and machine learning strategies, and 7) multimodal data integration for model development. Longitudinal biospecimens will be shared with the PCDC to support the Signature Cohorts, and de-identified stored images (MRI, EUS, digitized pathology) will be available for collaborative consortium efforts.