A novel sensor platform for early detection of pancreatic cancer

该项目旨在开发一种用于胰腺癌早期发现的即时诊断工具。预计未来几年胰腺癌将成为癌症死亡的第二大原因。它大多无症状，目前80%的病例在晚期被诊断出来。晚期发现导致极差的生存预后，诊断后平均生存期约为5个月。一种导致早期诊断的分析方法，如本文所提出的，将对胰腺癌的生存产生深远的积极影响。当体内存在肿瘤时，会产生指示疾病进展的特定蛋白质并出现在血液中。这些蛋白质被称为生物标志物，是疾病的指标。在胰腺癌中，有两种重要的生物标志物：CA 19 -9和CEA。不幸的是，这些生物标志物不能单独用于在一般人群中进行诊断。然而，大量其他蛋白质已被鉴定并与疾病进展有关。对于我们的项目，已经仔细选择了30种指示胰腺癌的生物标志物。目前鉴定关键生物标志物的方法缺乏检测早期胰腺癌所必需的灵敏度和特异性，运行耗时，并且需要熟练的操作人员来确保结果可靠。因此，需要一种新的方法来实现早期胰腺癌的检测。有效的即时诊断将大大减少可预防的病例。在这里，我们建议开发一个集成的传感器平台，使用新型传感器进行指示生物标志物存在的测量。然后，它将利用机器学习方法将这些测量结果与二级数据联合收割机结合起来，以加强诊断。二级数据将包括来自患者病史的“风险”因素，例如患有糖尿病。为了在胰腺癌发病时以非常低的水平检测生物标志物，我们建议设计一种新型的灵敏和选择性的基于晶体管的传感器系统。通常，晶体管通过直接向其沟道施加电信号来操作。这里，传感器系统将基于晶体管阵列，晶体管阵列通过使用放置在其通道上的“适体”被调谐用于检测特定的生物标志物。检测过程依赖于与适体结合的特异性生物标志物，其作用类似于输入信号以改变整体晶体管电特性，其随后可以被测量。为了确保有效和可靠的生物标志物检测，必须以一致的方式构建晶体管传感器，特别是关于加载适体的通道构建，因为这极大地影响操作。为了实现这一目标，将使用3D生物打印机以自动化的方式将受控体积的适体输送到晶体管。研究的实验阶段将从简单到复杂的检测任务系统地进行。第1阶段将专注于创建用于初始表征研究的传感器。在第二阶段，传感器检测已知生物标志物浓度的能力将决定传感器的灵敏度和检测限。最后阶段3将检查处于疾病进展不同阶段的胰腺癌患者以及健康对照的血清样本中的生物标志物检测。传感器数据将与风险生活因素相结合，以训练机器学习系统来检测胰腺癌的存在。最后，我们将评估传感器作为诊断工具的性能，以预测早发性胰腺癌。

This project aims to develop a point-of-care diagnostic tool for the early-onset detection of pancreatic cancer. Pancreatic cancer is predicted to become the second leading cause of cancer-death in the next few years. It is mostly asymptomatic, and currently 80% of cases are diagnosed at an advanced stage. Late detection leads to an extremely poor survival prognosis, with average survival of around 5 months after diagnosis. An analysis methodology that leads to early diagnosis, such as that proposed here, will have a profoundly positive effect on pancreatic cancer survival. When tumours are present in the body, specific proteins, indicative of disease progression, are produced and appear in the blood. These proteins, referred to as biomarkers, are indicators of disease. In pancreatic cancer there are two significant biomarkers: CA19-9 and CEA. Unfortunately, these biomarkers alone cannot be used to make a diagnosis in the general population. However, a large number of other proteins have been identified and linked to disease progression. For our project, 30 biomarkers indicative of pancreatic cancer have been carefully selected. Current methods to identify key biomarkers lack the necessary sensitivity and specificity for the detection of early-stage pancreatic cancer, are time consuming to run, and require skilled operators to ensure results are reliable. Therefore, a new approach is needed to achieve early onset pancreatic cancer detection. Effective point-of-care diagnosis will significantly reduce preventable cases. Here we propose to develop an integrated sensor platform that makes measurements indicating the presence of biomarkers using novel sensors. It will then make use of machine leaning approaches to combine these measurements with secondary data, to enhance diagnosis. The secondary data will include 'risk' factors from patient medical history, such as having diabetes. To detect biomarkers at the very low levels they manifest themselves at pancreatic cancer onset, we propose to design a novel sensitive and selective transistor-based sensor system. Normally transistors are operated by directly applying an electrical signal to their channels. Here the sensor system will be based upon an array of transistors which are tuned for the detection of specific biomarkers by using "aptamers" placed onto their channels. The detection process relies upon on specific biomarkers binding to aptamers, which acts like an input signal to change the overall transistor electrical characteristics, which can be subsequently measured. To ensure effective and reliable biomarker detection, it is essential the transistor sensors are built in a consistent fashion, especially regarding aptamer-loaded channel construction, since this greatly affects operation. To achieve this, a 3D-bioprinter will be used to deliver controlled volumes of aptamers to the transistor in an automated fashion. The experimental phase of the research will systematically progress from simple to complex detection tasks. Phase 1 of will focus on creating the sensors for initial characterisation studies. In the second phase, sensor capacity to detect known biomarker concentrations will determine the sensor sensitivity and detection limits. The final stage 3 will examine biomarker detection in serum samples from patients with pancreatic cancer at varying stages of disease progression, as well as healthy controls. Sensor data will be combined with risk-life factors to train a machine learning system to detect the presence of pancreatic cancer. Finally, we will evaluate sensor performance as a diagnostic tool to predict early-onset pancreatic cancer.