Multiethnic radio genomics reveals low-abundancy microRNA signature in plasma-derived extracellular vesicles for early diagnosis and subtyping of pancreatic cancer

WENJIE SHI,Jianying Xu,Yi Zhu,Chao Zhang,Julia Nagelschmitz,Maximilian Doelling,Sara Al-Madhi,Ujjwal Mukund Mahajan,Maciej Pech,Georg Rose,Roland Siegfried Croner,Guoliang Zheng,Christoph Kahlert,Ulf Dietrich Kahlert

DOI: https://doi.org/10.1101/2024.11.22.24317764

2024-11-29

Abstract:Purpose Currently, there is a lack of effective methods to accurately detect pancreatic cancer. In our study, we develop a liquid biopsy signature of EV miRNAs based on associated radiomics features of patients' tumors in order to provide new insights for the early diagnosis of pancreatic cancer. Experimental Design A total of eight datasets were enrolled in this study, featuring clinical and imaging data from different benign pancreatic lesions and malignant pancreatic cancers as well as small RNAseq data from the cargo of plasma extracellular vesicles of tumor patients. Radiomics feature extraction and different features analysis were performed using limma packages. Feature selection was performed by Boruta algorithms and the radionics-related signature model was build and validated by lasso regression algorithms. Radiomic signature related to low abundance EV miRNAs was analyzed by weighted gene co-expression network analysis. Ten machine-learning algorithms validate the diagnosis ability of the above miRNA. The shared target of candidate miRNAs were predicted and clustered followed by subsequently probing for predicting survival benefit of the patient, drug sensitivity of tumor cells and functional differences. Results A total of 88 significant radiologic features demonstrate differences between benign lesion and pancreatic cancer. Three radionics factor-related signatures related to a plasma EV-miRNAs triplet possess high accuracy in the diagnosis of cancer from benign lesions. Moreover, clustering miRNA and their predicted molecular signaling partners in tumor tissue identified tow molecular subtypes of pancreatic cancer. Cluster stratification separates low-risk tumors in terms of severely prolonged overall survival time of patients, higher sensitivity to immune therapies. We also propose the potential of purposing selected targeted drugs to specifically target the molecular activation markers in high-risk tumor clusters. Conclusion Our three radiogenomics-related blood plasma extracellular vesicle microRNA signature is a useful liquid biopsy tool for early diagnosis and molecular subtyping of pancreatic cancer, which might treatment decision making.