EXOSOMAL PROTEIN PROFILING FOR PREDICTING THE PROGNOSIS OF INTRADUCTAL PAPILLARY MUCINOUS NEOPLASMS

Background and Aims: Intraductal papillary mucinous neoplasm (IPMN) is one of the well-known precancerous lesions. Predicting the malignant risk of IPMN is important but challenging. This study aimed to find out key exosomal proteins that distinguish IPMN-associated carcinoma (IPMC) from IPMN with low risk (IPMN-LR).
Methods: Circulating exosomes in plasma from patients were purified using a multi-dimensional technology comprised of metal-based exosome precipitation and size-exclusion chromatography. After isolating exosomes in high purity, we performed a three-step analysis. In the discovery session, we performed protein profiling for enrolled 20 patients (6 IPMC and 14 IPMN-LR). In the verification session, we enrolled 14 patients (7 IPMC and 7 IPMN-LR) and performed Parallel Reaction Monitoring-Mass Spectrometry (PRM-MS) analysis to verify the biomarkers for IPMC. In the validation session, 24 additional patients (12 IPMC and 12 IPMN-LR) were enrolled independently, and PRM-MS was performed with blinding patients' information. To elucidate the functional relevance of validated proteins, lentiviral knockdown and overexpression systems were established in ASAN-PaCa IPMC cell lines. We performed cell titer glo, transwell, and three-dimensional (3D) dissemination assays for the validated proteins to investigate their biological functions on cell proliferation, migration, and invasion in the ASAN-PaCa cell lines.
Results: In the discovery session, we identified 75 potential IPMC biomarker candidates through exosomal protein profiling. Through the PRM-MS in the verification session and a pancreatic cancer survival rate analysis, we selected PDLIM7, LYPLA1, PDIA6, and HK1 as the biomarker candidates to be analyzed in the subsequent validation session. In the independent validation cohort, only PDLIM7 and LYPLA1 showed significantly different expression. The predictive performance of both candidates was remarkable with a high value of area under the curve (PDLIM7: 0.875, LYPLA1: 0.885, combination of both candidates: 0.944). PDLIM7 and LYPLA1 showed excellent predictive performance even in patients with normal levels of serum carbohydrate antigen 19-9. In an in vitro study, silencing of PDLIM7 and LYPLA1 expression significantly suppressed the cell proliferation rate and impaired the cell invasion ability. Notably, the knockdown of PDLIM7 and LYPLA1 expression resulted in a reduced invasive phenotype and a uniform round shape on 3D spheroids to mimic the in vivo environment. Similarly, PDLIM7- and LYPLA1-overexpressed ASAN-PaCa dramatically promoted cell mobility 3D cancer dissemination.
Conclusions: Through exosome proteomic analysis, PDLIM7 and LYPLA1 appear to be promising biomarkers to predict the malignant potential of IPMN.