Enhanced Diagnosis of Antibody-Mediated Kidney Rejection by Machine Learning and Hybrid Targeted-Shotgun Proteomics

点击翻译按钮获取中文摘要

Kidney transplantation offers the best quality of life for patients with chronic kidney failure. Antibody and T-cell mediated rejection (ABMR and TCMR) are key factors that determine graft survival. Currently, the diagnosis and differential diagnosis of rejection relies on histopathologic examination which has known limitations such as subjective interpretations, limited reproducibility, and the need for expert transplant pathologists. There is an unmet need to develop more specific and quantitative molecular tests that can complement and enhance conventional histologic assessment. Among the molecular assays, proteome profiling is more attractive than genomic and transcriptomic profiling which are subjected to numerous post-translational and epigenetic regulatory mechanisms. Moreover, morphologic changes form the basis of classifying different allograft diseases. Therefore, the transplant community needs to invest in biopsy-based assays in addition to the blood/urine-based assays that are being developed by others. This study is aimed to fully map the proteomic changes in routinely processed formalin fixed paraffin embedded (FFPE) biopsies using a liquid chromatography–tandem mass spectrometry (LC-MS/MS) platform. To meet the needs of personalized medicine, this platform uses a novel strategy and machine learning to simultaneously measure the absolute expression levels of a panel of targeted biomarkers as well as thousands of untargeted proteins. The central hypothesis is that LC-MS/MS can be used to define disease-specific biomarkers using a discovery data set, which can then be followed up by a validation data set to determine if LC-MS/MS based tests can be implemented in clinical practice. In the current project, we will focus on developing molecular assays for ABMR since antibody contributes to graft loss in 60% of patients. Two Aims are proposed. In Aim #1, quantitative proteomic strategies will be used to map proteome-level changes in a discovery set of biopsies with ABMR and its mimics, such as acute tubular injury (ATI), TCMR, BK virus nephropathy (BKVN), interstitial fibrosis/tubular atrophy (IFTA), and stable renal function (STA). The goal is to identify potential protein biomarkers that can distinguish ABMR from its mimics. In Aim #2, the potential ABMR biomarkers obtained in Aim #1 will be validated and optimized in an independent validation data set. Using a hybrid proteomic platform combing targeted, shotgun proteomics and machine learning, information on absolute quantitation of potential protein biomarkers and thousands of other proteins will be collected to build a kidney transplant Protein Atlas for assay development. Successful completion of this study has great potential to be translated into clinical tests that will enhance the diagnosis of ABMR from other diseases that can mimic that pathology. It will also serve as a model for developing a new generation of clincal diagnositc tests that will use routinely fixed biopsy materials. This will eliminate the need for intense biobanking efforts, which have hampered the more widespread implementation of molecular diagnostics into the transplant clinic.