Revolutionizing Renal Histopathology: Integrating Multiplexed Immunofluorescence and ML/AI Image-Based Biomarkers for Precision Diagnosis and Patient Stratification

Harvard Medical School, Boston MA

Investigators

Abstract

The accurate evaluation of kidney biopsy tissues is critical for diagnosing and managing individuals with kidney diseases, which affect hundreds of millions of people globally. However, conventional histopathology techniques have major limitations due to the reliance on manual interpretation and the lack of multi-marker quantitative biomarker analysis, hindering precise diagnosis and treatment stratification. This proposal aims to transform renal diagnostics by developing an advanced imaging and analysis platform based on Orion multiplexed immunofluorescence (IF) imaging of up to 20 antibodies and same-section H&E staining. The innovative platform will generate whole slide scans, allowing for quantitative assessment of protein levels and their spatial co-location and relationships within tissues. By integrating advanced machine learning (ML) and artificial intelligence (AI) methods, we will create a suite of multi-variate image-based biomarkers, that will significantly enhance diagnostic precision and patient stratification. The platform is designed for broad implementation across medical centers and research settings, promising immediate improvements in renal diagnostics and patient management and catalyzing sustained biomarker innovation and implementation by facilitating the collection of digital images and quantitative metrics on a large scale in data repositories. A key element of this approach is the direct involvement of renal pathologists throughout the development process. The reagents, methods, visualization, and analysis tools we create will be tailored to address current clinical challenges, ensuring that renal pathologists become equipped with firsthand experience in advanced quantitative renal imaging. This will foster their active participation in iteratively refining the assays and biomarkers, maximizing clinical benefit and supporting new initiatives in precision medicine clinical trials. We have strategically de-risked this project in several ways: (1) our team has extensive experience with the Orion multiplexing platform, having co-developed it through a successful SBIR grant; (2) we have already developed many of the primary analysis and visualization tools through our participation in the NCI Human Tumor Atlas Network (HTAN), which we will now adapt for renal diagnostics; (3) we have achieved proof-of concept imaging of kidney biopsies demonstrating the feasibility and positive reception of multiplexed kidney imaging among renal pathologists and nephrologists (see LOS). Aim 1 will focus on developing 18-marker multiplexed IF panels and a same section H&E protocol, enabling more comprehensive evaluation of common and rare kidney diseases. We will create automated analysis workflows and standardized methods for protein quantification and data normalization, as well as antibody cycling protocols to expand capabilities and the depth of molecular phenotyping. Aim 2 will develop open-source analysis software for single and multi-marker quantification, and new multiparameter metrics using ML/AI. This software will facilitate the evaluation of complex biomarker data and drive evolution of kidney precision medicine.