Membranous nephropathy in the era of computational pathology

Abstract Membranous nephropathy (MN) is a leading cause of nephrotic syndrome in adults and its diagnosis has historically relied on kidney biopsy. Recent advances in mass spectrometry have enabled the identification of disease-specific antigens, such as PLA2R, reshaping diagnostic strategies and refining patient classification. In parallel, computational pathology has emerged as a promising field to address challenges inherent to histological diagnosis, particularly in early or late disease stages and in cases of secondary MN with overlapping lesions. Artificial intelligence models trained on histopathological images have demonstrated high accuracy in distinguishing MN from other glomerular disorders, even within complex morphological contexts, while uncertainty-aware approaches enhance reliability. Furthermore, advanced imaging modalities, such as hyperspectral analysis, expand the potential to differentiate MN subtypes at a molecular level, supporting etiological diagnosis. Beyond diagnosis, computational systems can integrate histological, clinical, and serological data to predict prognosis and therapeutic response. These approaches highlight the capacity of computational tools to reduce inter-observer variability, improve reproducibility, and accelerate the translation of pathological findings into prognostic and therapeutic decision-making. Despite these advances, significant barriers remain to their routine application, including dataset representativeness, image heterogeneity, integration with omics platforms, and regulatory constraints. This review discusses how computational pathology, combined with molecular discoveries and novel imaging strategies, is reshaping the diagnostic and prognostic landscape of MN, with the potential to enable precision medicine.