Multidimensional Data Integration Identifies Tumor Necrosis Factor Activation in Nephrotic Syndrome: A Model for Precision Nephrology

Abstract

Background: Classification of nephrotic syndrome relies on clinical presentation and descriptive patterns of injury on kidney biopsies. This approach does not reflect underlying disease biology, limiting the ability to predict progression or treatment response. Methods: Systems biology approaches were used to categorize patients with minimal change disease (MCD) and focal segmental glomerulosclerosis (FSGS) based on kidney biopsy tissue transcriptomics across three cohorts and assessed association with clinical outcomes. Patient- level tissue pathway activation scores were generated using differential gene expression. Then, functional enrichment and non-invasive urine biomarker candidates were identified. Biomarkers were validated in kidney organoid models and single nucleus RNA-seq (snRNAseq) from kidney biopsies. Results: Transcriptome-based categorization identified three subgroups of patients with shared molecular signatures across independent North American, European and African cohorts. One subgroup demonstrated worse longterm outcomes (HR 5.2, p = 0.001) which persisted after adjusting for diagnosis and clinical measures (HR 3.8, p = 0.035) at time of biopsy. The molecular profile of this subgroup was largely (48%) driven by tissue necrosis factor (TNF) activation and could be predicted based on levels of TNF pathway urinary biomarkers TIMP-1 and MCP-1 and clinical features (correlation 0.63, p <0.001 for predicted vs observed score). Kidney organoids confirmed TNF-dependent increase in transcript and protein levels of these markers in kidney cells, as did snRNAseq from NEPTUNE biopsy samples. Conclusions: Molecular profiling identified a patient subgroup within nephrotic syndrome with poor outcome and kidney TNF pathway activation. Clinical trials using non-invasive biomarkers of pathway activation to target therapies are currently being evaluated.