Micro-fragmented adipose tissue reprogrammes the systemic molecular landscape of inflammatory knee osteoarthritis

Abstract

Knee osteoarthritis affects an estimated 374 million people and has no approved disease-modifying treatment. In our double-blind randomised trial (ISRCTN88966184), intra-articular micro-fragmented adipose tissue (MFAT) outperformed hyaluronic acid (HA) on pain, function and cartilage composition -- but clinical endpoints alone cannot say whether the disease changed or only the symptoms. Here we answer that question from the blood. Using an interpretable AI data-fusion framework based on non-negative matrix tri-factorisation, we integrated longitudinal plasma proteomics, N-glycomics and miRNA transcriptomics with patient genetics and prior molecular networks at baseline, one and six months, resolving 93 gene and 30 miRNA pathways, almost all significantly enriched in Gene Ontology Biological Process and Reactome annotations. By six months the two injections had left sharply distinct molecular fingerprints. HA remained dominated by canonical OA pathogenesis, including the cartilage-degrading effectors MMP13 and LIMK2 and persistent markers of synovial inflammation. MFAT instead shifted the systemic landscape towards chondroprotection, anti-inflammatory signalling and bone-cartilage homeostasis, prioritising effectors including SIRT7 and NDUFC1. These are, to our knowledge, the first in vivo human data to explain why a regenerative therapy outperforms viscosupplementation in knee osteoarthritis, and they place MFAT on the disease-modifying rather than the symptomatic side of that divide.