Human knee osteoarthritis patient-specific cartilage-on-a-chip model captures donor differences to stressors and treatments

Knee Osteoarthritis (KOA) is a progressive whole-joint disease without approved disease modifying OA drugs (DMOADs). Effective treatments have been hindered by multiple layers of heterogeneity, including diverse disease etiology and patient-to-patient variability. Here, we present a scalable, KOA patient-derived (PD) cartilage-on-a-chip (CartChip) model integrating end-stage KOA cartilage tissue explants on a microengineered platform that, under mechanical overloading and hyperinflammatory stressors, mimics different KOA etiologies. These stressors drove distinct multivariable model features, including changes in a curated panel of anabolic and catabolic genes, extracellular matrix protein and soluble factors. Exploratory analysis of coordinated model readouts identified stressor-agnostic and -specific KOA disease signatures. Despite using KOA tissue, the model showed improvements to dexamethasone, a symptom-modifying, anti-inflammatory KOA treatment. The model responses to dexamethasone were dependent on both stressor and donor heterogeneity. Exploratory groupings of coordinated model readouts provided proof-of-concept for predicting categories of patient responsiveness to test therapeutics. Annotating patient data provided additional donor-dependent contexts for interpreting model responsiveness. PD-CartChip provides a powerful research platform to potentially surmount the donor and stressor heterogeneity barrier in developing DMOADs.