The study presents a scalable, patient-derived cartilage-on-a-chip model (PD-CartChip) designed to address key barriers in developing disease-modifying osteoarthritis drugs (DMOADs). The authors note that Knee Osteoarthritis (KOA) is a progressive whole-joint disease lacking approved DMOADs and complicated by multiple layers of heterogeneity, including diverse etiologies and patient-to-patient variability. PD-CartChip aims to integrate clinical tissue and controlled microengineering to reproduce different KOA-relevant stressors and measure donor-specific responses.
PD-CartChip incorporates explanted cartilage tissue derived from end-stage KOA patients onto a microengineered platform. The approach uses intact, patient-derived tissue rather than isolated cells, positioning the model to retain donor-specific biological context. The platform is described as scalable, supporting comparative studies across multiple donors and conditions.
To mimic different KOA etiologies, the platform applies controlled mechanical overloading and hyperinflammatory stressors. These two categories of perturbation are intended to reproduce mechanical and inflammatory drivers of KOA, respectively, allowing the model to probe how different disease inputs shape tissue responses. The abstract emphasizes that these stressors produce distinct downstream changes, reflecting heterogeneity in KOA pathobiology.
Model responses were characterized using a curated panel of readouts spanning molecular and biochemical domains. Reported measures include changes in anabolic and catabolic gene expression, alterations in extracellular matrix (ECM) proteins, and profiles of soluble factors released from tissue. These coordinated readouts create a multivariable dataset that captures tissue-level consequences of applied stressors and potential treatment effects.
Exploratory analyses of coordinated model features revealed both stressor-agnostic and stressor-specific KOA disease signatures. In other words, some patterns of change were common across different insults, while other signatures were unique to mechanical overloading or hyperinflammatory conditions. The detection of both shared and distinct signatures supports the model's capacity to reflect multiple KOA etiologies within a controlled experimental framework.
Despite using tissue from end-stage KOA patients, PD-CartChip demonstrated measurable improvements to dexamethasone, identified here as a symptom-modifying, anti-inflammatory KOA treatment. Importantly, the magnitude and nature of the model's response to dexamethasone depended on both the applied stressor and donor heterogeneity. This donor- and stressor-dependent variation indicates the platform can capture clinically relevant differences in therapeutic responsiveness among patients.
The authors used exploratory groupings of coordinated model readouts to provide proof-of-concept for predicting categories of patient responsiveness to test therapeutics. These groupings integrated multiple readouts to stratify patterns of response, suggesting the PD-CartChip could contribute to preclinical decision-making by identifying subsets of donors likely to respond to specific interventions. The abstract presents this as an initial demonstration rather than a validated predictive tool.
Model readouts were further interpreted using annotated patient data, which supplied additional donor-dependent context. Linking experimental responses to donor information helped interpret heterogeneity in treatment responsiveness and disease signatures. The abstract indicates that such annotation is an important step toward understanding sources of variability in model outcomes.
PD-CartChip is proposed as a powerful research platform with potential to address the donor and stressor heterogeneity that has hindered DMOAD development. By combining patient-derived tissue, controlled mechanical and inflammatory stimuli, and coordinated multivariable readouts, the platform aims to improve mechanistic understanding of KOA and to provide a system for early therapeutic assessment that accounts for patient variability.
The authors disclosed potential competing interests and funding sources. Specific interests and funders were reported in the source; the abstract states these do not compete with the presented work. Funding support included institutional and grant-based awards. No additional experimental details, quantitative outcomes, or specific algorithms for prediction were reported in the abstract.