AUTHOR=Soheili Marzieh , Gilzad Kohan Hamed 
  
TITLE=A mechanistic quantitative systems pharmacology framework for glucosamine sulfate in knee osteoarthritis: linking exposure, cartilage biology, and clinical outcomes
  
JOURNAL=Journal of Pharmacy & Pharmaceutical Sciences
  
VOLUME=Volume 29 - 2026
  
YEAR=2026
  
URL=https://www.frontierspartnerships.org/journals/journal-of-pharmacy-pharmaceutical-sciences/articles/10.3389/jpps.2026.16471
  
DOI=10.3389/jpps.2026.16471
  
ISSN=1482-1826
  
ABSTRACT=BackgroundGlucosamine is widely used for osteoarthritis management, yet clinical trial results remain controversial, despite robust preclinical anti-inflammatory effects that fail to translate consistently into clinical benefit. We hypothesized that a mechanistic QSP framework could reconcile these contradictory findings by quantifying how differences in drug exposure at the target site translate into variable treatment effects.ObjectiveTo develop and validate a quantitative systems pharmacology (QSP) model that mechanistically integrates glucosamine pharmacokinetics, cartilage matrix dynamics, and clinical outcomes in knee osteoarthritis, and to identify pharmacokinetic determinants of clinical efficacy.MethodsA systematic literature review identified 177 records, from which five landmark studies were selected (three for calibration, one for external validation, and one for exploratory comparison). The QSP model comprises three modules: pharmacokinetics (steady-state synovial exposure), cartilage matrix dynamics (GAG turnover with anti-catabolic drug effect), and clinical outcomes (joint space width [JSW], WOMAC pain). The model implements a primarily anti-catabolic mechanism with minor synthesis stimulation (Emax,syn = 0.15). Five parameters were estimated using multi-start optimization against 3-year JSW data from Reginster et al. and Pavelka et al., and 24-week pain data from GAIT (2006). External validation used the GUIDE 2007 trial. Virtual population simulations (N = 500) characterized interindividual variability, and translational simulations explored dose-response, bioavailability enhancement, treatment duration, and patient stratification.ResultsThe model was calibrated to reproduce the 3-year JSW treatment effect (0.211 mm predicted vs. 0.240 mm observed; 12% error), a constrained fit conditional on the assumed anti-catabolic potency. The anti-catabolic effect parameter (Imax,deg) was estimated at its upper bound of 1.00. However, a wide-range re-calibration analysis showed that this predicted benefit is contingent on the assumed anti-catabolic potency: when IC50,deg was varied across the range at which glucosamine’s effects are actually observed in vitro (≈10–1,000 μg/mL), the predicted structural benefit fell below the minimal clinically important difference and the model fit degraded substantially. External validation against GUIDE 2007 reproduced both arms’ pain trajectories (mean absolute error 3.8 WOMAC points; all observations within the 90% prediction intervals) but under-predicted the between-arm treatment effect (−1.4 vs. −5.0 WOMAC points), indicating the model captures overall pain magnitude better than the glucosamine-attributable difference; systematic overprediction of pain reduction further suggests population-specific heterogeneity in placebo response. Under the optimistic potency anchor, translational simulations predicted a minimum effective dose of approximately 750 mg/day, that 83% of patients achieve the 0.1 mm MCID at the standard dose under real-world variability (98.2% show positive benefit, treatment effect >0, under calibration conditions), 73% efficacy enhancement through bioavailability improvement from 22% to 40%, continued structural benefit through 5 years without plateau, and consistent treatment effects (NNT = 2) across disease severity stages. All of these predictions are contingent on the same anti-catabolic potency assumption; under in vitro–consistent potency, the predicted benefit falls below clinical significance (see Conclusions and “Re-calibration across the biologically realistic potency range”).ConclusionThis QSP model provides a mechanistic framework for examining the conditions under which glucosamine could exert disease-modifying effects in knee osteoarthritis. Critically, the model’s predicted structural benefit is contingent on optimistic assumptions about both anti-catabolic potency and oral bioavailability; under values consistent with the available in vitro and pharmacokinetic evidence, the predicted benefit at achievable joint concentrations is minimal. This dependence offers a mechanistic explanation for the inconsistent and frequently null results of independent glucosamine trials. However, the assumption that salt form determines bioavailability is challenged by recent pharmacokinetic and crystallographic evidence suggesting that commercially available glucosamine sulfate formulations may not differ fundamentally from glucosamine hydrochloride in their active moiety. Model predictions are contingent on the calibration data from industry-sponsored European trials, and should be interpreted in light of the failure of independent trials to replicate these structural benefits. This work exemplifies how Model-Informed Drug Development approaches can quantify the sensitivity of clinical outcomes to pharmacokinetic parameters, informing future trial design and dose optimization.