A quantitative systems pharmacology model integrating four mechanistic layers — metabolic/pharmacodynamic responses, adverse-event dynamics, aging states (damage accumulation, repair capacity, frailty, biological age gap), and biomarkers including GDF15, cystatin C, and adiponectin — simulated the effects of semaglutide, SGLT2 inhibitors, metformin, and rapamycin across aging trajectories. Calibrated against STEP trial endpoints and validated with Bayesian hierarchical meta-analysis, the model identified two mechanistically distinct optimal combinations: GLP-1 agonist + SGLT2 inhibitor + metformin for metabolic improvement, versus GLP-1 agonist + SGLT2 inhibitor + rapamycin for aging-specific benefit. Sensitivity analysis confirmed that metabolic and aging parameter spaces are largely orthogonal — meaning optimizing one does not automatically optimize the other.
This divergence is the genuinely provocative finding. Clinical medicine has long assumed metabolic improvement and biological aging deceleration are tightly coupled — improve insulin sensitivity, reduce inflammation, slow aging. This model challenges that assumption computationally, suggesting rapamycin's mTOR-mediated repair-capacity effects occupy a different mechanistic lane than metformin's AMPK-driven metabolic normalization. For longevity-focused clinicians, this reframes combination therapy selection around explicit treatment goals rather than surrogate endpoints like HbA1c.
Critical caveats apply: this is a simulation study, not a clinical trial. QSP models are only as reliable as their underlying parameter estimates, and aging biomarkers like GDF15 remain imperfect proxies. Rapamycin's immunosuppressive profile makes clinical translation of the longevity combination non-trivial. The authors appropriately flag these as hypothesis-generating predictions requiring longitudinal external validation — an honest and necessary disclaimer for what is otherwise a methodologically sophisticated computational framework.
