A Bistable ODE Model Predicts Durable Off-Treatment Resolution of Chronic Plaque Psoriasis via Tissue-Resident Memory T Cell Depletion: An In Silico Multi-Mechanism Protocol

Background.
Chronic plaque psoriasis (CPP) affects an estimated 125 million people worldwide and currently has no approved curative therapy. Modern biologic therapies targeting interleukin-17A (IL-17A), interleukin-23 (IL-23) or tumour necrosis factor alpha (TNF-α) suppress disease activity during administration but fail to produce durable remission upon withdrawal because they do not eliminate the tissue-resident memory T cell (TRM) compartment that maintains site memory in clinically uninvolved skin.
Methods.
We developed a twenty-variable bistable ordinary differential equation (ODE) model of the psoriatic immune–keratinocyte–mechanical circuit, calibrated against published single-cell transcriptomic, kinetic and physiologically-based pharmacokinetic (PBPK) data, and implemented on top of the open-source CureNet simulation platform [19]. The model was used to screen 5,757 drug combinations across 21 agents and to execute three independent computational validation rungs: (i) a multi-compartment PBPK simulation of topical andrographolide (AGP) dermal delivery, (ii) a Hill-kinetic dose-response validation of berberine (BBR) fatty acid oxidation (FAO) inhibition, and (iii) an allometrically scaled mouse-specific ODE with Monte Carlo statistical power analysis.
Results.
A five-component protocol comprising narrowband ultraviolet B (NB-UVB) phototherapy three times weekly, concentration-matched AGP topical gel (5% on perilesional skin and 10% on thick lesional plaques), oral BBR 1500 mg/day, oral methotrexate (MTX) 15 mg/week and oral rapamycin (RAPA) 0.5 mg/week is predicted to achieve a permanent clearance rate of 85–87% at twelve months post-treatment depending on patient phenotype, at a monthly direct-drug cost of approximately USD 345–405. Leave-one-out (LOO) analysis identified NB-UVB, AGP gel and BBR as mechanistically non-negotiable: removal of any one collapses the predicted cure rate to below 5%. The mouse validation rung returned Cohen’s d = 4.17, giving 99% statistical power at n = 5 per arm.
Conclusions.
The computational framework delivers a pharmacokinetically self-consistent, sensitivity-tested, statistically tractable candidate protocol for the first genuinely curative intervention in chronic plaque psoriasis. We present the full mathematical formulation, the sensitivity landscape, the comprehensive absorption, distribution, metabolism, excretion, toxicity (ADMET) profile of the five active components, the proposed dosing cycle, and the wet-lab verification pipeline required for clinical translation.