The Patch as a Spatial Dose-Segregation Device: Free Local Drug Concentration as the Primary Efficacy Variable for Skin-Targeted JAK Inhibition — A Pre-Registered Experiment, with Tofacitinib as the Worked Example

Experiment-forward preprint (v3). For a drug whose target lives in the skin, the decision variable is the free (unbound) local concentration at the indication-relevant depth, traded against systemic exposure. The missing piece is not the principle (established prior art) but a decision-grade, depth-resolved, free-versus-total measurement with a pre-registered go/no-go gate. v3 corrects the bounding model to a piecewise steady-state form — an avascular viable epidermis (linear, no clearance sink) over a clearing dermis (exponential, λ=√(D/k)), matched at the junction. Positive and limiting-case controls pass (the piecewise model collapses exactly to the v2 homogeneous model as the epidermis thickness a→0). Result: spatial dose-segregation is reachable for superficial interfollicular targets (under ~1 mm) and out of range for the deep hair bulb without an active follicular route; the deep prediction is parameter-sensitive and not over-read. The paper specifies the experiment — intra-subject crossover, dermal microdialysis (free) alongside open-flow microperfusion (total), MALDI spatial mapping, and a pre-registered gate — with tofacitinib as the worked example. Scope: a specified, not-yet-run protocol; AI-assistance disclosed in the manuscript. Not a clinical claim, not medical advice, no dosing recommendation.