Androgen deprivation therapy (ADT) removes the testosterone that drives androgen-receptor (AR) signalling in prostate-cancer cells. AR activation saturates with a Michaelis–Menten term:
A(T) = T / (T + Km), Km ≈ 250 ng/dL
Androgen-dependent (AD) cells need AR signalling above half-maximal to keep growing; below it they regress:
dN_AD/dt = r_max · (2A - 1) · N_AD · (1 - N/K)
A rare androgen-independent (castration-resistant, CR) subclone — expanded AR gene copies, AR splice variants, or a bypass pathway — keeps a slow constant growth rate regardless of testosterone:
dN_CR/dt = r_CR · N_CR · (1 - N/K)
- GnRH Agonist (leuprolide-style) first over-stimulates the pituitary, producing a 1–2 week testosterone flare before receptor downregulation suppresses it.
- GnRH Antagonist (degarelix-style) blocks the receptor directly — testosterone falls immediately, no flare.
- Suppression dose sets how far testosterone is pushed toward the castrate floor (~15 ng/dL).
- Continuous vs Intermittent — intermittent ADT lets testosterone recover during scheduled "off" windows once PSA falls, which is thought to slow — not stop — selection for the resistant clone.
- PSA here is a live proxy for total AR-driven output:
k·(N_AD·A + N_CR·A_CR), where CR cells keep producing a low, T-independent baseline via AR amplification. Watch it fall with therapy, then climb again once the resistant clone dominates — the hallmark of clinical progression to castration-resistant prostate cancer (CRPC).
Note on this 2D adaptation: the original 3D engine's math was re-derived and checked numerically (Node script, not shipped) against the growth-rate sign flip at A=0.5 (T = Km = 250 ng/dL) and the PSA-proxy normalisation at t=0; both were internally consistent, so the equations above are carried over unchanged — this rebuild only replaces the InstancedMesh sphere cluster with a 2D packed-disk cross-section plus a live strip chart of T / PSA / resistant-fraction.