Integrated Quantitative Systems Pharmacology model of hypereosinophilic syndrome, linking molecular-genotype-specific eosinophilopoiesis — constitutively-active fusion tyrosine kinase signaling in myeloid/ neoplastic HES (FIP1L1-PDGFRA, PDGFRB-rearranged, JAK2-rearranged) or clonal-T-cell IL-5 overproduction in lymphocytic-variant HES — to circulating/tissue eosinophil burden, degranulation-mediated organ damage (cardiac Loeffler endocarditis as the anchor severity endpoint, plus pulmonary/dermatologic/neurologic/GI involvement and a hypercoagulable state), and genotype-matched pharmacologic response (imatinib, corticosteroids, anti-IL-5-axis biologics [mepolizumab, benralizumab], hydroxyurea, ruxolitinib).

Hypereosinophilic syndrome is defined by persistent peripheral blood eosinophilia (≥1500/µL on ≥2 occasions ≥1 month apart) with eosinophil- mediated organ damage after secondary causes are excluded (Valent 2012 JACI). Roughly 10-20% of cases carry a cryptic del(4)(q12q12) that fuses FIP1L1 to PDGFRA, producing a constitutively active tyrosine kinase that drives clonal eosinophil-progenitor expansion via STAT5/ PI3K-AKT/RAS-MAPK signaling and is exquisitely sensitive to low-dose imatinib (Cools 2003 NEJM) — a paradigm subsequently extended to rarer PDGFRB and FGFR1/JAK2 rearrangements (myeloid/neoplastic HES, M-HES). A distinct lymphocytic variant (L-HES) arises from an aberrant clonal CD3-CD4+ T-cell population that chronically secretes IL-5 (and IL-4/IL-13), driving eosinophilopoiesis through the IL-5 receptor/JAK2/STAT5 axis without a kinase-fusion driver (Simon 1999 NEJM); many cases remain idiopathic. Whatever the driver, marrow-derived eosinophils enter the circulation, home to tissue via the eotaxin/CCR3 axis, and degranulate — releasing major basic protein, eosinophil cationic protein, eosinophil peroxidase, and eosinophil-derived neurotoxin that directly damage endothelium and parenchyma. The most feared consequence is Loeffler endocarditis: acute eosinophilic endocardial necrosis progresses through mural thrombus formation to endomyocardial fibrosis, entrapping the papillary muscles/chordae to produce valvular regurgitation and restrictive cardiomyopathy (Parrillo 1979 Am J Med); pulmonary infiltrates, urticaria/angioedema, peripheral neuropathy, eosinophilic gastroenteritis, and a hypercoagulable thromboembolic state (including intracardiac-thrombus embolization to the CNS) complete the multi-organ picture (Ogbogu 2009 JACI). Treatment is genotype-matched: imatinib achieves rapid, durable hematologic and molecular remission in FIP1L1-PDGFRA+/PDGFRB-rearranged M-HES (resistance emerging via the T674I kinase-domain mutation); corticosteroids remain first-line for idiopathic/lymphocytic-variant disease via eosinophil apoptosis induction and IL-5 transcriptional suppression, but relapse on taper is common; mepolizumab (anti-IL-5) and benralizumab (anti-IL-5Rα, afucosylated Fc driving NK-cell ADCC eosinophil depletion) provide steroid-sparing options for steroid-refractory/dependent disease; and hydroxyurea, ruxolitinib (JAK2-rearranged HES), interferon-α, or alemtuzumab are reserved for refractory or kinase-inhibitor-resistant cases.

1. Etiology & molecular classification (diagnostic criteria, FIP1L1-PDGFRA/PDGFRB/FGFR1/JAK2/KIT, L-HES, idiopathic, familial, overlap)
2. Bone marrow eosinophilopoiesis & cytokine drivers (CD34+ progenitor, GATA-1/PU.1/C/EBPα, IL-5/IL-3/GM-CSF, JAK2-STAT5)
3. Myeloid/neoplastic kinase signaling (constitutive FIP1L1-PDGFRA/PDGFRB/FGFR1 TK, STAT5/PI3K-AKT/RAS-MAPK, T674I resistance)
4. Lymphocytic-variant (L-HES) T-cell pathway (CD3-CD4+ clone, TCR rearrangement, Th2 skew, IL-5/IL-4/IL-13, lymphoma-transformation risk)
5. Eosinophil activation & tissue effector mediators (eotaxin/CCR3, adhesion/transmigration, MBP/ECP/EPO/EDN degranulation, ROS, tissue factor)
6. Cardiac involvement — Loeffler endocarditis (necrosis → mural thrombus → fibrosis → valvular regurgitation/restrictive cardiomyopathy)
7. Pulmonary, dermatologic & neurologic involvement (infiltrates/fibrosis, urticaria/angioedema, mononeuritis multiplex, CNS embolism/encephalopathy)
8. GI/hepatosplenic involvement & hypercoagulable state (eosinophilic gastroenteritis, hepatosplenomegaly, platelet activation, VTE risk)
9. Drug PK/PD — imatinib (ATP-competitive PDGFRA/B kinase inhibition, hematologic/molecular response, T674I resistance)
10. Drug PK/PD — corticosteroids (glucocorticoid receptor, NF-κB inhibition, eosinophil apoptosis, IL-5 suppression, taper relapse)
11. Drug PK/PD — anti-IL-5-axis biologics (mepolizumab IL-5 neutralization; benralizumab IL-5Rα blockade + ADCC depletion)
12. Drug PK/PD — hydroxyurea / ruxolitinib / IFN-α / alemtuzumab (nonselective cytoreduction, JAK1/2 inhibition, T-cell immunomodulation/depletion)
13. Clinical endpoints, biomarkers & monitoring (AEC, ECP, IL-5, tryptase, composite symptom/flare score, organ-damage composite, survival)

• Drug PK (6 drug classes, 12 compartments) — imatinib (gut + central), prednisone/prednisolone (gut + central), mepolizumab (SC depot + central), benralizumab (SC depot + central), hydroxyurea (gut + central), ruxolitinib (gut + central).
• Disease/PD network (9 compartments) — constitutive kinase-signaling index (genotype-dependent baseline, imatinib/ruxolitinib-inhibitable), endogenous IL-5 index (corticosteroid-suppressible, mepolizumab- neutralizable), marrow eosinophil-production drive, circulating absolute eosinophil count (AEC; corticosteroid apoptosis + benralizumab ADCC depletion), tissue eosinophil burden, cardiac (Loeffler) damage index (slow, partly irreversible fibrotic accrual), cardiac troponin, serum ECP, and a composite HES symptom/flare score.
• A GENOTYPE flag (0=idiopathic/L-HES, 1=FIP1L1-PDGFRA+, 2=PDGFRB- rearranged, 3=JAK2-rearranged) sets the baseline kinase-signaling vs. IL-5 drive and gates drug sensitivity exactly as in clinical practice (imatinib/ruxolitinib act only on kinase-driven genotypes; anti-IL-5- axis biologics and corticosteroids act on the IL-5/apoptosis axis regardless of genotype).

1. Patient profile — genotype/mutation flags, age/weight/baseline AEC.
2. Drug PK — log-scale plasma concentrations for the six tracked drugs.
3. Kinase/IL-5/marrow PD — constitutive kinase-signaling index, endogenous IL-5, marrow production drive.
4. Eosinophil & tissue burden — circulating AEC (log scale, diagnostic threshold line) and tissue eosinophil burden.
5. Cardiac & clinical endpoints — Loeffler damage index, composite symptom/flare score, flare-event table.
6. Biomarkers — serum ECP and cardiac troponin.
7. Scenario comparison — runs all 10 scenarios with the chosen profile.
8. References — key citations and link to the full bibliography.

# 1) Render the mechanistic map
dot -Tsvg hes_qsp_model.dot -o hes_qsp_model.svg
dot -Tpng -Gdpi=150 hes_qsp_model.dot -o hes_qsp_model.png

# 2) Simulate scenarios in R
install.packages(c("mrgsolve","dplyr","tidyr","ggplot2","shiny",
                   "shinydashboard","DT"))
library(mrgsolve)
mod <- mread_cache("hes_mrgsolve_model.R")
out <- mod %>% param(GENOTYPE=1) %>%
  mrgsim(ev(amt=100, cmt="GUT_IMA", ii=24, addl=180), end=4320, delta=1)
plot(out, c("AEC","TISSUE","CARDIAC","SYMPTOM"))

# 3) Launch the dashboard
shiny::runApp("hes_shiny_app.R")

• Designed for research, education, and hypothesis generation — not clinical decision support, hematology/immunology practice guidance, or a substitute for validated organ-damage risk-prediction tools.
• The kinase-signaling and IL-5 indices are lumped arbitrary-unit surrogates for genotype-specific molecular drive, not mechanistic receptor-binding/transcriptional kinetic models; real FIP1L1-PDGFRA/ PDGFRB/JAK2 signaling involves additional feedback loops (e.g., mTOR, negative regulators of cytokine signaling) not explicitly represented here.
• Cardiac (Loeffler) damage is modeled as a single lumped fibrotic- accrual index; it does not resolve the distinct necrotic/thrombotic/ fibrotic histopathologic stages or chamber-specific (mitral vs. tricuspid) valvular mechanics.
• Inter-individual variability (genotype penetrance, organ-damage susceptibility, CYP3A4 polymorphism affecting imatinib clearance) needs omega() blocks for population simulations; all parameters here are typical-value point estimates.

Inherits the repository LICENSE.