Skip to content

restless-legs-syndrome

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

restless-legs-syndrome

README.md

Restless Legs Syndrome (RLS / Willis-Ekbom Disease) — QSP Model

Comprehensive Quantitative Systems Pharmacology (QSP) model integrating brain-iron deficiency, dopaminergic, adenosinergic (A1R), α2δ, opioid, and glutamatergic signaling that drive sensorimotor symptoms and periodic limb movements in sleep (PLMS) of restless legs syndrome — and the response to first- and second-line pharmacotherapies.

분류 · 만성 신경계 | 약자 · RLS / WED | 최근 갱신 · 2026-06-30

Quick links

File Purpose
rls_qsp_model.dot · .svg · .png 100+ node mechanistic map (12 subgraph clusters)
rls_mrgsolve_model.R 19-compartment ODE / PK-PD QSP model
rls_shiny_app.R 8-tab interactive dashboard
rls_references.md 50+ PubMed-indexed references

Pathophysiology (mechanistic summary)

GWAS risk loci (BTBD9, MEIS1, MAP2K5/SKOR1, PTPRD, TOX3)
    ↓ ↘
Low serum / CSF ferritin → ↓ brain iron in SN, putamen, thalamus  (Connor 2003 Brain)
    ↓
↓ tyrosine hydroxylase activity (Fe²⁺ cofactor) + paradoxical ↑ DA synthesis with ↓ post-syn D2/D3
    ↓
↓ A1 adenosine tone (Ferré 2018) → disinhibited glutamate + dopamine release at striatum/spine
    ↓
A11 → spinal dorsal horn dopamine tone drops at night → sensorimotor hyperexcitability
    ↓
Periodic limb movements in sleep (PLMS) + urge-to-move + IRLS ↑
    ↓
Augmentation hazard rises with chronic high-dose DA agonist exposure

Model components

1. Mechanistic map (.dot)

  • 12 clusters: genetics · iron · dopamine · adenosine · glutamate/GABA · opioid · spinal cord/PNS · circadian/sleep · augmentation · drug PK · pharmacotherapies · clinical endpoints
  • >100 nodes with edge annotations citing primary literature (Earley, Connor, Allen, Ferré, Trenkwalder)
  • All current and emerging drug classes mapped to their molecular targets

2. mrgsolve ODE model (.R)

  • 19 ODE compartments: 6 drug-class PKs + IV iron PK + ferritin + brain iron + DA augmentation + IRLS + PLMS
  • 6 drug classes included: pramipexole, ropinirole, rotigotine (patch), gabapentin enacarbil, pregabalin, oxycodone/naloxone PR, IV ferric carboxymaltose
  • 8 pre-built treatment scenarios: untreated · pramipexole 0.25/0.5 mg · rotigotine 2 mg · gabapentin enacarbil 600 mg · pregabalin 300 mg · oxycodone/naloxone PR · IV FCM · augmentation scenario (1 yr)
  • Calibration anchors: Allen 2003 (IRLS validation), Trenkwalder 2008 (rotigotine), Garcia-Borreguero 2010 (gabapentin enacarbil PIVOT-RLS), Allen 2014 (pregabalin head-to-head vs pramipexole), Trenkwalder 2013 (oxycodone-naloxone RELOXYN), IRLSSG augmentation criteria (Garcia-Borreguero 2016)
  • Patient covariates: weight, eGFR (gabapentinoid CL), sex, pregnancy, ESRD, SSRI trigger, daily caffeine

3. Shiny app (.R)

8 tabs:

  1. Patient profile — covariate inputs & risk snapshot
  2. Drug PK — concentration-time curves
  3. Iron / ferritin — serum ferritin & brain iron trajectories
  4. Network tones — DA · A1 · α2δ · MOR effects
  5. IRLS / PLMS / Sleep — symptom dynamics
  6. Augmentation — cumulative hazard
  7. Scenario comparison — side-by-side IRLS / endpoints
  8. Biomarkers & QoL — CGI-I, ICD hazard, constipation surrogate

4. References (.md)

50+ PubMed-indexed references grouped by:

  • Epidemiology & IRLSSG criteria
  • GWAS / risk loci
  • Brain iron / ferritin
  • Dopaminergic & adenosinergic pathophysiology
  • Drug-specific RCTs (DA agonists, α2δ, opioids, IV iron)
  • Augmentation / safety / impulse-control disorders
  • Special populations (ESRD, pregnancy)
  • Pharmacokinetic primary sources

Clinical / regulatory anchors

Trial Drug Endpoint Effect
Trenkwalder 2008 Lancet Neurol Rotigotine 3 mg/24h patch ΔIRLS −13.7 vs −7.4 placebo @ 6 mo
Garcia-Borreguero 2010 (PIVOT-RLS) Gabapentin enacarbil 1200 mg ΔIRLS −13.2 vs −8.8 @ 12 wk
Allen 2014 NEJM Pregabalin 300 mg vs pramipexole 0.5 mg ΔIRLS @ 12 wk · augmentation @ 52 wk −14.6 vs −12.7 · 1.7% vs 7.7%
Trenkwalder 2013 (RELOXYN) Oxycodone/Naloxone PR ΔIRLS −16.5 vs −9.4 @ 12 wk
Allen 2011 IV FCM 1000 mg IRLS responders >50% vs ~30% placebo

How to run

# (1) Compile the model
source("rls_mrgsolve_model.R")

# (2) Launch the Shiny dashboard (R ≥ 4.3, mrgsolve ≥ 1.5, shiny, DT)
shiny::runApp("rls_shiny_app.R")

Notes / limitations

  • Brain iron pool is modeled as a phenomenological index (au, 100 = healthy reference); volume-of-interest MRI R2* could be substituted.
  • The augmentation hazard accumulates monotonically with effective DA-agonist tone above a threshold; clinical augmentation is multifactorial and individual susceptibility varies.
  • A1 adenosine tone is included for emerging dipyridamole/A1 PAM strategies but is not yet a routine therapeutic target.
  • Sleep efficiency is a linear surrogate; full polysomnographic architecture (REM/N3 staging) is not simulated.

License

Same as the parent repository (MIT) — see ../LICENSE.