Beyond the Liver: 10-Compartment PBPK for GalNAc-siRNA Distribution to Peripheral Nerve in CIDP

Published PBPK models for GalNAc-conjugated siRNA — cemdisiran, inclisiran, givosiran — focus on liver distribution. The GalNAc ligand drives hepatic uptake through ASGR1 with a liver:plasma AUC ratio of approximately 2.5, which is exactly what these drugs need to knock down a liver-expressed target. But for CIDP, the therapeutic target is the peripheral nerve. GalNAc-siRNA distribution to nerve is unmodeled in the literature. BioMate's pbpk_peripheral_nerve workflow extends standard PBPK to 10 compartments — adding peripheral nerve, spinal cord, DRG, NMJ, and CSF — with CIDP blood-nerve barrier disease modifiers.

Why peripheral nerve PBPK matters for CIDP

Chronic inflammatory demyelinating polyneuropathy (CIDP) is an autoimmune peripheral neuropathy driven by T-cell and antibody-mediated attack on the myelin sheath and axonal membrane proteins (neurofascin-155, contactin-1, paranodal proteins). The site of pathology is the peripheral nerve, not the liver or bloodstream.

In healthy tissue, the blood-nerve barrier (BNB) restricts macromolecule access to the endoneurial compartment. The BNB is analogous to the blood-brain barrier but less restrictive — tight junctions on endoneurial capillaries limit passive paracellular transport of molecules above approximately 500 Da. GalNAc-siRNA at 13,500 Da is well above this threshold.

In CIDP, BNB disruption is a documented pathological feature. Gadolinium enhancement of peripheral nerve on MRI correlates with CIDP activity, and endoneurial edema indicates increased vascular permeability. Quantitative estimates from the literature place BNB permeability 2–5× higher in inflamed nerve segments compared to healthy controls (Csurhes et al. 2005, Journal of the Peripheral Nervous System).

Whether a GalNAc-siRNA reaches therapeutic concentrations in CIDP nerve — even with 3× BNB permeability — had not been modeled. This workflow provides that estimate.

The 10-compartment model

Standard GalNAc-siRNA PBPK uses 5 compartments: blood, liver, kidney, muscle, and fat. The pbpk_peripheral_nerve workflow adds 5 new compartments representing the peripheral nervous system:

• Peripheral nerve — endoneurial compartment, accessed via blood-nerve barrier permeability (Papp_BNB)
• Spinal cord — blood-spinal cord barrier (BSCB), less restrictive than the blood-brain barrier but tighter than peripheral BNB
• Dorsal root ganglia (DRG) — sensory neuron cell bodies, which have fenestrated capillaries and are relatively accessible to macromolecules
• Neuromuscular junction (NMJ) — junction between motor axon and muscle fiber, relevant to MG and motor neuropathies
• Cerebrospinal fluid (CSF) — blood-CSF barrier at the choroid plexus, relevant for intrathecal delivery modeling

The ODE system is integrated using SciPy's solve_ivp with the LSODA integrator, which handles the stiff kinetics of multi-compartment PBPK at the timescales relevant for oligonucleotides (hours to weeks).

SC depot (3 mg/kg, MW 13,500 Da)
     │ ka (MW-based SC absorption rate)
     ▼
┌──────────┐  CL_liver    ┌──────────────────┐
│  Blood   │─────────────▶│  Liver (ASGR1)   │ AUC ratio ~2.5×
│ (central │◀─────────────│  primary uptake  │
│   PK)    │              └──────────────────┘
│          │  Papp_renal  ┌──────────────────┐
│          │─────────────▶│  Kidney (renal)  │
└────┬─────┘              └──────────────────┘
     │
     ├── Muscle (Papp_muscle = 5e-8 cm/s)
     ├── Fat    (Papp_fat = 2e-8 cm/s)
     │
     ├── Peripheral Nerve  (Papp_BNB × 3.0 in CIDP)  → AUC ratio ~0.03–0.05
     │   Papp_BNB healthy  = 1.0e-7 cm/s
     │   Papp_BNB CIDP     = 3.0e-7 cm/s
     │
     ├── Spinal Cord (Papp_BSCB = 4e-8 cm/s)         → AUC ratio ~0.02
     │
     ├── DRG (fenestrated; Papp_DRG × 1.8 vs BNB)   → AUC ratio ~0.08
     │
     ├── NMJ (local microvascular Papp_NMJ)
     │
     └── CSF (Papp_BCSFB, choroid plexus)             → AUC ratio ~0.005

Simulation: 672 hours (28 days) · Integrator: SciPy LSODA

The CIDP disease modifier — 3× BNB permeability

The BNB permeability modifier of 3× is grounded in clinical pathology data. Gadolinium enhancement in CIDP peripheral nerve indicates breakdown of the tight junction barrier between endoneurial capillaries and the endoneurial space. Quantitative estimates from in vivo rodent models of experimental autoimmune neuritis — the CIDP analog — suggest permeability increases of 2–6× at peak inflammation.

The model uses Papp_BNB = 1.0 × 10⁻⁷ cm/s in healthy nerve and 3.0 × 10⁻⁷ cm/s in CIDP. This is conservative — pathologically active nerve segments may have higher permeability — but represents a reasonable central estimate for a population-average PBPK model.

Even with this 3× increase, the simulation predicts a nerve:plasma AUC ratio of 0.03–0.05 for a 13,500 Da molecule at 3 mg/kg SC. The liver:plasma ratio under the same conditions is approximately 2.5 (driven by active ASGR1-mediated endocytosis). The difference is five orders of magnitude in the driving mechanism: passive permeability across the BNB versus active receptor-mediated uptake in hepatocytes.

Clinical interpretation: what this means for therapeutic development

The model output answers a specific clinical question: will a GalNAc-siRNA dosed at the standard therapeutic range (1–5 mg/kg SC) achieve pharmacologically relevant concentrations in CIDP peripheral nerve?

The predicted nerve Cmax for cemdisiran-like molecules at 3 mg/kg SC in CIDP (3× BNB) is approximately 0.1 ng/mL — well below the effective concentration for target gene knockdown in nerve tissue, which for most siRNA-addressable targets requires low µg/mL tissue concentrations. The liver Cmax under the same dosing is approximately 2–4 µg/mL, in the therapeutic range for liver targets.

To achieve hepatic-equivalent concentrations in peripheral nerve via systemic dosing, the model predicts doses of 60–90 mg/kg SC would be required — clinically infeasible. This finding has a direct implication: if GalNAc-siRNA is to be developed for a peripheral nerve indication like CIDP, local delivery (intrathecal, perineural injection, nerve cuff devices) or nerve-targeted formulation strategies would be required.

Conversely, for DRG targets — relevant to sensory neuropathies and pain — the AUC ratio of ~0.08 is more favorable. DRG neurons have fenestrated capillaries rather than tight-junction BNB, making them considerably more accessible to macromolecules. A therapeutic siRNA targeting a DRG-expressed pain mediator could achieve target tissue exposure at systemic doses 3–5× lower than for endoneurial nerve targets.

References