# Celiac Target Tolerability & Modality Scoring Rubric

## Why celiac needs a different target profile than EoE

The safety bar is set by the **benefit-risk denominator**, not patient temperament:

- **EoE:** no safe, effective baseline existed, so an injectable biologic (dupilumab)
  competed against untreated disease — favorable math even for a systemic agent.
- **Celiac:** a safe and effective (if burdensome) baseline already exists — the
  gluten-free diet. Per FDA draft guidance, the GFD is *"the only treatment for CeD"*
  and drugs are developed as an **adjunct to** it, not a replacement. Any drug therefore
  competes against a safe diet, so its own safety/tolerability must be very high.

**Verified patient-preference signal** (Branchi et al. 2016, *Digestion* 93(2):160–166,
n=372 Italian Celiac Association members):
- GFD accepted by **88%**, yet **65%** still report a need for an alternative drug therapy.
- Preferred modalities are **low-systemic-exposure**: enzymes (145 subjects) > vaccine
  (111 subjects).
- Patients wanting a drug had a *lower* QoL gain on GFD (p=0.003) — i.e. demand
  concentrates in those the diet serves least well.

**Consequence:** for the large diet-adherent population, prioritize **oral / locally-acting /
gut-restricted / low-side-effect** mechanisms — ideally **repurposable** approved oral drugs.

## Two target populations → two tiers

| | Tier-1 (repurposing-first) | Tier-2 (novel biology) |
|---|---|---|
| Population | Diet-adherent; protect villi from *accidental* gluten | Refractory / persistent villous atrophy despite GFD |
| Safety bar | Very high — competes with a safe diet | Higher risk acceptable — active disease, unmet need |
| Preferred modality | Approved **oral** drug on target/pathway; gut-restricted | Gut-restricted biologic OK; systemic if justified |
| Regulatory anchor | FDA: adjunct-to-GFD, histologic endpoint | FDA: first adjunctive patients have persistent villous atrophy |

## The villous-destruction axis (what we mine the omics for)

FDA histologic eligibility defines the damage we want to prevent:
**villous atrophy + crypt hyperplasia + intraepithelial lymphocytosis** (modified
Marsh-Oberhuber). The effector biology is well-characterized and is the target space:
- **IL-15** — drives intraepithelial-lymphocyte (IEL) survival/activation and licenses cytotoxicity.
- **NKG2D (KLRK1) — MIC (MICA/MICB) stress-ligand axis** — IEL-mediated killing of stressed enterocytes.
- **IFN-γ** from gluten-specific CD4 T cells — the dominant cytokine of the lesion.
- **Cytotoxic IEL programs** (granzyme/perforin) — the proximal effectors of villous destruction.

## Scoring axes (each target scored 0–3)

1. **Mechanistic centrality to villous destruction** — dysregulation magnitude &
   reproducibility in *villous/mucosal* tissue (from the meta-analysis) and expression in
   the destroying compartment (IEL/epithelium, from single-cell).
2. **Local actionability** — can it be hit at the mucosa without systemic immunosuppression?
   luminal/gut-restricted (3) > surface on gut-resident cells (2) > systemic-only (1).
3. **Oral druggability / repurposing** — approved **oral** drug on the target or its
   immediate pathway (3) > oral drug in trials (2) > only injectable/biologic (1) >
   undrugged (0). This is the axis that operationalizes the patient-preference evidence.

**Tier assignment:** Tier-1 = high centrality **AND** (oral-repurposing ≥2 **or**
local-actionability =3). Tier-2 = high centrality but requires a novel/systemic modality —
carried for the refractory population and flagged by required modality.

## Guardrails carried from the pMHC stream
- **Presentation ≠ pathology; dysregulation ≠ causation.** Omics nominates; functional/genetic
  evidence confirms. Every target is a prior.
- **HLA is necessary but not sufficient** — burden/expression is risk/capacity, not diagnosis.
- All modality/repurposing calls are hypotheses for review, not clinical recommendations.