Primidone
| 證據等級: L5 | 預測適應症: 10 個 |
目錄
Primidone: From Epilepsy/Essential Tremor to Trigeminal Nerve Neoplasm
One-Sentence Summary
Primidone is a barbiturate-class prodrug (metabolized to phenobarbital + PEMA) used as an antiepileptic/antitremor agent, acting via GABA-A receptor potentiation and high-frequency sodium-channel blockade. The TxGNN model's top-ranked prediction is Trigeminal Nerve Neoplasm, but this candidate currently has 0 clinical trials and 0 publications supporting it, and the evidence pack's own mechanistic rationale flags the association as a likely knowledge-graph artefact ("trigeminal" term-adjacency confusion) rather than a genuine pharmacological link.
Quick Overview
| Item | Content |
|---|---|
| Original Indication | Not documented in evidence pack (no market authorization on file); drug class per rationale text: barbiturate-derivative antiepileptic/antitremor agent |
| Predicted New Indication | Trigeminal Nerve Neoplasm |
| TxGNN Prediction Score | 99.99% |
| Evidence Level | L5 |
| New Zealand Market Status | ✗ Not Marketed |
| Number of Authorizations | 0 |
| Recommended Decision | Hold |
Why is This Prediction Reasonable?
Detailed mechanism-of-action data (original_moa) is formally marked as a data gap in this evidence pack. However, the model's own rationale text supplies working pharmacology: primidone is a barbiturate prodrug, metabolized to phenobarbital and PEMA, and acts through GABA-A receptor potentiation combined with high-frequency sodium-channel blockade — the classical mechanism underlying anticonvulsant and antitremor effects.
For the top-ranked candidate, Trigeminal Nerve Neoplasm, no plausible mechanistic bridge exists between this GABA-A/sodium-channel pathway and tumour growth pathways in trigeminal nerve sheath tumours. The evidence pack explicitly notes this: the high TxGNN score "may reflect knowledge-graph term-adjacency confusion (the word 'trigeminal') rather than a true pharmacological relationship." This is a case where a high similarity score should not be read as strong biological plausibility.
By contrast, several lower-ranked candidates in this pack have a far more coherent mechanistic story: reflex/stimulus-triggered epilepsies (audiogenic seizures, startle epilepsy, micturition-induced seizures, reading seizures, thinking/eating seizures) sit squarely within primidone's established anticonvulsant pharmacology, and trigeminal neuralgia has a historical precedent of barbiturate use predating carbamazepine. These are discussed further below.
Clinical Trial Evidence
Currently no related clinical trials registered.
Literature Evidence
Currently no related literature available.
New Zealand Market Information
Primidone currently has no market authorization on file (0 licenses); no product/dosage-form records are available.
Other Predicted Indications (Ranked Candidates)
The evidence pack scores 10 candidate indications for primidone. Ranked by TxGNN score, only the reflex-epilepsy cluster and trigeminal neuralgia reach L4 (preclinical/mechanistic literature, no controlled trials); the remainder are L5 (model prediction only) and held.
| Rank | Predicted Indication | TxGNN Score | Evidence Level | Decision Stage | Recommendation | Note |
|---|---|---|---|---|---|---|
| 1 | Trigeminal nerve neoplasm | 99.99% | L5 | S0 | Hold | Likely KG term-confusion, no evidence |
| 2 | Orgasm-induced seizures | 99.99% | L5 | S0 | Hold | No literature or trials |
| 3 | Audiogenic seizures | 99.99% | L4 | S1 | Research Question | 12 publications; classic AED animal-screening model, primidone/phenobarbital textbook evidence, but human data limited to case reports |
| 4 | Startle epilepsy | 99.99% | L4 | S1 | Research Question | 1 case report only |
| 5 | Micturition-induced seizures | 99.99% | L4 | S1 | Research Question | 15 publications, incl. NEJM comparative trial (PMID 3925335, carbamazepine/phenobarbital/phenytoin/primidone), but not disease-specific |
| 6 | Eating seizures | 99.99% | L5 | S0 | Hold | Single veterinary (canine) case report — not human data |
| 7 | Thinking seizures | 99.99% | L5 | S0 | Hold | 1 indirect geriatric-AED cohort study |
| 8 | Reading seizures | 99.99% | L4 | S1 | Research Question | 9 publications on reflex-epilepsy family, no disease-specific trial |
| 9 | Trigeminal neuralgia | 99.98% | L4 | S1 | Research Question | Historical barbiturate precedent (1957) pre-dating carbamazepine; most literature is on carbamazepine, not primidone directly |
| 10 | Beta-ketothiolase deficiency | 99.96% | L5 | S0 | Hold | No mechanistic overlap (mitochondrial acetyl-CoA disorder); no evidence |
Safety Considerations
Please refer to the package insert for safety information.
Conclusion and Next Steps
Decision: Hold
Rationale: The top-ranked candidate (Trigeminal Nerve Neoplasm) has zero supporting trials or literature, and the evidence pack itself attributes the high score to likely knowledge-graph noise rather than a real pharmacological signal. Across all 10 candidates in this pack, none exceed L4 (case reports, animal models, and general AED reviews); no candidate has a controlled human trial specific to the predicted indication.
To proceed, the following is needed:
- TFDA/regulatory package insert warnings and contraindications (currently blocking — DG001)
- Confirmed DrugBank mechanism-of-action and formal original-indication text (currently a high-severity gap — DG002)
- If pursuing further inquiry, re-scope toward the reflex-epilepsy cluster (audiogenic seizures, startle epilepsy, reading seizures) or trigeminal neuralgia, which have comparatively stronger mechanistic and literature support than the top-scored candidate
- Drug-drug interaction data (currently not found)
Disclaimer
This content is for research purposes only and does not constitute medical advice. Clinical validation is required before any clinical application.