Ammonia

證據等級: L5

預測適應症: 6 個

Ammonia: From No Established Therapeutic Indication to Acrodermatitis Chronica Atrophicans

One-Sentence Summary

Ammonia (NH₃, DrugBank DB11118) is a simple inorganic compound with no established therapeutic indication in available US regulatory records. The TxGNN model predicts it may be effective for Acrodermatitis Chronica Atrophicans (ACA) with a prediction score of 99.70%; however, no clinical trials and no published literature currently support this direction, and the mechanistic rationale is considered biologically implausible across all 6 predicted indications.

Quick Overview

Item	Content
Original Indication	No established therapeutic indication on record
Predicted New Indication	Acrodermatitis Chronica Atrophicans
TxGNN Prediction Score	99.70%
Evidence Level	L5
US Market Status	Not Marketed
Number of NDAs	0
Recommended Decision	Hold

Why is This Prediction Reasonable?

Detailed mechanism of action data is not currently available for Ammonia (DB11118). From a biochemical standpoint, ammonia is a simple nitrogenous compound that arises endogenously as a byproduct of amino acid catabolism. Under normal physiological conditions it is rapidly converted to urea in the liver via the urea cycle. At elevated concentrations, ammonia is neurotoxic and disrupts cellular energy metabolism — properties that define it primarily as a metabolic intermediate and toxin, not a therapeutic agent.

Acrodermatitis chronica atrophicans (ACA) is a late-stage skin manifestation of Lyme disease caused by Borrelia burgdorferi infection. Its hallmark is progressive dermal atrophy and it requires prolonged antibiotic therapy (typically doxycycline) to eradicate the underlying infection. There is no established biological mechanism by which exogenous ammonia administration would inhibit Borrelia replication, suppress the associated inflammatory cascade, or reverse atrophic skin changes. The same absence of mechanistic plausibility applies across all 6 predicted indications, which span autoimmune dermatomyositis variants, fibroproliferative acne keloid, childhood interstitial lung disease, and a rare EBV-linked photosensitivity disorder — disease categories whose standard-of-care therapies (immunosuppressants, antibiotics, targeted biologics) share no overlap with ammonia’s known biochemistry.

This pattern strongly suggests a model false positive arising from ammonia’s ubiquitous representation in metabolic pathways within the TxGNN knowledge graph, creating spurious co-occurrence associations with diverse disease nodes. The high raw prediction scores (all >99.5%) paradoxically reflect network topology artifacts rather than true biological signal. This prediction should not be interpreted as a credible repurposing opportunity without an independent mechanistic hypothesis.

Clinical Trial Evidence

Currently no related clinical trials registered for Ammonia in any of the 6 predicted indications.

Literature Evidence

Currently no related literature available for Ammonia in any of the 6 predicted indications.

US Market Information

No NDA or marketing authorization found for Ammonia in the United States. The drug has no current US market presence.

Safety Considerations

Please refer to the package insert for safety information.

Important context: Ammonia is a recognized respiratory irritant and systemic toxin. Inhalation of concentrated ammonia causes chemical burns to the airways, and systemic exposure at supratherapeutic levels produces hyperammonemia with encephalopathy. Any experimental administration in humans would require extensive toxicological justification.

Conclusion and Next Steps

Decision: Hold

Rationale: All 6 TxGNN-predicted indications for Ammonia are rated L5 with zero supporting clinical trials or published literature; each mechanistic rationale provided in the Evidence Pack independently concludes that the prediction lacks biological plausibility, and the drug’s inherent toxicity profile creates an unfavorable risk baseline that cannot be offset without a credible therapeutic hypothesis.

To proceed, the following is needed:

Expert triage review: Confirm whether these high-score predictions are systematic knowledge-graph artifacts for metabolic compounds in TxGNN, and consider flagging Ammonia for exclusion from active repurposing pipelines
Mechanistic hypothesis generation: Before any experimental consideration, a peer-reviewable mechanistic rationale connecting ammonia to at least one predicted indication must be established
Full toxicological safety profile: Retrieve complete package insert, occupational exposure limits, and NOAEL/LOAEL data before advancing to any preclinical study design
Model calibration note: The pattern of high scores across biologically unrelated indications warrants review of how endogenous metabolites are represented in the TxGNN training graph, to reduce false-positive burden in future prediction runs
Disclaimer

This content is for research purposes only and does not constitute medical advice. Clinical validation is required before any clinical application.