Tenecteplase

Evidence Level: L5 Predicted Indications: 50

Quick Overview

Item Value
Drug Name Tenecteplase
DrugBank ID DB00031
Brand Names (EU) Metalyse
Evidence Level L5
Predicted Indications 50
Top Prediction Score 99.91%

Approved Indication (EMA)

Metalyse is indicated for the thrombolytic treatment of suspected myocardial infarction with persistent ST elevation or recent left-bundle-branch block within six hours after the onset of acute-myocardial-infarction symptoms. Metalyse is indicated in adults for the thrombolytic treatment of acute ischaemic stroke (AIS) within 4.5 hours from last known well and after exclusion of intracranial haemorrhage.


Predicted New Indications

TxGNN model predictions for potential drug repurposing:

Rank Indication Score Source
1 myocardial infarction 99.91% DL
2 posteroinferior myocardial infarction 99.87% DL
3 posterolateral myocardial infarction 99.87% DL
4 septal myocardial infarction 99.85% DL
5 coronary thrombosis 99.80% DL
6 hemoglobinopathy 99.69% DL
7 congenital coronary artery anomaly 99.61% DL
8 coronary stenosis 99.53% DL
9 myocardial infarction (disease) 99.49% DL
10 partial deletion of the short arm of chromosome 16 99.34% DL
11 beta-thalassemia with other manifestations 99.29% DL
12 hemolytic anemia due to glucophosphate isomerase deficiency 99.26% DL
13 pyropoikilocytosis, hereditary 99.20% DL
14 pyruvate kinase deficiency of red cells 99.20% DL
15 heparin cofactor 2 deficiency 96.66% DL
16 antithrombin deficiency type 2 95.48% DL
17 factor 5 excess with spontaneous thrombosis 95.33% DL
18 thrombophilia 94.76% DL
19 pulmonary embolism (disease) 91.10% DL
20 acute coronary syndrome 87.44% DL

Showing top 20 of 50 predictions.


About TxGNN Predictions

Prediction Sources

Source Description
KG Knowledge Graph - Network topology-based associations
DL Deep Learning - Neural network score prediction

Evidence Levels

Level Definition
L1 Multiple Phase 3 RCTs / Systematic Reviews
L2 Single RCT or multiple Phase 2 trials
L3 Observational studies / Large case series
L4 Preclinical / Mechanistic / Case reports
L5 AI prediction only (current)

Clinical Validation Needed

Research Use Only: These predictions are computational hypotheses that require clinical validation. They should NOT be used for clinical decision-making.

Next Steps for Validation

  1. Literature Review: Search PubMed for existing evidence
  2. Clinical Trial Search: Check ClinicalTrials.gov for ongoing studies
  3. Mechanistic Analysis: Evaluate biological plausibility
  4. Preclinical Studies: Conduct in vitro/in vivo validation
  5. Clinical Trials: Design and conduct human studies

Data Access


Citation

If using this data, please cite:

@article{huang2023txgnn,
  title={A foundation model for clinician-centered drug repurposing},
  author={Huang, Kexin and others},
  journal={Nature Medicine},
  year={2023},
  doi={10.1038/s41591-023-02233-x}
}

Disclaimer: This report is for research purposes only and does not constitute medical advice. Drug repurposing predictions require rigorous clinical validation before any therapeutic application.

Copyright © 2026 EuTxGNN Project. For research purposes only. Not medical advice.