Sotatercept

證據等級: L5

預測適應症: 10 個

Sotatercept: From Pulmonary Arterial Hypertension to Acute Lymphoblastic Leukemia

One-Sentence Summary

Sotatercept (DB12118) is an ActRIIA-Fc fusion protein currently marketed for pulmonary arterial hypertension (PAH), acting as an activin/TGF-β superfamily signaling trap. The TxGNN model predicts it may be effective for Acute Lymphoblastic Leukemia, but this prediction is currently supported by 0 clinical trials and 0 publications — it is a pure model-topology signal with no corroborating evidence.

Quick Overview

Item	Content
Original Indication	Not available in structured data (licenses empty); drug is known globally as treatment for pulmonary arterial hypertension per rationale text
Predicted New Indication	Acute Lymphoblastic Leukemia (disease)
TxGNN Prediction Score	99.78%
Evidence Level	L5 (model prediction only, no clinical or literature support)
Norway Market Status	✗ Not Marketed
Number of Authorizations	0
Recommended Decision	Hold

Why is This Prediction Reasonable?

Currently, detailed mechanism of action data is not available in the structured drug record (original_moa = Data Gap). Based on contextual information embedded in the evidence pack's rationale text, sotatercept is described as an ActRIIA-Fc fusion protein that traps activin/TGF-β superfamily ligands, primarily regulating erythrocyte maturation and vascular remodeling — mechanisms consistent with its known approved use in pulmonary arterial hypertension (PAH).

There is no known pathophysiological connection between PAH and acute lymphoblastic leukemia. The evidence pack itself states plainly: "與 ALL 之間無已知病理生理學連結，TxGNN 高分僅反映 knowledge graph 拓撲相似性，非機轉證據" — i.e., the high TxGNN score reflects graph-topology similarity in the knowledge graph rather than any biological mechanism linking the two conditions.

Notably, 9 of the 10 predicted indications in this evidence pack (severe diabetic retinopathy, drug-induced osteoporosis, HER2+ breast carcinoma, multiple rare urothelial carcinoma subtypes, etc.) show the same pattern: high TxGNN scores with zero supporting trials or literature. The repeated clustering of urothelial carcinoma subtypes (ranks 7–10) in particular suggests a systematic model artifact — likely node sparsity in the knowledge graph for these rare cancer subtypes — rather than a genuine repurposing signal. None of the 10 candidates currently rise above L5 evidence.

Clinical Trial Evidence

Currently no related clinical trials registered

Literature Evidence

Currently no related literature available

Norway Market Information

Sotatercept is currently not marketed in Norway — no product authorizations are on record (total_licenses = 0).

Safety Considerations

Please refer to the package insert for safety information.

(Note: taiwan_regulatory warnings/contraindications and TFDA label data are flagged as a Blocking data gap (DG001) in this evidence pack — safety review cannot proceed to Stage S1 until this is resolved.)

Conclusion and Next Steps

Decision: Hold

Rationale: The top-ranked prediction (acute lymphoblastic leukemia) — and all 9 other candidates in this pack — rest entirely on TxGNN topological scoring (L5) with zero clinical trials and zero literature support. There is no plausible mechanistic bridge between sotatercept's known activin-trap/PAH pharmacology and hematologic malignancy. Several lower-ranked candidates further suggest model noise from sparse graph nodes (e.g., repeated rare urothelial carcinoma subtypes) rather than genuine signal.

To proceed, the following is needed:

Resolve DG001 (Blocking): obtain TFDA/regulatory label warnings and contraindications before any S1 safety screening can begin
Resolve DG002 (High): obtain confirmed mechanism of action (MOA) from DrugBank API to properly evaluate mechanistic plausibility
If pursuing the ALL hypothesis further, commission a targeted literature/preclinical search specifically on activin/TGF-β signaling in leukemic bone marrow niches before any trial-stage investment
Given the pattern of clustered, unsupported urothelial carcinoma predictions, consider flagging this drug's KG neighborhood for review of potential data sparsity artifacts in the underlying TxGNN model
Disclaimer

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