Pembrolizumab
| 證據等級: L5 | 預測適應症: 10 個 |
目錄
Pembrolizumab: From Unspecified Indication to Fibromatosis, Gingival
One-Sentence Summary
Pembrolizumab’s original approved indication data is not available in this Evidence Pack, and the drug is currently marked as not marketed with zero registered NDAs. The TxGNN model’s top-ranked prediction is Fibromatosis, Gingival, with a raw similarity score of 99.40% — but this candidate has zero supporting clinical trials and zero literature, and the model’s own rationale flags it as lacking any biological basis. Across all 10 predicted indications reviewed in this pack, none reach a credible evidence tier, and several are explicitly identified as disease-term mismatches rather than genuine drug-disease signals.
Quick Overview
| Item | Content |
|---|---|
| Original Indication | Not available — original_indications is empty and no approved indication text was returned |
| Predicted New Indication | Fibromatosis, Gingival |
| TxGNN Prediction Score | 99.40% (rank #13,775 by raw score ordering — near the bottom of the candidate pool, not a top hit) |
| Evidence Level | L5 (model prediction only, no supporting trials or literature) |
| US Market Status | Not Marketed |
| Number of NDAs | 0 |
| Recommended Decision | Hold |
Why is This Prediction Reasonable?
Currently, detailed mechanism of action data is not available for pembrolizumab in this Evidence Pack (original_moa is unrecorded, flagged as a High-severity data gap, DG002).
Based on the evidence embedded elsewhere in this pack (literature attached to other candidate indications), pembrolizumab is a PD-1 immune checkpoint inhibitor that restores T-cell antitumor activity in malignancies with immune evasion phenotypes (e.g., NSCLC, melanoma, hepatocellular carcinoma, MSI-H/dMMR tumors). This mechanism is well-documented but was not the mechanism invoked for the top-ranked prediction.
For the actual top candidate — gingival fibromatosis — the model’s own rationale states there is no known mechanistic relationship: gingival fibromatosis is a benign connective-tissue overgrowth condition (associated with SOS1 mutations/collagen metabolism), unrelated to PD-1/PD-L1 blockade. The high TxGNN score appears to reflect knowledge-graph embedding similarity rather than a real pharmacological or clinical signal. This is consistent with the pattern seen across all 10 candidates in this pack (see note below).
Clinical Trial Evidence
Currently no related clinical trials registered.
Literature Evidence
Currently no related literature available.
US Market Information
No NDA or marketing authorization records are present in this Evidence Pack (total_licenses = 0, market_status = 未上市/Not Marketed). No product/dosage-form/indication table can be constructed from available data.
Cytotoxicity (Antineoplastic Drugs Only)
Pembrolizumab is an antineoplastic (immune checkpoint inhibitor), based on literature embedded in this pack describing its approved use in NSCLC, melanoma, hepatocellular carcinoma, head-and-neck squamous cell carcinoma, and MSI-H/dMMR solid tumors.
| Item | Content |
|---|---|
| Cytotoxicity Classification | Immunotherapy (anti-PD-1 checkpoint inhibitor) — not a conventional cytotoxic agent |
| Myelosuppression Risk | Low — checkpoint inhibitors are not primarily myelosuppressive; toxicity is immune-mediated rather than cytotoxic |
| Emetogenicity Classification | Low |
| Monitoring Items | Immune-related adverse event (irAE) monitoring: thyroid function, liver function tests, renal function, pituitary/adrenal axis (hypophysitis reported), pulmonary status (pneumonitis), skin, and baseline CBC |
| Handling Protection | Standard IV monoclonal antibody infusion precautions; no cytotoxic hazardous-drug handling protocol required (distinct from conventional chemotherapy) |
Note: this summary is derived from literature attached to other candidate indications in this pack, not from a dedicated pembrolizumab toxicity dataset.
Safety Considerations
Please refer to the package insert for safety information. (key_warnings, contraindications, and DDI data are all unavailable in this Evidence Pack; DG001 — TFDA label warnings/contraindications — is flagged as a Blocking data gap.)
Additional Note: Pattern Across All 10 Candidates
This Evidence Pack evaluated 10 TxGNN-predicted indications for pembrolizumab (ranks #13,775–#16,101 by score, i.e., far outside the model’s highest-confidence range). All 10 received an evidence level of L4 or L5 and a Hold recommendation. Several are worth flagging explicitly:
- Rank 4 (lung hilum carcinoma) and Rank 8 (pulmonary sulcus neoplasm) are anatomic subtypes of NSCLC — mechanistically plausible given pembrolizumab’s approved NSCLC indication — but supporting literature consists only of adverse-event case reports, not efficacy data.
- Ranks 6, 9, and 10 returned literature/trials that, on manual review, are disease-term mismatches: retrieved records concern pembrolizumab’s already-approved malignant indications (NSCLC, melanoma, colorectal, hepatocellular) rather than the benign or unrelated conditions actually predicted (e.g., “lung benign neoplasm,” rare syndromes). These appear to be false positives from evidence-retrieval keyword overlap, not genuine support.
- Ranks 1, 2, 3, 5, 7 have zero clinical trial or literature evidence and no mechanistic rationale.
No candidate in this batch meets the bar for progression beyond S0.
Conclusion and Next Steps
Decision: Hold
Rationale: The top-ranked candidate (gingival fibromatosis) has no supporting evidence and no plausible mechanism, and this pattern holds across all 10 reviewed candidates — most are either evidence-free or rely on mismatched/false-positive literature. There is currently no credible repurposing signal for pembrolizumab in this Evidence Pack.
To proceed, the following is needed:
- Resolve DG001 (TFDA/US label warnings, contraindications) — currently Blocking
- Resolve DG002 (mechanism of action) — currently High severity
- Re-run or audit the evidence-matching pipeline, given the high rate of disease-term mismatches identified in ranks 6, 9, and 10
- Re-examine TxGNN candidates at higher confidence ranks (closer to rank #1 by score) rather than this batch (#13,775–#16,101), which sits far outside the model’s top predictions
- If gingival fibromatosis specifically is to be pursued, obtain preclinical/mechanistic data connecting PD-1 blockade to connective-tissue overgrowth pathways before any further evaluation
Disclaimer
This content is for research purposes only and does not constitute medical advice. Clinical validation is required before any clinical application.