Paliperidone

Evidence Level: L5 Predicted Indications: 10

Table of Contents

  1. Paliperidone
  2. Paliperidone: From Schizophrenia to Retinal Dystrophy with or without Extraocular Anomalies
    1. One-Sentence Summary
    2. Quick Overview
    3. Why is This Prediction Reasonable?
    4. Clinical Trial Evidence
    5. Literature Evidence
    6. EU Market Information
    7. Safety Considerations
    8. Conclusion and Next Steps
    9. Disclaimer

## Pharmacist Assessment Report

Paliperidone: From Schizophrenia to Retinal Dystrophy with or without Extraocular Anomalies

One-Sentence Summary

Paliperidone is an atypical antipsychotic drug (per supporting literature in this evidence pack, it is used for schizophrenia and related psychotic disorders). The TxGNN model predicts it may be effective for Retinal Dystrophy with or without Extraocular Anomalies, but this prediction is currently supported by 0 clinical trials and only general ophthalmology literature with no direct relevance to the drug — the model’s rationale itself flags this as likely a knowledge-graph proximity artifact rather than a true pharmacological signal.


Quick Overview

Item Content
Original Indication Not available — no marketing authorization or original-indication data in this evidence pack (drug not marketed in this jurisdiction); literature elsewhere in this pack indicates an existing schizophrenia indication
Predicted New Indication Retinal Dystrophy with or without Extraocular Anomalies
TxGNN Prediction Score 99.92%
Evidence Level L5
EU 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 for paliperidone in this evidence pack. Based on other information contained in the pack (see the treatment-refractory schizophrenia candidate below), paliperidone is a D2/5-HT2A receptor antagonist belonging to the second-generation (atypical) antipsychotic class, and its efficacy in schizophrenia is well established.

For the top-ranked prediction — retinal dystrophy with or without extraocular anomalies — there is no identified biological or pharmacological link to paliperidone’s known mechanism. The model’s own rationale states that this high score likely reflects proximity within the knowledge graph rather than a genuine pharmacological relationship. The attached literature (orbital infection, diplopia, congenital ptosis, lens anomalies, etc.) covers general ophthalmology topics and does not discuss paliperidone or antipsychotic pharmacology at all.

Given the absence of mechanistic rationale, clinical trials, or targeted literature, this prediction should be treated as a pure model output requiring substantial further validation before any clinical consideration.


Clinical Trial Evidence

Currently no related clinical trials registered.


Literature Evidence

PMID Year Type Journal Key Findings
9416661 1997 Review Semin Ultrasound CT MR General review of orbital infections; no mention of paliperidone or retinal dystrophy
20127583 2010 Review Semin Neurol Diagnostic approach to diplopia; general neuro-ophthalmology, not drug-specific
22241537 2012 Review Klin Monbl Augenheilkd Overview of congenital ptosis and extraocular muscle fibrosis
38249493 2023 Review Taiwan J Ophthalmol Congenital anomalies of lens shape; anterior segment dysgenesis
109006 1979 Case Report Am J Ophthalmol Case series of unilateral cryptophthalmia with orbital malformation
24413161 2014 Case Report J Neuroophthalmol Isolated case of congenital trochlear-oculomotor synkinesis
7035111 1981 Review Doc Ophthalmol Wagner-Stickler syndrome: vitreoretinal degeneration and extraocular features
38321238 2024 Review Pediatr Radiol Imaging review of pediatric ocular pathologies including congenital lesions
19826317 2009 Case Report Optom Vis Sci Case of congenital extraocular muscle fibrosis with variable divergence
33447730 2020 Review Ther Adv Ophthalmol Review of ocular involvement in inherited metabolic disorders

Note: None of the above literature discusses paliperidone, antipsychotic pharmacology, or retinal dystrophy treatment directly — all are general ophthalmology/neuro-ophthalmology background references.


EU Market Information

No marketing authorizations found. This drug is not currently marketed in the covered jurisdiction, so no product/dosage-form/indication data is available.


Safety Considerations

Please refer to the SmPC for safety information.

(Note: key warnings, contraindications, and drug-interaction data are flagged as a Blocking data gap in this evidence pack — see Conclusion below.)


Conclusion and Next Steps

Decision: Hold

Rationale: The top-ranked prediction (retinal dystrophy with or without extraocular anomalies) has no clinical trials, no directly relevant literature, and no plausible mechanistic link — it is an L5, model-only signal (S0 stage) that the model’s own rationale flags as likely a knowledge-graph artifact rather than a real pharmacological relationship.

To proceed, the following is needed:

  • TFDA/regulatory label data (warnings, contraindications) — currently a Blocking gap that prevents any S1 safety assessment
  • Confirmed mechanism of action via DrugBank (currently unavailable)
  • Independent mechanistic or preclinical rationale connecting paliperidone to retinal/ophthalmic pathology before any further evaluation
  • Portfolio note: a lower-ranked candidate in this same pack — treatment-refractory schizophrenia (score 99.80%, L2 evidence, “Proceed with Guardrails”) — has a direct pharmacological rationale (D2/5-HT2A antagonism), a completed Phase 4 observational study, and ongoing Phase 4 trials. This is a substantially stronger candidate and may warrant prioritization over the top TxGNN-ranked prediction.

    Disclaimer

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



Copyright © 2026 藥提醒科技有限公司 (yao.care). For research purposes only. Not medical advice.

This site uses Just the Docs, a documentation theme for Jekyll.