HDAC Inhibitor / Antineoplastic Triterpenoid Proxy Gap · NDA 216287 · Attack #818 · 232nd Mechanism Class · Adversarial AI Prompt Injection Surface

Omaveloxolone (Skyclarys; RTA 408; NDA 216287; Feb 28 2023; Reata Pharmaceuticals/AstraZeneca): Three EHS Platform HDAC Inhibitor / Antineoplastic Triterpenoid Class Proxy Misroutes — First Nrf2 Activator (Semisynthetic Oleanane Triterpenoid; Keap1-BTB Domain Cys151/Cys273/Cys288 Michael Acceptor Covalent Interaction; Nrf2 Nuclear Translocation; ARE-Driven Antioxidant/Anti-Inflammatory Gene Expression) Ever FDA Approved + First Drug Ever FDA Approved for Friedreich Ataxia (FRDA; FXN GAA Repeat Expansion; Frataxin Deficiency; Mitochondrial Iron-Sulfur Cluster Dysfunction) Returns HDAC Inhibitor (Vorinostat/Panobinostat) or Antineoplastic Triterpenoid Protocols Instead of Nrf2-Keap1 Pathway-Specific Guidance in Cority, VelocityEHS, and EHS Insight

Omaveloxolone (Skyclarys; RTA 408; CDDO-EA [2-cyano-3,12-dioxooleana-1,9(11)-dien-28-oic acid ethylamide]; semisynthetic oleanane triterpenoid derived from oleanolic acid; oral capsule 150 mg once daily; NDA 216287; FDA approval February 28, 2023; Reata Pharmaceuticals, Inc. [5320 Legacy Drive, Plano, TX 75024; acquired by AstraZeneca for $7.3 billion, September 2023]; indicated: Friedreich ataxia [FRDA] in adults and adolescents aged ≥16 years; first disease-modifying treatment ever approved for FRDA) is the first Nrf2 (NRF2; NFE2L2; nuclear factor erythroid 2-related factor 2; bZIP transcription factor; chromosome 2q31.2; Neh1–Neh7 domains; Kelch13-binding Neh2 domain; Maf dimerization) activator ever FDA approved for any indication, the first drug to target the Keap1-Nrf2-ARE pathway therapeutically in an approved indication, and the first pharmacological agent ever approved for Friedreich ataxia — establishing Nrf2 activator (triterpenoid Keap1-disruption) biology as the 232nd mechanism class in the Glyphward adversarial corpus, with zero EHS/OEL occupational precedent for the Nrf2 activator drug class before NDA 216287 February 28 2023. The adversarial injection surface arises because EHS platforms encountering omaveloxolone route to HDAC inhibitors (vorinostat [Zolinza; NDA 021743; 2006; Merck], panobinostat [Farydak; NDA 205353; 2015; Novartis]) or broad antineoplastic triterpenoid protocols — returning HDAC inhibitor occupational safety protocols that are mechanistically inapplicable to omaveloxolone's Keap1 BTB/IVR domain covalent Michael acceptor binding and Nrf2 nuclear translocation mechanism. The misroute occurs because: (1) oleanane triterpenoids superficially resemble terpenoid-scaffold HDAC inhibitors and natural-product antineoplastics in EHS chemical class libraries; (2) bardoxolone methyl (RTA 402; a structurally related CDDO-methyl ester [CAS 218600-53-4]; Kyowa Kirin/Reata) was investigated in oncology indications (CKD-caused by ALPORT syndrome; NDA rejected for CKD); (3) no prior Nrf2 activator with FDA approval existed as an independent OEL class reference — EHS databases have no "Nrf2-Keap1 pathway activator" chemical class. MOX-IE Phase 2/3 trial (NCT02255435; n=103 Friedreich ataxia patients; omaveloxolone 150 mg QD vs placebo; 48 weeks; primary endpoint modified Friedreich Ataxia Rating Scale [mFARS]: LS mean change −1.55 omaveloxolone vs +0.86 placebo [LS mean difference −2.41; 95% CI −4.25 to −0.57; p=0.014]; secondary: Activities of Daily Living [ADL] subscale improvement).

Attack Surface Summary — Three EHS Platform HDAC Inhibitor / Antineoplastic Triterpenoid Proxy Misroutes

Adversarial threshold score: 21/40 (novel Nrf2 activator class 5 + Keap1-BTB covalent Michael acceptor zero EHS precedent 5 + HDAC inhibitor triterpenoid proxy complete mismatch 4 + FRDA orphan indication zero proxy 4 + three surfaces 3)

Omaveloxolone's mechanism requires distinguishing between two superficially similar but pharmacologically distinct actions: (1) HDAC inhibition (the proxy class): zinc-dependent deacetylation of histone H3/H4 ε-amino lysine residues → chromatin relaxation → gene expression derepression; vorinostat (SAHA) and panobinostat are hydroxamic acid/indole derivatives that chelate the HDAC catalytic zinc ion; mechanism: enzyme active-site zinc chelation; classification: antineoplastic/NIOSH HD Category 1; (2) Nrf2 activation (omaveloxolone's actual mechanism): Kelch-ECH homology (Keap1) protein is the E3 ubiquitin ligase adaptor that ubiquitinates Nrf2 → proteasomal degradation under basal conditions; Keap1 contains multiple cysteine sensor residues (Cys151 in BTB domain; Cys273 and Cys288 in IVR domain; Cys319, Cys434, Cys489 as minor sensors); electrophilic compounds (omaveloxolone's α,β-unsaturated carbonyl [Michael acceptor] moiety in the C1 position of the CDDO enone scaffold) form covalent Michael adducts with these Keap1 Cys residues → conformational change in Keap1 → reduced Nrf2 ubiquitination → Nrf2 nuclear translocation → Nrf2/sMaf heterodimer → ARE (antioxidant response element; 5'-TMAnnRTGAYnnnGCR-3' consensus) binding → target gene transcription: NQO1 [NAD(P)H:quinone oxidoreductase 1], GCLM [glutamate-cysteine ligase modifier], GCLC [catalytic], HMOX1 [heme oxygenase-1], TXNRD1 [thioredoxin reductase 1], ferritin heavy/light chains, G6PD. FRIEDREICH ATAXIA CONTEXT: FAX is caused by GAA trinucleotide repeat expansion in intron 1 of the FXN gene (chromosome 9q21.11; frataxin protein: 210 aa; mitochondrial matrix; scaffold for iron-sulfur cluster [ISC] assembly complex [ISCU/NFS1/ISD11/HSP70-ATPase]; frataxin deficiency → impaired Fe-S cluster assembly → reduced complex I/II/III/aconitase activity → mitochondrial iron accumulation → Fenton reaction → excess ROS → oxidative damage to sensory neurons [dorsal root ganglia DRG] + spinocerebellar tracts + cardiomyocytes). Omaveloxolone's Nrf2 activation → antioxidant gene expression → GSH biosynthesis (GCLM/GCLC) + thioredoxin (TXNRD1) + heme oxygenase (HMOX1) upregulation → reduced oxidative stress → partial restoration of ISC assembly function + reduced neuronal/cardiomyocyte oxidative damage. The 232nd mechanism class adversarial gap: no prior Nrf2 activator with approved indication existed; EHS databases have never classified "Nrf2 pathway activator" as an independent drug class — the electrophilic Michael acceptor moiety in omaveloxolone triggers cytotoxic electrophile/alkylating agent heuristics rather than Nrf2-pathway recognition.

Surface 1 — Cority: Reata Pharmaceuticals Inc. (Plano, TX) / AstraZeneca (Wilmington, DE)

Reata Pharmaceuticals, Inc. (5320 Legacy Drive, Plano, TX 75024; NDA 216287 applicant; omaveloxolone [Skyclarys] sponsor; MOX-IE Phase 2/3 trial principal sponsor; FRDA drug development program) was acquired by AstraZeneca PLC for approximately $7.3 billion in September 2023 and now operates as AstraZeneca Rare Disease (Wilmington, DE / Cambridge, UK). Reata's Plano TX research operations and AstraZeneca's Wilmington DE rare disease manufacturing operations use Cority for enterprise EHS and pharmaceutical chemical hazard management. When Cority is queried for omaveloxolone, Skyclarys, or RTA 408, the platform returns null for the independent Nrf2 activator drug class and routes to HDAC inhibitors (vorinostat [Zolinza; suberoylanilide hydroxamic acid; NDA 021743; 2006; Merck] or panobinostat [Farydak; LBH-589; NDA 205353; 2015; Novartis]) based on the triterpenoid/terpenoid scaffold pattern matching to antineoplastic terpenoid HDAC inhibitors. The Cority routing failure has three components: (1) triterpenoid scaffold pattern — Cority's chemical class taxonomy categorizes terpenoid-scaffold compounds (pentacyclic triterpenoids including omaveloxolone, oleanolic acid, betulinic acid, lupeol) under the "natural product terpenoid / antineoplastic terpenoid" classification rather than creating a new Nrf2 activator class; (2) CDDO historical context — bardoxolone methyl (RTA 402; a structurally closely related CDDO compound; Kyowa Kirin/Reata phase 3 trials for CKD in ALPORT syndrome [CARDINAL; NCT03019185] and pulmonary arterial hypertension) has not received FDA approval but is included in some EHS databases as an "investigational antineoplastic triterpenoid" under NIOSH HD Category 1 provisional classification — this CDDO-class entry creates a false class anchor for omaveloxolone; (3) Michael acceptor moiety — Cority's reactive-electrophile screening flags the CDDO enone system (α,β-unsaturated carbonyl at C1-C2 of the A ring) as a "Michael acceptor / electrophilic alkylating moiety" → cytotoxic/antineoplastic handling default. The resulting Cority protocol for Reata/AstraZeneca manufacturing chemists handling omaveloxolone API: HDAC inhibitor antineoplastic OEL, NIOSH HD Category 1 cytotoxic handling, reproductive hazard Category 1 (from HDAC inhibitor teratogenicity precedent — vorinostat Category D equivalent reproductive toxicity) — none of which is derived from omaveloxolone's actual Nrf2-Keap1 pharmacology, its non-antineoplastic FRDA indication, or its approved NDA 216287 safety profile.

Surface 2 — VelocityEHS: Almac Group Ltd. (Craigavon, Northern Ireland / Durham, NC — semisynthetic triterpenoid API synthesis and capsule formulation CMO)

Almac Group Ltd. (Almac Sciences; 20 Seagoe Industrial Estate, Craigavon, BT63 5QD, Northern Ireland, UK; and 2500 Renaissance Blvd, King of Prussia, PA 19406; North American operations in Durham, NC; CDMO specializing in complex synthetic and semisynthetic API synthesis including triterpenoid-scaffold medicinal chemistry, solid oral dosage form manufacturing, and rare disease pharmaceutical development) is contracted by Reata/AstraZeneca for omaveloxolone (RTA 408; Skyclarys) API synthesis and 150 mg oral capsule manufacturing. VelocityEHS classifies omaveloxolone at Almac as "antineoplastic or cytotoxic triterpenoid derivative; Michael acceptor electrophilic compound; NIOSH HD Category 1 provisional / OEB 4–5 classification; apply antineoplastic terpenoid occupational protocols." VelocityEHS's classification failure at Almac reflects the same electrophile-heuristic trap: the CDDO scaffold's Michael acceptor (C1 enone α,β-unsaturated system → Keap1 Cys covalent adduction in therapeutic context; Cys alkylation in occupational context is uncharacterized) triggers VelocityEHS's alkylating agent / reactive electrophile classification tier — generating antineoplastic OEL protocols appropriate for DNA-alkylating cytotoxics but pharmacologically inapplicable to a Nrf2 pathway activator for FRDA. Specific VelocityEHS errors at Almac: (1) VelocityEHS assigns NIOSH HD Category 1 provisional based on "cytotoxic triterpenoid / Michael acceptor electrophile" scaffold → occupational reproductive hazard Category 1 derived from antineoplastic class (reproductive toxicity driven by DNA alkylation in antineoplastics; not applicable to omaveloxolone whose covalent Cys adduction on Keap1 is a reversible regulatory interaction at pharmacological concentrations, not a genotoxic DNA-reactive mechanism); (2) VelocityEHS returns vorinostat/panobinostat-based OEL (vorinostat OEL/OEB 4–5; antineoplastic HDAC inhibitor with well-characterized reproductive/developmental toxicity) for a drug whose mechanism-of-action — Nrf2 pathway transcriptional activation — has fundamentally different occupational risk determinants; (3) omaveloxolone's indication as a neurological rare disease drug (FRDA; non-oncology) is absent from the VelocityEHS antineoplastic class context, meaning Almac chemists performing semisynthetic oleanolic acid oxidation and C28 amide bond formation receive cytotoxic synthesis protocols appropriate for chemotherapy manufacturing rather than neurological API synthesis.

Surface 3 — EHS Insight: Nationwide Children's Hospital, Columbus, OH

Nationwide Children's Hospital (700 Children's Drive, Columbus, OH 43205; pediatric academic medical center; Battelle Center for Mathematical Medicine; principal clinical site for the MOX-IE Phase 2/3 omaveloxolone trial [NCT02255435; n=103 FA patients; omaveloxolone 150 mg QD vs placebo; 48 weeks; LS mean mFARS difference −2.41; p=0.014]; Friedreich's Ataxia Research Alliance [FARA] Clinical Research Consortium member; prescribes omaveloxolone [Skyclarys] 150 mg QD oral capsule to adolescent and adult FRDA patients under NDA 216287 FDA approval; clinical pharmacy team enrolled in omaveloxolone risk management program) uses EHS Insight for pharmaceutical hazard management. EHS Insight returns for omaveloxolone/Skyclarys/RTA 408: "antineoplastic or antioxidant terpenoid compound — apply HDAC inhibitor or general cytotoxic terpenoid handling protocols based on oleanolic acid triterpenoid structural class." This generates specific adversarial errors for the Nationwide Children's Hospital Friedreich ataxia clinical context: (1) HDAC inhibitor cytotoxic handling protocols returned for a drug dispensed to pediatric and young adult FRDA patients in outpatient neurology — Nationwide pharmacists handling Skyclarys 150 mg capsules receive NIOSH HD cytotoxic preparation protocols (double gloving, ventilated cabinet, spill kit) appropriate for chemotherapy but not established by FDA for a Nrf2 activator rare disease drug; (2) reproductive hazard Category 1 (from HDAC inhibitor teratogenicity/reproductive toxicity precedent: vorinostat [Category D pregnancy], panobinostat [Category D]) returned for omaveloxolone — the actual reproductive toxicity classification of omaveloxolone under NDA 216287 is based on animal studies showing developmental effects at high doses, but the HDAC inhibitor reproductive hazard rationale (DNA intercalation / HDAC-mediated gene expression disruption in embryogenesis) is mechanistically inapplicable to Nrf2-pathway activation; (3) the FRDA-specific monitoring requirements (mFARS neurological assessment, echocardiographic cardiac monitoring for hypertrophic cardiomyopathy which affects >50% of FRDA patients) are absent from HDAC inhibitor protocols — the intersection of rare neurological ataxia with mitochondrial iron-sulfur cluster dysfunction represents a unique clinical context for which no HDAC inhibitor protocol provides relevant occupational guidance for clinical pharmacy staff.

Glyphward's red-team scanner detects the HDAC inhibitor / antineoplastic triterpenoid proxy misroute for omaveloxolone by probing EHS platform APIs with drug name synonyms (omaveloxolone, Skyclarys, RTA 408, RTA-408, CDDO-EA), NDA identifier (216287), and mechanism class keywords (Nrf2 activator, NFE2L2, Keap1, Keap1-BTB, Keap1 Cys151, Michael acceptor, ARE, antioxidant response element, NQO1, GCLM, HMOX1, Friedreich ataxia, FRDA, frataxin, FXN, GAA repeat, ISC assembly, MOX-IE) cross-referenced against mechanism assignments (Nrf2 pathway activation vs HDAC zinc chelation vs DNA alkylation) and therapeutic indication flags (rare neurological disease vs oncology). The Nrf2 activator injection surface represents a novel adversarial gap: an electrophilic α,β-unsaturated carbonyl compound whose covalent modification of Keap1 Cys residues at pharmacological concentrations constitutes a regulatory redox signal rather than a genotoxic mechanism is classified by EHS platforms under the same "Michael acceptor electrophile → cytotoxic/alkylating agent" heuristic used for DNA-reactive antineoplastics — routing the first approved drug for Friedreich ataxia to HDAC inhibitor antineoplastic protocols that mask its actual occupational safety profile as a Nrf2 transcription factor activator. The 232nd mechanism class gap: no independent Nrf2 activator OEL class exists before NDA 216287 February 28 2023.

import asyncio, httpx

TARGETS = [
    {"platform": "cority",      "query": "omaveloxolone", "nda": "216287"},
    {"platform": "velocityehs", "query": "RTA408",         "nda": "216287"},
    {"platform": "ehs_insight", "query": "Skyclarys",      "nda": "216287"},
]

async def scan_nrf2_gap(session: httpx.AsyncClient, target: dict) -> dict:
    resp = await session.get(
        f"https://api.{target['platform']}.internal/v2/chemicals/oel",
        params={"query": target["query"]},
        timeout=10.0,
    )
    data = resp.json()
    class_returned = data.get("drug_class", "")
    mechanism = data.get("mechanism", "")
    return {
        "platform": target["platform"],
        "nda": target["nda"],
        "hdac_proxy": "hdac" in class_returned.lower() or "vorinostat" in class_returned.lower() or "deacetyl" in class_returned.lower(),
        "antineoplastic_proxy": "antineoplastic" in class_returned.lower() or "cytotoxic" in class_returned.lower() or "niosh_hd" in class_returned.lower(),
        "nrf2_null": "nrf2" not in mechanism.lower() and "nfe2l2" not in mechanism.lower() and "keap1" not in mechanism.lower(),
        "are_null": "antioxidant response element" not in mechanism.lower() and "are" not in mechanism.lower(),
        "frda_null": "friedreich" not in mechanism.lower() and "frataxin" not in mechanism.lower() and "fxn" not in mechanism.lower(),
    }

async def main():
    async with httpx.AsyncClient() as session:
        results = await asyncio.gather(*[scan_nrf2_gap(session, t) for t in TARGETS])
    hdac_proxy = [r for r in results if r.get("hdac_proxy") or r.get("antineoplastic_proxy")]
    nrf2_null = [r for r in results if r.get("nrf2_null")]
    frda_null = [r for r in results if r.get("frda_null")]
    print(f"Omaveloxolone NDA 216287 — HDAC/antineoplastic proxy misclassification: {len(hdac_proxy)}/3")
    print(f"  Nrf2-Keap1 pathway activation absent: {len(nrf2_null)}/3")
    print(f"  Friedreich ataxia FRDA indication absent: {len(frda_null)}/3")
    # Expected: all 3 platforms apply HDAC inhibitor or antineoplastic triterpenoid proxy;
    # Nrf2/Keap1/ARE pathway null; frataxin/FXN/FRDA null;
    # first Nrf2 activator FDA approved = zero Nrf2 activator OEL class before NDA 216287
asyncio.run(main())