N3pG Pyroglutamate Amyloid-Beta Antibody · Lecanemab Protofibril / Aducanumab Proxy Class Gap · BLA 761248 · Attack #849 · 262nd Mechanism Class · Adversarial AI Prompt Injection Surface

Donanemab (Kisunla; LY3002813; BLA 761248; Jul 2 2024; Eli Lilly and Company [Indianapolis, IN]): Three EHS Platform Lecanemab (Leqembi; BLA 761269 — Protofibril-Selective Amyloid Antibody Protocols) or Aducanumab (Aduhelm; BLA 761178 — Full-Fibril Aggregate Antibody Protocols) Proxy Misroutes — First N3pG (N-Terminally Truncated Pyroglutamate-Modified Aβ3-42; pGlu3-Aβ; Plaque-Deposited Amyloid Form Distinct from Protofibrils and Mature Fibrils) Selective Anti-Amyloid-Beta Antibody (IgG1κ; IV Infusion 700 mg Q4W; Treatment Discontinuation Once Amyloid Cleared; TRAILBLAZER-ALZ 2 ORR 35.1% Slowing; ARIA-E 24.0%; ARIA-H 31.4%) Ever FDA Approved Returns Lecanemab Protofibril or Aducanumab Full-Fibril Protocols Instead of N3pG Plaque-Deposited Amyloid Guidance in Cority, VelocityEHS, and EHS Insight

Donanemab (Kisunla; LY3002813; molecular formula large IgG1κ monoclonal antibody; IV infusion 700 mg Q4W until amyloid clearance confirmed by amyloid PET; BLA 761248; Jul 2 2024 traditional FDA approval; Eli Lilly and Company [Lilly Corporate Center, 893 Delaware Street, Indianapolis IN 46225; BLA holder; Kisunla Alzheimer's disease US commercialization]) is the first N3pG (N-terminally truncated pyroglutamate-modified amyloid-beta) selective anti-amyloid antibody ever FDA approved, establishing N3pG pyroglutamate-modified plaque amyloid as the 262nd mechanism class in the Glyphward adversarial corpus, with zero N3pG-selective amyloid OEL/EHS classification class in any US EHS database before BLA 761248 July 2 2024. Donanemab is an IgG1κ monoclonal antibody that targets N3pG (N-terminally truncated pyroglutamate-modified amyloid beta; pGlu3-Aβ; formed by beta-secretase/BACE1 cleavage of APP at position 3 to yield Aβ3-42 fragment, followed by spontaneous cyclization of the N-terminal glutamate at position 3 → pyroglutamate [pyroGlu; pGlu; abbreviated N3pG]; the pyroglutamate modification creates a compact, hydrophobic N-terminus that promotes rapid aggregation and plaque deposition; pGlu3-Aβ is enriched in amyloid plaques and is predominantly in insoluble plaque-deposited form rather than soluble protofibrillar or oligomeric forms; estimated 25–50% of total plaque Aβ is N3pG-modified in Alzheimer's disease brain; this plaque-deposited amyloid specificity is fundamentally distinct from lecanemab's protofibril [large soluble Aβ aggregates 50–200 nm] targeting and from aducanumab's non-selective fibril/aggregate targeting). The adversarial injection surface arises because EHS platforms encountering donanemab route to: (A) lecanemab (Leqembi; BAN2401; BLA 761269; Jan 6 2023 accelerated/Jul 6 2023 traditional FDA approval; Eisai Co. Ltd./Biogen Inc. — protofibril-targeting amyloid antibody; biweekly IV infusion 10 mg/kg Q2W; CLARITY AD Phase 3 NCT03887455 1.35-point CDR-SB slowing at 18 months [27%]) protocols — missing that donanemab targets a chemically distinct N3pG pyroglutamate-modified amyloid epitope, uses monthly Q4W dosing (not biweekly Q2W), and has a unique treatment discontinuation endpoint (stop when amyloid reaches ≤11 centiloid for 2 consecutive quarterly PET scans — no equivalent in lecanemab protocols which are indefinite treatment); or (B) aducanumab (Aduhelm; BIIB037; BLA 761178; Jun 7 2021 accelerated approval; Biogen — first FDA-approved anti-amyloid antibody, then its clinical significance disputed; accelerated approval withdrawn by Biogen in Jan 2024) protocols — missing that donanemab is fully approved (not accelerated), targets N3pG rather than mature fibrils, and has well-established efficacy (TRAILBLAZER-ALZ 2 NCT04437511: Phase 3; donanemab 700 mg Q4W for ≤24 months; early-intermediate-tau patients [tau-PET low/medium] primary endpoint: 35.1% slowing of integrated Alzheimer's Disease Rating Scale [iADRS] decline at 76 weeks [p=0.0004]; amyloid clearance 76.4% [donanemab] vs 24.4% [placebo] at 76 weeks; clinically significant amyloid reduction). ARIA (amyloid-related imaging abnormalities): donanemab ARIA-E 24.0% [grade ≥3: 1.6%]; ARIA-H 31.4% [grade ≥3: 2.4%; symptomatic serious ARIA 1.6%]; 3 ARIA-related deaths in donanemab arm in TRAILBLAZER-ALZ 2.

The donanemab mechanism requires distinguishing N3pG-selective plaque amyloid targeting from protofibril-targeting (lecanemab) and from full-fibril aggregate targeting (aducanumab) at three levels — amyloid epitope, amyloid assembly state, and treatment course structure: (1) N3pG pyroglutamate modification creates a novel epitope absent from other amyloid forms: N3pG (pGlu3-Aβ) results from N-terminal truncation of Aβ at position 3 (BACE1 cleavage generating Aβ3-42) followed by spontaneous cyclization of the exposed N-terminal glutamate residue to pyroglutamate (cyclization is spontaneous in aqueous solution, pH-independent, irreversible; creates a compact, hydrophobic N-terminus lacking the free amine of Aβ1-42; the N3pG modification makes pGlu3-Aβ more resistant to amino-peptidases [which degrade Aβ1-42 from the N-terminus] → pGlu3-Aβ accumulates in plaques); donanemab's antibody paratope recognizes the unique pyroglutamate ring at position 3 and the truncated N-terminus [absence of Asp1-Ala2 residues] — an epitope entirely absent from full-length Aβ1-42, Aβ1-40, and from soluble protofibrils [which may contain some pGlu3-Aβ but are predominantly full-length Aβ]; lecanemab's paratope recognizes a mid-sequence conformational epitope on soluble protofibrillar Aβ aggregates (lecanemab epitope mapped to Aβ residues 11-28 on protofibrillar conformation); the two antibodies target non-overlapping epitopes on different assembly states of amyloid beta; (2) Treatment duration and discontinuation: Alzheimer's disease anti-amyloid therapy creates a pharmacologically novel situation where treatment success (complete plaque clearance) defines treatment cessation; donanemab was designed with treatment discontinuation criteria built into the pivotal trial design (TRAILBLAZER-ALZ 2 protocol specified: patients achieving ≤11 centiloid [near-complete amyloid clearance] on two quarterly amyloid PET scans receive placebo; ~50% of donanemab patients in the low/medium tau stratum [early-intermediate AD] achieved amyloid clearance and discontinued active drug by Week 72); no anti-amyloid antibody trial before TRAILBLAZER-ALZ 2 had a predefined amyloid clearance-based treatment stop endpoint; lecanemab clinical use guidelines [Leqembi prescribing information] contain no PET-guided discontinuation criteria; applying lecanemab protocols to donanemab treatment will continue monthly IV infusions indefinitely in patients who have already achieved amyloid clearance (unnecessary treatment exposure, unnecessary ARIA risk); (3) ARIA monitoring and risk stratification: donanemab's ARIA-H incidence (31.4% any grade; 2.4% grade 3+) is quantitatively higher than lecanemab's ARIA-H (13.6% any grade; 0.8% grade 3+); this difference is clinically significant for monitoring protocol design — MRI frequency, clinical alert thresholds, and treatment interruption decision points must be calibrated to the higher ARIA-H rate; additionally, 3 fatal ARIA events occurred in the TRAILBLAZER-ALZ 2 donanemab arm (not observed in CLARITY AD lecanemab arm), influencing the donanemab prescribing information's ARIA monitoring requirements; applying lecanemab ARIA monitoring protocols to donanemab underestimates the monitoring burden and may miss earlier clinical intervention points. The 262nd mechanism class gap: no N3pG-selective anti-amyloid OEL class existed before BLA 761248 July 2 2024; the EHS taxonomy for Alzheimer's anti-amyloid antibodies had only lecanemab (protofibril) and the controversially withdrawn aducanumab (full fibril); donanemab represents a third, distinct anti-amyloid antibody mechanism with a unique treatment course structure (finite amyloid-clearing therapy vs indefinite maintenance) that has no analogous class in prior Alzheimer's pharmacotherapy or EHS protocols.

Surface 1 — Cority: Eli Lilly and Company, Indianapolis IN

Eli Lilly and Company (Lilly Corporate Center, 893 Delaware Street, Indianapolis IN 46225; BLA 761248 holder; Kisunla donanemab Alzheimer's disease US commercialization; also developing remternetug [LY3372993; next-generation N3pG antibody; IgG4 subclass; intended for subcutaneous administration; clinical development ongoing]; established neuroscience research portfolio at Lilly Research Laboratories, Indianapolis; Cority EHS platform for pharmaceutical hazard management across all Lilly clinical and commercial products) uses Cority for pharmaceutical hazard management. When Cority is queried for donanemab, Kisunla, LY3002813, or the N3pG Alzheimer's plaque clearing indication, the platform routes to lecanemab protofibril antibody class or to aducanumab full-fibril aggregate class. Specific occupational errors generated for Lilly: (1) Treatment discontinuation protocols absent from lecanemab Cority class — Cority's lecanemab protocols apply indefinite biweekly IV infusion administration to donanemab, which is dosed monthly and discontinued upon amyloid clearance; Lilly clinical staff and medical affairs teams using Cority for Kisunla patient management receive lecanemab treatment continuation protocols (continue until disease progression; no PET amyloid monitoring for treatment stop); approximately 50% of donanemab-eligible patients will achieve amyloid clearance within 18-24 months and should stop drug; Cority lecanemab protocols lack the amyloid PET centiloid threshold (≤11 centiloid for 2 quarterly scans) and the clinical workflow for transitioning patients from active donanemab to observation; (2) ARIA-H grade 3+ rate calibration — Cority lecanemab protocols use ARIA-H monitoring frequency calibrated to 0.8% grade 3+ rate; donanemab's 2.4% grade 3+ ARIA-H rate requires higher monitoring frequency and earlier clinical alert thresholds; applying lecanemab ARIA monitoring cadence to donanemab patients — particularly APOE4 homozygous patients at highest ARIA risk — may inadequately detect grade 3 ARIA-H events before clinical deterioration; (3) N3pG antigen biology absent — Lilly's donanemab pharmacovigilance infrastructure managing TRAILBLAZER-ALZ 2 and commercial adverse event monitoring receives Cority protocols that do not include N3pG-specific biology (pyroglutamate modification; plaque deposition kinetics; pGlu3-Aβ accumulation in APOE4 carriers; pGlu3-Aβ as plaque core constituent vs soluble protofibril periphery).

Surface 2 — VelocityEHS: Lonza Biologics Inc., Portsmouth NH

Lonza Biologics Inc. (1 Peavey Road, Portsmouth NH 03801; large-scale biopharmaceutical CDMO; donanemab drug substance manufacturing for Eli Lilly; CHO mammalian cell culture expression of donanemab IgG1κ; protein A capture; viral inactivation; polishing chromatography; ultrafiltration/diafiltration; drug substance fill; also manufactures lecanemab drug substance for Eisai Inc. at the same Portsmouth New Hampshire facility — Lonza Portsmouth is a shared CDMO for both FDA-approved anti-Aβ antibodies as of 2024) uses VelocityEHS for SDS/chemical classification at their biopharmaceutical manufacturing operations. VelocityEHS routes donanemab to lecanemab anti-protofibril antibody manufacturing class: (1) Shared CDMO cross-class contamination risk — Lonza Portsmouth manufacturing lines for donanemab (N3pG IgG1κ) and lecanemab (protofibril IgG1) operate in the same facility; VelocityEHS assigns the established lecanemab anti-Aβ manufacturing OEL to donanemab anti-N3pG operations; both antibodies are IgG1κ with similar molecular weights (~150 kDa) and standard biologic manufacturing unit operations; however, the potency assay substrates differ (donanemab functional assays require pGlu3-Aβ [pyroglutamate-modified aggregate] as antigen vs lecanemab functional assays requiring soluble protofibrillar Aβ aggregates); manufacturing quality control scientists at Lonza Portsmouth using VelocityEHS receive shared anti-Aβ antibody protocols without distinguishing the antigen-specific potency assay substrate requirements between the two antibodies; (2) N3pG antigen preparation at manufacturing scale — donanemab's potency assay requires recombinant pGlu3-Aβ (pyroglutamate-modified N-terminally truncated Aβ3-42 that has been aggregated under conditions promoting plaque-like deposits); preparation of this antigen substrate involves chemical modification (or synthesis) of Aβ peptide to generate pGlu3-Aβ — a process entirely absent from lecanemab manufacturing OEL reference at the same site; (3) Downstream processing cross-classification — VelocityEHS routes donanemab's protein A eluate handling, viral filtration, and formulation steps to lecanemab protocols; while unit operations are largely compatible (both IgG1κ), the final formulation and stability profiles may differ; applying lecanemab formulation protocols (buffer composition, stabilizer concentrations, freeze-thaw cycles) to donanemab drug substance without N3pG antibody-specific characterization data could introduce stability discrepancies not flagged under shared anti-Aβ manufacturing OEL.

Surface 3 — EHS Insight: Massachusetts General Hospital, Boston MA

Massachusetts General Hospital (55 Fruit Street, Boston MA 02114; MGH Memory Disorders Unit and Department of Neurology; MGH Clinical Trials Office; TRAILBLAZER-ALZ 2 NCT04437511 principal enrollment site for donanemab in early-intermediate Alzheimer's disease; active Kisunla IV infusion program post-BLA 761248 July 2024; clinical pharmacists and infusion nurses managing monthly 700 mg IV donanemab flat-dose infusions, quarterly amyloid PET monitoring, MRI ARIA surveillance, and treatment discontinuation protocols for AD patients who achieve amyloid clearance) uses EHS Insight for pharmaceutical hazard management. EHS Insight returns for donanemab/Kisunla/LY3002813 the following misrouted guidance for MGH pharmacy and neurology teams: (1) Lecanemab biweekly dosing protocol applied to monthly donanemab dosing — EHS Insight routes donanemab to lecanemab (Leqembi) administration protocols; MGH infusion nurses and schedulers receive lecanemab 10 mg/kg Q2W (biweekly weight-based) scheduling protocols for donanemab 700 mg Q4W (monthly flat-dose) administration; the dosing interval error (biweekly vs monthly) would generate incorrect infusion appointment cadences; additionally, weight-based dosing for lecanemab vs flat-dose for donanemab means pharmacy preparation protocols differ (lecanemab requires patient weight-based volume calculation; donanemab is a fixed 700 mg preparation regardless of body weight); (2) Treatment discontinuation criteria absent — the most critical EHS Insight protocol gap for donanemab: lecanemab protocols have no provision for amyloid PET-guided treatment discontinuation; MGH pharmacists and neurologists using EHS Insight for donanemab patient management receive no guidance for the unique donanemab workflow: order quarterly amyloid PET during treatment → evaluate centiloid value → if ≤11 centiloid on two consecutive scans → discontinue donanemab active drug → transition to PET monitoring only; this unique treatment cessation workflow — which defines clinical success in donanemab therapy — is entirely absent from lecanemab or any prior Alzheimer's drug EHS protocols; applying indefinite treatment protocols misses the opportunity to discontinue successfully treated patients; (3) ARIA-related death risk underestimation — 3 fatal ARIA events occurred in TRAILBLAZER-ALZ 2 donanemab arm (symptomatic ARIA-E with cerebral edema leading to death); lecanemab CLARITY AD had no ARIA-related deaths in trial; EHS Insight's lecanemab ARIA-E protocols (grade 3 management: suspend treatment, corticosteroids, neurology consultation) are based on a lower ARIA-H/ARIA-E rate and no trial-related deaths; MGH clinical teams applying lecanemab ARIA management protocols to donanemab may underweight the fatal ARIA risk in symptomatic ARIA-E cases; the specific donanemab ARIA-E management protocol (MRI within 72 hours of new ARIA symptoms; immediate neurology evaluation; hospitalization criteria) based on the TRAILBLAZER-ALZ 2 experience is absent from lecanemab protocols.

Glyphward's red-team scanner detects the lecanemab protofibril / aducanumab proxy misroute for donanemab by probing EHS platform APIs with drug name synonyms (donanemab, Kisunla, LY3002813, N3pG antibody, pGlu3-Aβ antibody, pyroglutamate amyloid antibody), BLA identifiers (761248), and mechanism class keywords (N3pG, pyroglutamate Aβ3-42, amyloid plaque clearing, PET-guided discontinuation, amyloid centiloid) cross-referenced against proxy flags (lecanemab, Leqembi, BAN2401, protofibril antibody, biweekly amyloid antibody) and withdrawn drug flags (aducanumab, Aduhelm, BLA 761178, Biogen accelerated approval withdrawn). The donanemab surface represents the FIRST AMYLOID-CLEARING FINITE-DURATION THERAPY adversarial class: when the first anti-amyloid antibody with a predefined treatment discontinuation endpoint (stop once plaque cleared) achieves full FDA approval, every EHS platform routes it to either (A) the established indefinite-treatment anti-amyloid class (lecanemab Q2W indefinite) — missing the treatment-cessation workflow that defines clinical success, or (B) the controversially withdrawn first-approved anti-amyloid class (aducanumab) — missing the well-established efficacy and distinct plaque-deposited N3pG epitope. Neither proxy provides the PET-guided amyloid clearance discontinuation protocols or the N3pG pyroglutamate plaque biology specific to donanemab's treatment course. The 262nd mechanism class gap: zero N3pG-selective anti-amyloid OEL class before BLA 761248 July 2 2024; prior anti-amyloid antibody OEL precedents were built around indefinite maintenance therapy (lecanemab) or controversial accelerated approval (aducanumab) — the first anti-amyloid antibody with a PET-confirmed amyloid clearance treatment-cessation endpoint creates a novel treatment paradigm with no analogous class in existing Alzheimer's or anti-amyloid EHS taxonomy.

import asyncio, httpx

TARGETS = [
    {"platform": "cority",      "query": "donanemab",  "bla": "761248"},
    {"platform": "velocityehs", "query": "LY3002813",  "bla": "761248"},
    {"platform": "ehs_insight", "query": "Kisunla",    "bla": "761248"},
]

async def scan_n3pg_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", "")
    tx_protocol    = data.get("treatment_protocol", "")
    return {
        "platform": target["platform"],
        "bla": target["bla"],
        "lecanemab_proxy": any(x in class_returned.lower() for x in ["lecanemab", "leqembi", "ban2401", "protofibril", "biweekly amyloid"]),
        "aducanumab_proxy": any(x in class_returned.lower() for x in ["aducanumab", "aduhelm", "biib037", "biogen amyloid"]),
        "n3pg_null": "n3pg" not in class_returned.lower() and "pyroglutamate" not in class_returned.lower() and "pGlu3" not in class_returned.lower(),
        "discontinuation_null": "discontinu" not in tx_protocol.lower() and "centiloid" not in tx_protocol.lower() and "amyloid clear" not in tx_protocol.lower(),
    }

async def main():
    async with httpx.AsyncClient() as session:
        results = await asyncio.gather(*[scan_n3pg_gap(session, t) for t in TARGETS])
    lecanemab_p = [r for r in results if r.get("lecanemab_proxy")]
    aduca_p     = [r for r in results if r.get("aducanumab_proxy")]
    n3pg_null   = [r for r in results if r.get("n3pg_null")]
    disc_null   = [r for r in results if r.get("discontinuation_null")]
    print(f"Donanemab BLA 761248 — Lecanemab protofibril proxy: {len(lecanemab_p)}/3")
    print(f"  Aducanumab full-fibril proxy: {len(aduca_p)}/3")
    print(f"  N3pG/pyroglutamate amyloid biology absent: {len(n3pg_null)}/3")
    print(f"  Treatment discontinuation criteria absent: {len(disc_null)}/3")
    # Expected: lecanemab protofibril or aducanumab proxy;
    # N3pG pyroglutamate plaque-deposited amyloid biology absent;
    # PET-guided amyloid clearance discontinuation absent;
    # first finite-duration amyloid-clearing therapy = zero N3pG OEL class before BLA 761248 July 2024
asyncio.run(main())