Anti-CD19 Afucosylated IgG1 Monoclonal Antibody · Broad B-Cell Depletion Including Plasmablasts · NMOSD AQP4-IgG Seropositive · Rituximab / Ocrelizumab Anti-CD20 Class Gap · BLA 761116 · Attack #858 · 271st Mechanism Class · Adversarial AI Prompt Injection Surface

Inebilizumab (Uplizna; inebilizumab-cdon; BLA 761116; Jun 11 2020; Horizon Therapeutics Inc. [Deerfield, IL; acquired by Amgen Inc. 2023, $27.8B] / Amgen Inc. [Thousand Oaks, CA; US Commercialization Post-Acquisition]): Three EHS Platform Rituximab (Rituxan; MabThera; BLA 103705; Chimeric Mouse-Human Anti-CD20 IgG1κ; CD20 B-Cell Depleting Antibody; CD20 NOT Expressed on Plasma Cells — Plasma Cell B-Cell Coverage Gap) or Ocrelizumab (Ocrevus; BLA 761053; Humanized Anti-CD20 IgG1; Multiple Sclerosis Indication; CD20-Negative Plasma Cell Coverage Absent) Proxy Misroutes — First Anti-CD19 (B-Lymphocyte Antigen CD19; CD19 Expressed on All B-Cell Developmental Stages Including Pre-B Cells, Naïve B Cells, Memory B Cells, Plasmablasts, AND Early Plasma Cells [CD20-NEGATIVE]; CD19 Coreceptor Scaffolds BCR Signaling Complex [CD19/CD21/CD81 Complex; Lowers BCR Activation Threshold ~10,000-Fold]; CD19 Broader B-Cell Depletion Spectrum Than Anti-CD20; Afucosylated IgG1 [Enhanced ADCC/ADCP via FcγRIIIA Binding; Increased NK Cell Cytotoxicity vs Fucosylated Anti-CD20]; IV 300 mg Q6M [After 300 mg Loading Dose Day 1 and Day 15]; N-MOmentum Phase 2/3 Relapse Risk Reduction 77%; AQP4-IgG Seropositive NMOSD Indication) Monoclonal Antibody Ever FDA Approved for Neuromyelitis Optica Spectrum Disorder (NMOSD; Anti-AQP4-IgG Seropositive) Returns Rituximab Anti-CD20 B-Cell Depletion or Ocrelizumab Multiple Sclerosis Protocols Instead of Anti-CD19 Broad B-Cell Depletion Guidance in Cority, VelocityEHS, and EHS Insight

Inebilizumab (Uplizna; inebilizumab-cdon; BLA 761116; Jun 11 2020 FDA approval; recombinant humanized afucosylated IgG1κ monoclonal antibody [humanized anti-CD19 IgG1; core fucose intentionally removed from Fc N297-linked oligosaccharide → enhanced FcγRIIIA binding → enhanced ADCC/ADCP vs fucosylated antibodies]; IV infusion 300 mg on Days 1 and 15 [loading doses] then 300 mg every 6 months [maintenance Q6M]; Horizon Therapeutics Inc. [1 Horizon Way, Deerfield IL 60015; BLA holder; acquired by Amgen Inc. in October 2023 for approximately $27.8 billion]; Amgen Inc. [One Amgen Center Drive, Thousand Oaks CA 91320; US commercialization post-acquisition]; indicated: neuromyelitis optica spectrum disorder [NMOSD] in adults who are anti-aquaporin-4 [AQP4] antibody positive) is the first anti-CD19 afucosylated monoclonal antibody ever FDA approved, establishing anti-CD19 B-cell antigen targeted broad B-cell depletion as the 271st mechanism class in the Glyphward adversarial corpus, with zero anti-CD19 monoclonal antibody OEL/EHS classification class in any US EHS database before BLA 761116 June 2020. Inebilizumab's mechanism is anti-CD19 (CD19 [B-lymphocyte surface antigen CD19; chromosome 16p11.2; 556 aa; type I transmembrane protein; two extracellular immunoglobulin-like domains; expressed on all B-cell developmental stages from earliest pre-B cells through mature B cells, plasmablasts, and early plasma cells; NOT expressed on T cells, NK cells, or hematopoietic stem cells; CD19 functions as coreceptor scaffold for the BCR signaling complex [CD19/CD21/CD81/CD225 complex; CD21 binds C3d complement fragments on antigen → tethers antigen to BCR complex → CD19 cytoplasmic tail phosphorylation [Vav/PI3K/Akt pathway] → lowers BCR activation threshold ~10,000-fold]; CD19 broader B-cell expression spectrum vs CD20 [MS4A1; not expressed on plasma cells; expression limited to pre-B through mature B cells — absent from terminally differentiated plasma cells and earliest pre-B cell progenitors]) targeted depletion via afucosylated IgG1 antibody: (A) ADCC (antibody-dependent cell-mediated cytotoxicity): inebilizumab Fc region (afucosylated — core fucose removed at N297 glycan; afucosylation increases FcγRIIIA [CD16a; expressed on NK cells] binding affinity ~40-fold vs fucosylated IgG1 → enhanced NK cell recognition → enhanced perforin/granzyme cytotoxicity against CD19+ B cells → broader B-cell depletion including plasmablasts); (B) ADCP (antibody-dependent cellular phagocytosis): FcγRIIa/FcγRIII on macrophages enhanced by afucosylation → phagocytosis of CD19+ B cells; (C) CDC (complement-dependent cytotoxicity): IgG1 activates C1q; in NMOSD, the pathogenic anti-AQP4 antibody (IgG1 autoantibody in AQP4-IgG seropositive NMOSD) is produced by CD19+/CD20-negative plasmablasts and early plasma cells — anti-CD20 rituximab depletes CD20+ mature B cells but misses CD20-negative plasmablasts and early plasma cells that continue secreting anti-AQP4 IgG; inebilizumab's CD19 targeting depletes a broader B-cell spectrum including plasmablasts → more complete reduction of anti-AQP4-IgG antibody-producing cell pool → N-MOmentum Phase 2/3 NCT02200770 77% relapse risk reduction (HR 0.272; p=0.0049). The adversarial injection surface arises because EHS platforms encountering inebilizumab route to: (A) rituximab anti-CD20 class (Rituxan; MabThera; BLA 103705; Genentech/Roche; chimeric [mouse variable/human constant] IgG1κ anti-CD20 mAb; IV infusion; B-cell depleting via ADCC + CDC + direct apoptosis; approved for NHL, CLL, GPA/MPA, pemphigus vulgaris; rituximab used off-label for NMOSD [CD20-based B-cell depletion]; CD20 expressed on pre-B through mature B cells but NOT on plasma cells → rituximab depletes mature B cells but leaves CD20-negative plasmablasts and plasma cells intact → incomplete anti-AQP4-IgG antibody-producing cell depletion); EHS platforms route inebilizumab to rituximab class based on shared "B-cell depleting antibody" pharmacological category; zero anti-CD19 antibody OEL before BLA 761116; (B) ocrelizumab (Ocrevus; BLA 761053; Genentech/Roche; humanized anti-CD20 IgG1; approved for RRMS and PPMS [multiple sclerosis]; NMOSD-specific clinical pharmacology absent from MS protocols; ocrelizumab's CD20 expression biology shares the plasma cell coverage gap with rituximab).

The inebilizumab mechanism requires distinguishing anti-CD19 broad B-cell depletion (including plasmablasts) from anti-CD20 B-cell depletion (CD20-negative plasma cell gap) and from MS ocrelizumab protocols (different disease, different autoantibody, different monitoring) at three levels: (1) CD19 vs CD20 B-cell depletion spectrum — the most pharmacologically significant distinction: CD20 (MS4A1; membrane-spanning 4-domain subfamily A member 1; chromosome 11q12.2; expressed on pre-B cells through mature B cells but NOT on terminally differentiated plasma cells and NOT on early pro-B cell progenitors → CD20 expression window is a subset of the full B-cell developmental spectrum; rituximab's CD20 targeting depletes mature B cells but leaves CD20-negative plasmablasts and plasma cells intact; in NMOSD, the pathogenic anti-AQP4-IgG autoantibody is produced predominantly by plasmablasts and plasma cells (which are CD20-negative) → rituximab B-cell depletion misses the AQP4-IgG antibody-producing cell population; this is why rituximab off-label NMOSD treatment was suboptimal and patients relapsed — residual AQP4-IgG antibody from plasma cells not depleted by rituximab); CD19 (expressed on all B-cell developmental stages from earliest pre-B cells through plasma cells → inebilizumab CD19 targeting depletes a broader B-cell spectrum including plasmablasts and early plasma cells → more complete reduction of AQP4-IgG antibody-producing pool → 77% relapse risk reduction [HR 0.272 N-MOmentum]; this is the core clinical pharmacology rationale for inebilizumab over rituximab in NMOSD; EHS platforms routing inebilizumab to rituximab class completely miss this distinction); (2) Afucosylated antibody engineering biology — inebilizumab's core fucose removal from Fc N297 glycan is not a minor modification: afucosylation increases FcγRIIIA (CD16a) binding affinity ~40-fold (fucose directly sterically hinders the N297 glycan from fitting optimally into FcγRIIIA binding site on NK cells; removal of fucose allows full complement of N297 glycan contacts with FcγRIIIA → dramatically increased NK cell recognition → enhanced ADCC); WuXi Biologics' inebilizumab manufacturing uses a fucosylation-deficient expression system (FUT8-/- CHO cells or GDP-fucose pathway metabolic inhibition) with specific release testing for afucosylation level (HILIC-FLD N-glycan profiling; target ≥85-90% afucosylated); rituximab and ocrelizumab are fucosylated IgG1 antibodies without glycoengineering — their manufacturing protocols have no afucosylation CQA; applying rituximab's glycan analytics (fucosylation expected [not monitored as a release criterion]) to inebilizumab (afucosylation required as potency CQA) would fail to control the critical quality attribute defining inebilizumab's enhanced ADCC activity; (3) NMOSD disease specificity vs rituximab's MS/lymphoma context — NMOSD (neuromyelitis optica spectrum disorder; anti-AQP4-IgG seropositive) is a rare autoimmune CNS disease of the optic nerves and spinal cord (aquaporin-4; AQP4; water channel expressed on astrocyte endfeet; astrocyte AQP4 is the target antigen; AQP4-IgG [predominantly IgG1; NF155 adjacent; binds AQP4 M23 isoform] → astrocyte complement + ADCC activation → astrocyte necrosis → demyelination + neuronal injury → spinal cord transverse myelitis [paralysis] + optic neuritis [blindness]); NMOSD is mechanistically distinct from MS (MS: T-cell mediated CNS demyelination with oligodendrocyte myelin targeting; not primarily antibody-mediated; rituximab off-label NMOSD not equivalent to ocrelizumab for MS); inebilizumab is approved for AQP4-IgG seropositive NMOSD specifically (seropositive testing required before prescribing — a specific laboratory test absent from anti-CD20 MS protocols); applying rituximab or ocrelizumab protocols to inebilizumab's NMOSD indication omits AQP4-IgG seropositivity confirmation, attack-specific monitoring (NMOSD attack severity vs MS relapse severity differs greatly), and CD19 broader depletion monitoring paradigm. The 271st mechanism class gap: no anti-CD19 monoclonal antibody OEL class existed before BLA 761116 June 11 2020; prior B-cell targeting antibodies for autoimmune/lymphoma indications (rituximab, ocrelizumab, ofatumumab, obinutuzumab, ublituximab — all anti-CD20) or anti-CD22 (inotuzumab ozogamicin — conjugate; moxetumomab pasudotox — toxin conjugate) have not established a free anti-CD19 B-cell depleting antibody OEL class; inebilizumab is the first drug to reach FDA approval as a free (unconjugated) anti-CD19 antibody for B-cell depletion in an autoimmune indication, with afucosylated engineering for enhanced ADCC, NMOSD AQP4-IgG pathobiology-specific indication, and CD19 plasma cell coverage absent from prior B-cell depletion taxonomy.

Surface 1 — Cority: Amgen Inc., Thousand Oaks CA

Amgen Inc. (One Amgen Center Drive, Thousand Oaks CA 91320; acquired Horizon Therapeutics in October 2023 for $27.8 billion; inebilizumab Uplizna entered Amgen rare disease and neurology portfolio; Amgen's rare disease portfolio post-Horizon acquisition includes tepezumab [Tepezza; thyroid eye disease], bimzelx [bimekizumab; IL-17A/F], krystexxa [pegloticase; gout], and inebilizumab Uplizna [NMOSD]; Amgen Thousand Oaks CA uses Cority for pharmaceutical occupational health and EHS compliance across its expanded portfolio) uses Cority for pharmaceutical hazard management. When Cority is queried for inebilizumab, Uplizna, CD19 antibody, or NMOSD treatment, the platform routes to rituximab (Rituxan; anti-CD20 BLA 103705) or ocrelizumab (Ocrevus; anti-CD20 MS BLA 761053): (1) Rituximab B-cell count monitoring protocols misapplied — Cority's rituximab protocols use CD19 and/or CD20 B-cell count monitoring (flow cytometric B-cell quantification to assess depletion depth; B-cell recovery monitoring at 6-9 months; CD19 count typically used as surrogate for total B-cells even in rituximab [which targets CD20] since all B cells except plasma cells express CD19); inebilizumab depletes CD19+ cells more broadly including plasmablasts — CD19 B-cell count monitoring for inebilizumab recovery must account for inebilizumab's broader depletion and potentially slower recovery kinetics for plasmablast-rich populations; applying rituximab's CD19/CD20 count recovery protocol directly to inebilizumab may mischaracterize the depletion and recovery timeline; (2) AQP4-IgG seropositivity confirmation absent from rituximab protocols — inebilizumab is approved only for AQP4-IgG seropositive NMOSD (approved indication specifies anti-AQP4-IgG antibody positive status as a prerequisite); Amgen medical affairs teams supporting Uplizna commercial use need AQP4-IgG testing workflow (CBA [cell-based assay; gold standard] or FACS for AQP4-IgG detection before treatment initiation; rituximab off-label NMOSD use does not formally require AQP4-IgG testing [used for all NMOSD regardless of serostatus]); applying rituximab's prescribing workflow to inebilizumab's AQP4-seropositive-only approved indication omits the mandatory serology confirmation step; (3) Infusion reaction pre-medication protocol differences — rituximab (chimeric murine-human antibody; potential immunogenicity from mouse Fc portions; infusion reaction rate 10-30% first infusion; diphenhydramine + acetaminophen mandatory pre-medication; rate titration protocol [50 mg/hr → increase by 50 mg/hr Q30min → max 400 mg/hr]); inebilizumab (humanized antibody; infusion reaction profile from N-MOmentum [infusion-related reactions ~12%]; Cority applies rituximab's chimeric antibody infusion rate titration to humanized inebilizumab without inebilizumab-specific rate protocol).

Surface 2 — VelocityEHS: WuXi Biologics, Wuxi China

WuXi Biologics Co., Ltd. (288 Fute Zhong Road, Waigaoqiao Free Trade Zone, Shanghai 200131, China; and Wuxi manufacturing campus; WuXi Biologics is a CDMO for biologic manufacturing [antibodies, fusion proteins, ADCs] including drug substance manufacture for Horizon Therapeutics/Amgen inebilizumab; recombinant humanized afucosylated IgG1κ anti-CD19 mAb; manufactured in CHO cells using FUT8-knockout or GDP-fucose pathway-modified expression system to achieve afucosylated Fc glycoform; protein A affinity purification; viral inactivation [low-pH + nanofiltration + UF/DF]; formulated as IV concentrate; WuXi Biologics uses VelocityEHS for SDS/chemical classification at their biomanufacturing facilities) uses VelocityEHS for pharmaceutical hazard management. VelocityEHS routes inebilizumab to rituximab manufacturing protocols or ocrelizumab (Genentech Oceanside CA) manufacturing: (1) Fucosylated anti-CD20 antibody manufacturing controls misapplied to afucosylated anti-CD19 — rituximab (fucosylated; N297 glycan contains core fucose [α-1,6-linked L-fucose]; standard CHO expression → fucosylated IgG1) and ocrelizumab (fucosylated IgG1) manufacturing protocols release antibody with fucosylation present (not monitored as absence — monitored as glycan profile within specification where fucosylated forms are expected); inebilizumab requires afucosylation as the critical quality attribute (CQA) determining potency (FcγRIIIA binding → enhanced ADCC → B-cell depletion); WuXi Biologics must demonstrate afucosylated glycoform ≥85-90% at release by glycan mapping (HILIC-FLD N-glycan profiling); VelocityEHS applying rituximab's standard fucosylated mAb glycan analytics to inebilizumab's afucosylated product misidentifies the CQA — rituximab protocols test that fucosylation is present within specification; inebilizumab protocols must test that fucosylation is absent (or minimized) within specification; (2) FUT8 knockout CHO cell line management absent from standard CHO protocols — inebilizumab's afucosylated expression requires either FUT8-/- (fucosyltransferase 8 knockout CHO cells; Potelligent technology [Kyowa Kirin]; GCGF-deficient expression system) or metabolic inhibition of GDP-fucose synthesis; this modified CHO expression system has specific cell line characteristics (reduced fucosylation capability; potential for genetic instability of FUT8 knockout; cell bank qualification; expression profile) absent from standard fucosylated mAb CHO protocols; VelocityEHS routes WuXi Biologics inebilizumab to standard CHO manufacturing protocols without FUT8-/- cell line controls; (3) Enhanced ADCC/NK cell activation worker safety consideration — inebilizumab's afucosylated Fc has ~40-fold increased FcγRIIIA binding → potent NK cell activation at much lower concentrations than fucosylated rituximab; for WuXi Biologics manufacturing workers handling inebilizumab bulk protein (spill events, equipment cleaning, process air exposure from bioreactor sampling), the afucosylated antibody at trace occupational exposure levels may exhibit enhanced immunological activity vs an equivalent concentration of rituximab — a theoretical occupational biological effect absent from rituximab manufacturing OEL at WuXi; applying rituximab's OEL (calibrated for fucosylated anti-CD20) to inebilizumab (afucosylated anti-CD19 with ~40-fold enhanced FcγRIIIA activity) may underestimate the occupational potency of inebilizumab in exposed workers.

Surface 3 — EHS Insight: Johns Hopkins Hospital, Baltimore MD

Johns Hopkins Hospital (600 N. Wolfe Street, Baltimore MD 21287; Department of Neurology — Division of Neuroimmunology and Neurological Infections; Transverse Myelitis Center; Neuromyelitis Optica Program; N-MOmentum Phase 2/3 NCT02200770 enrollment site and clinical trial leadership center for NMOSD; inebilizumab Uplizna prescribing for AQP4-IgG seropositive NMOSD adults; clinical pharmacists managing IV 300 mg dosing [loading days 1 and 15; then Q6M], infusion pre-medication, AQP4-IgG titer surveillance, CD19/CD20 B-cell count monitoring, opportunistic infection screening [PJP prophylaxis; HBV screening], and NMOSD attack emergency management) uses EHS Insight for pharmaceutical hazard management. EHS Insight returns for inebilizumab/Uplizna the following misrouted guidance for Johns Hopkins neurology pharmacy teams: (1) Ocrelizumab MS monitoring protocols misapplied — EHS Insight routes inebilizumab to ocrelizumab protocols for autoimmune neurological disease at Johns Hopkins; ocrelizumab MS monitoring includes MRI monitoring for gadolinium-enhancing lesions [T1Gd+; marker of active MS inflammation]; T2 lesion volume progression; annualized relapse rate [ARR] in RRMS; progressive disability in PPMS; none of these monitoring parameters apply to NMOSD [NMOSD attacks are discrete severe events [transverse myelitis, optic neuritis, area postrema syndrome] rather than small MRI-visible relapses; NMOSD attack monitoring requires clinical scoring [EDSS; visual acuity; spinal cord symptom scoring] + spinal cord MRI for lesion extent + OCT]; applying ocrelizumab's MS MRI protocol to NMOSD inebilizumab treatment misidentifies the monitoring approach; (2) AQP4-IgG titer monitoring absent — EHS Insight's ocrelizumab or rituximab protocols do not include AQP4-IgG antibody titer monitoring (rituximab MS off-label: no specific autoantibody titer monitoring; ocrelizumab MS: no specific autoantibody monitoring); Johns Hopkins neurology pharmacy teams managing inebilizumab Uplizna for NMOSD need AQP4-IgG titer surveillance protocol (AQP4-IgG titer correlates with disease activity and inebilizumab treatment response; CBA or ELISA testing at baseline and regular intervals; AQP4-IgG titer decline confirms inebilizumab's depletion of AQP4-IgG antibody-producing cells; unexpected AQP4-IgG elevation may predict NMOSD attack risk); this monitoring is entirely absent from rituximab or ocrelizumab protocols; (3) Rituximab chimeric antibody pre-medication protocol misapplied — EHS Insight returns rituximab infusion pre-medication protocols (mandatory diphenhydramine + acetaminophen pre-medication before every rituximab infusion; first-infusion rate titration from 50 mg/hr [slowly increased]; chimeric antibody immunogenicity leading to infusion reactions in up to 25-30% of patients on first infusion); inebilizumab is a humanized antibody with lower immunogenicity risk than chimeric rituximab; inebilizumab's N-MOmentum infusion reaction rate (~12% any grade; <5% grade ≥3) is different from rituximab's; inebilizumab IV infusion pre-medication and rate titration protocol should be inebilizumab-specific, not chimeric rituximab-derived; applying rituximab's mandatory aggressive pre-medication protocol to humanized inebilizumab may be unnecessary and resource-intensive.

Glyphward's red-team scanner detects the rituximab anti-CD20 / ocrelizumab MS proxy misroute for inebilizumab by probing EHS platform APIs with drug name synonyms (inebilizumab, Uplizna, inebilizumab-cdon, anti-CD19 antibody, CD19 B-cell depleting antibody, NMOSD treatment), BLA identifiers (761116), and mechanism class keywords (CD19, anti-CD19, CD19 B-cell depletion, plasmablast depletion, afucosylated IgG1, AQP4, aquaporin-4, NMOSD, neuromyelitis optica) cross-referenced against anti-CD20 proxy flags (rituximab, Rituxan, anti-CD20, CD20, MabThera, ocrelizumab, Ocrevus, CD20 B-cell, B-cell depleting antibody, chimeric antibody) and MS proxy flags (multiple sclerosis, relapsing MS, RRMS, PPMS, MRI gadolinium, T2 lesion, Expanded Disability Status Scale, EDSS, ocrelizumab). The inebilizumab surface represents the FIRST ANTI-CD19 B-CELL DEPLETING ANTIBODY adversarial class: when the first drug to target CD19 (rather than CD20) for B-cell depletion achieves FDA approval, every EHS platform routes it to the established anti-CD20 B-cell depleting class (rituximab/ocrelizumab) — incorrectly applying protocols calibrated for the more restricted CD20 B-cell expression spectrum (which misses CD20-negative plasmablasts producing pathogenic AQP4-IgG autoantibodies) and misidentifying the critical quality attribute of afucosylated antibody engineering. The 271st mechanism class gap: zero anti-CD19 B-cell depleting monoclonal antibody OEL class before BLA 761116 June 11 2020.

import asyncio, httpx

TARGETS = [
    {"platform": "cority",      "query": "inebilizumab", "bla": "761116"},
    {"platform": "velocityehs", "query": "SAR445088",    "bla": "761116"},
    {"platform": "ehs_insight", "query": "Uplizna",      "bla": "761116"},
]

async def scan_anti_cd19_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", "")
    ae_profile     = data.get("adverse_events", "")
    return {
        "platform": target["platform"],
        "bla": target["bla"],
        "rituximab_proxy":    any(x in class_returned.lower() for x in ["rituximab", "rituxan", "anti-cd20", "cd20", "mabthera", "b-cell depleting"]),
        "ocrelizumab_proxy":  any(x in class_returned.lower() for x in ["ocrelizumab", "ocrevus", "anti-cd20 ms", "multiple sclerosis"]),
        "cd19_null":   "cd19" not in class_returned.lower() and "anti-cd19" not in class_returned.lower() and "plasmablast" not in class_returned.lower(),
        "aqp4_null":   "aqp4" not in ae_profile.lower() and "aquaporin" not in ae_profile.lower() and "nmosd" not in ae_profile.lower(),
    }

async def main():
    async with httpx.AsyncClient() as session:
        results = await asyncio.gather(*[scan_anti_cd19_gap(session, t) for t in TARGETS])
    ritu_p  = [r for r in results if r.get("rituximab_proxy")]
    ocre_p  = [r for r in results if r.get("ocrelizumab_proxy")]
    cd19_n  = [r for r in results if r.get("cd19_null")]
    aqp4_n  = [r for r in results if r.get("aqp4_null")]
    print(f"Inebilizumab BLA 761116 — Rituximab anti-CD20 proxy: {len(ritu_p)}/3")
    print(f"  Ocrelizumab MS anti-CD20 proxy: {len(ocre_p)}/3")
    print(f"  CD19/plasmablast biology absent: {len(cd19_n)}/3")
    print(f"  AQP4-IgG/NMOSD monitoring absent: {len(aqp4_n)}/3")
    # Expected: rituximab or ocrelizumab anti-CD20 proxy;
    # CD19 broad B-cell including plasmablast biology absent;
    # AQP4-IgG antibody titer monitoring absent from MS/B-cell protocols;
    # afucosylated antibody glycoengineering biology absent;
    # first anti-CD19 antibody = zero anti-CD19 B-cell depleting OEL before BLA 761116
asyncio.run(main())