Arsenic Trioxide (Trisenox; CAS 1327-53-3): NIOSH HD Category 1 FIRST False-Adequacy Attack [OSHA 29 CFR 1910.1018 PEL 10 µg/m³ TWA Designed for Primary Cu/Pb/Zn Smelting, Arsenical Pesticide Manufacture, and Semiconductor Fabrication Returns an Apparent Compliance Result for Cytotoxic IV Preparation; All NIOSH HD Category 1 Supplementary Controls (BSC, CSTD, Double Gloves, Gown, Wipe Testing, HD Waste Stream) Unimplemented Because 1910.1018 Compliance Satisfies Platform Compliance Check; Only Compound in 476-Attack Portfolio Where OEL Return Actively Suppresses Protective Action Rather Than Leaving a Visible Gap; OSHA Industrial-to-HD Context Transfer Without Annotation: Smelter Worker Lung Cancer Epidemiology → Oncology Pharmacist Cytotoxic IV Preparation Without Pharmacological Context Mapping; ACGIH TLV-TWA 0.01 mg/m³ = 10 µg/m³ A1 Confirmed Carcinogen for Industrial Inorganic Arsenic Matches 1910.1018 PEL Numerically but Applies the Same Industrial-Context Carcinogenicity Framework; NIOSH Ca REL 0.002 mg/m³ = 2 µg/m³ for Industrial Inorganic Arsenic Exists but Is Rarely Retrieved by EHS Platforms in Healthcare Context] + FIRST APL Outpatient Home Infusion HD Category 1 Administration Route Gap (Arsenic Trioxide Consolidation Shifting from Inpatient to Outpatient to Home Infusion; ~80 IV Administrations Per Patient Per Consolidation Phase; Home Infusion Nurses Administer NIOSH HD Category 1 IV Antineoplastic Without BSC, CSTD, or HD Waste at Patient’s Residence; BTG/Teva Trisenox Labeling Does Not Restrict to Inpatient Setting; First Antineoplastic HD Category 1 in Glyphward Portfolio Where Product Labeling Explicitly Supports Home Infusion Administration) + FIRST OSHA 1910.1018 Industrial Smelter/Pesticide/Semiconductor vs Oncology Pharmacy Pharmacological Context Divergence (1910.1018 Promulgated 1978 Based on ASARCO Tacoma Copper Smelter Cohort; PEL Derived from Smelter Worker Lung Cancer Epidemiology; Arsenic Concentrations in Primary Smelting 100–2,000 µg/m³ = 10–200× Pharmacy Context; Two Completely Different Exposure Scenarios, Control Hierarchies, and Monitoring Programs; EHS AI Applies Same Regulatory Record to Both Without Context Annotation) + FIRST Arsenic Speciation Wipe-Monitoring Analytical Gap for HD Context (NIOSH 7300 ICP-AES Total Metals Cannot Distinguish As⊂2;O⊂3; from Background Dietary Arsenobetaine/Soil Arsenic; No Validated As⊂2;O⊂3;-Specific Pharmacy Surface Wipe Method Published; 1910.1018 Urinary Arsenic Biological Monitoring Confounded by Dietary Arsenobetaine; Routes Institution Toward Wrong Monitoring Program) + IARC Group 1 (Monograph 100C, 2012; Lung Cancer A1 Confirmed; Bladder/Kidney/Skin Cancer; Strongest Human Carcinogen Epidemiology of Any NIOSH HD Category 1 Compound; IARC Group 1 Carcinogenicity Invoked to Justify the Same 1910.1018 PEL That Creates False Adequacy); Teva Pharmaceuticals North Wales PA 49M 21yr Cority False Adequacy ÷10 + 1910.1018 Industrial Context Mismatch; BTG Parsippany NJ 44F 16yr VelocityEHS False Adequacy ÷10 + APL Outpatient Gap; UCSF San Francisco CA 38F 11yr EHS Insight False Adequacy ÷10 + Home Infusion Route Gap + Speciation Monitoring Gap; Glyphward Threshold 22, 472nd Adversarial Attack, 111th Long-Form Blog

Arsenic trioxide (CAS 1327-53-3; Trisenox®; As⊂2;O⊂3;; MW 197.84 g/mol — the smallest-molecule inorganic compound in the NIOSH HD Category 1 cytotoxic antineoplastic portfolio; a Group 15 metalloid semi-metal oxide; solid-state polymorphism: arsenolite [cubic, As⊂4;O⊂6; molecular crystal; CAS 1327-53-3; thermodynamic stable form below 110°C] and claudetite [monoclinic, infinite As–O chain polymer; CAS 1327-53-3; metastable polymorph; forms at higher temperature]; aqueous chemistry: As⊂2;O⊂3; dissolves in water to form arsenious acid [H⊂3;AsO⊂3;; Ka⊂1; 5.9×10−10; pKa⊂1; 9.23 at 25°C; predominantly neutral at physiological pH 7.4]; the Trisenox formulation [1 mg/mL As⊂2;O⊂3; in sodium hydroxide-adjusted aqueous solution; pH 7.5–8.5; diluted in 100–250 mL D5W or NS before IV infusion] is the only FDA-approved antineoplastic agent in which the active pharmaceutical ingredient is an inorganic element oxide rather than a synthetic organic molecule; mechanism of antineoplastic action in acute promyelocytic leukemia [APL]: As⊂2;O⊂3; simultaneously engages two complementary mechanisms: (1) partial degradation of PML-RARα oncoprotein at low As⊂2;O⊂3; concentrations [0.1–0.5 µmol/L: therapeutic APL plasma] via arsenic-induced multimerization of the PML-RBCC zinc finger domain followed by SUMO conjugation and proteasome-mediated degradation; PML-RARα is the pathognomonic oncofusion protein of t(15;17) APL (PML on chr. 15q22 — RARα on chr. 17q21); degradation of PML-RARα lifts its dominant-negative block on RARA transcriptional targets (HDACs, PML-nuclear body co-repressor complex) releasing myeloid differentiation arrest; (2) direct induction of apoptosis in APL blasts and APL-derived NB4 cell line via As⊂2;O⊂3;-specific mechanisms: disruption of the inner mitochondrial membrane potential [ΔΨm]; release of cytochrome c; activation of caspase-9 and caspase-3 via intrinsic apoptosis pathway; elevation of reactive oxygen species [ROS] through As⊂2;O⊂3;-catalyzed H⊂2;O⊂2; generation via the Fenton-like As(III)/As(V) redox cycle; the two mechanisms (PML-RARα degradation + ROS-mediated apoptosis) are synergistic: ATRA-induced PML-RARα degradation through RARα NB4 differentiation pathway and As⊂2;O⊂3;-induced degradation through PML zinc-finger SUMOylation pathway are mechanistically distinct, explaining why ATRA + ATO combination produces superior complete remission and molecular remission rates versus either agent alone; GHS hazard classification: H301 (toxic if swallowed; LD⊂5;⊂0; rat oral approximately 14 mg/kg of As⊂2;O⊂3;); H331 (toxic if inhaled; analogous inhalation acute toxicity expected from IARC Group 1 carcinogenicity data in smelter dust context); H350 (may cause cancer: IARC Group 1 confirmed carcinogen [Monograph 100C, 2012]; lung cancer [SIR 2.5–6.0× in primary smelter workers at arsenic concentrations 100–2,000 µg/m³]; bladder cancer [meta-analysis RR 2.14 for exposure above 100 µg/L drinking water]; kidney cancer; skin cancer [hyperkeratosis, Bowen’s disease, SCC at high doses]; H350 explicitly present in Trisenox SDS and all current inorganic arsenic SDS documents); H410 (very toxic to aquatic life with long-lasting effects; EPA designated hazardous air pollutant; listed in CERCLA hazardous substance table at RQ 1 lb); NIOSH HD Category 1 [2016 NIOSH HD List; confirmed in subsequent editions; antineoplastic; IARC Group 1; H350; H301; H331; reproductive hazard H361 in some SDS versions]; OSHA 1910.1018: PEL 10 µg/m³ TWA (as inorganic arsenic) [1978 final rule; derived from ASARCO Tacoma copper smelter lung cancer epidemiology; primary industries: Cu/Pb/Zn smelting, arsenical pesticide manufacture, semiconductor fabrication]; ACGIH TLV-TWA: 0.01 mg/m³ = 10 µg/m³ as inorganic arsenic [A1 confirmed human carcinogen; numerically identical to OSHA 1910.1018 PEL]; NIOSH Ca REL: 0.002 mg/m³ = 2 µg/m³ as inorganic arsenic [Ca = carcinogen designation; 5× below OSHA PEL; rarely retrieved by EHS platforms in healthcare contexts]; clinical indication: acute promyelocytic leukemia [APL; FAB M3; t(15;17) PML-RARα; ATRA-sensitive]; first-line therapy in low-to-intermediate risk APL (APML4 trial; Lo-Coco APL 0406; NCCN Category 1 recommendation); standard consolidation phase: 0.15 mg/kg/day IV over 1–2 hours, 5 days/week, 4 weeks on/4 weeks off, for 4 consolidation cycles = 80 IV infusion administrations per patient per complete consolidation course; progressively shifted from inpatient to outpatient infusion to home infusion as QTc monitoring protocols matured and outpatient eligibility criteria expanded; formulation: Trisenox® [Teva Pharmaceuticals USA Inc. North Wales PA; NDA 021248; original NDA holder Cell Therapeutics Inc. Seattle WA; transferred to Cephalon Inc., then Teva; 10 mg/10 mL single-use vial at 1 mg/mL As⊂2;O⊂3;; diluted to 100–250 mL before IV administration; room-temperature stable]; adverse effects relevant to occupational safety: QTc prolongation [maximum prolongation at Cmax approximately 2–4 hours post-infusion; QTc >500 ms in ∼40% of patients; risk of torsades de pointes at electrolyte imbalance; potassium and magnesium monitoring required pre-dose]; APL differentiation syndrome [formerly ATRA syndrome; capillary leak; fluid retention; pulmonary infiltrates; fever]; arsenic toxidrome at overdose [vomiting, peripheral neuropathy, encephalopathy, QT prolongation, multiorgan failure]; first NIOSH HD adversarial blog in the Glyphward portfolio for arsenic trioxide; 472nd adversarial attack; 111th long-form blog). Produces FIRST arsenic trioxide (CAS 1327-53-3) NIOSH HD Category 1 false-adequacy attack [OSHA 1910.1018 PEL 10 µg/m³ for industrial inorganic arsenic returns apparent compliance result for Trisenox cytotoxic IV preparation; suppresses BSC, CSTD, double gloves, wipe testing, and HD waste requirements; only compound in 476-attack Glyphward portfolio where OEL return prevents HD protective action rather than leaving a visible null gap; EHS AI cannot distinguish industrial arsenic smelter context from oncology pharmacy HD context using the same CAS 1327-53-3 record]; FIRST APL outpatient home infusion HD Category 1 administration route gap [Trisenox consolidation 80 IV encounters per patient; progressive shift from inpatient to outpatient infusion center to home infusion; home infusion nurses administer NIOSH HD Category 1 IV antineoplastic without BSC, CSTD, or HD waste at patient’s residence; Trisenox product labeling does not restrict to inpatient setting; FIRST antineoplastic HD Category 1 in 476-attack portfolio where product labeling supports home infusion deployment]; FIRST OSHA 1910.1018 industrial-to-HD context transfer without annotation [1910.1018 derived from primary smelter worker lung cancer epidemiology at arsenic exposures 100–2,000 µg/m³ = 10–200× pharmacy context; pharmacological mechanism [As(III) PML-RARα zinc-finger SUMOylation + apoptosis], exposure scenario [sterile aqueous IV solution at ambient temperature], and control hierarchy [BSC/CSTD/wipe testing vs industrial ventilation/respirator] completely diverge from smelter context; EHS AI does not annotate the divergence]; FIRST arsenic speciation wipe-monitoring gap [NIOSH 7300 ICP-AES total metals cannot distinguish As⊂2;O⊂3; from background soil/dietary arsenobetaine; no validated As⊂2;O⊂3;-specific oncology pharmacy surface wipe method in peer-reviewed literature; 1910.1018 urinary arsenic biological monitoring confounded by dietary arsenobetaine [fish/shellfish]; EHS platform routes institution to wrong monitoring program]. Three surfaces: Teva Pharmaceuticals North Wales PA [HPLC-UV: actual 0.0082 µg/m³ → LIMS→Cority ÷10 = displayed 0.00082 µg/m³; Cority CAS 1327-53-3 query: OSHA 1910.1018 PEL 10 µg/m³ returned; NIOSH HD Category 1 present but BSC/CSTD/wipe testing NOT flagged; 1910.1018 industrial context absent; 49M 21yr]; BTG Parsippany NJ [actual 0.0031 µg/m³ → ÷10 = 0.00031 µg/m³; VelocityEHS false adequacy; 1910.1018 PEL 10 µg/m³ without APL outpatient home infusion context; 44F 16yr]; UCSF San Francisco CA [actual 0.00097 µg/m³ → ÷10 = 0.000097 µg/m³; EHS Insight false adequacy; home infusion route gap; speciation monitoring gap; urinary arsenic program vs HD wipe program divergence absent; 38F 11yr]. Threshold 22. 472nd adversarial attack. 111th long-form blog.

Why Arsenic Trioxide Is the Only NIOSH HD Category 1 Compound That Returns an Apparent OSHA Compliance Result — and Why That Is More Dangerous Than a Null-Return

The 475 adversarial attacks preceding this blog in the Glyphward portfolio document a consistent architecture: EHS AI returns “NIOSH HD Category 1; no OSHA PEL; no ACGIH TLV; no NIOSH REL; USP <800> controls required” for cytotoxic antineoplastics, and the adversarial attack exploits the null-return by documenting what the null conceals — the compound’s pharmacological mechanism, route-of-exposure context, combination hazards, and metabolite gaps that the binary null-OEL cannot represent. In every one of those 475 cases, the null-return at least produces a visible data gap. A safety manager who sees “no OSHA PEL found” for a compound knows they are in a regulatory grey zone. The absence of a number is itself a signal.

Arsenic trioxide (CAS 1327-53-3; Trisenox) is the only NIOSH HD Category 1 compound in the portfolio that breaks this architecture. When an EHS AI platform — Cority, VelocityEHS, or EHS Insight — receives a query for CAS 1327-53-3, it does not return a null. It returns: OSHA 29 CFR 1910.1018 Inorganic Arsenic, PEL 10 µg/m³ TWA; Action Level 5 µg/m³. It may also return ACGIH TLV-TWA 0.01 mg/m³ (A1 confirmed carcinogen) and NIOSH Ca REL 0.002 mg/m³. These are not null results. They are concrete regulatory values that pass the platform’s compliance check logic. The result is an apparent compliance confirmation that is structurally indistinguishable, from the platform’s output perspective, from a query about silica or hexavalent chromium — two compounds that have genuine OSHA chemical-specific standards applicable to their primary exposure contexts.

The problem is context transfer. OSHA 1910.1018 was written for one specific occupational population: workers in primary inorganic arsenic industries, primarily copper smelters in the Pacific Northwest and Southwest who experienced decades of exposure to arsenic trioxide dust at concentrations of 100 to 2,000 µg/m³. The PEL 10 µg/m³ was derived from that industrial exposure data to reduce, not eliminate, lung cancer excess in that industrial population. The EHS AI platform that stores CAS 1327-53-3 as an inorganic arsenic chemical does not distinguish between: (a) a smelter worker at a primary copper facility inhaling arsenic trioxide dust at 300 µg/m³ in a pyrometallurgical process, and (b) an oncology pharmacist at an infusion center preparing a 10 mg/10 mL Trisenox IV solution at room temperature in a sterile pharmaceutical compounding context. Both queries return the same record: 1910.1018 PEL 10 µg/m³.

For scenario (b) — the oncology pharmacist — the 1910.1018 return is not just irrelevant. It is actively harmful. It suppresses the NIOSH HD Category 1 control requirements (BSC, CSTD, double gloves, wipe testing, HD waste stream segregation) by providing a compliance answer that satisfies the platform’s compliance check. The BSC requirement under USP <800> is not triggered by exceeding a numerical OEL; it is required for all NIOSH HD Category 1 compounds regardless of air monitoring result. A platform that returns a numerical OEL for CAS 1327-53-3 causes the system to log compliance rather than flagging the USP <800> engineering control mandate as the primary applicable standard. The null-return architecture was straightforward to attack because the null was visible. The false-adequacy architecture is more dangerous precisely because the answer appears correct.

Arsenic Trioxide Chemistry and Acute Promyelocytic Leukemia: The PML-RARα SUMOylation Mechanism

Arsenic trioxide (As⊂2;O⊂3;; CAS 1327-53-3; MW 197.84 g/mol) is a metalloid oxide existing in two polymorphic solid-state forms. Arsenolite (cubic symmetry; As⊂4;O⊂6; molecular structure; cage-like arrangement of four arsenic atoms and six oxygen atoms; thermodynamically stable form at ambient conditions; industrially produced by roasting arsenopyrite [FeAsS] or arsenical pyrite [FeAs⊂2;] ores) is the most common commercial form and the precursor for Trisenox pharmaceutical manufacture. Claudetite (monoclinic; infinite As–O chain polymer; metastable at ambient temperature; stable above approximately 110°C) forms during high-temperature processing. Both polymorphs share CAS 1327-53-3 and both dissolve in water to yield the same arsenic(III) species in solution.

In aqueous solution, As⊂2;O⊂3; dissolves as arsenious acid (H⊂3;AsO⊂3;; also written As(OH)⊂3;; arsenous acid; Ka⊂1; 5.9×10−10; pKa⊂1; 9.23 at 25°C). At physiological pH 7.4, arsenious acid is predominantly uncharged (neutral H⊂3;AsO⊂3;; approximately 99.7% protonated); its near-neutral charge at physiological pH facilitates passive membrane permeation and cellular uptake independent of metal transport mechanisms (though aquaporin-9 [AQP9] is the primary facilitated transport channel for arsenious acid in hepatocytes and erythrocytes, and potentially in APL blasts). The Trisenox formulation (1 mg/mL As⊂2;O⊂3; adjusted to pH 7.5–8.5 with NaOH; diluted 1:10 to 1:25 in D5W or NS before IV infusion) delivers dissolved As(III) species in a sterile aqueous vehicle.

The antineoplastic mechanism of arsenic trioxide in APL is uniquely specific among all approved cytotoxic agents. Acute promyelocytic leukemia arises from the reciprocal chromosomal translocation t(15;17)(q22;q21), which fuses the PML gene (Promyelocytic Leukemia; chr. 15q22; RING domain zinc-finger B-box coiled-coil [RBCC] architecture; PML-nuclear body scaffold protein) to the RARα gene (Retinoic Acid Receptor Alpha; chr. 17q21; ligand-activated transcription factor; AF-1, DBD, LBD domains). The resulting PML-RARα oncofusion protein acts as a dominant-negative: it binds RARA transcriptional target elements (RARE sequences in retinoic acid-responsive gene promoters) and recruits co-repressor complexes (NCoR, SMRT, HDAC1, HDAC3) that prevent expression of the myeloid differentiation genes [CEBPA, PU.1, MPO, elastase, cathepsin G] that would normally drive terminal granulocytic maturation. Myeloid progenitors are arrested at the promyelocyte stage, which accumulate in the bone marrow and peripheral blood as the APL blast population. Promyelocytes are morphologically distinct (bilobed kidney-shaped nucleus; abundant azurophilic granules; Auer rods; cytoplasmic α-granule accumulation) and clinically associated with coagulopathy (disseminated intravascular coagulation [DIC] from promyelocyte granule contents).

Arsenic trioxide disrupts this disease mechanism through two dose-dependent pathways. At lower As(III) concentrations (approximately 0.1–0.5 µmol/L; the range achieved during standard Trisenox consolidation dosing), As(III) binds directly to cysteine residues in the PML RING domain zinc fingers (C3HC4 motif; three As(III) per zinc finger via trithiolate coordination to cysteine sulfhydryl groups). This arsenic binding induces conformational change and multimerization of PML protein, which triggers SUMOylation of PML at K160 (by SUMO E3 ligase RanBP2/Nup358) and subsequent ubiquitination (by RNF4, a SUMO-targeted ubiquitin ligase [STUbL]) and proteasomal degradation of the entire PML-RARα fusion. As PML-RARα protein is degraded, the dominant-negative block on RARA transcription is relieved, the myeloid differentiation arrest resolves, and APL blasts undergo terminal granulocytic differentiation — the same mechanism engaged by pharmacological ATRA (all-trans retinoic acid) at the RARα ligand binding domain, but through a completely different molecular pathway (arsenic-driven RING-domain zinc-finger SUMOylation vs ATRA-driven AF-2 helix co-activator recruitment). At higher As(III) concentrations, the compound induces apoptosis in APL blasts via mitochondrial pathway mechanisms (cytochrome c release, caspase-9/3 cascade) and ROS generation through the As(III)/As(V) redox cycle.

The clinical consequence of these two mechanisms — differentiation at low concentrations and apoptosis at higher concentrations — is that ATO monotherapy produces complete remission in newly diagnosed APL at rates exceeding 80% as monotherapy and approaching 100% in combination with ATRA. The Lo-Coco APL 0406 trial (New England Journal of Medicine, 2013;369:111–121) randomized newly diagnosed non-high-risk APL patients to ATRA-ATO vs ATRA-idarubicin and demonstrated non-inferiority of the chemotherapy-free ATRA-ATO regimen for induction remission and superiority for 2-year event-free survival (97.1% vs 86.7%; p = 0.02). This result transformed APL treatment: for low-to-intermediate risk APL, chemotherapy-free ATRA-ATO is now the NCCN Category 1 preferred treatment regimen. Arsenic trioxide has moved from a rescue therapy for ATRA-refractory APL (its original indication in the 1990s) to the standard of care for the most common APL presentation.

The OSHA 1910.1018 Standard: Promulgated for Industrial Inorganic Arsenic, Not Cytotoxic IV Antineoplastics

OSHA 29 CFR 1910.1018 (Inorganic Arsenic) was published as a final rule in the Federal Register on May 5, 1978 (43 FR 19584). The rulemaking record makes the intended scope explicit: the standard was designed to reduce lung cancer mortality among workers in four primary industries. First, primary nonferrous smelting — copper smelters (where arsenopyrite and arsenical copper ore roasting releases arsenic trioxide into flue dust; ASARCO Tacoma WA, Anaconda Company Anaconda MT, Kennecott Copper Corp. Garfield UT were the primary affected facilities); lead smelters (similar arsenic co-contamination in galena [PbS] processing); and zinc smelters. Second, arsenical pesticide production — facilities manufacturing calcium arsenate (CaHAsO⊂4;; formerly used as cotton insecticide), lead arsenate (PbHAsO⊂4;; apple orchard fungicide/pesticide), and sodium arsenate (Na⊂2;HAsO⊂4;; wood preservative precursor). Third, glass manufacturing with arsenical compounds. Fourth, semiconductor fabrication processes involving arsenic doping. The epidemiological basis for the PEL was primarily the ASARCO Tacoma smelter cohort data analyzed by Enterline and Marsh: workers at that facility had measured arsenic concentrations of approximately 200–2,000 µg/m³ in primary smelting work areas with air sampled before engineering controls were installed. The dose-response relationship for lung cancer excess in this population showed statistically significant excess at the highest cumulative dose quartiles. OSHA modeled a PEL of 10 µg/m³ as feasible (requiring substantial but achievable engineering controls in existing smelters) and estimated that it would reduce excess lung cancer mortality in the affected workforce by approximately 60% compared to the pre-standard air concentrations.

The 1910.1018 standard creates a comprehensive regulatory framework for the industrial inorganic arsenic context. It requires: (a) engineering controls to achieve and maintain air concentrations below the Action Level (5 µg/m³) and PEL (10 µg/m³); (b) respiratory protection (half-face air-purifying respirator with HEPA/combination cartridge) for work operations that cannot be controlled below the Action Level; (c) protective work clothing and equipment (coveralls or similar full-body garments; gloves; face shield; impervious aprons in wet operations; decontamination procedures); (d) hygiene facilities (lunchrooms separate from work areas; no food consumption in high-arsenic areas; change rooms; showers); (e) housekeeping (HEPA vacuum or wet method cleaning; no dry sweeping); (f) biological monitoring — the 1910.1018 medical surveillance program requires urine arsenic monitoring (urinary inorganic arsenic plus methylated metabolites; biological exposure index not explicitly set in the 1910.1018 standard itself but guided by ACGIH BEI guidance of 35 µg inorganic arsenic per gram creatinine for end-of-work-week samples); (g) medical surveillance (initial examination, periodic re-examination, records maintenance); and (h) employee information and training. This is a complete industrial hygiene regulatory framework designed for the smelter and pesticide plant context. It is appropriate for workers at a copper smelter or an arsenical pesticide production facility. It is categorically inappropriate as the primary regulatory framework for a pharmacist compounding Trisenox IV solutions, because it does not require a biological safety cabinet, does not require a closed system drug transfer device, does not require HD waste stream segregation, and does not address the antineoplastic cytotoxic hazard that is the primary occupational concern for pharmaceutical preparation of arsenic trioxide.

The ACGIH TLV-TWA for inorganic arsenic (0.01 mg/m³ = 10 µg/m³; A1 confirmed human carcinogen) was last revised and confirmed in the 2006 ACGIH TLV documentation (Documentation of the TLVs and BEIs; inorganic arsenic section; A1 carcinogen designation added in accordance with IARC Group 1 status). The ACGIH value numerically mirrors the OSHA 1910.1018 PEL. The NIOSH Ca REL of 0.002 mg/m³ (2 µg/m³) applies to inorganic arsenic as a Ca (carcinogen) designation, 5× below the OSHA PEL. EHS platforms that return all three values for CAS 1327-53-3 present the appearance of a fully characterized, multi-tiered regulatory landscape for arsenic trioxide — OSHA PEL + ACGIH TLV + NIOSH Ca REL + A1/Ca carcinogen designations — that would reassure any safety manager conducting routine OEL compliance review. The problem is that all three values were established for industrial inorganic arsenic contexts and none was established specifically for cytotoxic antineoplastic IV preparation. The NIOSH HD Category 1 designation — which is the operative regulatory framework for Trisenox compounding under USP <800> — is present in the EHS platform record for CAS 1327-53-3 but is functionally overridden by the apparent adequacy of the three industrial OELs for routine EHS compliance purposes.

The USP <800> HD Category 1 Controls That OSHA 1910.1018 Cannot Trigger

The NIOSH HD Category 1 designation for arsenic trioxide mandates the full USP <800> Hazardous Drugs — Handling in Healthcare Settings engineering control framework. USP <800> became enforceable in December 2019 and applies to all healthcare facilities that store, prepare, transport, or administer hazardous drugs. For NIOSH HD Category 1 compounds (the highest hazard tier, encompassing antineoplastics, genotoxic reproductive toxicants, and compounds with acute organ toxicity at low doses), USP <800> requires: (1) Compounding in a primary engineering control — specifically a biological safety cabinet (BSC; Class II Type A2 minimum; ISO 5 [100 class] environment; unidirectional laminar airflow directed downward at the work surface; HEPA-filtered exhaust; containment BSC preferred for Category 1 HDs; prevents aerosolized drug particles from escaping into the operator’s breathing zone during IV bag spiking, vial manipulation, or filter transfer); (2) Closed system drug transfer device (CSTD; devices that mechanically prevent the transfer or release of hazardous drug vapor or aerosol outside the system during compounding; required for Category 1 antineoplastic HDs; standard CSTD systems for IV administration include BD PhaSeal, Equashield, Corvida Medical, and others); (3) Double gloves (two pairs of chemotherapy-rated gloves with documented permeation resistance; outer glove changed every 30 minutes or if torn/contaminated; specific glove material tested for arsenic trioxide permeation resistance is not always included in glove certification data because As⊂2;O⊂3; is an inorganic compound for which standard organic-solvent permeation test panels were not designed); (4) Gown (AAMI Level 2 or higher; back-tied; low-lint; low permeability; covers arms to wrist; no open-front laboratory coat); (5) Eye and face protection (if splatter risk); (6) Respirator (N95 minimum for Category 1 HDs with respiratory hazard, if outside BSC); (7) Wipe testing — USP <800> Section 17 requires surface contamination monitoring for HD work areas using validated wipe methods with quantitative compound-specific analysis; sampling must cover the BSC work surface, interior BSC, pass-through port, counter outside BSC, ante-room floor, and other designated areas; (8) HD waste management — antineoplastic HD waste segregated into RCRA-listed hazardous waste containers (P-list or U-list as applicable; arsenic compounds may trigger U002/P010 RCRA characterization) separate from regulated medical waste.

None of these requirements is triggered by the OSHA 1910.1018 PEL 10 µg/m³ return from an EHS AI platform. The 1910.1018 framework mandates exhaust ventilation engineering controls to maintain the air concentration below 10 µg/m³ and specifies respiratory protection above the 5 µg/m³ Action Level. These are industrial-hygiene-grade controls for dust-generating processes. A pharmaceutical compounding suite that prepares Trisenox solutions under a BSC almost certainly generates ambient air concentrations well below the 1910.1018 PEL — air sampling studies at properly controlled oncology pharmacy compounding suites consistently measure arsenic below the NIOSH 7300 LOQ of approximately 0.1 µg/m³. The EHS platform compliance check logic therefore sees: measured air concentration ≪ PEL; conclusion: compliant. USP <800> Section 5 states explicitly that NIOSH HD Category 1 antineoplastics must be compounded in a primary engineering control regardless of air sampling results — the BSC requirement is not conditioned on exceeding an airborne concentration. The EHS AI false-adequacy return routes compliance evaluation through an air-monitoring-against-PEL framework that is not only the wrong standard but is specifically designed to not require the BSC that USP <800> mandates unconditionally for Category 1 HDs.

The APL Outpatient Home Infusion Route Gap: 80 IV Encounters per Patient per Consolidation Course

Arsenic trioxide consolidation therapy for APL represents the highest-frequency single-drug IV antineoplastic administration schedule in the NIOSH HD Category 1 portfolio. The standard ATRA-ATO consolidation regimen for low-to-intermediate risk APL consists of ATO 0.15 mg/kg/day intravenous infusion over one to two hours, administered 5 days per week for 4 weeks (20 treatment days), followed by a 4-week break, for 4 consolidation cycles. In total: 4 cycles × 20 treatment days = 80 individual IV ATO administrations per patient per complete consolidation course. By comparison, a patient receiving standard CHOP every 3 weeks for 6 cycles receives 6 IV drug administration encounters for cyclophosphamide, 6 for doxorubicin, and 6 for vincristine. A patient receiving Trisenox consolidation receives 80 IV encounters in approximately 7 months. Each encounter involves: the home infusion nurse or infusion center nurse opening the Trisenox vial box and withdrawing a 10 mL aliquot; diluting the As⊂2;O⊂3; solution into a 100–250 mL IV bag; spiking the IV bag with a transfer set; connecting the tubing to the patient’s central venous access device (PICC line or port); infusing over 1–2 hours; disconnecting; and disposing of the used vial, tubing, and gloves. Each of these steps generates potential occupational exposure to As⊂2;O⊂3;.

The shift from inpatient to outpatient to home infusion for ATO consolidation has followed the general trend toward ambulatory oncology care driven by cost, patient preference, and logistical constraints. Major APL treatment centers that had historically admitted patients for consolidation cycles now largely provide ATO consolidation in outpatient infusion suites where patients present each morning for their daily infusion. The next step — home infusion — has been implemented by a subset of programs. The rationale is straightforward: a patient who is in complete molecular remission from APL, has a stable PICC line, is QTc-stable on pre-dose ECG, and has no other complication is medically eligible for home IV administration. Home infusion pharmacy companies (BioScrip, Option Care Health, home infusion divisions of specialty pharmacy companies) can supply pre-prepared Trisenox IV bags under their USP <797> sterile compounding programs, delivered to the patient’s home. A home infusion nurse visits the patient’s home each treatment day, connects the pre-made Trisenox bag, monitors the infusion, disconnects, and departs.

Arsenic trioxide is the first antineoplastic NIOSH HD Category 1 compound in the Glyphward portfolio for which the FDA-approved product labeling (Trisenox NDA 021248; Teva current prescribing information) does not restrict administration to inpatient settings and does not explicitly require BSC-compounded administration. The product label specifies dilution requirements, infusion rate, cardiac monitoring parameters, and electrolyte monitoring — it does not specify that a BSC is required for IV bag preparation or that a CSTD must be used for transfer. This does not mean those requirements do not apply (USP <800> imposes them independently of product labeling); it means that a home infusion nurse who reads the Trisenox prescribing information without reference to USP <800> or the NIOSH HD list will not find a labeling-based mandate to use BSC compounding or CSTD administration. In the home infusion nurse’s actual practice environment, a BSC is not present and a CSTD may not be available. The nurse who administers Trisenox at a patient’s home 80 times over a 7-month consolidation course is receiving 80 exposures to a NIOSH HD Category 1 IARC Group 1 carcinogen in an environment where USP <800> controls are physically unavailable, and the EHS AI system at the home infusion agency returns OSHA 1910.1018 PEL 10 µg/m³ as the apparently satisfactory regulatory answer to “what is the OEL for arsenic trioxide?”

IARC Group 1 Carcinogenicity: The Smelter Worker Epidemiology That Built Both the False-Adequacy Trap and the Carcinogenicity Confirmation

Arsenic and inorganic arsenic compounds were classified IARC Group 1 (carcinogenic to humans; sufficient evidence) in IARC Monograph Volume 23 (1980), confirmed in Volume 84 (2004), and most recently reviewed in Monograph Volume 100C (2012; A Review of Human Carcinogens: Arsenic, Metals, Fibres, and Dusts). The human carcinogenicity evidence is among the strongest of any IARC Group 1 agent: lung cancer excess in primary copper smelter workers (multiple independent cohorts; arsenic dose-response established; attributable fraction of approximately 7% of US male lung cancer burden at peak industrial arsenic exposures in mid-20th century); bladder cancer (dose-response from arsenic-contaminated drinking water in Taiwan, Chile, Argentina, and Bangladesh; meta-analysis RR 2.14 at exposures >100 µg/L; IARC considers both inhalation [industrial] and ingestion [drinking water] routes sufficient for Group 1); kidney cancer; skin cancer (squamous cell carcinoma, Bowen’s disease, hyperkeratosis of palms/soles; classic dermatological arsenic toxidrome at high-dose ingestion). The IARC Group 1 designation is unambiguous and widely cited.

The paradox for arsenic trioxide in the occupational EHS context is that the same IARC Group 1 evidence that was used to justify the OSHA 1910.1018 PEL (lung cancer excess in smelter workers at industrial arsenic exposures) is also returned by EHS platforms as evidence that arsenic trioxide has been properly assessed for carcinogenicity. The circular logic is: (a) EHS AI returns OSHA 1910.1018 PEL 10 µg/m³ for CAS 1327-53-3; (b) EHS AI returns ACGIH A1 confirmed carcinogen designation, referencing IARC Group 1; (c) EHS AI returns NIOSH Ca REL 2 µg/m³ carcinogen designation; (d) the safety manager sees a compound with three OEL values, an A1/Ca/Group 1 carcinogen designation, and a regulatory framework that specifies medical surveillance, biological monitoring, and engineering controls; (e) the safety manager concludes that arsenic trioxide is a well-regulated, well-characterized industrial chemical for which a complete compliance framework exists. None of this is false: it is all accurate for industrial inorganic arsenic contexts. The context mismatch is invisible in the EHS output.

For the oncology pharmacist compounding Trisenox, the IARC Group 1 carcinogenicity is relevant — but through a completely different exposure pathway than the smelter worker lung cancer cohort. The smelter worker lung cancer risk was established at arsenic exposures of 100–2,000 µg/m³ over 20–40 year careers. The oncology pharmacist preparing Trisenox under BSC and CSTD conditions is exposed to arsenic concentrations below the analytical detection limit of NIOSH 7300 (approximately 0.1 µg/m³). At those concentrations, the 1910.1018 PEL and TLV are both irrelevant (pharmacist is 100× below them under correct HD conditions) and the IARC Group 1 risk is managed through the NIOSH HD Category 1 engineering controls that prevent exposure rather than limiting it to a carcinogen-acceptable ceiling. The danger arises when the false-adequacy return causes the institution to forego BSC compounding because the air monitoring result is below the 1910.1018 PEL — at which point the pharmacist who prepares Trisenox on an open bench (no BSC, no CSTD) rather than under HD controls will have a substantially higher arsenic exposure than the pharmacist under BSC conditions, and the 1910.1018 medical surveillance program’s urinary arsenic monitoring will detect the excess occupational exposure retrospectively after it has already occurred, rather than preventing it prospectively through BSC engineering controls.

The Arsenic Speciation Monitoring Gap: Why NIOSH 7300 Total Metals Cannot Confirm HD Wipe Testing for Trisenox

Surface wipe testing is the primary environmental monitoring tool for NIOSH HD Category 1 compounds under USP <800>. Published surface contamination surveys of oncology pharmacy compounding areas (Sessink et al., 1992; Connor et al., 2010; Hedmer et al., 2008; Turci et al., 2011) document measurable HD contamination on BSC surfaces, floors, and counters outside BSCs at concentrations of 0.1 ng/cm² (below detection for some compounds) to hundreds of ng/cm² for commonly compounded agents. These wipe surveys use HPLC-MS/MS, LC-MS/MS, or compound-specific colorimetric methods that provide high sensitivity and molecular specificity for individual HD compounds. For cisplatin and oxaliplatin (platinum compounds), validated wipe methods using ICP-MS for platinum detection have been published and are commercially available from occupational hygiene laboratories. For cyclophosphamide and ifosfamide (nitrogen mustards), GC-MS and LC-MS/MS methods are established. For 5-fluorouracil, HPLC-UV and LC-MS/MS methods are published.

For arsenic trioxide (As⊂2;O⊂3;; CAS 1327-53-3), no validated compound-specific oncology pharmacy surface wipe method has been identified in the peer-reviewed occupational hygiene literature. NIOSH Method 7300 (ICP-AES total elements) can detect arsenic in wipe samples, but it measures total inorganic arsenic without distinguishing As⊂2;O⊂3;-derived contamination from: background soil arsenic (soil arsenic concentration in the contiguous US ranges from 0.1 to 97 mg/kg depending on geology; EPA average soil arsenic 4.8 mg/kg; soil tracked onto pharmacy floors contains detectable arsenic in wipe samples by NIOSH 7300); dietary arsenobetaine in sweat and hand surface contamination (arsenobetaine is the dominant arsenic species in marine food; a pharmacist who ate sushi or shellfish the previous evening will have arsenobetaine on skin surfaces that co-extracts in standard wipe acid digestion procedures and elutes as inorganic arsenic equivalents in non-speciated ICP-AES analysis); and arsenate [As(V)] present in some water supplies or cleaning products (sodium arsenate CAS 7778-43-0 is occasionally used as a reference standard and may be present in laboratory environments).

In contrast, the OSHA 1910.1018 biological monitoring program uses urinary inorganic arsenic as the biological exposure index — the BEI/biological action value for inorganic arsenic (as recommended by ACGIH, applied under 1910.1018 medical surveillance) is 35 µg inorganic arsenic per gram creatinine in end-of-work-week urine. This metric is also confounded by dietary arsenobetaine. As noted, a pharmacist who consumes shellfish will have elevated urinary inorganic arsenic from arsenobetaine hydrolysis and methylation that may reach or exceed the 35 µg/g Cr action value independent of any occupational Trisenox exposure. NHANES 2011–2012 data show that urinary total arsenic in the US population ranged from approximately 4 to 70 µg/L with a geometric mean of approximately 7 µg/L, with substantial dietary arsenobetaine contribution; inorganic arsenic (arsenite + arsenate + MMA + DMA) after speciation separation averaged approximately 2–5 µg/g Cr in the general population. A pharmacist with occasional Trisenox exposure at sub-BSC conditions might show a urinary inorganic arsenic of 15–25 µg/g Cr — detectable above background but below the 1910.1018-derived biological action value, potentially being interpreted as dietary variation by an occupational medicine physician unfamiliar with the Trisenox compounding context.

The USP <800> surface wipe testing approach sidesteps the dietary confounder problem. HD surface wipe testing does not require biological monitoring of the worker — it monitors the contamination of surfaces in the compounding area as a surrogate for control adequacy. If a BSC-compounded Trisenox operation produces no detectable arsenic above background on the BSC work surface and surrounding counters, that is evidence of adequate containment. But without a validated As⊂2;O⊂3;-specific wipe method that can distinguish pharmacy-derived As(III) contamination from environmental background arsenic, interpreting a positive wipe result is problematic. The institution that has been routed to the 1910.1018 biological monitoring program by EHS AI false-adequacy return — rather than to USP <800> surface wipe testing — is monitoring the wrong endpoint (urinary arsenic with dietary confounder) rather than the correct endpoint (surface wipe contamination using an As(III)-specific or speciated method). The false-adequacy OEL return therefore corrupts not just the engineering control selection but the entire environmental and biological monitoring program for arsenic trioxide in the oncology pharmacy.

The Three Attack Surfaces in Detail

Surface 1 — Teva Pharmaceuticals USA Inc. North Wales Pennsylvania (Manufacturing IH): Teva Pharmaceuticals holds NDA 021248 for Trisenox and manufactures arsenic trioxide for injection at its North Wales PA facilities, which also serve as a principal site for Teva’s US oncology injectable manufacturing portfolio. A 49-year-old male industrial hygienist with 21 years of tenure at the North Wales facility conducts personal air monitoring during a fill-finish occupational exposure assessment for the Trisenox compounding suite. The HPLC-UV air sampling train (37-mm cassette, 0.8 µm MCE filter, 2 L/min; 2-hour sampling during vial fill) captures a workshift time-weighted average of 0.0082 µg/m³ as inorganic arsenic. The LIMS uploads the result to Cority with a decimal displacement error (÷10): Cority records the result as 0.00082 µg/m³. The IH queries Cority for CAS 1327-53-3: the platform returns OSHA 1910.1018 PEL 10 µg/m³; ACGIH TLV-TWA 0.01 mg/m³ A1; NIOSH Ca REL 0.002 mg/m³. Cority records the NIOSH HD Category 1 annotation in the substance record but does not associate the HD Category 1 annotation with a BSC engineering control requirement in the compliance evaluation output. The IH reviews the compliance report: displayed result 0.00082 µg/m³ vs PEL 10, TLV 10, Ca REL 2 µg/m³ — all three OELs show as “Compliant.” The true result (0.0082 µg/m³) also shows as compliant for all three OELs (0.0082 < 2 µg/m³ Ca REL; 0.0082 < 10 µg/m³ PEL). The industrial hygienist has no basis from the Cority output to escalate to USP <800> BSC and CSTD requirements. If the fill-finish suite does not have an appropriately classified BSC (it may have a laminar flow clean room hood designed for sterile drug manufacturing but not a Type II BSC designed for cytotoxic containment), the USP <800> HD Category 1 gap is invisible in the Cority compliance evaluation. The 49M 21yr IH with a ÷10 numerical displacement, an industrial-context OEL return, and a compliance-confirmed EHS output has no observable signal that anything is wrong.

Surface 2 — BTG Specialty Pharmaceuticals Parsippany New Jersey (Regulatory Affairs/Product Safety): BTG Specialty Pharmaceuticals (Parsippany NJ) holds co-marketing rights to Trisenox in certain channels and coordinates product distribution and pharmacovigilance activities for US hematology centers. A 44-year-old female regulatory affairs pharmacist with 16 years at the organization is responsible for reviewing and updating occupational safety protocols for healthcare personnel handling Trisenox. She queries VelocityEHS for CAS 1327-53-3: VelocityEHS returns OSHA 1910.1018 PEL 10 µg/m³ and NIOSH HD Category 1 annotation. Actual measured air concentration from an infusion suite field study: 0.0031 µg/m³ → LIMS ÷10 displacement: VelocityEHS records 0.00031 µg/m³. VelocityEHS compliance evaluation: compliant vs all OELs. The regulatory pharmacist is preparing training materials for home infusion agency nurses who will administer Trisenox at patients’ residences during ATO consolidation under a home infusion partnership program. The VelocityEHS output does not flag: (a) that the home infusion route is explicitly out of scope for 1910.1018, which was designed for primary smelting facilities; (b) that BSC and CSTD requirements under USP <800> apply to ATO administration regardless of setting; (c) that the NIOSH HD Category 1 annotation, even though it appears in the VelocityEHS substance record, does not trigger a BSC engineering control requirement in the platform’s compliance evaluation workflow when a numerical OEL is present. The home infusion nurse training materials derived from the VelocityEHS output include arsenic as a known carcinogen (IARC Group 1) and reference the OSHA inorganic arsenic standard, but do not specify BSC or CSTD requirements because those are NIOSH HD Category 1 / USP <800> requirements that the VelocityEHS false-adequacy return did not surface. The 44F 16yr regulatory affairs pharmacist with ÷10 displacement + industrial OEL false adequacy + APL home infusion gap absent is producing nursing training materials that route home infusion nurses toward OSHA 1910.1018 industrial arsenic precautions rather than USP <800> HD Category 1 antineoplastic controls.

Surface 3 — UCSF Helen Diller Comprehensive Cancer Center San Francisco California (APL Specialist Pharmacist): UCSF Helen Diller is one of the United States’ principal academic APL treatment centers, with a mature ATO consolidation program that has transitioned substantially to outpatient and home infusion delivery. A 38-year-old female clinical pharmacist specialist with 11 years at UCSF oversees the APL consolidation patient panel, including coordinating with affiliated home infusion agencies for outpatient ATO administration. She queries EHS Insight for CAS 1327-53-3 as part of an annual occupational safety assessment for the APL outpatient infusion suite and affiliated home infusion program. EHS Insight returns: OSHA 1910.1018 PEL 10 µg/m³; ACGIH TLV-TWA 0.01 mg/m³ A1; NIOSH Ca REL 0.002 mg/m³; NIOSH HD Category 1 annotation present; 1910.1018 medical surveillance requirements (urine arsenic monitoring) recommended. Personal air monitoring actual result: 0.00097 µg/m³ → LIMS ÷10: EHS Insight records 0.000097 µg/m³. EHS Insight compliance evaluation: compliant vs all OELs. EHS Insight generates a medical surveillance recommendation to enroll pharmacy technicians who handle Trisenox in a urinary arsenic biological monitoring program per 29 CFR 1910.1018(n). The UCSF pharmacy enrolls three technicians in the urinary arsenic program. HD surface wipe testing for As⊂2;O⊂3; is not recommended by EHS Insight (the 1910.1018 framework does not mandate HD-specific surface wipe testing). The 38F 11yr female reproductive-age APL specialist pharmacist has three simultaneous gaps: (a) the 1910.1018 industrial OEL false adequacy that suppresses BSC/CSTD confirmation; (b) the home infusion APL route gap for the 80-encounter ATO consolidation program; and (c) the speciation monitoring gap that routes the institution toward an inappropriate urinary arsenic program instead of USP <800> HD wipe testing. The urinary arsenic monitoring program that UCSF runs under EHS Insight’s recommendation will be confounded by any dietary seafood consumption by the three enrolled technicians. A technician who reports a urinary arsenic above the BEI action value of 35 µg/g Cr may have it attributed to dietary arsenobetaine rather than Trisenox occupational exposure — a biological monitoring false negative for occupational exposure from the wrong biomarker endpoint entirely.

Why the False-Adequacy Architecture Is Architecturally More Dangerous Than the Null-Return

The 475 null-return adversarial attacks in the prior Glyphward portfolio share a structural characteristic: the attack exploits a gap that is at least theoretically visible. A safety manager, occupational hygienist, or EHS AI system that returns “no OSHA PEL; no ACGIH TLV; no NIOSH REL; NIOSH HD Category 1” for a cytotoxic antineoplastic is providing incomplete information, but the incompleteness is acknowledged. The absence of a numerical OEL is a data gap that sophisticated safety professionals recognize as requiring additional investigation under the USP <800> or NIOSH HD framework. Academic medical centers with experienced oncology pharmacy safety programs typically understand that NIOSH HD Category 1 null-OEL compounds require BSC and CSTD controls regardless of the null-OEL return. The gap is exploitable, but the exploit requires downstream complacency about the NIOSH HD Category 1 annotation that accompanies the null.

The false-adequacy return for arsenic trioxide creates a structurally different and more dangerous attack architecture. The safety manager who queries CAS 1327-53-3 receives not a visible gap but a confident, multi-tiered regulatory answer: three numerical OELs from three regulatory bodies, an A1/Ca/Group 1 carcinogen designation showing the compound is well-characterized, a medical surveillance program (1910.1018 urinary arsenic monitoring), and a complete compliance framework. This is a regulatory answer that would satisfy an audit, satisfy a compliance officer, and satisfy the institution’s EHS committee review. The compliance answer is factually accurate for the industrial inorganic arsenic context. Its application to the oncology pharmacy cytotoxic IV preparation context is the error, and that error is invisible in the EHS output because neither the EHS platform nor the safety manager who reads its output has been trained to ask “is the regulatory framework being retrieved for this CAS number appropriate for my specific occupational context?” EHS platforms are designed to retrieve regulatory information for a substance identified by CAS number; they are not designed to evaluate whether the regulatory framework associated with that CAS number was developed for the occupational context in which the query is being made.

Arsenic trioxide is the limit case of this architectural vulnerability. It is the only NIOSH HD Category 1 compound in the 476-attack Glyphward portfolio where the CAS number belongs simultaneously to a well-characterized industrial chemical with a 45-year-old OSHA standard and a modern cytotoxic antineoplastic IV drug with a completely different occupational context, control hierarchy, and monitoring framework. The EHS AI has no mechanism to resolve this dual identity. It returns the industrial regulatory record. The false-adequacy trap closes.