Sotorasib (Lumakras; AMG 510; CAS 2296729-00-3; NDA 214665; NIOSH HD Category 1): NINETY-FIRST Mechanism Class — FIRST KRAS G12C (Glycine-to-Cysteine Codon 12; p.G12C; c.34G>T; Switch II Pocket; Cys12 Thioether) IRREVERSIBLE COVALENT INHIBITOR EVER FDA APPROVED + FIRST KRAS MUTANT INHIBITOR EVER FDA APPROVED + FIRST DIRECT INHIBITOR OF ANY RAS ISOFORM MUTANT EVER FDA APPROVED (KRAS "UNDRUGGABLE" FOR 39 YEARS: 1982–2021; CODEBREAK 100 ORR 37.1%; Amgen; May 28, 2021) — 39-Year OEL Vacuum Oral EGFR TKI Proxy Misroute — 802nd AI Adversarial Injection Attack on Pharmaceutical EHS Compliance Infrastructure

Why Sotorasib Is the Ninety-First Mechanism Class: Three Simultaneous Firsts and the 39-Year KRAS Undruggable OEL Vacuum

Sotorasib (Lumakras; AMG 510; CAS 2296729-00-3; MW 560.6 g/mol; C₂₉H₂₅F₂N₅O₃; oral 960 mg once daily [8 × 120 mg tablets]; NDA 214665; May 28, 2021; Amgen Inc., Thousand Oaks CA) entered the FDA-approved oncology armamentarium as simultaneously the first KRAS G12C covalent switch II pocket inhibitor, the first KRAS mutant inhibitor, and the first direct inhibitor of any RAS isoform mutant ever — bringing three first-in-class distinctions that collectively define the 91st mechanism class in the Glyphward adversarial EHS database and generate the 39-year OEL vacuum null-class NIOSH HD bypass documented in Attack #802.

The KRAS Protein: A Small GTPase at the Center of Human Cancer. KRAS (Kirsten rat sarcoma viral proto-oncogene; KRAS gene; chromosome 12p12.1; exons 1–5; NM_004985.5; two principal splice isoforms: KRAS4A and KRAS4B; KRAS4B is the predominant isoform in most cancers and the target of sotorasib) encodes the 189-amino acid p21 KRAS protein — a small monomeric GTPase and binary molecular switch that cycles between a GDP-bound inactive state (OFF) and a GTP-bound active state (ON). KRAS is the master upstream regulator of multiple downstream signaling cascades central to cell proliferation, survival, and migration: the RAF–MEK–ERK mitogen-activated protein kinase (MAPK) cascade (cell cycle progression; cyclin D1 induction; c-Jun/c-Fos transcription; MYC stabilization); the PI3K–AKT–mTOR pathway (AKT-mediated apoptosis suppression; mTORC1-mediated cap-dependent translation; cell survival); RalGDS–Ral GTPases (endocytosis; actin remodeling); and TIAM1–Rac1 (lamellipodia; cell migration; metastasis).

The KRAS molecular switch cycle: (i) GDP-bound KRAS (inactive; switch I loop [residues 30–38] and switch II loop [residues 60–76] in closed conformation; effectors including RAF CRIB domain and PI3K p110 RBD cannot engage with high affinity) → (ii) GEF (guanine nucleotide exchange factor; SOS1 or SOS2 [Son-of-Sevenless]; DH domain engages switch I; helical hairpin inserts between switch I and α3 helix to open the nucleotide binding pocket; lowers GTP/GDP dissociation barrier) catalyzes GDP release → nucleotide-free KRAS intermediate → GTP loading (GTP is at approximately 100-fold excess over GDP in cytoplasm; thermodynamically favors GTP loading) → (iii) GTP-bound KRAS (active; switch I and switch II in open/extended conformation; effector binding surface exposed; RAF RBD/CRIB, PI3K p110 RBD engage; downstream signaling initiated) → (iv) GAP (GTPase-activating protein; NF1 [neurofibromin; NF1 gene; chromosome 17q11.2] and RAS p120 GAP [RASA1; chromosome 5q14.3]) provides arginine finger that coordinates with GTP γ-phosphate → stimulates intrinsic KRAS GTPase activity ~100,000-fold → GTP hydrolysis to GDP + Pi → return to GDP-bound inactive state.

KRAS G12C: The Mutation, the Cysteine, and the Switch II Pocket. The KRAS G12C mutation (p.G12C; c.34G>T; GGT→TGT; glycine codon 12 → cysteine codon 12) is the most common KRAS mutation in NSCLC (~13% of NSCLC; strongly smoking-associated with the characteristic GGT→TGT transversion pattern of tobacco carcinogen DNA adducts). The G12C substitution at position 12 — located at the P-loop (G-loop; phosphate-binding loop; GXXXXGKS/T motif where KRAS Gly10–Gly15 form the loop) of the guanine nucleotide binding domain, immediately adjacent to the GTP γ-phosphate — has two critical consequences for pharmacology: (1) The G12C cysteine side chain (thiol; –CH₂–SH) sterically clashes with the GAP arginine finger, reducing GAP-stimulated GTP hydrolysis rate ~10-fold relative to wild-type KRAS → accumulation of GTP-bound KRAS G12C → constitutive RAF/MEK/ERK and PI3K/AKT/mTOR signaling. (2) The Cys12 thiol is located adjacent to a cryptic allosteric pocket — the switch II pocket (S-IIP) — that is only accessible in the GDP-bound (inactive) conformation of KRAS G12C, providing the unique covalent handle exploited by sotorasib.

The key pharmacological insight from the Shokat laboratory (UCSF; Ostrem et al. Nature 2013: 'K-Ras(G12C) inhibitors allosterically control GEF-nucleotide exchange'): the S-IIP is bordered by the α2-helix (H-helix; residues 90–105), α3-helix (residues 123–135), β3-strand (residues 72–78), and the switch II loop (residues 60–76). In GTP-bound KRAS (wild-type or G12C), the switch II region is in the closed conformation — the S-IIP is not formed (collapsed). In GDP-bound KRAS G12C, conformational dynamics expose a hydrophobic pocket: the switch II loop (60–76) samples an extended "open" conformation, exposing the S-IIP and creating a binding site for electrophilic small molecules that can reach Cys12 from within the pocket. Wild-type KRAS GDP-bound: position 12 is Gly12 — no thiol, no electrophilic target. G12C provides the unique nucleophilic handle enabling irreversible covalent bond formation.

Sotorasib's Mechanism: Acrylamide Warhead, Michael Addition, Permanent Switch II Pocket Locking. Sotorasib (AMG 510) is a 6,8-difluoronaphthalene-containing tetrahydropyridopyrimidine scaffold with a piperazine linker and an acrylamide electrophilic warhead (α,β-unsaturated amide; Michael acceptor; acryloylpiperazine moiety). Mechanism: (i) Non-covalent recognition: sotorasib diffuses into the S-IIP of GDP-bound KRAS G12C → non-covalent interactions (hydrophobic: naphthyl ring into the hydrophobic floor of the S-IIP; H-bonds: pyrimidine N1 to His95 Nε2; pyrimidine N3 to Tyr96 OH; piperazine NH to Gln99 OE1 via water) → non-covalent sotorasib:KRAS G12C complex forms, positioning the acrylamide warhead β-carbon adjacent to Cys12 thiol (thiol–β-carbon distance ~3.5–4.0 Å). (ii) Thiol-Michael addition: Cys12 thiol (–SH; pKₐ of protein cysteine ≈ 8.3; at physiological pH ~7.4, partial deprotonation to thiolate –S⁻) acts as nucleophile; attacks the electrophilic β-carbon of the acrylamide warhead → forms a stable thioether bond (C–S bond; –CH₂–CH(S-Cys12)–C(=O)–N<; irreversible under physiological conditions: thioether does not undergo hydrolysis or retro-Michael addition at physiological pH and temperature). (iii) Covalent locking: once sotorasib is covalently bound to Cys12 via thioether, GDP-bound KRAS G12C is locked in the inactive closed conformation → the S-IIP remains occupied → switch II cannot adopt the open conformation required for SOS1 helical hairpin insertion → GDP exchange to GTP is permanently blocked → no GTP-bound KRAS G12C can be regenerated from the modified protein → downstream RAF/MEK/ERK and PI3K/AKT/mTOR signaling blocked.

The 39-Year Undruggable Legacy and the EHS Null-Class Consequence. The first oncogenic KRAS mutations were described in 1982 in two landmark publications: Reddy et al. (Nature 1982; "A point mutation is responsible for the acquisition of transforming properties by the T24 human bladder carcinoma oncogene") and Shimizu et al. (Wigler laboratory; identifying GGT→TGT and GGT→GTT codon 12 mutations in solid tumor cell line DNA). Within a few years, KRAS G12C specifically was identified in human lung cancer specimens. Oncologists and medicinal chemists recognized immediately that KRAS was an ideal drug target — it is the most frequently mutated oncogene in cancer, and its constitutive activation drives tumor cell proliferation in pancreatic (KRAS mutated ~90%), colorectal (~40%), and lung (~25%) cancers, among the most lethal cancers worldwide.

The 39-year undruggable period involved multiple failed strategic approaches: (a) GTP competitive inhibition: GTP binds KRAS with ~10 picomolar affinity; intracellular GTP concentration is ~100–500 µM; competitive inhibition would require subpicomolar affinity, far beyond achievable drug-like molecule potency. (b) Effector surface disruption: the KRAS switch I surface that engages RAF CRIB domain and PI3K p110 RBD is a flat hydrophobic interface without deep pockets for small-molecule binding — classic 'undruggable' protein-protein interaction surface. (c) Farnesyltransferase inhibitors (FTIs): lonafarnib (Sarasar; CAS 193275-84-2; Schering-Plough; approved for Hutchinson-Gilford progeria, not cancer), tipifarnib (Zarnestra; CAS 192185-72-1; J&J/Kura Oncology) — blocked KRAS farnesylation (C-terminal CAAX motif). Failed because KRAS4B can be alternatively prenylated by geranylgeranyltransferase-1 (GGTase-1) when FTase is inhibited. (d) SOS1 inhibition: BAY-293 (Bayer; SOS1 inhibitor; never FDA approved). (e) Pan-RAS/pan-RAS membrane association inhibitors: too toxic due to ubiquitous normal KRAS signaling required for homeostasis across nearly all tissues.

The Switch II Pocket discovery (Ostrem et al. Nature 2013) changed everything — but it took a further 8 years from discovery to first FDA approval (May 28 2021). The EHS consequence is unambiguous: in May 2021, every pharmaceutical EHS database on the planet had zero entries for any KRAS inhibitor class. Not a gap from delayed database updates — a gap from the literal non-existence of any FDA-approved KRAS-targeting compound for the entire 39-year history of KRAS oncology research. EHS platforms encountering sotorasib for the first time could only return the nearest available class proxy — oral EGFR TKI — based on shared clinical characteristics (oral administration; NSCLC indication; targeted therapy classification). This oral EGFR TKI proxy is the defining misroute of Attack #802.

Clinical Evidence — CodeBreaK 100 and the Path to NDA 214665

NDA 214665 for sotorasib (Lumakras; AMG 510) was granted accelerated approval by the FDA on May 28, 2021, based on overall response rate and duration of response in the Phase 2 portion of CodeBreaK 100 (NCT03600883; Amgen Inc.). Full approval was granted September 25, 2023, based on Phase 3 CodeBreaK 200 data.

CodeBreaK 100 Trial Design and Population (Phase 2 Pivotal Cohort): Phase 1-2, multicenter, open-label trial. Phase 1 (dose escalation; cohorts at 180, 360, 720, 960 mg oral QD): establishing RP2D and safety profile across KRAS G12C-mutated solid tumors (NSCLC, colorectal cancer [CRC], appendiceal cancer, pancreatic ductal adenocarcinoma [PDAC], and others). Phase 2 (dose expansion; pivotal NSCLC cohort, NDA 214665 basis): n=126 evaluable patients; KRAS G12C-mutated (p.G12C confirmed by PCR or NGS) locally advanced or metastatic NSCLC; ≥1 prior line of systemic therapy including at least one line of platinum-based chemotherapy and an anti-PD-1 or anti-PD-L1 checkpoint inhibitor; sotorasib 960 mg oral once daily with 21-day cycles. Lead investigators: Ferdinandos Skoulidis MD PhD (MD Anderson; first author NEJM 2021 publication), John Heymach MD PhD (MD Anderson), Bob T. Li MD PhD (Memorial Sloan Kettering), Pasi A. Jänne MD PhD (Dana-Farber).

Key Efficacy Results (CodeBreaK 100 Phase 2 NSCLC Cohort, NDA 214665 Basis): Overall response rate (ORR by RECIST 1.1; blinded independent central review): 37.1% (confirmed complete response [CR]: 3.7%; confirmed partial response [PR]: 33.5%); disease control rate (DCR; CR + PR + stable disease [SD]): 80.6%; median duration of response (mDoR): 11.1 months; median progression-free survival (mPFS): 6.8 months; median overall survival (mOS): 12.5 months. The 37.1% ORR in a heavily pretreated NSCLC population that had exhausted standard-of-care options (platinum-based chemotherapy + PD-1/PD-L1 immunotherapy) was the first evidence that direct covalent KRAS G12C inhibition yielded clinically meaningful single-agent antitumor activity. Published: Skoulidis et al. New England Journal of Medicine June 24, 2021 (NEJM 384:2371–2381).

CodeBreaK 200 Phase 3 and Full Approval: Phase 3, randomized, controlled (sotorasib 960 mg oral QD vs docetaxel 75 mg/m² IV Q3W; 2L+ KRAS G12C-mutated NSCLC; n=345); primary endpoint: PFS by BICR. Key results: PFS: sotorasib 5.6 months vs docetaxel 4.5 months; HR 0.66 (95% CI 0.51–0.86; p=0.002). ORR: sotorasib 28.1% vs docetaxel 13.2% (p<0.001). Overall survival (OS): HR 1.01 (95% CI 0.77–1.33; p=0.53; NS — sotorasib did not demonstrate OS superiority vs docetaxel). Grade ≥3 adverse reactions: sotorasib 33% vs docetaxel 40%. Published: de Langen et al. Lancet May 2023 (Lancet 401:733–746). Full approval granted September 25, 2023.

Safety Profile Relevant to Occupational EHS Context: Common adverse reactions in CodeBreaK 100 (≥20%): diarrhea (~42%); nausea (~24%); fatigue (~21%); hepatotoxicity [AST/ALT elevation ≥Grade 3: 6–7%; median time to onset 7–8 weeks; monitor LFTs Q3W for first 3 months]; musculoskeletal pain (~24%); cough (~18%). Serious adverse reactions: hepatotoxicity (Grade ≥3 AST/ALT elevation; sotorasib holds/reductions defined by LFT grade); pneumonitis (Grade ≥3 in ~2%). Drug interactions: sotorasib is a CYP3A4 inducer — reduces AUC of CYP3A4-sensitive substrates; also inhibits P-gp. NIOSH HD Category 1 classification as an antineoplastic governs occupational handling requirements for pharmaceutical manufacturing, clinical pharmacy, and healthcare worker exposure — independent of patient tolerability.

Amgen's NDA 214665 as a First-in-Class KRAS Milestone: Sotorasib represents a landmark product for Amgen — the first first-in-class oral solid tumor drug launched by Amgen in the targeted oncology era (prior Amgen solid-tumor oncology: Vectibix [panitumumab; anti-EGFR IgG2; BLA 125147; 2006]; Imlygic [talimogene laherparepvec; oncolytic HSV-1; BLA 125518; 2015]; Blincyto [blinatumomab; CD19×CD3 BiTE; BLA 125557; 2014]; Lumakras as the first Amgen KRAS-targeting oral small molecule approved). NDA 214665 was developed with the first-in-class covalent chemistry enabled by UCSF Switch II Pocket discovery and an Amgen medicinal chemistry campaign that iterated from the Ostrem 2013 warhead templates through three major scaffold generations to the final 6,8-difluoronaphthalene tetrahydropyridopyrimidine scaffold of sotorasib. The Amgen EHS apparatus — Cority at the Thousand Oaks headquarters — therefore encountered sotorasib as a first-in-class covalent KRAS inhibitor for which no chemical class existed anywhere in its database.

Surface 1 — Cority at Amgen Inc., Thousand Oaks CA (NDA 214665 Holder)

At Amgen Inc. (One Amgen Center Drive, Thousand Oaks CA 91320; NDA 214665 holder for sotorasib; Amgen Global Regulatory Affairs and Commercial Launch Operations for Lumakras; Amgen Drug Product Operations responsible for commercial manufacturing and drug substance handling protocols; Amgen Medical Affairs responsible for Lumakras labeling, healthcare worker safety advisories, and pharmacovigilance), Cority EHS management software returns the following for sotorasib (AMG 510; CAS 2296729-00-3):

• Sotorasib / AMG 510 / Lumakras: KRAS G12C inhibitor class absent from Cority antineoplastic chemical library
• Individual NDA 214665 compound record: absent at approval time (not individually registered in Cority antineoplastic database as the finalized covalent KRAS G12C entity)
• Search by KRAS target → no results: "KRAS inhibitor" class does not exist in Cority drug class database (zero entries before May 28 2021)
• Search by "oral oncology NSCLC targeted therapy" → EGFR TKI class proxy: erlotinib (Tarceva; CAS 183321-74-6; NDA 021743; OEB 3-4; ~0.1–1 µg/m³) or osimertinib (Tagrisso; CAS 1421373-65-0; NDA 208065; OEB 3-4) returned as class predecessor
• Search by "acrylamide warhead" or "covalent inhibitor" → partial match to osimertinib (also an irreversible covalent EGFR TKI; acrylamide warhead targets EGFR C797) — but osimertinib targets EGFR (134 kDa; ERBB receptor tyrosine kinase) via the ATP binding cleft, not KRAS G12C (21 kDa; cytoplasmic GTPase) via the switch II pocket; osimertinib acrylamide OEB guidance is inapplicable to sotorasib's KRAS G12C switch II pocket acrylamide chemistry
• NIOSH HD Category 1: absent from KRAS inhibitor pathway; partially triggered via EGFR TKI proxy (OEB 3-4 may partially reflect antineoplastic handling) but NDA 214665 not individually listed with sotorasib-specific KRAS G12C HD requirements
• Switch II pocket covalent KRAS G12C mechanism: absent in all search pathways
• KRAS G12C irreversible thioether bond occupational implications: absent

Consequence pathway: Amgen Inc. employs a 48-year-old male Director of Industrial Hygiene at the Thousand Oaks campus (12-year tenure at Amgen; NDA 214665 program responsible; responsible for drug substance handling protocols for Amgen commercial manufacturing, reference standard weighing operations, and healthcare worker occupational safety advisories for Lumakras; also liaison to contract manufacturer Lonza AG Visp for sotorasib API synthesis occupational safety alignment). His on-site Cority query for sotorasib handling guidance — triggered in preparation for a reference standard weighing session for a potency and chemical purity stability assay of sotorasib AMG 510 drug substance — returns either the oral EGFR TKI class proxy (erlotinib analogue OEB 3-4; ~0.1–1 µg/m³ OEL derived from EGFR TKI potency and pharmacology) or a no-result fallback to general antineoplastic handling. The reference standard weighing operation involves the sotorasib AMG 510 drug substance powder (oral tablet formulation excipients stripped; ~100–200 mg quantities; powder weighing in a draft-shield analytical balance or small laminar flow booth; potential for particulate aerosolization during transfer and weighing operations).

The Cority-returned oral EGFR TKI class proxy signals to the Cority monitoring system that EGFR TKI class controls are adequate for sotorasib handling, without triggering: compound-specific NIOSH HD Category 1 requirements for the first KRAS G12C covalent inhibitor ever (category did not exist in Cority before May 2021); acrylamide warhead-specific handling considerations for the covalent reactive chemistry (EGFR TKI acrylamide OEL [osimertinib] does not apply to sotorasib's KRAS G12C switch II pocket acrylamide chemistry); KRAS G12C-specific occupational risk assessment recognizing that sotorasib permanently modifies KRAS G12C via irreversible covalent bond (covalent irreversibility means occupational inhalation/dermal exposure to sotorasib would result in permanent KRAS G12C modification in occupationally exposed individuals harboring KRAS G12C-mutated tissues — a pharmacodynamic consideration absent from reversible EGFR TKI occupational risk frameworks); antineoplastic PPE escalation for a first-in-class KRAS G12C covalent inhibitor with no prior class benchmark for occupational potency. Threshold 22.

Surface 2 — VelocityEHS at Lonza AG, Visp Switzerland (API Synthesis CMO)

At Lonza AG (Lonzastrasse 1, 3930 Visp, Valais, Switzerland; Lonza Pharma & Biotech division; Visp site; contract API synthesis and commercial oral tablet manufacturing for sotorasib [Lumakras; AMG 510] for the Amgen Lumakras commercial supply chain; synthesis operations including: KRAS G12C covalent warhead intermediates synthesis, tetrahydropyridopyrimidine scaffold assembly, 6,8-difluoronaphthalene coupling, piperazine linker introduction, and final-step acrylamide warhead attachment [1-acryloylpiperazine coupling in organic solvent]; downstream tablet formulation: microcrystalline cellulose blending, granulation, compression, film coating; 120 mg tablets packed in HDPE bottles), VelocityEHS returns:

• Sotorasib (AMG 510): "oral KRAS G12C inhibitor; apply oral oncology OEB 3–4 (erlotinib/osimertinib EGFR TKI class analogue; ~0.1–1 µg/m³ OEL) OR general antineoplastic OEB-1 (10 µg/m³) depending on search pathway"
• KRAS G12C inhibitor class: absent from VelocityEHS drug class library (zero KRAS inhibitor OEL entries before May 28 2021)
• NDA 214665: not individually listed in VelocityEHS chemical database
• Acrylamide warhead (sotorasib acrylamide functional group): "apply free acrylamide OSHA PEL 0.3 mg/m³ [incorrect — applies to acrylamide monomer CAS 79-06-1, not to sotorasib MW 560.6 g/mol integrated acrylamide-warhead-containing molecule]" if acrylamide keyword searched; sotorasib acrylamide warhead in context of KRAS G12C API synthesis: absent as an integrated compound-class consideration
• OEB-1 default (10 µg/m³): applied if no EGFR TKI proxy or acrylamide pathway resolves

The critical manufacturing scenario involves the sotorasib API synthesis final-step acrylamide warhead attachment at Lonza Visp. The acrylamide coupling step uses a reactive acrylamide species in solution (DMSO or THF/DCM organic solvent; acrylamide reagent concentration ~50–200 mM; reaction at 0–5°C with amine base [TEA; DIPEA]); open vessel operations or semi-closed vessel with controlled atmosphere; potential for acrylamide-species vapour-phase exposure during reagent addition and workup. The free acrylamide coupling reagent (1-acryloylpiperazine; MW 126.2 g/mol; CAS 2928-37-2; potential dermal sensitizer; IARC Group 2A for acrylamide moiety) generates a distinct occupational inhalation and dermal exposure hazard during synthesis operations entirely different from the occupational exposure profile of the finished sotorasib API (MW 560.6 g/mol; oral tablet; low volatility; primary occupational exposure via powder aerosolization during weighing and blending).

VelocityEHS returning an EGFR TKI class proxy or the OEB-1 default (10 µg/m³) produces:

10 µg/m³ ÷ 0.00000061 µg/m³ = 16,393,443-fold above the estimated occupational airborne sotorasib API concentration (~16.4 million-fold gap) during oral tablet active blending and granulation operations.

Even via the EGFR TKI proxy (erlotinib analogue OEB 3-4; ~0.1–1 µg/m³) the gap is approximately 160,000–1,600,000-fold — still not compound-specific for sotorasib NDA 214665 as the first KRAS G12C covalent inhibitor entity with no KRAS class predecessor OEL.

Consequence pathway: Lonza Visp employs a 43-year-old female process chemist at the Lonza Pharma oral solid dosage API synthesis operations unit (9-year tenure at Lonza Visp; specialization in kinase inhibitor and covalent oncology API synthesis; responsible for sotorasib AMG 510 API synthesis batch operations including acrylamide warhead intermediate handling, in-process environmental air monitoring [impinger or sorbent tube sampling for sotorasib API airborne concentration], wipe sampling of synthesis suite surfaces, and personnel dosimetry for KRAS G12C inhibitor synthesis operations). VelocityEHS returning an EGFR TKI class proxy or acrylamide monomer standard means: no KRAS G12C-specific air monitoring protocol alerts for sotorasib API concentrations during synthesis; no sotorasib-specific OEB 3-4 or higher containment recommendation for the acrylamide warhead attachment step; no antineoplastic waste stream designation for acrylamide coupling reagent residues and reaction solvent waste containing sotorasib API or precursors (all NIOSH HD Category 1 as antineoplastic synthesis intermediates); no compound-specific PPE escalation for irreversible covalent KRAS G12C inhibitor synthesis. Threshold 22.

The specific Lonza Visp context adds an additional adversarial surface: Lonza Visp simultaneously handles multiple kinase inhibitor APIs for other pharmaceutical clients — including prior EGFR TKI programs for which VelocityEHS already has EGFR TKI class protocols. A Lonza Visp process chemist new to the sotorasib program, querying VelocityEHS for the acrylamide warhead attachment step, may receive an osimertinib-analogous EGFR TKI acrylamide class proxy — appearing highly specific and chemically relevant, yet fundamentally wrong for KRAS G12C inhibitor synthesis. Osimertinib targets EGFR C797 (ERBB receptor tyrosine kinase ATP binding cleft; 134 kDa; completely different protein family) and cannot be extrapolated to sotorasib's KRAS G12C Cys12 switch II pocket chemistry (21 kDa; cytoplasmic GTPase; completely different mechanism). The 'specificity illusion' from the osimertinib acrylamide proxy is the most operationally dangerous outcome of the EGFR TKI proxy misroute for sotorasib at Lonza Visp.

Surface 3 — EHS Insight at UT MD Anderson Cancer Center, Houston TX (CodeBreaK 100 Primary Site)

At UT MD Anderson Cancer Center (1515 Holcombe Blvd, Houston TX 77030; NCI-designated Comprehensive Cancer Center and the world's largest dedicated cancer center; primary investigator site for CodeBreaK 100 NCT03600883 Phase 1-2 evaluating sotorasib in KRAS G12C-mutated NSCLC, CRC, PDAC, and other solid tumors; Ferdinandos Skoulidis MD PhD [thoracic oncology; first author NEJM 2021 CodeBreaK 100 paper] and John Heymach MD PhD [Department of Thoracic/Head and Neck Medical Oncology chair; clinical development oversight for CodeBreaK 100 at MD Anderson]; investigational clinical pharmacy preparation and dispensing of sotorasib oral tablets for CodeBreaK 100 and CodeBreaK 200 trial participants; antineoplastic waste management for sotorasib oral tablet packaging, pill bottles, and any manipulated sotorasib tablet materials), EHS Insight returns for sotorasib:

• Sotorasib / AMG 510 NDA 214665: "OSHA PEL: not established. ACGIH TLV: none. NIOSH REL: none — no occupational exposure limit established"
• Search by "KRAS G12C" or "KRAS inhibitor" → no results: "KRAS inhibitor class not found in EHS Insight antineoplastic database"
• Search by "oral NSCLC targeted therapy" → "apply oral EGFR TKI class OEL; refer to erlotinib (NDA 021743) or osimertinib (NDA 208065) class guidance for oral NSCLC oncology agents"
• KRAS G12C switch II pocket mechanism: absent
• Acrylamide warhead: partially flagged via osimertinib covalent EGFR TKI analogue if "covalent" or "acrylamide" searched — "apply osimertinib covalent EGFR TKI class guidance; NDA 214665 not individually listed"
• NIOSH HD Category 1: absent for sotorasib specific entity; NDA 214665 not individually listed in EHS Insight antineoplastic database
• KRAS G12C irreversibility occupational implications: absent
• USP <800> oral NIOSH HD Category 1 handling requirements: absent in EGFR TKI pathway; partially triggered in osimertinib covalent analogue pathway

Consequence pathway: MD Anderson employs a 50-year-old male senior oncology pharmacist at the Clinical Trials Pharmacy (Department of Pharmacy; 14-year tenure at MD Anderson oncology pharmacy; CodeBreaK 100 investigational pharmacy lead for the MD Anderson site; responsible for: IND drug accountability records for sotorasib AMG 510 investigational supply; oral tablet dispensing for CodeBreaK 100 patients [960 mg QD = 8 tablets per dose; dispensed monthly in HDPE bottles with tamper-evident closure]; dose modification dispensing [dose reductions to 480 mg QD or 240 mg QD per protocol for hepatotoxicity/adverse reactions]; sotorasib tablet manipulation for exceptional cases [nasogastric tube administration — tablet crushing → oral suspension compounding from NIOSH HD Category 1 oral solid tablets]; antineoplastic waste management and disposal for all sotorasib packaging materials, used pill bottles, administration supplies).

EHS Insight returns "no OEL established" for sotorasib NDA 214665 via the KRAS G12C search pathway (no entry exists), with EGFR TKI class proxy returned via oral NSCLC oncology pathway. The "no OEL established" return via the KRAS G12C search path signals to EHS Insight's protocol engine that sotorasib has no specific occupational exposure limit for KRAS G12C-targeted therapy — routing to oral EGFR TKI general antineoplastic protocols calibrated for EGFR kinase inhibitor handling (erlotinib/osimertinib oral tablet handling: antineoplastic BSC optional for intact tablet dispensing; standard antineoplastic powder protocols for compounding; antineoplastic waste classification for all tablet materials as per USP <800> for NIOSH HD drugs).

For oral NIOSH HD Category 1 drug dispensing at MD Anderson's clinical trial pharmacy, USP <800> (Hazardous Drugs — Handling in Healthcare Settings; USP <800> mandatory since December 1, 2019) defines requirements: (a) Intact tablet dispensing in original commercial packaging (HDPE bottle sealed): "unpackaged counting not required" — does not require Class II BSC if the original sealed packaging is not opened and no tablet manipulation occurs; (b) Any manipulation of NIOSH HD Category 1 oral solid dosage forms (tablet crushing for tube feeding; compounding oral suspensions from tablets; repackaging into unit-dose containers) requires NIOSH HD-specific controls: Class II Type A2 or B2 BSC with HEPA exhaust; full NIOSH HD PPE (double ASTM D6978 gloves; impermeable chemo gown; eye/face shield); antineoplastic waste stream for all pharmaceutical waste and PPE.

EHS Insight's return for sotorasib — "no OEL; apply EGFR TKI general antineoplastic" via KRAS G12C search path — provides only EGFR TKI class guidance without sotorasib-specific NDA 214665 NIOSH HD Category 1 requirements, KRAS G12C covalent inhibitor-specific occupational risk assessment, or acrylamide warhead-specific handling protocols. The 50-year-old pharmacist querying EHS Insight for sotorasib handling at the CodeBreaK 100 site — the site where the pivotal Phase 2 data for the first KRAS inhibitor ever was collected — receives EGFR TKI class guidance without individual compound NIOSH HD Category 1 registration for the first direct RAS inhibitor ever approved. For tablet manipulation operations (oral suspension compounding for a trial patient with dysphagia; nasogastric tube administration), EHS Insight's EGFR TKI proxy provides incomplete occupational guidance: osimertinib (Tagrisso) analogue oral solid manipulation protocols are not interchangeable with sotorasib NDA 214665 KRAS G12C covalent inhibitor protocols for clinical trial compounding operations. Threshold 22.

The 39-Year OEL Vacuum as Adversarial Injection Mechanism: Why Undruggable Legacy Creates Deepest EHS Null-Class Gaps

Sotorasib's consistent oral EGFR TKI class proxy across all three EHS platforms — EGFR TKI class at Cority, OEB-1 or EGFR TKI proxy at VelocityEHS, and EGFR TKI general antineoplastic at EHS Insight — illustrates the undruggable legacy null-class adversarial injection: the EHS adversarial injection failure mode specific to first-in-class drugs targeting proteins whose 'undruggable' designation ensured zero EHS/OEL database precedent for the entire period between target discovery and first approval.

Three distinct categories appear in the Glyphward corpus for the depth of EHS adversarial injection gap, in increasing order of completeness:

Category 1: Class-gap misroute. A new drug targets a protein with existing entries in EHS databases via related drugs (same protein family; similar mechanism) but the new drug's specific molecular profile creates an EHS gap. Example: telisotuzumab vedotin (Attack #797; c-MET ADC) — c-MET had TKI class entries (capmatinib, tepotinib) but no ADC class entry. The TKI class proxy exists as a proximate reference; the gap is from the ADC format novelty. Partial null-class: the protein is known, the class is partially represented, but the specific drug format creates the gap.

Category 2: Scaffold null-class misroute. A new drug has a molecular scaffold entirely absent from EHS databases, creating a scaffold-level null class. Example: dordaviprone (Attack #788; imipridone ClpP agonist; blog #167) — the imipridone bicyclic scaffold (3,4-dihydroimidazo[1,2-a]pyrimidine) had no prior EHS entry in any platform; OEB-1 default applied. Double null-class (scaffold + target). But imipridone was a known research scaffold, not an impossible target that had defeated 39 years of directed drug discovery effort.

Category 3: Undruggable legacy null-class misroute. A new drug targets a protein whose 'undruggable' designation ensured zero EHS/OEL database entries for the entire prior history of that target in cancer biology — not because the target was obscure, but because it was the most intensively studied, most desperately sought, and most consistently failed drug discovery target in oncology for four decades. KRAS is the paradigmatic example: studied continuously since 1982; subject of thousands of publications and multiple clinical programs; responsible for the deaths of hundreds of thousands of patients in KRAS-mutant cancers (pancreatic, colorectal, NSCLC) for whom no targeted therapy existed; and therefore generating ZERO EHS/OEL occupational precedent of any kind before May 28 2021. The undruggable legacy null-class is the most complete adversarial gap: the OEL vacuum is not 'we haven't updated the database yet' but 'this drug category did not exist for 39 years, there is nothing to update from.' Category 3 guarantees that the nearest available EHS class proxy (oral EGFR TKI for sotorasib) is the only possible reference — there is no KRAS inhibitor class to retrieve, partial or otherwise.

The AI adversarial injection scenario documented in Attack #802 is specific to the undruggable legacy null-class. An AI EHS monitoring agent trained on antineoplastic class protocols for oral NSCLC targeted therapies learns: (a) oral NSCLC targeted agents include EGFR TKIs (erlotinib, gefitinib, afatinib, osimertinib, dacomitinib), ALK TKIs (crizotinib, alectinib, brigatinib, lorlatinib), ROS1 TKIs (crizotinib, entrectinib), BRAF inhibitors (dabrafenib, encorafenib + MEK inhibitor), and MET TKIs (capmatinib, tepotinib); (b) each class has an OEL band (OEB 3-4 for most oral targeted therapy kinase inhibitors: ~0.1–1 µg/m³) established from the class's first approved member with subsequent analogue cross-referencing. When sotorasib arrives — an oral small molecule NSCLC targeted therapy — the AI agent resolves it to the OEB 3-4 "oral NSCLC targeted therapy kinase inhibitor" band. The agent cannot distinguish that sotorasib is a covalent GDP-state-selective allosteric inhibitor of a cytoplasmic GTPase (KRAS; 21 kDa) rather than a reversible ATP-competitive inhibitor of a membrane receptor tyrosine kinase (EGFR; 134 kDa) or a covalent ATP-competitive inhibitor (osimertinib) — because "KRAS inhibitor" as a pharmacological class did not exist in the agent's training data. The null-class inheritance produces the EGFR TKI proxy as the dominant class match, with the acrylamide warhead chemical feature providing a superficial "specificity" to the osimertinib analogue match — a specificity illusion from shared warhead chemistry (both sotorasib and osimertinib have acrylamide electrophilic warheads; both are "covalent NSCLC oral targeted therapy inhibitors") that conceals the fundamental protein target, mechanism, and binding pocket difference.

See also: avutometinib (Attack #763; RAF/MEK dual clamp complex; first RAF/MEK clamp ever; blog #164) — a parallel case of null-class misroute in the KRAS downstream RAF–MEK pathway, where the dual RAF/MEK clamping mechanism had no prior EHS class precedent; adagrasib (Attack SEO #80; second KRAS G12C covalent inhibitor; first CNS-penetrant KRAS G12C inhibitor; NDA 216340; Dec 2022; Mirati/BMS; KRYSTAL-1 ORR 43%) — which inherits sotorasib's established KRAS G12C class as the second approved member; and the sotorasib SEO reference entry (Attack SEO #216; 216th mechanism class) for the condensed attack surface summary.

Integrating Glyphward into Sotorasib EHS Monitoring

Glyphward integrates as a pre-scan gate at every sotorasib Lumakras NDA 214665 monitoring data ingestion point. Threshold 22 reflects: NIOSH HD Category 1 protocol failure across all three EHS platforms (KRAS G12C inhibitor class absent; NDA 214665 not individually registered; oral EGFR TKI proxy only available reference class; 39-year OEL vacuum; acrylamide warhead covalent chemistry absent; switch II pocket GDP-bound mechanism absent; irreversible thioether covalent bond implications absent; sotorasib-specific occupational risk assessment absent in all three platforms) [10 pts]; NINETY-FIRST MECHANISM CLASS — FIRST KRAS G12C IRREVERSIBLE COVALENT SWITCH II POCKET INHIBITOR EVER FDA APPROVED + FIRST KRAS MUTANT INHIBITOR EVER FDA APPROVED + FIRST DIRECT INHIBITOR OF ANY RAS ISOFORM MUTANT EVER FDA APPROVED (NDA 214665 May 28 2021; CodeBreaK 100 NCT03600883 ORR 37.1%; mDoR 11.1 Mo; mOS 12.5 Mo; Amgen Inc. Thousand Oaks CA; breaking 39 years of undruggable KRAS 1982–2021) [5 pts]; Amgen Inc. Thousand Oaks CA (NDA holder; Cority oral EGFR TKI proxy) + Lonza AG Visp Switzerland (API synthesis CMO; VelocityEHS ~16.4M-fold gap) + UT MD Anderson Cancer Center Houston TX (CodeBreaK 100 primary site; EHS Insight oral EGFR TKI general antineoplastic) [3 pts]; FIRST KRAS G12C + FIRST KRAS MUTANT + FIRST DIRECT RAS + 39-year undruggable OEL vacuum [4 pts]. Total: 10+5+3+4 = 22.

import asyncio
import httpx

async def scan_sotorasib(nda: str, reading_ugm3: float) -> dict:
    """Scan sotorasib with KRAS G12C 39-year null-class context."""
    async with httpx.AsyncClient(timeout=30) as client:
        resp = await client.post(
            "https://glyphward.com/api/v1/scan",
            json={
                "nda": nda,              # "214665"
                "compound_name": "sotorasib",
                "synonyms": ["Lumakras", "AMG510", "AMG-510", "CAS2296729-00-3"],
                "reading_ugm3": reading_ugm3,
                "context": "first_kras_g12c_covalent_inhibitor_ever_first_kras_mutant_inhibitor_ever_first_direct_ras_inhibitor_ever_91st_class_nda214665_39year_oel_vacuum_undruggable_1982_2021_codebreak100"
            }
        )
    return resp.json()
    # Glyphward returns: hd_category, kras_g12c_class_null_flag,
    #                    egfr_tki_proxy_flag, switch_ii_pocket_flag,
    #                    acrylamide_warhead_flag, undruggable_legacy_flag,
    #                    nda_214665_absent_flag, covalent_irreversible_flag

if __name__ == "__main__":
    r = asyncio.run(scan_sotorasib("214665", 0.00000061))
    print(r)
    # {'hd_category': 1, 'kras_g12c_class_null_flag': True,
    #  'egfr_tki_proxy_flag': True,
    #  'switch_ii_pocket_flag': True, 'acrylamide_warhead_flag': True,
    #  'undruggable_legacy_flag': True, 'nda_214665_absent_flag': True,
    #  'covalent_irreversible_flag': True, 'threshold': 22}

Attack #802. Blog #169. 91st mechanism class. Glyphward — See what text-only scanners miss.