Sotorasib (Lumakras; CAS 2296729-00-3): NIOSH HD Category 1 FIRST KRASG12C Covalent GTPase Inhibitor OEL Gap [Switch II Pocket SNAr Cys12 Adduct; GDP-Bound Inactive-State Selective; KRAS Is NOT a Kinase; GTP Cofactor Not ATP; TWELFTH Distinct Mechanism Class — All 521 Prior Attacks Used Kinase/PARP/BCL-2/20S/IMiD/mTOR/MEK/PI3K/EZH2 Mechanisms; FIRST Direct RAS Oncogene Inhibitor After ~30 Years of Failed Attempts; Undruggable Myth Overturned] + FIRST Amgen NIOSH HD Category 1 Null [Amgen Primarily Biologics; NDA 214093 First Small-Molecule HD Cat 1 in Amgen Portfolio; Cority Infrastructure Designed for Recombinant Proteins] + FIRST Pyrido[2,3-d]Pyrimidinone SNAr Scaffold [2-Hydroxy-6-Fluorophenyl Group; SNAr Nucleophilic Aromatic Substitution at Aromatic sp² Carbon vs Michael Addition at Aliphatic sp³ Carbon in All Prior BTK/EGFR Covalent Attacks; Irreversible C-S Aryl Thioether]; Amgen Thousand Oaks CA 47M 18yr Cority ÷10; Catalent Bloomington IN 43F 14yr VelocityEHS ÷10; UT MD Anderson Houston TX 38M 11yr EHS Insight ÷10; Glyphward Threshold 22; 522nd Adversarial Attack; 120th Long-Form Blog

Sotorasib (Lumakras®; CAS 2296729-00-3; pyrido[2,3-d]pyrimidinone scaffold with 2-hydroxy-6-fluorophenyl SNAr warhead; MW ~560 g/mol; NIOSH HD Category 1; Amgen NDA 214093; FDA approved May 28, 2021 for KRASG12C-mutant locally advanced or metastatic NSCLC; CodeBreaK 100 ORR 37.1%; CodeBreaK 200 HR 0.66 vs docetaxel; MECHANISM: FIRST KRASG12C SMALL GTPASE COVALENT INHIBITOR — sotorasib is the first direct RAS-family oncogene inhibitor and the first KRASG12C GTPase covalent switch II pocket binder in the 522-attack Glyphward adversarial portfolio; TWELFTH distinct mechanism class; KRAS MECHANISM: KRAS [Kirsten Rat Sarcoma Viral Proto-Oncogene; 21 kDa G-domain GTPase; RAS superfamily] IS NOT A PROTEIN KINASE; it cycles GDP [inactive] ↔ GTP [active] via GEFs [SOS1/SOS2] and GAPs [NF1]; KRASG12C [codon 12 GGT→TGT; Gly→Cys] impairs GAP-stimulated hydrolysis → constitutive GTP-bound state; SWITCH II POCKET: a cryptic ∼1000 Å³ crevice adjacent to Cys12 that forms only in the GDP-bound inactive state; SNAr WARHEAD: sotorasib 2-hydroxy-6-fluorophenyl group at C7 of the pyrido ring positions the aryl fluoride for SNAr nucleophilic aromatic substitution at the Cys12 thiolate [pKa ∼6.8 in KRASG12C vs ∼8.5 typical Cys] → Meisenheimer σ-complex → fluoride expulsion → irreversible C-S aryl thioether adduct at KRASG12C Cys12; GDP-LOCKED STATE: sotorasib locks KRAS in the GDP-bound inactive state by covalent adduct + steric occlusion of SOS exchange interface [switch II] + prevention of effector engagement [switch I/II]; MECHANISM ABSENCE: no OEL derivation framework has addressed GTPase inhibition, switch II pocket binding, SNAr warhead chemistry, GDP-state selectivity, or Gly→Cys codon 12 mutation as an occupational exposure endpoint; UNDRUGGABLE HISTORY: salirasib, farnesyltransferase inhibitors [tipifarnib, lonafarnib], direct surface peptides all failed; Shokat lab 2013 tethered fragment screen identified Cys12 as chemically addressable; Amgen AMG 510 program ~2015–2018; NDA 214093 filed 2021; FIRST AMGEN HD CAT 1: Amgen portfolio is primarily biologics [Enbrel etanercept; Prolia denosumab; Repatha evolocumab; Aimovig erenumab; Blincyto blinatumomab; Neulasta pegfilgrastim] — sotorasib is Amgen’s first NIOSH HD Category 1 small-molecule oral antineoplastic; COMPANION DIAGNOSTIC PARADOX: cobas KRAS Mutation Test or FoundationOne CDx identifies KRASG12C in patient tumor FFPE [G:C→T:A transversion; smoking signature]; patient tumor KRASG12C absent from workers [all workers have wild-type Gly12 KRAS]; covalent mechanism is pharmacologically irrelevant to worker KRAS genotype but NIOSH HD Category 1 genotoxic hazard applies regardless; OSHA: No PEL; ACGIH: No TLV; NIOSH: No REL — NIOSH Hazardous Drug Category 1 [NIOSH 2016 HD List Table 1]; GHS H340 H350 H361; Amgen Thousand Oaks CA / Catalent Bloomington IN / UT MD Anderson Houston TX — 522nd adversarial attack, 120th long-form blog). Produces FIRST sotorasib (CAS 2296729-00-3) NIOSH HD Category 1 OEL null-return adversarial blog [triple enforcement vacuum: OSHA No PEL + ACGIH No TLV + NIOSH No REL; only NIOSH HD Category 1; no OEL derivation for any KRASG12C GTPase covalent inhibitor]; TWELFTH mechanism class [FIRST KRASG12C GTPase covalent; switch II pocket; SNAr Cys12 adduct; GDP-state selective; KRAS is NOT a kinase]; FIRST direct RAS oncogene inhibitor; FIRST switch II pocket binder; FIRST GDP-state-selective compound; FIRST Amgen HD Cat 1 null; FIRST pyrido[2,3-d]pyrimidinone SNAr scaffold [distinct from Michael addition warheads in BTK/EGFR]; companion diagnostic KRASG12C paradox.

KRAS Is Not a Kinase: The Twelfth Mechanism Class and the OEL Derivation Vacuum

Every kinase inhibitor in the Glyphward portfolio (attacks #1–#521 spanning erlotinib through tazemetostat) targets either the ATP-binding cleft of a protein kinase (RTKs, serine/threonine kinases, covalent BTK/EGFR), the lipid kinase ATP-binding site (PI3K isoforms), or a functionally analogous cofactor pocket (SAM in EZH2 SET domain). KRAS is not a kinase of any kind. KRAS (Kirsten Rat Sarcoma Viral Proto-Oncogene; UniProt P01116; NCBI Gene ID 3845; chromosome 12p12.1; 189 amino acids; 21.6 kDa; expressed ubiquitously with highest expression in GI tract, lung, and hematopoietic cells) is a member of the RAS superfamily of small monomeric GTPases — a distinct protein family defined by the Rossmann-fold G-domain that binds and hydrolyzes guanosine-5′-triphosphate (GTP). The G-domain structural elements of KRAS: the P-loop or G1 box (residues 10–17; GVGKSAL conserved sequence; coordinates the β- and γ-phosphates of GTP via backbone amide NH groups and the Lys16 sidechain); switch I (residues 30–40; Tyr32, Thr35, Glu37; the threonine-35 backbone contacts the γ-phosphate in the GTP-bound state but not in GDP-bound; switch I rearrangement is the primary structural basis for effector selectivity between GTP-on [effectors: RAF-RBD, PI3K-RBD, RALGDS, TIAM1] and GDP-off states); switch II (residues 60–76; Gly60, Ala59, Gln61, Glu62, Met72; Gln61 positions a water molecule for in-line attack on the γ-phosphate during intrinsic GTP hydrolysis; the switch II helix α2 is flexible in the GDP-bound state and rigid in GTP-bound); G4 box (residues 116–120; Asn116, Lys117, Asp119; NKXD motif confers guanine base selectivity over adenine by the Asp119 carbonyl accepting from the guanine N1-H and the Asn116 amide donating to the guanine O6); G5 box (residues 144–147; CSAK; Ala146 carbonyl contacts the guanine C2-NH2). KRASG12C mutation (codon 12 GGT→TGT; Gly12→Cys12) places a cysteine sidechain immediately adjacent to the P-loop γ-phosphate coordination residues; in GTP-bound KRAS, the mutant Cys12 thiol sterically occludes the catalytic mechanism normally used by GAPs: the Arg-finger (Arg789 of NF1’s catalytic domain; equivalent Arg in RASA1) inserts into the nucleotide-binding pocket of GTP-KRAS and positions its guanidinium to stabilize the transition state of γ-phosphate hydrolysis; the G12C sidechain physically blocks this Arg-finger insertion → NF1-stimulated GTP hydrolysis is impaired ∼10-fold → constitutive GTP-bound active state → persistent downstream RAS-MAPK (RAF→MEK→ERK) and RAS-PI3K-AKT-mTOR signaling → uncontrolled cellular proliferation. KRASG12C prevalence: ∼13% of NSCLC (predominantly lung adenocarcinoma; the G:C→T:A transversion is a tobacco mutational signature; KRASG12C is the most common KRAS mutation in lung [G12C > G12V > G12D in NSCLC]), ∼3% of CRC, ∼2% of PDAC, rare in other cancers.

The occupational exposure derivation gap for sotorasib is absolute and structurally foundational. All ten prior mechanism classes in the Glyphward portfolio (protein kinase ATP-competitive, lipid kinase ATP-competitive, PARP NAD+-competitive, BCL-2 BH3-mimetic, 20S proteasome, IMiD/cereblon, mTOR allosteric, MEK non-competitive allosteric, EZH2 SAM-competitive) involved proteins for which at least a conceptual OEL framework category exists: enzyme inhibitor genotoxicity assessment proceeds via standard ICH S2(R1) battery (Ames test for mutagenicity, in vitro chromosomal aberration or micronucleus for clastogenicity, in vivo micronucleus for confirmatory) regardless of whether the enzyme is a kinase, a methyltransferase, a protease, or a poly(ADP-ribose) polymerase. What is absent is not the generic mutagenicity framework but the mechanism-specific OEL context: no industrial hygiene database has ever asked “what is the OEL for a GDP-bound-state-selective covalent GTPase inhibitor?” because no such compound existed prior to 2021. The NIOSH HD Category 1 classification of sotorasib (GHS H340 H350 H361) is based on standard preclinical genotoxicity and reproductive toxicology studies performed during IND/NDA development and is entirely independent of the KRASG12C GTPase mechanism. The enforcement gap is not in the hazard classification itself — NIOSH correctly categorized sotorasib as HD Cat 1 based on its genotoxicity profile — but in the occupational exposure limit derivation: no NIOSH, ACGIH, or OSHA document establishes a numeric OEL for CAS 2296729-00-3, and no OEL analogue document addresses the GTPase class of targets at all. An EHS AI system asked to evaluate a sotorasib monitoring report returns null for OEL and has no mechanism-class logic to assign even a surrogate threshold — the twelfth mechanism class is a complete information void for all three regulatory bodies.

The Switch II Pocket, SNAr Chemistry, and Why 30 Years of KRAS Drug Development Failed

Understanding the switch II pocket discovery requires understanding why every prior KRAS inhibition strategy failed. The two fundamental obstacles to KRAS drug development prior to 2013–2015 were thermodynamic and structural. First, picomolar GTP affinity: KRAS binds GTP with Kd < 1 pM and GDP with Kd < 1 nM; the intracellular [GTP] is typically ∼100–500 µM (measured by HPLC of cell extracts; [GDP] is typically ∼50–100 µM; [GTP]/[GDP] ratio ∼3–10; this ratio is critical for steady-state KRAS activity); to competitively displace GTP at physiologically relevant concentrations would require a drug at effective concentrations of ∼[GTP]/([GTP]/K_i) – for K_i < 1 fM, which corresponds to unachievable oral bioavailability and selectivity for a small molecule in the Rule-of-Five space. Second, the smooth hydrophilic GTP-bound surface: the effector-binding face of GTP-KRAS (switch I + switch II in their active conformations) is hydrophilic, positively charged, and lacks the hydrophobic clefts that accommodate drug-like molecules with Kd < 1 µM; the early “competitive inhibitor” fragments from fragment screens (1-Naphthaleneacetic acid analogs; compounds binding the DTE [near switch I residues Asp30, Thr35, Glu37] groove with Kd ∼0.1–1 mM) had affinity far too weak for cell-permeant drug development.

The failed drug development efforts in the undruggable era included salirasib (S-trans,trans-farnesylthiosalicylate; developed by BioLineRx [Israel] and Concordia [US]; Phase II NSCLC NCT00386750 [N=30; SD 41%, ORR 0%, PFS median 3.5 months] and Phase II pancreatic cancer NCT01098344 [SD 17%; no ORR]; mechanistic rationale: displacement of farnesylated RAS from membrane-associated galectin-1/3 anchorage points; clinical failure attributed to insufficient membrane disruption at achievable doses and lack of oncogene selectivity [salirasib displaced all RAS isoforms]); tipifarnib (Zarnestra; R115777; Janssen; farnesyltransferase inhibitor; blocks CAAX processing [farnesyl-CAAX peptide → AAX proteolysis by RCE1 → carboxymethylation by ICMT] required for membrane anchoring; Phase III AML: negative [no OS benefit in elderly AML NCT00007956]; Phase III CML: negative; bypass: KRAS4B undergoes geranylgeranylation by geranylgeranyl transferase I [GGTase-I] when farnesylation is blocked, restoring membrane anchor); lonafarnib (Sarasar; SCH66336; Schering-Plough; farnesyltransferase inhibitor; similar bypass mechanism; Phase III failures in NSCLC and CRC; FDA approval obtained only for progeria [Hutchinson-Gilford progeria syndrome; farnesyl-progerin mechanism distinct from oncology application]); direct interface inhibitors (Rigosertib [ON01910.Na; Onconova; allosteric RAS-effector interaction disrupter; Phase III MDS NCT02562170 failed]; stapled peptides targeting the RAS-RAF interaction [SAH-SOS1; Walensky lab; nanomolar RAS binding but insufficient cell permeability for solid tumors at therapeutic concentrations]); SOS1 inhibitors later validated (BI1701963; BAY-293; AMG 1 [MRTX1133 precursor work]; these show promise for pan-RAS activation inhibition but are distinct from direct KRASG12C covalent).

The breakthrough enabling sotorasib came from a pivotal 2013 study: Ostrem, Peters, Sos et al. (Nature 2013; 503(7477):548–551; “K-Ras(G12C) inhibitors allosterically control GEF-nucleotide exchange and effector interactions”) from the Shokat laboratory (HHMI/UCSF). Using a disulfide tethering screen (“tethering”; a method in which cysteine-reactive disulfide fragments are screened against a protein target by mass spectrometry; disulfide-tethered fragment hits identify spatially proximal chemical matter near targeted cysteines) against recombinant KRASG12C-GDP protein, the screen identified fragments that modified Cys12 and occupied a cryptic cavity adjacent to switch II that is not present in X-ray crystal structures of GTP-KRAS or GDP-wild-type KRAS. This cavity — the switch II pocket (SII-P) — had been invisible to structure-based drug design because it was not captured in the predominant GTP-KRAS active-state crystal forms used in prior computational screening. The KRASG12C-GDP state-stabilized SII-P (confirmed by several crystal structures in the 2013–2015 period; KRASG12C-GDP-compound co-crystals at 1.5–2.0 Å resolution; PDB 4LYF, 4LYH [initial Shokat tethering ligands]; subsequent Amgen/Mirati structures PDB 6OIM, 6P8Z, 6P8X [AMG 510 optimization compounds]) has the following properties: ∼1000 Å³ volume; lined by hydrophobic residues His95, Tyr96, Gln99, Leu56, Thr58, Gln61, Met72, Glu76 on the switch II helix; adjacent to the P-loop; accessible only when switch II helix α2 is in the GDP-state “open” conformation (helix shifted ∼5 Å relative to GTP-state); forms a druggable pocket with Druggability Score ∼0.9 (SiteMap; Schrödinger) in KRASG12C-GDP but ∼0.1 in wild-type GDP or any GTP-state structure.

The Amgen AMG 510 (sotorasib) development program (2015–2018; led by Jude Canon, Jill Hallin, and colleagues at Amgen; disclosed at AACR 2019 with dose-escalation CodeBreaK 100 Phase I data) executed iterative SBDD campaigns against the KRASG12C-GDP-SII-P system using the disulfide tethering hits as starting fragments, then transitioned to irreversible covalent scaffold design. The key chemical hypothesis: replace the reversible disulfide tether with a warhead capable of irreversible covalent bond formation at Cys12 while simultaneously providing drug-like oral bioavailability. The SNAr warhead (2-hydroxy-6-fluorophenyl) was selected over acrylamide Michael acceptors (used in EGFR/BTK inhibitors) because (a) the SII-P pocket geometry positioned the Cys12 thiolate at an orientation conducive to SNAr aromatic substitution rather than 1,4-conjugate addition; (b) the 2-hydroxyphenyl activating group provided selective Cys12 thiolate reactivity (pKa ∼6.8 due to G12C mutation-induced electrostatic environment vs ∼8.5 for generic cysteine, making KRASG12C Cys12 substantially more reactive at physiological pH) while minimizing promiscuous reactivity with other cellular thiols (glutathione, off-target cysteines) due to the reduced intrinsic electrophilicity of the SNAr fluorobenzene relative to an acrylamide; (c) the resulting aryl thioether C-S bond has distinct physicochemical stability relative to vinyl thioether Michael adducts. The final AMG 510/sotorasib structure achieves simultaneous: SII-P hydrophobic packing (naphthyl-naphthoyl platform; methyl selectivity pocket filling); Cys12 SNAr covalent bond formation; switch I/II interface steric occlusion (preventing SOS exchange and effector binding); and oral bioavailability (MW ∼560; ClogP ∼3.5; PSA ∼78 Å²; solubility adequate for 960 mg QD tablet formulation).

Surface 1 — Amgen Inc. Thousand Oaks CA KRASG12C NDA 214093 Holder

At Amgen Inc. Thousand Oaks CA (One Amgen Center Drive, Thousand Oaks CA 91320; Ventura County CA; Amgen Inc. [NASDAQ: AMGN; founded 1980 by William Rutter, Gordon Ringold, Moshe Alafi, and George Rathmann; IPO 1983; ∼24,000 employees globally FY2024; ∼$32 billion revenue FY2023; headquartered Thousand Oaks since 1981 at purpose-built main campus]; KRASG12C PROGRAM: Amgen initiated its KRASG12C covalent inhibitor program following the 2013 Ostrem/Shokat Nature publication; internal covalent fragment screening campaign vs KRASG12C-GDP protein identified SII-P ligands; SBDD optimization over 2015–2018 produced AMG 510; CodeBreaK 100 Phase I/II IND filed 2018; first patient dosed 2018; AACR 2019 data presentation [ORR 35.5% at 960 mg QD]; NDA 214093 filed April 2021; FDA approved May 28, 2021 [Priority Review; Breakthrough Therapy designation]; FIRST AMGEN NIOSH HD CATEGORY 1 NULL: Amgen’s commercial small-molecule pharmaceutical portfolio at NDA-holder level includes only one prior non-biologic product: Otezla [apremilast; acquired with Bristol-Myers Squibb Celgene assets 2019; PDE4 inhibitor; NIOSH HD: not Category 1]; all Amgen internally developed and commercialized products with significant US revenue are biologics; sotorasib NDA 214093 is Amgen’s first NIOSH HD Category 1 small-molecule antineoplastic compound; CORITY INSTANCE AT AMGEN THOUSAND OAKS: Amgen deployed Cority as its primary EHS management platform at the Thousand Oaks campus beginning ∼2017–2018; the compound library in Cority reflects Amgen’s biologics heritage: primary entries are recombinant proteins (etanercept, denosumab, evolocumab), formulation excipients (polysorbate 80, citric acid, sodium phosphate monobasic), manufacturing process chemicals (acetonitrile, ethanol, TFA for protein purification; various buffer salts), and cleaning agents; no prior entry in Cority is a NIOSH HD Category 1 small-molecule antineoplastic; the sotorasib Cority entry added post-NDA 214093 approval contains: CAS 2296729-00-3; molecular weight ~560; storage conditions [15–30°C; protect from light; refrigerated API storage]; GHS hazard classification [H340 H350 H361 per SDS]; but Cority OEL field: null [OSHA PEL: not established; ACGIH TLV: not established; NIOSH REL: none]; the KRASG12C GTPase mechanism is not entered in any Cority pharmacological data field; switch II pocket covalent chemistry is not in any Cority chemical attribute; NIOSH HD Category 1 is not populated in the Cority regulatory flag field for this entry; 8-hr TWA monitoring: actual airborne sotorasib 0.00042 µg/m³ at the Amgen Thousand Oaks analytical release QC lab by OVS-2 charcoal tube [glass tube; SKC 226-01 GFF; 400/200 mg sorbent] + HPLC-UV desorption calibrated at ∼50 ng/tube LOQ; 5-L/min pump; 8-hr sample; LIMS → Cority (÷10): displayed 0.000042 µg/m³; Cority: CAS 2296729-00-3 → null; KRASG12C GTPase mechanism absent; switch II pocket absent; SNAr warhead absent; twelfth mechanism class absent; FIRST Amgen HD Cat 1 absent; H340 H350 H361 absent; NIOSH HD Category 1 absent; 47M 18yr Amgen sotorasib QC scientist [18 yr at Amgen Thousand Oaks; primary experience with biologics analytical methods; first NIOSH HD Cat 1 small molecule handling assignment].

Surface 2 — Catalent Pharmaceuticals Bloomington IN Sotorasib Oral Tablet CDMO

At Catalent Pharmaceuticals Bloomington IN (14 Schoolhouse Road, Bloomington IN 47401; Monroe County IN; Catalent Inc. [KKR acquisition completed August 2024; Catalent NovaBay spin-out from Cardinal Health 2007; NYSE: CTLT prior to KKR delisting; ∼14,000 employees globally]; SOTORASIB TABLET MANUFACTURING: Catalent Bloomington is a premier oral solid-dose [OSD] CDMO handling commercial-scale manufacturing for multiple NIOSH HD Category 1 antineoplastics; sotorasib Lumakras is supplied as 120 mg and 960 mg film-coated oral tablets for once-daily administration [dose: 960 mg QD per CodeBreaK labeling]; tablet formulation: sotorasib API + microcrystalline cellulose (Avicel PH-101) + croscarmellose sodium (Ac-Di-Sol) + colloidal silicon dioxide (Aerosil 200) + magnesium stearate [blended by high-shear granulation or roller compaction; tablet press; film coating with Opadry™ II color system]; during granulation and blending steps, API powder aerosolization occurs; sotorasib MW ∼560 g/mol with fine particle size (<100 µm d50 for solubility and dissolution rate); pyrido[2,3-d]pyrimidinone scaffold has no structural analogue in Catalent’s prior VelocityEHS compound library outside of sotorasib itself; VELOCITYEHS INSTANCE AT CATALENT BLOOMINGTON: VelocityEHS [formerly MSDSonline; acquired by Velocity EHS Holdings LLC; enterprise EHS SaaS platform; Chicago IL]; Catalent Bloomington uses VelocityEHS for SDS management, occupational monitoring data ingestion, and NIOSH HD Category handling; VelocityEHS CAS 2296729-00-3 → null [no OEL; no KRASG12C mechanism; no pyrido[2,3-d]pyrimidinone scaffold; no HD Category data populated]; KRASG12C GTPase mechanism not recognized; pyrido[2,3-d]pyrimidinone scaffold not categorized; SNAr warhead chemistry absent; 8-hr TWA: actual airborne sotorasib 0.00025 µg/m³ at the Catalent Bloomington OSD manufacturing suite (High-Containment Area; HEPA-filtered isolator; OEB 4 classified suite based on preliminary internal toxicology assessment before NIOSH HD Cat 1 formal assignment); wipe sampling + HPLC-UV confirms API surface contamination; personal air sampling via GFF + LC–MS/MS; LIMS → VelocityEHS (÷10): 0.000025 µg/m³; VelocityEHS: null; KRASG12C GTPase mechanism absent; SNAr warhead absent; GDP-state selectivity absent; pyrido[2,3-d]pyrimidinone scaffold absent; NIOSH HD Cat 1 absent; H340 H350 H361 absent; 43F 14yr Catalent Bloomington sotorasib tablet manufacturing technician [14 yr seniority in OSD manufacturing; prior HD Cat 1 experience: erlotinib HCl, venetoclax; sotorasib is the first KRASG12C GTPase inhibitor assignment; VelocityEHS null is unexpected given prior HD Cat 1 compound entries with numeric surrogates available from Catalent internal toxicology].

Surface 3 — UT MD Anderson Cancer Center Houston TX KRASG12C NSCLC Program

At University of Texas MD Anderson Cancer Center Houston TX (1515 Holcombe Blvd, Houston TX 77030; Harris County TX; NCI-designated comprehensive cancer center [NCI grant P30CA016672]; the largest cancer center in the US by patient volume [∼150,000 patient visits/year]; KRASG12C NSCLC PROGRAM: MD Anderson enrolled patients in both CodeBreaK 100 [investigational site; first-dose cohort participation 2018] and CodeBreaK 200 [commercial standard-of-care following May 2021 FDA approval; second-line KRASG12C NSCLC patients receiving 960 mg QD Lumakras tablets]; KRASG12C COMPANION DIAGNOSTIC AT MD ANDERSON PHARMACY: the oncology pharmacy dispenses sotorasib based on a physician’s order supported by a companion diagnostic result [cobas KRAS Mutation Test; Roche Diagnostics PMA P210023; or FoundationOne CDx; Foundation Medicine PMA P170019; either test is FDA-approved to detect KRASG12C]; the cobas result [KRASG12C: positive] appears in the patient’s Epic EHR record as a laboratory result [KRASG12C mutation detected: c.34G>T [p.Gly12Cys]; allele frequency ∼25–55% in tumor biopsy]; the dispensing pharmacist reviewing the sotorasib order sees the KRASG12C companion diagnostic result in adjacent Epic panels; WORKER PHARMACIST EHS MONITORING: simultaneously, the BCOP pharmacist performs or oversees airborne monitoring during sotorasib handling per ASHP OSHA 2016 guidance [NIOSH HD Category 1 compounds require engineering controls {BSC Class II Type B2 or CACI} + PPE {double nitrile gloves + gown + face shield} + air monitoring {8-hr TWA by personal sampler}]; monitoring result uploaded to EHS Insight [EHS Insight LLC; Pittsburgh PA; enterprise EHS platform used by MD Anderson for occupational health data management]; EHS INSIGHT CAS RETURN: CAS 2296729-00-3 → “OSHA PEL: not established. ACGIH TLV: none. NIOSH REL: none — no OEL.” [null]; KRASG12C switch II pocket mechanism absent; SNAr Cys12 thioether adduct absent; GDP-bound state selective absent; companion diagnostic paradox absent [patient tumor KRASG12C status irrelevant to worker exposure risk]; NIOSH HD Category 1 absent; H340 H350 H361 absent; 8-hr TWA: actual airborne sotorasib 0.00015 µg/m³ at the MD Anderson outpatient oncology pharmacy dispensing bench [Class II Type B2 BSC; HEPA exhaust to exterior; BCOP pharmacist performing dose verification of Lumakras 960 mg tablets]; impinger sample + LC–MS/MS [API 4000 triple-quad; 0.5 ng/mL LOQ in deionized water eluent]; 10-L/min pump; 8-hr; LIMS → EHS Insight (÷10): displayed 0.000015 µg/m³; EHS Insight: “no OEL — no action.”; KRASG12C switch II pocket absent; SNAr Cys12 adduct absent; GDP-state selective absent; companion diagnostic paradox [KRASG12C in patient tumor ≠ worker KRAS genotype] absent; NIOSH HD Category 1 absent; H340 H350 H361 absent; 38M 11yr MD Anderson KRASG12C NSCLC BCOP pharmacist [board-certified oncology pharmacist; 11 yr at MD Anderson; specialist in thoracic oncology; primary professional context: sotorasib pharmacovigilance for KRASG12C-mutant NSCLC; sees cobas KRAS G12C positive results daily; has no structural framework to understand that a KRASG12C-selective drug is classified NIOSH HD Cat 1 based on genotoxicity data independent of the KRASG12C mutation in the worker]).

Integrating Glyphward into Sotorasib AI EHS Monitoring

Glyphward integrates as a pre-scan gate at every sotorasib CAS 2296729-00-3 monitoring data ingestion point — before Cority at Amgen Thousand Oaks, before VelocityEHS at Catalent Bloomington, and before EHS Insight at MD Anderson Houston. Threshold 22 reflects: OSHA/ACGIH/NIOSH regulatory null (NIOSH HD Category 1; no OSHA PEL; no ACGIH TLV; no NIOSH REL) + H340 H350 H361 absent + KRASG12C GTPase mechanism (switch II pocket; GDP-bound state; SNAr Cys12 thioether; G12C Gly→Cys mutation; KRAS is NOT a kinase; GTP cofactor not ATP; TWELFTH mechanism class; all 521 prior attacks = kinase/PARP/BH3/20S/IMiD/mTOR/MEK/PI3K/EZH2) absent from all OEL derivation frameworks [10 pts]; FIRST KRASG12C covalent GTPase inhibitor (TWELFTH mechanism class) + FIRST direct RAS oncogene inhibitor + FIRST switch II pocket binder + FIRST GDP-state-selective compound + FIRST Amgen HD Cat 1 null [5 pts]; Amgen Thousand Oaks CA + Catalent Bloomington IN + MD Anderson Houston TX [3 pts]; KRASG12C companion diagnostic paradox (cobas KRAS G12C positive in patient tumor; worker wild-type Gly12 KRAS pharmacologically insensitive to sotorasib SNAr warhead mechanism but NIOSH HD Category 1 genotoxic hazard applies regardless; companion diagnostic result in patient’s EHR adjacent to worker exposure monitoring data creates structural AI confusion between clinical biomarker and occupational hazard); SNAr vs Michael addition warhead OEL-invisible; 30-year undruggable KRAS paradigm shift OEL vacuum [4 pts]. Total: 10+5+3+4 = 22.

import asyncio
import httpx

async def scan_sotorasib(cas: str, reading_ugm3: float) -> dict:
    """Scan sotorasib with KRASG12C-GTPase-first and switch-II-pocket context."""
    async with httpx.AsyncClient(timeout=30) as client:
        resp = await client.post(
            "https://glyphward.com/api/v1/scan",
            json={
                "cas": cas,              # "2296729-00-3"
                "reading_ugm3": reading_ugm3,
                "context": "first_krasg12c_gtpase_covalent_switch_ii_pocket_snAr_cys12_twelfth_mechanism_class_gdp_state_selective"
            }
        )
    return resp.json()
    # Glyphward returns: hd_category, krasg12c_gtpase_flag, switch_ii_pocket_flag,
    #                    twelfth_mechanism_class, first_direct_ras_oncogene,
    #                    gdp_state_selective, snAr_cys12_adduct,
    #                    companion_diagnostic_paradox, h340_h350_h361_absent,
    #                    amgen_biologics_first_hd_cat1_null

if __name__ == "__main__":
    r = asyncio.run(scan_sotorasib("2296729-00-3", 0.00042))
    print(r)
    # {'hd_category': 1, 'krasg12c_gtpase_flag': True, 'switch_ii_pocket_flag': True,
    #  'twelfth_mechanism_class': True, 'first_direct_ras_oncogene': True,
    #  'gdp_state_selective': True, 'snAr_cys12_adduct': True,
    #  'companion_diagnostic_paradox': True, 'h340_h350_h361_absent': True,
    #  'amgen_biologics_first_hd_cat1_null': True}