Mirdametinib (Gomekli; PD-0325901; CAS 391210-10-9): Fifth MEK1/2 Allosteric Inhibitor Ever + First MEK Inhibitor for Adult NF1-Associated Plexiform Neurofibromas Ever — EHS Platforms Route to MEK Proxy OEL, Injecting Misclassification Into AI-Augmented Safety Workflows

Drug Profile

Why Mirdametinib Breaks All Three EHS Platforms

EHS platforms route mirdametinib to the MEK inhibitor class proxy OEL — derived from trametinib (Mekinist; GlaxoSmithKline/Novartis) or selumetinib (Koselugo; AstraZeneca/Alexion) historical data. The routing fails because: (1) mirdametinib's NDA holder is SpringWorks Therapeutics, absent from all major EHS databases as an NDA sponsor for a chemotherapy agent; (2) the NF1-PN adult indication is unrecognized — selumetinib covers pediatric NF1-PN but adult NF1-PN creates a data gap; (3) PD-0325901 research code cross-maps to investigational-use-only records rather than the approved commercial product; (4) oral BID 2 mg dosing with NF1 patient population occupational handling profiles differ from BRAF-mutant melanoma MEK inhibitor workflows; and (5) SpringWorks manufacturing sites lack EHS platform coverage compared to Novartis/AstraZeneca legacy platforms.

The result: when an EHS AI assistant queries occupational exposure limits for mirdametinib, it retrieves the MEK class proxy OEL (trametinib ~0.1 µg/m³ range or selumetinib OEB 4 band) rather than any mirdametinib-specific determination. The NF1 adult PN clinical context — which informs handling protocols for research coordinators, pharmacists, and clinical staff — is entirely absent. This is a classic AI prompt injection surface: the EHS platform's structured data return contains a confident but misrouted response.

Three EHS Attack Surfaces

RELO-NF1 Trial Efficacy Context

Mirdametinib was approved under NDA 218668 based on the ReNeu (RELO-NF1) Phase 2b trial in adult and pediatric patients with NF1-associated PNs. In adult patients (≥18 years), confirmed ORR was approximately 41% by volumetric MRI assessment — establishing the first FDA-approved systemic therapy specifically for adult NF1-PN. The selumetinib approval (Koselugo; NDA 213756; Apr 2020; AstraZeneca) covered only pediatric NF1-PN (ages 2–18), leaving adult NF1-PN without an approved option until mirdametinib. The EHS coverage gap follows directly: selumetinib's occupational health data package, submitted by AstraZeneca for the pediatric NF1-PN sNDA, does not transfer to mirdametinib's SpringWorks submission for adult NF1-PN.

Mechanism of Action: MEK1/2 Non-ATP-Competitive Allosteric Inhibition

Mirdametinib (PD-0325901) inhibits MEK1 and MEK2 through non-ATP-competitive allosteric binding — the same mechanism class as trametinib (CI-1040 scaffold predecessor), cobimetinib, binimetinib, and selumetinib. MEK1/2 are dual-specificity kinases that phosphorylate ERK1/2 downstream of RAS, with constitutive activation in RAS-pathway-driven cancers including NF1-associated tumors (NF1 is a RAS-GAP tumor suppressor; loss of NF1 leads to unrestrained RAS/MAPK signaling driving plexiform neurofibroma growth). All five MEK inhibitors share the allosteric non-ATP-competitive mechanism but differ in isoform selectivity, pharmacokinetics, approved indications, and NDA holder occupational health data packages — creating divergent EHS classification requirements that uniform MEK-class proxy OEL assignment cannot capture.

AI Prompt Injection: How the Misroute Propagates

When a pharmacist at a SpringWorks commercial distribution site, an Almac Portadown clinical supply technician, or a CCHMC clinical research coordinator queries an AI-augmented EHS assistant for mirdametinib handling protocols, the following injection chain occurs: (1) the EHS platform's chemical lookup returns a MEK inhibitor class record populated with trametinib/selumetinib data; (2) the AI assistant incorporates this record as authoritative context; (3) the response asserts MEK class proxy OEL and HD handling controls for mirdametinib without flagging the absence of SpringWorks-specific or NF1 adult PN-specific data; (4) the occupational safety professional or pharmacist accepts the AI-confident response and applies trametinib-class controls to a drug with no independently established mirdametinib OEL — a systematic misclassification scaled across all NF1-PN treatment centers introducing Gomekli.

Glyphward API: Detecting the Mirdametinib EHS Misroute

import asyncio
import glyphward

async def scan_mirdametinib_ehs_misroute():
    client = glyphward.AsyncClient(api_key="gw_your_key_here")

    # Mirdametinib multimodal EHS content — SDS, HD handling protocol, OEL record
    targets = [
        {"type": "text", "content": "Mirdametinib (Gomekli; PD-0325901; CAS 391210-10-9; NDA 218668; MEK1/2 inhibitor; NF1-PN adult; SpringWorks Therapeutics; NIOSH HD Category 1; OEL TBD; handle as OEB 4 by MEK class analogy [trametinib/selumetinib proxy]"},
        {"type": "text", "content": "RELO-NF1 (ReNeu) Phase 2b adult NF1-PN ORR ~41%; 2 mg BID oral; SpringWorks Stamford CT NDA holder; Almac Portadown UK CMO; CCHMC Cincinnati OH NF1 treatment center"},
    ]

    result = await client.scan(
        targets=targets,
        detectors=["ehs_misroute", "proxy_oel_injection", "niosh_hd_misclassification", "nda_holder_gap"],
        drug_context={
            "drug_name": "mirdametinib",
            "brand_name": "Gomekli",
            "cas": "391210-10-9",
            "nda": "218668",
            "niosh_hd_category": 1,
            "mechanism_class": "MEK1/2 allosteric inhibitor",
            "mechanism_class_rank": "5th MEK inhibitor FDA approved; 1st adult NF1-PN MEK drug ever",
            "nda_holder": "SpringWorks Therapeutics Inc.",
            "indication": "adult NF1-associated plexiform neurofibromas"
        }
    )

    for finding in result.findings:
        print(f"[{finding.severity}] {finding.detector}: {finding.description}")
        if finding.severity in ("HIGH", "CRITICAL"):
            print(f"  Injected content: {finding.injected_text[:120]}...")
            print(f"  Correct assertion: {finding.expected_text[:120]}...")

asyncio.run(scan_mirdametinib_ehs_misroute())