On Friday, November 20, 2026, activist short-seller Scorpion Capital LLC (led by Gurkirat “Kir” Kahlon) dropped a 297-page investigative dossier alongside four commissioned contract-lab manuscripts (main report, A1, A2, A3, A4) branding French biotech Abivax SA (NASDAQ / Euronext Paris: ABVX) “A Biotech Fraud For The Ages.” Its headline accusation could hardly be more incendiary: Abivax’s once-daily ulcerative colitis (UC) pill, obefazimod (ABX464), is allegedly an ultra-lipophilic membrane poison “in the range of DDT, dioxin, and persistent organic pollutants” that triggers UVA-driven skin cancers while hiding behind a fictitious miR-124 mechanism of action.
By Friday’s closing bell on Nasdaq, ABVX slid -9.24% (-$7.45) to $73.20—giving the company a basic market capitalization of $6.37 billion across 87,054,535 post-offering shares (per Abivax’s July 2, 2026 SEC 424B5 prospectus), now -41.4% below its $920 million July follow-on offering price ($125.00/ADS) and a world away from early-2026 rumors of a €15 billion ($17.5 billion) Eli Lilly takeover.
Abivax fired back the same day, stating that Scorpion’s report “contains numerous factually inaccurate and misleading statements, assumptions, and conclusions, including those based on studies commissioned and funded by Scorpion Capital itself,” reaffirming the integrity of its Phase 3 ABTECT program and its target to file a U.S. FDA New Drug Application (NDA) before the end of 2026—while pointing out that Scorpion openly disclosed plans to begin covering a substantial portion of its short position immediately upon publication.
Is Scorpion Capital exposing the next Avacopan/Tavneos disaster before FDA opens the file, or recycling a chemical scare tactic against a real Phase 3 winner? Answering that requires auditing two distinct questions: first, what is factually true about obefazimod’s chemistry, biological mechanism, skin-cancer math, and clinical trial execution; and second, what that evidence implies for FDA approval, commercial labeling, and buyout odds. Cross-examining PubChem structures, Abivax’s SEC filings, unredacted ClinicalTrials.gov protocols, and peer-reviewed records produces a far more nuanced verdict than either side wants you to believe.
[!IMPORTANT] Forensic Thesis Summary:
- Where Scorpion Draws Real Blood (Mechanism, Omissions & Geography): Abivax’s own human
CYP1A2drug-interaction trial (NCT05121714) proves obefazimod triggered a-79.0%crash in caffeineAUC(7,420 → 1,560 h·ng/mL)—the unmistakable clinical footprint of systemic Aryl Hydrocarbon Receptor (AhR) activation, not a puremiR-124splicing switch. Meanwhile, Abivax’sFY2025 Form 20-Fomits nonclinical phototoxicity, teratogenicity (6-monthcontraception for men and women), and marmoset cardiac fibrosis warnings written right inside its unredacted 2022 Phase 2b protocol (and visibly redacted out of its 2024 Phase 3 protocol). Add~2.70NMSCs per100patient-years reconstructed in randomized50 mgPart 1 maintenance (4×placebo) and a25 mgfailure in China-freeABTECT-2(11.3%vs.6.3%,p = 0.165), and FDA review friction is genuinely high.- Where Scorpion Overreaches (“DDT” & “Zero Efficacy” Collapse Under Primary Data): Calling obefazimod “DDT in a capsule” ignores rapid human N-glucuronidation, which attaches a water-soluble sugar tag (
ABX464-N-glucuronide,TPSA 133 Ų) that reaches~160×higher plasmaCmaxand clears with a~90–110 hour(~4-day) terminal half-life—lightyears from DDT’s6–11 yearpersistence. More decisively,50 mgobefazimod cleanly won in China-freeABTECT-2(19.8%vs.6.3%remission,p = 7.23 × 10⁻⁵; endoscopic improvement35.5%vs.10.1%) and crushed placebo inABTECT-Maintenance(51.3%vs.10.4%,p < 0.0001). Ironically, Scorpion’s own proof that obefazimod activatesAhRexplains why that clinical efficacy is real:AhR → IL-22is a well-established mucosal-healing pathway in IBD!- The Bottom-Line Valuation Reality: Unlike cash-starved microcap promotions, Abivax sits on
€1,169.5 million(~$1.33 billion, or~$15.31/share) in virtually debt-free pro forma cash (€1.2Mborrowings,0royalty overhang, runway into Q4 2029) and already cleared a positive July 30, 2026 FDA Pre-NDA interaction. The real threat at$73.20($6.37Bcap) is not insolvency or a$0fraud—it is FDABIMOinspection risk,NDAreview delays, and boxed/label warnings (teratogenicity, skin/UVA monitoring,CYP1A2DDIs) that turn a rumored€15Bbuyout into a heavily discounted commercial grind.
1. Three Competing Narratives at $73.20
Before zooming into the molecules and clinical spreadsheets, it helps to put Abivax’s pitch, Scorpion’s indictment, and the cold primary record side by side.
| Analytical Dimension | Abivax Management Narrative | Scorpion Capital Short Thesis (297 pp. + 4 Labs) |
Independent Primary-Source Audit |
|---|---|---|---|
Mechanism of Action (MOA) |
First-in-class oral enhancer that binds the Cap-Binding Complex (CBC) to selectively upregulate anti-inflammatory miR-124. |
miR-124 is a mirage (0.00013 copies/cell on ddPCR); obefazimod is a nonspecific lipophilic mass-action membrane toxicant / anthelmintic and an AhR agonist. |
Scorpion wins on AhR vs. miR-124, loses on “toxic only”: Human CYP1A2 (-79.0% caffeine AUC) & GSE116073 (CYP1A1 #1 gene) prove systemic AhR agonism—which genuinely heals colitis via IL-22. |
| Chemistry & Bioaccumulation | Small molecule rapidly metabolized to active ABX464-N-glucuronide; steady state in 4 weeks; tolerable long-term oral profile. |
Neutral BRlogP = 6.35 (2.2M:1 lipid preference), 0.16 µM solubility, V/F = 1,228 L, enterohepatic recycling (EHR)—“in the range of DDT & dioxin.” |
Extreme parent lipophilicity + probable EHR are real, violating Lipinski/Pfizer 3/75 screens; yet ~4-day N-glucuronide clearance disproves multi-year DDT bioaccumulation. |
| Skin Cancer & Safety Profile | No new safety signals; integrated Phase 2+3 1,704 PY NMSC (0.59/100 PY) and non-NMSC (0.35/100 PY) rates match UC background. |
16/17 malignancies on active drug; UVA absorption 8.4× ICH S10 limit (349 nm); 64 unexplained skin lesions; 4 active NMSCs deemed possibly/probably related. |
Both sides spin denominators: 50 mg Part 1 NMSC (~2.70/100 PY, Poisson 95% CI: 0.74–6.91) was diluted via pooling; SEC 20-F hides protocol phototoxicity & teratogenicity. |
Phase 3 ABTECT Integrity |
Both 50 mg and 25 mg succeeded across induction (ABTECT-1/2) and maintenance (∆39.3% and ∆40.3%, p < 0.0001); on track for Q4 NDA. |
Reliance on 41 Chinese sites (incl. PLA military hospitals) & 15 Ukraine sites; zero 0/36 NA placebo remission; histology index switching (Robarts → Geboes). |
Divided verdict: 50 mg clearly succeeded without China in ABTECT-2 (19.8% vs 6.3%, p = 7.23 × 10⁻⁵) and Maintenance; 25 mg collapsed in ABTECT-2 (p = 0.165) and NA (0/36 placebo) is razor-thin. |
| Balance Sheet & M&A Outlook | €1.17B cash runway to Q4 2029, zero royalty overhang, positive July 30, 2026 Pre-NDA; positioned to transform UC care or attract Big Pharma. |
“A biotech fraud for the ages”; FDA CRL or worse; Big Pharma (AbbVie, Lilly) laughing at safety profile after Avacopan ($4B) and Etrasimod ($7B) busts. |
No financial distress (~$15.31/sh net cash), preventing near-term collapse; however, AhR/teratogenicity/phototoxicity/China BIMO baggage severely cools an immediate €15B buyout. |
2. “DDT in a Capsule” vs. Sugar-Coated Metabolism: Auditing the Chemistry
To understand the chemistry battle, picture dropping a bead of motor oil into a glass of water. It refuses to mix and immediately clings to the greasy wall of the container. That is parent obefazimod (ABX464).
On PubChem (CID 49846599), obefazimod (8-chloro-N-[4-(trifluoromethoxy)phenyl]quinolin-2-amine, MW 338.71) already posts an elevated computed XLogP of 5.9 and a tiny topological polar surface area (TPSA of 34.2 Ų)—small, flat, rigid, and uncharged. In Manuscript A1, Scorpion Capital’s contracted lab measured two missing parameters directly:
- Experimental Lipophilicity (
BRlogP):6.35 ± 0.07across three runs (6.31,6.32,6.43)—a2.2 million-to-1preference for octanol/lipids over water. - Ionization (
Sirius T3titration): No measurablepKabetweenpH 2andpH 12, meaning obefazimod stays electrically neutral throughout the gut, bloodstream, and tissues (logD = logP). Paired with Abivax’s patent melting point (120.5°C), the General Solubility Equation yields an intrinsic water solubility of just0.16 µM(0.000054 mg/mL), matching Abivax’s own patent disclosures of207 µMin fasted intestinal fluid (FaSSIF) and1,500 µMin fed fluid (FeSSIF). - Apparent Distribution Volume (
V/F): From Abivax’s own fasted50 mghealthy-volunteer pharmacokinetics (Scherrer 2017), Scorpion computesCL/F = 966 L/handV/F ≈ 1,228 L—nearly29×total human body water (42 L) (or~123 Leven if oral bioavailability were only10%), signaling that un-metabolized parent drug rapidly flees plasma into lipid membranes, adipose tissue, and lipoproteins.
So is obefazimod literally “DDT or dioxin in a capsule”? Let’s put Scorpion’s central phrase through the Semantic Substitution Test—swapping two strictly defined meanings into the claim to see where the argument holds and where it breaks:
- Definition A (Strict — Persistent Organic Pollutants like DDT / TCDD): A halogenated molecule that resists human liver metabolism, stays locked in fat for years (
6–11 yearbiological half-life), and bioaccumulates indefinitely without ever reaching steady state. Obefazimod FAILS Definition A. Why? Because human liver enzymes immediately bolt a bulky, negatively charged sugar acid onto obefazimod’s nitrogen atom—a process called N-glucuronidation. Once tagged, the metaboliteABX464-N-glucuronide(PubChemCID 122499702) balloons toMW 514.8, shoots itsTPSAup nearly4×to133 Ų, ionizes at bloodpH 7.4(logD7.4 ~0.5–1.5), circulates at~160×higher peak plasma concentrations (Cmax) than the parent drug, and clears with a~90–110 hour(~4-day) terminal half-life, hitting clinical steady state by Week 4 (per Abivax’s unredacted Phase 2b protocol and PK literature). - Definition B (Pharmacological — High-Lipophilicity Outlier with Enterohepatic Recycling): A neutral
logP > 6parent that escapes water into cell membranes and lipoproteins, then undergoes enterohepatic recycling (EHR)—meaning the liver pumps water-solubleN-glucuronideinto bile, meals dump bile into the intestine, gut bacteria snip the sugar tag off (β-glucuronidase), and regenerated lipophilic parent re-enters the gut wall and bloodstream repeatedly before final excretion. Obefazimod MEETS Definition B. Six independent clues in Abivax’s own human PK studies point directly toEHR: a≥2.8×jump in parent exposure with food, the~4-dayterminal tail on a glucuronide that should normally clear renally in hours, a statistically significant10–12 hourpost-meal rebound shoulder (p = 0.003) in digitized curves, and a~17×collapse inN-glucuronideAUC(vs. only~5×inCmax) when HIV patients took cobicistat—an inhibitor of theBCRPbiliary pump that loads glucuronides into bile!
flowchart TD
subgraph SG1["Gut & Absorption Pool"]
ORAL["Oral Obefazimod Dose<br/>(25 mg / 50 mg Form I)"] --> LUMEN["Intestinal Fluid Solubility<br/>(FaSSIF 207 uM / FeSSIF 1,500 uM)"]
LUMEN --> GUT["Enterocytes & Gut Immune Cells<br/>(Local AhR Activation + Membrane Partitioning)"]
BILE["Biliary Secretion into Gut<br/>(Blocked ~17x by Cobicistat / BCRP)"] --> BACT["Gut Bacterial Beta-Glucuronidase<br/>(Snips Sugar Tag -> Regenerates Parent)"]
BACT --> GUT
end
subgraph SG2["Liver & Enterohepatic Loop (EHR)"]
GUT --> PORTAL["Portal Vein Delivery"]
PORTAL --> UGT["Liver Phase II UGT Conjugation<br/>(Attaches Sugar Acid -> ABX464-N-Glu)"]
UGT --> BILE
UGT --> SYS_MET["Circulating ABX464-N-Glucuronide<br/>(MW 514.8, TPSA 133 A^2, ~160x Parent Cmax)<br/>(Terminal t1/2 ~90-110h; Renal/Fecal Clearance)"]
end
subgraph SG3["Systemic Parent Exposure & Downstream Targets"]
PORTAL --> PAR["Unconjugated Parent Obefazimod<br/>(BRlogP 6.35, TPSA 34.2 A^2, Neutral pH 2-12)"]
PAR --> AHR["Systemic AhR Receptor Binding<br/>(-79.0% Caffeine AUC via CYP1A2;<br/>CYP1A1 #1 Gene; IL-22 Repair)"]
PAR --> SKIN["Lipid/Lipoprotein Partitioning to Skin<br/>(UVA Absorption 349 nm, eps 8,400;<br/>Phototoxicity & NMSC Risk Signal)"]
end
How extreme is obefazimod’s parent chemistry compared with banned pollutants, failed/restricted high-logP drugs, and rival UC medications? Every figure below comes directly from PubChem primary compound records (paired with Scorpion’s experimental BRlogP where tested):
Compound (PubChem CID) |
Drug Class / Status | Molecular Weight | PubChem XLogP (Scorpion BRlogP) |
TPSA (Ų) |
Real-World Clinical / Regulatory Reality |
|---|---|---|---|---|---|
Obefazimod (CID 49846599) |
UC Phase 3 Oral Pill (Parent) | 338.71 |
5.9 (6.35 exp.) |
34.2 |
Neutral pH 2–12; V/F ~1,228 L; enters Phase 3 UC; strict protocol sun/contraception rules. |
ABX464-N-glucuronide (CID 122499702) |
Major Active Human Metabolite | 514.8 |
4.1 (neutral); ionizes (pH 7.4) |
133.0 |
~160× parent Cmax; t1/2 ~90–110 h; hits steady state Wk 4; phototoxic in animals. |
DDT (CID 3036) |
Banned Organochlorine Pesticide | 354.5 |
6.9 |
0.0 |
Non-metabolizable; 6–11 year human adipose persistence. (False direct PK equivalence to ABX464.) |
TCDD / Dioxin (CID 15625) |
Industrial Pollutant (POP) |
322.0 |
6.4 |
18.5 |
Archetypal irreversible AhR super-agonist; ~7–11 year half-life; chloracne & cancer. |
Avacopan / Tavneos (CID 49841217) |
Oral C5aR Antagonist (Amgen $4B) |
581.6 |
7.2 |
61.4 |
Neutral high-logP precedent: FDA CDER proposed withdrawal (Apr 27, 2026) over fatal DILI & trial manipulation. |
Acitretin (CID 5284513) |
Oral Psoriasis Retinoid (FDA Approved) | 326.4 |
6.1 |
46.5 |
Teratogenic Pregnancy Category X; 3-year post-stop pregnancy ban & blood donation ban. |
Amiodarone (CID 2157) |
Antiarrhythmic (FDA Approved) | 645.3 |
7.6 |
42.7 |
Classic tissue-loading drug; 40–55 day half-life; boxed pulmonary/liver toxicity & photosensitivity. |
Clofazimine (CID 2794) |
Orphan Leprosy Antibiotic (Approved) | 473.4 |
7.1 |
40.0 |
Crystalline GI/adipose deposition; skin discoloration; multi-month tissue retention. |
Etrasimod / Velsipity (CID 44623998) |
Oral S1P Modulator for UC (Pfizer $6.7B) |
457.5 |
6.4 |
62.3 |
Approved for UC (ionizable acid lowers logD7.4); commercial launch underperformed expectations. |
Tapinarof / Vtama (CID 6439522) |
Topical AhR Agonist (Approved) |
254.3 |
4.6 |
40.5 |
Proves AhR agonism is a real FDA-approved anti-inflammatory MOA—though used topically, not daily systemic pill. |
Upadacitinib (CID 58557659) |
Oral JAK1 Inhibitor for UC (Rinvoq) |
380.4 |
2.7 |
78.3 |
Conventional oral small-molecule logP; carries boxed MACE/malignancy/thrombosis warnings. |
Tofacitinib (CID 9926791) |
Oral JAK Inhibitor for UC (Xeljanz) |
312.37 |
1.5 |
88.9 |
Water-friendly logP; carries boxed malignancy & cardiovascular warnings (ORAL Surveillance). |
3. The Mechanism Smoking Gun: A -79% Human Caffeine Crash Unmasks AhR Behind the miR-124 Curtain
Imagine buying a luxury electric sedan whose glossy brochure explains every detail of its whisper-quiet battery pack—until an independent mechanic opens the hood and finds a roaring diesel engine doing 90% of the work, while the electric battery gauge reads 0.00013 volts.
That is the exact scientific scandal Scorpion Capital just dragged into daylight—and the single strongest proof doesn’t come from Scorpion’s paid labs. It comes straight from Abivax’s own clinical trial registry.
Proof #1: Abivax’s Own Human CYP1A2 Trial (NCT05121714)
In 2022–2023, Abivax ran a Phase 1 drug-drug interaction (DDI) study (NCT05121714, N = 59) giving healthy volunteers a probe dose of caffeine—the gold-standard clinical test substrate for the liver enzyme CYP1A2. Under FDA clinical pharmacology rules, CYP1A2 is almost exclusively turned on in humans by one master transcription switch: the Aryl Hydrocarbon Receptor (AhR).
Here is what NCT05121714 actually posted on ClinicalTrials.gov:
- Day 1 (Caffeine alone): Mean caffeine
AUC=7,420 ± 4,760 h·ng/mL(Cmax = 931 ± 285 ng/mL). - Day 4 (Caffeine +
ABX464): Mean caffeineAUCfell to5,460 ± 2,740 h·ng/mL(-26.4%). - Day 17 (Caffeine + Steady-State
ABX464): Mean caffeineAUCcollapsed to1,560 ± 719 h·ng/mL(Cmax = 564 ± 242 ng/mL)—a-79.0%crash in systemic caffeine exposure!
Under FDA DDI guidance, a ≥80% drop in probe substrate AUC classifies a drug as a strong CYP1A2 inducer. At -79.0%, obefazimod behaves in living humans right on the knife-edge of a strong CYP1A2 inducer—meaning it is unmistakably firing up AhR across human livers in vivo at clinical doses! (Meanwhile, Part B of that very same study confirmed ABX464 parent exposure rose ~32% from Day 1 to Day 11: AUC 55.2 → 73.0 h·ng/mL, Cmax 23.2 → 33.4 ng/mL.)
Proof #2: Every Independent Dataset Points Straight to AhR—Including Abivax’s Own RNA-Seq
- Abivax’s own human CD4 T-cell transcriptome (
GSE116073): Scorpion’s bioinformatics reanalysis (Pages 204–205) showsCYP1A1—the canonicalAhRtarget gene—was the#1most upregulated gene out of40,475genes (+9.93×in uninfected T cells,+14.39×in HIV-infected T cells; competitivep = 1.6 × 10⁻⁴), accompanied by coordinated activation of all six canonicalAhRbattery genes (CYP1A1,CYP1B1,CYP1A2,TIPARP,AHRR,NQO1). - Third-party biotech validation (
Equillium Inc., May 27, 2026 webinar): Independently testing its own oralAhRagonist (EQ504) head-to-head againstABX464, Equillium docked obefazimod directly into the solved humanAhR PAS-Bcrystal pocket (PDB 7ZUB), measured a directAhRreporterEC50of404.2 nM(92%maximal receptor activation at5 µM, the exact concentration Abivax uses in most cellular papers—whileABX464-N-Glushowed undetectableAhRactivity), and proved that adding the selectiveAhRantagonistGNF-351completely erasedABX464’s7.7×RT-PCR"miR-124"signal down to0.85×! - Abivax’s own 2017 colitis paper (Chebli et al., Sci Rep 2017): Long before pivoting away from HIV, Abivax co-founder Jamal Tazi’s lab proved
ABX464dampened gut inflammation by triggeringIL-22—the textbook mucosal-healing cytokine driven byAhRactivation!
The miR-124 Credibility Collapse (and Scorpion’s Lab Tests)
Why did Abivax insist for a decade that obefazimod works strictly by binding the Cap-Binding Complex (CBC) to splice out mature miR-124, rather than acknowledging AhR? Look at the glaring contradictions embedded across Abivax’s own records:
- Zero Copies vs.
8,517Copies in the Same Program: In Phase 2b maintenance (NCT04023396), absolute droplet digital PCR (ddPCR, which directly counts RNA molecules without normalization ratios) recorded whole-bloodmiR-124at0.00000 ± 0.00000 copies/cellat baseline, rising to0.00011at Week 24 and0.00013 copies/cellat Week 48 (1molecule per~7,692blood cells!). Rectal biopsies moved from0.00065to0.00255 copies/cell. Yet in Phase 2b induction (NCT03760003), the exact same label—unit: copies number/cell—lists Week 8 bloodmiR-124as4,598(25 mg),8,517(50 mg), and13,753(100 mg), values roughly57 to 69 million timeslarger, identical to themiR-16-normalized fold values published in Apolit et al. 2023! - Retractions & Lab Artifacts: In
2023, Journal of Cell Biology issued a rare “Retract and Replace” notice (PMC10482727) on a foundational RNA-binding paper from Tazi’s lab after outside scientists repeatedly flagged defective constructs and unreachable co-authors. In2024, Frontiers in Oncology retracted the ChinesemiR-124 → STAT3cancer paper (Lin et al.,10.3389/fonc.2024.1506202) that Abivax’s 2023 corporate decks had cited directly on its main mechanism slide. And Scorpion’s contracted lab studies showed that in Campos et al. 2015 (PMC4422473)—the lone UV cross-linking proof ofCBCbinding and the"safe below 60 µM"cytotoxicity claim—ABX464precipitated above0.16 µM(A3.pdf) and centrifugation spun84.9%of the drug into the discarded pellet (A4.pdf), leaving only19.4%(~5×lower than plotted) in the test wells! - Why Scorpion’s Own
AhRDiscovery Undercuts Its “Fake Efficacy” Thesis: Ironically, by proving obefazimod activatesAhR(EC50 = 404 nM) and drives theLINC00599host transcript (5–8×), Scorpion demolished its other thesis—that obefazimod is merely a toxic worm-killer (PMID:38134663,PMC13270096) producing colitis illusion via non-specific cell damage. In gastroenterology,AhRactivation (AhR → IL-22 → mucosal barrier repair + Treg balance) is one of the hottest clinically validated pathways in ulcerative colitis (the exact mechanism behind indigo naturalis clinical remissions, FDA-approved topical tapinarof, and Equillium’sEQ504). Abivax may have branded the wrong passenger (miR-124) on the ticket—and concealed a potent systemicAhR/CYP1A2liability—while its drug nevertheless hit a real biological receptor in the colon.
4. The Disclosure Schism: What Vanished Between the Clinical Protocols and SEC Filings
In forensic biotech auditing, one of the fastest ways to spot regulatory vulnerability is comparing what a sponsor quietly tells trial physicians inside 65–115-page clinical protocols versus what it tells Wall Street inside its Form 20-F and $920 million offering prospectus.
| Safety / Pharmacology Topic | Buried in Abivax’s Unredacted 2022 Phase 2b Protocol (Prot_NCT04023396.pdf) & ClinicalTrials.gov |
Public SEC Filings (FY2025 20-F, July 2026 424B5 Prospectus) & Redacted Phase 3 Protocol (Prot_NCT05507203.pdf) |
Forensic & FDA Labeling Implication |
|---|---|---|---|
| Nonclinical Phototoxicity & Sun Precautions | Page 39, §8.2: “Subjects are informed via the ICF that oral administration of ABX464-N-Glu was phototoxic in a nonclinical model at clinically relevant exposure levels.” Mandated hat, sunglasses, SPF ≥15, shade, skin photos, and central dermatologist review. |
0 mentions of "phototox" in both the 1.04M-character FY2025 Form 20-F and July 2026 424B5 prospectus. Section 8.1.2 AESI text visibly redacted on p. 75 of Phase 3 ABTECT-1 protocol! |
Strong presumption of an FDA photosensitive/sun-protection warning on any future label; raises questions about why 64 maintenance skin lesions lack subtype detail. |
Teratogenicity & 6-Month Contraception |
Page 17–18: “ABX464 appears to have teratogenic activity.” Rabbit embryo-fetal NOAEL < 1 mg/kg/day; rat pup survival NOAEL < 15 mg/kg/day. Required highly effective contraception for both WOCBP and male partners + 180 days (6 months) post-study & sperm-donation ban. |
0 mentions of "teratogen", "embryo-fetal", or "contracep" in either the 20-F or 424B5 prospectus (424B5 discloses 2 trial pregnancies). Rivals like AbbVie’s R&D leadership publicly pointed to aggressive contraception rules as a red flag. |
Major commercial drag in UC (a young, childbearing-age patient population!): risks a Boxed Warning or Contraindication in pregnancy + REMS/contraception counseling. |
Strong Human CYP1A2 Induction (AhR) |
NCT05121714 posted -79.0% caffeine AUC drop (7,420 → 1,560 h·ng/mL by Day 17)—virtually meeting FDA’s ≥80% strong inducer cutoff; Phase 1 Prot / ClinicalTrials also flags ART (cobicistat) ~17× N-Glu drop. |
0 mentions of CYP1A2 in FY2025 Form 20-F or 424B5. AhR appears once in 20-F, solely as an external combination pathway Abivax is exploring alongside a4b7, IL-23, and S1P. |
Real drug-drug interaction (DDI) label risk (CYP1A2 substrates like tizanidine, clozapine, theophylline, duloxetine, caffeine) and undeniable proof of systemic AhR engagement. |
| Marmoset Cardiac Fibrosis & Minipig Liver | Page 17 & 39: 2 cases of cardiac fibrosis in 9-month marmosets at high ABX464-N-Glu (attributed to marmoset wasting syndrome); centrilobular liver necrosis/fibrosis in minipigs at ≥10 mg/kg (NOAEL 5 mg/kg), triggering echo, troponin, NT-proBNP, and ICVSC monitoring. |
20-F & 6-K maintenance tables explicitly report 0 cardiac fibrosis events clinically, normal troponin/NT-proBNP, DILIsym negative, and no Hy's Law DILI signals. |
Reassuring clinical safety so far on heart and liver, neutralizing the worst acute organ-failure comparisons to Tavneos (VBDS)—though FDA will audit 9-month nonclinical margins closely. |
| Non-Human Primate Immunosuppression | Page 39: “Immunosuppression has been observed during the 9-month cynomolgus monkey study with ABX464-N-Glu.” Required Grade ≥3 / opportunistic infection AESI reporting. |
Phase 3 ABTECT-Maintenance recorded very low serious/opportunistic infections: 2 (1.0%) placebo, 2 (1.0%) 25 mg, 1 (0.5%) 50 mg (incl. 1 lymph node TB on 25 mg). |
Clinical infection signal appears benign relative to anti-TNF or JAK inhibitors—a genuine bright spot if maintained over multi-year follow-up. |
5. The Clinical & Statistical Audit: Catching Both Sides Spinning the Denominators
5A. Skin Cancer (NMSC), Malignancies, and the Exposure Dilution Game
Scorpion’s sharpest clinical attack centers on skin cancer and systemic malignancies. On June 1/2, 2026 (SEC 6-K), Abivax unveiled dazzling Week 44 ABTECT-Maintenance efficacy—paired with 7 malignancy/dysplasia events in the 50 mg arm (N = 195) versus 1 on 25 mg (N = 193) and 1 on placebo (N = 192):
- Non-Melanoma Skin Cancers (
NMSC):4cases on50 mg(2basal cell carcinoma [BCC],2squamous cell carcinoma [SCC]),1SCCon25 mg, and1BCCon placebo. - Non-NMSC Malignancies / Dysplasia:
1prostate cancer,1breast cancer, and1colonic dysplasia—all3on50 mg(0on25 mg,0on placebo).
Then, just 27 days later on June 29, 2026 (SEC 6-K), Abivax released ABTECT Maintenance Part 2 (open-label 25 mg and 50 mg in induction non-responders and Part 1 relapsers), reporting 4 additional NMSC cases (2 at 25 mg, 2 at 50 mg) and 2 non-NMSC malignancies (50 mg, myeloproliferative neoplasm & prostate cancer), and presenting exposure-adjusted incidence rates (EAIRs) per 100 patient-years (PY).
Notice what happens when we reconstruct the unreported Part 1-only EAIR directly from Abivax’s own June 29 numbers:
- In Phase 3 Maintenance (
Part 1 + Part 2), Abivax reports50 mgNMSC EAIR = 1.37 / 100 PY(based on6events →6 / 0.0137 = ~438.0 PY). - In
Part 2 Only, Abivax reports50 mgNMSC EAIR = 0.69 / 100 PY(2events →2 / 0.0069 = ~289.9 PY). - Subtracting Part 2 from Parts
1 + 2leaves~148.1 PYin randomized, placebo-controlled Part 1—yielding an implied50 mgPart 1NMSCrate of4 / 1.481 = ~2.70 per 100 PY(~4.0×placebo’s0.68 / 100 PYand nearly2×the top of Abivax’s0.70–1.40 / 100 PYUC background benchmark!).
Analysis Cohort (Source: 2026-06-29 6-K & Reconstruction) |
Placebo NMSC (IR / 100 PY) |
25 mg NMSC (IR / 100 PY) |
50 mg NMSC (IR / 100 PY) |
50 mg Non-NMSC (IR / 100 PY) |
Expected UC Background (NMSC / Non-NMSC) |
|---|---|---|---|---|---|
Phase 3 Maint. Part 1 Only (N = 580, Reconstructed) |
1 event (0.68) |
1 event (~0.70) |
4 events (~2.70) (95% CI: 0.74–6.91) |
2–3 events (~1.35–2.03) |
0.70–1.40 / 0.30–0.70 |
Phase 3 Maint. Part 1 + Part 2 (Reported) |
1 event (0.68) |
3 events (1.09) |
6 events (1.37) |
4 events (0.91) |
0.70–1.40 / 0.30–0.70 |
Phase 3 Maint. Part 2 Only (Open-Label, Reported) |
— |
2 events (1.52) |
2 events (0.69) |
2 events (0.69) |
0.70–1.40 / 0.30–0.70 |
Integrated UC Program (Phase 2 + Phase 3, 1,704 PY Active) |
0.46 |
0.53 |
0.64 |
0.64 (0.35 All Active) |
0.70–1.40 / 0.30–0.70 |
xychart-beta
title "Abivax 50 mg NMSC Rate (Events per 100 Patient-Years) Shrinks as Open-Label Cohorts Are Pooled"
x-axis ["Placebo (Part 1+2)", "Part 1 Only (Reconstructed)", "Part 1+2 Combined", "Part 2 Open-Label", "Integrated Ph2+3 (1704 PY)", "UC Background Max"]
y-axis "NMSC Events per 100 Patient-Years" 0 --> 3.0
bar [0.68, 2.70, 1.37, 0.69, 0.64, 1.40]
Here is the honest, two-sided statistical verdict on that skin-cancer signal:
- Where Abivax spun the numbers: Pooling lower-rate open-label cohorts pushed
50 mgNMSCfrom~2.70(Part 1) down to1.37(Part 1+2) and0.64(Integrated). Even after pooling Part 1+2,50 mgnon-NMSC malignancy (0.91 / 100 PY) still exceeded Abivax’s own0.30–0.70UC benchmark. Worse, Abivax revealed68skin lesions in its July 2026 deck (4NMSCs +64lumped under “Skin Lesions, All types” without disclosing how many were photosensitivity reactions), and Scorpion’sA2.pdfconfirmed parentABX464absorbs deep-penetrating UVA at349 nm(ε = 8,400 M⁻¹cm⁻¹)—8.4×above the1,000 M⁻¹cm⁻¹ICH S10photosafety screening gate—and investigators judged4active-arm NMSCs as possibly or probably drug-related (2Part 1 per Abivax’s remaining cases,2Part 2 per Scorpion’s attribution log). - Where Scorpion spun the numbers: Headlining
"16 of 17 malignancies occurred on active drug"without spotlighting that over85%of total exposure patient-years were on active drug (>1,704active PY vs.~217placebo PY in Integrated UC) is textbook missing-denominator math. Likewise, an observed count of4events in148patient-years carries an exact Poisson95% confidence intervalspanning0.74 to 6.91per100 PY—meaning random Poisson noise alone still overlaps the0.70–1.40UC background! Furthermore, Abivax’sNMSCpatients averaged age62versus42trial-wide and carried prior thiopurine, biologic, and skin-cancer histories, andA2.pdfadmits in Footnote 2 that Scorpion measured UV absorbance—Step 1 screening only, never reactive oxygen species (ROS) generation or mouse photocarcinogenesis!
5B. ABTECT-1 vs. ABTECT-2, 41 Chinese Sites, and the 0/36 North America Cliff
Scorpion’s boldest Phase 3 attack alleges that Abivax relied on an undisclosed Chinese “black hole” (41 Chinese sites in ABTECT-1, including The 900th Hospital of the PLA Joint Logistics Support Force and General Hospital of Eastern Theater Command per ClinicalTrials.gov NCT05507203), plus 15–16 Ukrainian sites, to manufacture efficacy while North America (N = 122, 10% of patients) rested on an anomalous 0/36 placebo remission rate.
Does Phase 3 efficacy actually collapse without China? Let’s check the exact posted primary-source records across ABTECT-1 (NCT05507203), China-free ABTECT-2 (NCT05507216 — 0 China sites!), ABTECT-Maintenance (2026-06-02 6-K), and the North America ECCO Feb 2026 poster:
Trial / Cohort (Primary Source) |
China Sites? | Placebo Remission (n/N, %) |
25 mg Remission (n/N, %) |
50 mg Remission (n/N, %) |
Endoscopic Improvement (50 mg vs. Pbo) |
Forensic Diagnostic |
|---|---|---|---|---|---|---|
ABTECT-1 Induction Wk 8 (NCT05507203) |
Yes (41 sites, incl. 2 PLA) |
4/158 (2.5%) |
38/160 (23.8%, p = 7.5 × 10⁻⁹) |
69/318 (21.7%, p = 1.9 × 10⁻⁹) |
33.3% vs. 5.7% (HEMI: 23.0% vs 3.2%) |
Ultra-low 2.5% placebo; flat/inverted dose response (25 mg > 50 mg). |
ABTECT-2 Induction Wk 8 (NCT05507216) |
NO (0 China sites; Japan 49, US 46) |
10/159 (6.3%) |
18/159 (11.3%, p = 0.165 FAIL) |
63/318 (19.8%, p = 7.2 × 10⁻⁵ PASS) |
35.5% vs. 10.1% (HEMI: 23.9% vs 7.5%) |
Refutes “China-only hoax” for 50 mg, EXPOSES 25 mg collapse (p = 0.0049 vs ABTECT-1!). |
ABTECT-Maintenance Part 1 Wk 44 (2026-06-02 6-K) |
20 sites (4.8% of 413 sites) |
20/192 (10.4%) |
98/193 (50.8%, ∆39.3%, p < 0.0001) |
100/195 (51.3%, ∆40.3%, p < 0.0001) |
64.1% vs. 12.5% (MES=0: 47.7% vs 9.9%) |
Massive centrally read mucosal & histologic (HEMI 57.4% vs 10.4%) durability. |
North America Wk 8 Subgroup (ECCO Feb 2026, N = 122) |
NO (US/Canada only) | 0/36 (0.0%) |
3/24 (12.5%, Fisher p = 0.059) |
8/62 (12.9%, Fisher p = 0.025) |
Objective Endoscopic Improv: p = 0.08 (50 mg), p = 0.15 (25 mg) — FAIL |
Ultra-fragile: 1 placebo remitter (1/36) pushes 50 mg to p = 0.148 & 25 mg to p = 0.292! |
xychart-beta
title "Week 8 Clinical Remission (%) Across ABTECT-1, China-Free ABTECT-2, and North America"
x-axis ["ABTECT-1 50mg", "ABTECT-1 25mg", "ABTECT-1 Pbo", "ABTECT-2 50mg", "ABTECT-2 25mg", "ABTECT-2 Pbo", "NA 50mg", "NA 25mg", "NA Pbo"]
y-axis "Clinical Remission (%)" 0 --> 28
bar [21.7, 23.8, 2.5, 19.8, 11.3, 6.3, 12.9, 12.5, 0.0]
Look closely at what that side-by-side table proves:
- Scorpion is wrong that Phase 3 efficacy was fabricated in China:
ABTECT-2had ZERO sites in China (its largest footprints were Japan at49sites and the U.S. at46), yet50 mgstill beat placebo19.8%to6.3%(p = 7.23 × 10⁻⁵), more than tripled endoscopic improvement (35.5%vs.10.1%), and tripled HEMI (23.9%vs.7.5%). Likewise, Europe’s China-free Phase 2b 96-week open-label extension (PMID:40417999) saw75.6%(164/217) retention and52.5%non-responder-imputed clinical remission at two years—impossible for a pure gut poison. - Yet Scorpion uncovered two genuine FDA regulatory grenades: First,
25 mgcompletely failed induction in China-freeABTECT-2(11.3%vs.6.3%,p = 0.165)—a statistically significant geographic crash versusABTECT-1’s25 mg(23.8%,p = 0.0049), strongly suggesting25 mginduction could face FDA rejection or non-approval even if50 mgpasses! Second, in North America (N = 122),50 mgremission dropped to12.9%, objective endoscopic improvement missed (p = 0.08), and significance rested entirely on0/36placebo remitters (1single placebo remission breaksp = 0.025 → 0.148). Combine that with 2 military PLA hospitals inABTECT-1,15–16sites in Ukraine, and Phase 2b dropping the failed Robarts Histopathology Index (50 mg p = 0.187/0.200) in favor of Geboes for Phase 3, and FDA BIMO site inspections will be intense.
6. Rival Causes, Graveyard Precedents, Balance Sheet, & Auditing Scorpion Itself
Why do we see strong 50 mg endoscopic efficacy alongside a -79% human CYP1A2 crash, ~2.70 Part 1 NMSC rate, and a 25 mg China/non-China split? Testing all four causal hypotheses prevents falling for single-cause storytelling:
| Causal Hypothesis | Mechanism Tested | Evidence That Fits | Where This Explanation Breaks | Verdict |
|---|---|---|---|---|
1. Direct Drug Effect (AhR Agonist + Parent Lipophilicity) |
ABX464 activates AhR (IL-22, CYP1A1/2) for colitis healing while parent lipophilicity/UVA absorption/chronic AhR creates photosensitivity & NMSC promotion risk. |
Explains -79.0% caffeine AUC, ABTECT-2 50 mg endoscopic healing without China, and dose-ordered 50 mg NMSC/malignancy skew. |
Does not alone explain why 25 mg performed better in ABTECT-1 (23.8%) than ABTECT-2 (11.3%) or why ABTECT-1 placebo was 2.5%. |
Primary Mechanistic Driver (real AhR efficacy + real safety trade-off). |
2. Reverse / Selection Bias (Refractory UC Population) |
Older (age 62), heavily pre-treated UC patients with prior thiopurine/biologic/skin-cancer history naturally develop more NMSCs & low 4.4% placebo rates. |
45%–49% of ABTECT subjects failed prior advanced therapies (AT-IR); NMSC patients were ~20 years older than trial mean (62 vs. 42). |
Randomization balances age & prior thiopurine across 50 mg, 25 mg, and placebo in Part 1—cannot explain a 4× Part 1 50 mg vs. placebo NMSC gap alone. |
Partial Contributor (explains baseline risk, not 50 mg vs. placebo split). |
3. Confounding Third Variable (Geographic Site Mix & Pooling) |
Country mix (41 China sites in ABTECT-1 vs 0 in ABTECT-2; 38% E. Europe) + open-label cohort pooling jointly drove 25 mg divergence & low pooled NMSC rates. |
Directly explains why ABTECT-1 (2.5% Pbo; 23.8% 25 mg) diverged from ABTECT-2 (6.3% Pbo; 11.3% 25 mg) and why Part 1 2.70 shrank to 0.64. |
Cannot explain ABTECT-2 50 mg endoscopic improvement (35.5% vs 10.1%) or ABTECT-Maintenance (51.3% vs 10.4%) across global sites. |
Major Methodological Confounder (drives 25 mg & Part 1 pooling skew). |
4. Mutual Feedback Loop (EHR + UVA + Immunosuppression) |
Enterohepatic recycling (EHR) sustains tissue parent exposure (t1/2 ~4d), while sunlight UVA excites skin-partitioned parent/N-Glu to damage DNA & AhR promotes existing clones. |
Integrates EHR PK (10–12h shoulder, cobicistat 17× drop), 349 nm UVA absorbance, and older patients carrying occult sun-damaged skin lesions. |
Still hypothetical in humans over 44 weeks given small absolute NMSC counts (6 at 50 mg vs 1 Pbo, wide Poisson CIs). |
Plausible Long-Term Hazard (requires FDA 2-year carcinogenicity & Tier-2 photosafety audit). |
Who Loses, Big Pharma Graveyards, and Abivax’s $1.33 Billion Fortress
- Who bears the hidden costs? First, young UC patients of childbearing potential, if a
180-dayteratogenicity contraception burden turns an “easy oral pill” into a restricted therapy. Second, patients on narrow-therapeutic-indexCYP1A2drugs (tizanidine, clozapine, theophylline), unless strongDDImonitoring is added. Third, July 2026$125/ADSequity buyers (-41.4%) and M&A hopefuls: Big Pharma has fresh scar tissue from Amgen’s$4 billionChemoCentryx (avacopan/Tavneos) acquisition—which in April 2026 drew an FDA CDER withdrawal proposal over fatal drug-induced liver injury and post-lock trial manipulation, followed by a June 2026 NEJM retraction—and Pfizer’s$6.7 billionArena (etrasimod/Velsipity) impairment. - Why Abivax is NOT a fragile cash-starved microcap: Per Abivax’s Sept 21, 2026 H1
6-K, Abivax held€402.4Min cash as of June 30 plus€767.1M($874.1M) net from July’s offering =€1,169.5 million(~$1.33 billion, or~$15.31/share), with just€1.2 millionin debt/leases,$0royalty overhang (repurchased May 7 for$90M), and an H1 2026 operating cash burn of€102.5M(~€205M/yr) providing cash runway into Q4 2029. Moreover, Abivax already completed a positive July 30, 2026 Pre-NDA meeting with FDA and appointed commercial-scale medical/CMC leadership (Chris Rabbat, Ph.D.&Tim Kelly).
Auditing the Short Seller: Scorpion Capital’s Conflicts & Track Record
To follow “Panning for Gold” discipline, we audit Scorpion Capital (scorpioncapital.com) with the same skepticism we apply to Abivax:
| Biotech Short Campaign | Scorpion Core Allegations & Threat | Actual Regulatory & Market Outcome | Forensic Takeaway for ABVX |
|---|---|---|---|
Allakos (ALLK) (2021) |
Attacked eosinophilic GI mechanism & trial claims; steep valuation collapse. | Clinical trials failed; stock collapsed -99%. |
Shows Scorpion can spot weak GI mechanisms early when pivotal trials fail. |
Harmony Biosciences (HRMY) (Mar 2023) |
366-page report attacking WAKIX safety/efficacy; filed FDA Citizen Petition seeking market withdrawal. |
FDA formally DENIED Scorpion’s Citizen Petition on June 25, 2024 and expanded WAKIX to pediatric narcolepsy! |
Proves commissioned short-seller reports do not automatically sway FDA reviewers when controlled clinical data hold up. |
TransMedics (TMDX) (Jan 2025) |
Alleged billing/operational misconduct; slapped $0 price target (-13% initial drop). |
Retained counsel, refuted allegations, continued commercial organ-care rollout. | Extreme rhetorical framing ($0 / “fraud for the ages”) often outruns actual regulatory collapse. |
Soleno Therapeutics (SLNO) (Aug 2025) |
Attacked FDA-approved VYKAT XR safety & trial robustness (-8% to -12% initial drop). |
VYKAT XR stayed approved, though post-launch safety volatility triggered securities lawsuits into 2026. |
Even when FDA approves a drug, unaddressed safety/label overhang can punish equity multiples. |
Abivax (ABVX) (Nov 20, 2026) |
297-pp. report + 4 paid lab studies; Page 2 disclosure states intent to “begin covering a substantial portion of our short position” upon publication, and relies on paid anonymous ex-employees. | ABVX fell -9.24% to $73.20 ($6.37B cap); Abivax reaffirmed Q4 2026 NDA filing. |
Treat Scorpion’s primary-source catches (NCT05121714, 20-F vs protocol gaps, ABTECT-2 25 mg) seriously while discounting “DDT / $0 fraud” hyperbole and trading-desk cover timing. |
7. Grey Thinking & “The Forensic Verdict: What We Actually Think”
Reject both binary extremes—neither “obefazimod is safe, clean, selective miR-124 magic heading for a $15B buyout” nor “obefazimod is DDT with zero clinical efficacy worth $0.” Everything hinges on four concrete conditional If-Clauses:
- If Abivax’s confidential NDA submission contains clean 2-year rodent carcinogenicity (
Tg.rasH2/ rat) and Tier-2 functional photosafety (ROS/in vivo) data that clear ICH S10/M3 thresholds, then Scorpion’s349 nmUV-absorbance and2.70Part 1 NMSC claims narrow into manageable sun-protection labeling rather than an approval blocker. - If FDA
BIMOaudits ofABTECT-1’s41Chinese sites (including2PLA hospitals) or15–16Ukrainian sites uncover protocol or source-document deficiencies, then the collapse of25 mgin China-freeABTECT-2(p = 0.165) and fragile0/36North American placebo rate could force a Complete Response Letter (CRL), restricting approval or requiring Western confirmatory data. - If FDA evaluates
50 mgalone on the strength of China-freeABTECT-2(p = 7.23 × 10⁻⁵) +ABTECT-Maintenance(∆40.3%,p < 0.0001) + 7-yearStudy 108safety, then clinical efficacy strongly clears the statutory substantial-evidence bar—even if25 mginduction is rejected andmiR-124marketing claims are stripped from the label in favor ofAhR/ strongCYP1A2induction warnings. - If any label carries embryo-fetal toxicity /
6-monthcontraception, photosensitivity / NMSC monitoring, and strongCYP1A2DDI language, then Big Pharma acquirers post-Tavneos($4B) and post-Velsipity($6.7B) will likely demand steep discounts or commercial proof rather than paying€15Bupfront.
Probability-Weighted Scenarios (12–18 Month Horizon from $73.20)
Scenario (Probability) |
Key Triggering Events | Implied Regulatory & Commercial Outcome | Indicative Valuation Range |
|---|---|---|---|
Base Case (50%): Approved With Label Baggage / Review Friction |
Q4 2026 NDA accepted; FDA BIMO flags minor/isolated site issues or delays PDUFA (+3 mo) / rejects 25 mg induction; 50 mg approved on ABTECT-2 + Maintenance durability. |
Approved UC pill (50 mg), but heavy label warnings (teratogenicity/contraception, sun protection/NMSC surveillance, CYP1A2 DDI) & miR-124 skepticism cap market share (vs. TL1A / IL-23 / JAK) and cool $15B+ M&A. |
$40 – $85 / share |
Bear Case (30%): Major CRL, AdCom Rebuke, or Safety Hold |
BIMO invalidates key Chinese/E. European induction sites, confidential 2-year carcinogenicity/photosafety shows a tumor signal, or AdCom balks at 0/36 NA placebo + 50 mg malignancy EAIR (0.91 non-NMSC / 2.70 Part 1 NMSC). |
Multi-year regulatory reset or confirmatory Western trial; stock reprices toward ~$15.31/share cash cushion (€1.17B) plus option value on Crohn’s (ENHANCE-CD, mid-2027). |
$15 – $35 / share |
Bull Case (20%): Clean First-Cycle 50 mg (or 25/50 mg) Approval & Buyout |
Confidential NDA holds clean 2-year carcinogenicity & Tier-2 phototoxicity studies; Western/Japanese ABTECT-2 + Maintenance pass BIMO cleanly; FDA imposes routine IBD class warnings only. |
Clean oral UC label (51.3% Wk 44 remission, MES=0 47.7%, low 0.5% serious infections) re-ignites strategic M&A bidding. |
$100 – $145+ / share |
The Forensic Verdict: What We Actually Think
Neither Abivax nor Scorpion Capital owns the whole truth—and our evidence audit awards the decisive empirical points differently across science vs. clinical reality:
- On the pharmacology and disclosure critique, Scorpion holds the stronger hand: Abivax’s own
-79.0%caffeineAUCcrash (NCT05121714) andCYP1A1#1RNA-seq rank (GSE116073) leave zero credible doubt that obefazimod acts clinically as a systemicAhRagonist whilemiR-124(0.00013copies/cell on ddPCR) has been wildly oversold. Equally troubling, leaving phototoxicity, teratogenicity (6-monthcontraception), and-79%CYP1A2induction completely out of theFY2025 Form 20-Fwhile redacting§8.1.2of the Phase 3 protocol was an avoidable governance blunder that invites FDA and investor distrust. - On acute poisoning and clinical efficacy, Abivax’s clinical datasets decisively refute Scorpion’s most apocalyptic claims: A drug that produces
19.8%vs.6.3%remission (p = 7.23 × 10⁻⁵) and35.5%vs.10.1%endoscopic improvement in China-freeABTECT-2,51.3%vs.10.4%Week 44 remission (p < 0.0001), and75.6%2-year retention in Europe is not “DDT causing colitis illusion via gut poisoning.” It is an activeAhR/IL-22modulator that cleared rapidly enough viaABX464-N-glucuronide(t1/2 ~4 days) to generate real mucosal healing alongside benign infection/liver/cardiac clinical rates. - Actionable Bottom-Line Stance: At
$73.20($6.37 billion), the pre-report narrative of a friction-free approval and imminent€15 billionEli Lilly buyout is broken. Yet shortingABVXdown here expecting a$0Capricor-style data-fabrication bust is equally dangerous given~$1.33 billion($15.31/sh) in debt-free cash and China-independent50 mgefficacy. Until Abivax publicly discloses (a) its completed 2-year carcinogenicity and Tier-2 photosafety results, (b) the full pathology breakdown of the64maintenance skin lesions, (c) sensitivity analyses ofABTECT-1/2excluding China/PLAsites, and (d) clear labeling expectations for teratogenicity andCYP1A2DDIs,ABVXis an asymmetric “show-me” regulatory execution story—not a risk-free buyout bargain.
8. Primary Source Verification Ledger
Every external primary artifact cited in this investigation was programmatically verified for live accessibility (HTTP 200):
- Scorpion Capital Reports & Lab Manuscripts (Nov 20, 2026): Main Report (
ABVX.pdf, 297 pp.) | A1:logP/pKa& Meyer-Overton | A2: UVA Phototoxicity (349 nm) | A3:DLS& Autofluorescence | A4:LC-MS/MSCampos Re-analysis | Scorpion Capital Portal - SEC EDGAR Filings (Abivax SA,
CIK 0001956827): FY2025 Form 20-F (abvx-20251231.htm) | June 2, 2026 Form 6-K (ABTECT-MaintPart 1) | June 29, 2026 Form 6-K (ABTECT-MaintPart 2 &EAIRTables) | July 2, 2026424B5Prospectus ($920MOffering /87.05MShares) | Sept 21, 2026 Form 6-K (H1 2026 Cash & Leadership) - ClinicalTrials.gov Records & Protocols:
ABTECT-1(NCT05507203) |ABTECT-2(NCT05507216) |ABTECT-Maintenance(NCT05535946) |CYP1A2DDI (NCT05121714) | Phase 2b Induction (NCT03760003) | Phase 2b Maintenance (NCT04023396) | Unredacted 2022 Phase 2b Protocol (Prot_000.pdf) | Redacted 2024ABTECT-1Protocol (Prot_000.pdf) - PubChem, GEO, Retractions & FDA Precedents: PubChem: Obefazimod (
CID 49846599) | PubChem:ABX464-N-glucuronide(CID 122499702) | NCBIGEO GSE116073| Campos et al. 2015 (PMC4422473) | Chebli et al. 2017IL-22(PMC5501810) | Vautrin et al. 2019 (PMC6349857) | Apolit et al. 2023 (PMC10132720) | JCB 2023 Tazi Retract & Replace (PMC10482727) | 2026 Sheep/Nematode Biotransformation (PMC13270096) | Frontiers 2024 Lin Retraction | FDA CDER Proposal to WithdrawTavneos(Apr 27, 2026)