Ziltivekimab: ZEUS, HERMES, ATHENA. Three BVCT NO-GOs, already on record.
BVCT locked NO-GO calls on ziltivekimab for ZEUS, HERMES and ATHENA (No. 1524–1526) on . ZEUS missed its primary endpoint on ; HERMES and ATHENA were discontinued for futility on . The dashboard classifies all three readouts as True Negative (TN).
Prospective · called before readout. Three BVCT NO-GOs, already on record. 859 days early. Every readout confirmed: the dashboard classifies all three readouts as True Negative (TN). All locked on one date: . All three Phase 3 calls landed spot-on — from Drug MoA and the publicly registered protocols alone.
Also sent to Novo Nordisk before the readout: CagriSema #1516 (REIMAGINE 4) and monlunabant #1683 (DKD) — slide 4.
859 days early = (BVCT #1524 locked) to (Novo Nordisk: target engagement did not translate into MACE risk reduction versus placebo, HR 0.99, 95% CI 0.88–1.11; primary endpoint not met, per Novo Nordisk’s announcement). #1525 and #1526 locked the same day; HERMES and ATHENA: further development discontinued on a data monitoring committee futility finding, (897 days after the lock). ATHENA study start per ClinicalTrials.gov NCT06200207 — after the lock.
ATHENA · called before the first patient was enrolled
Before the first patient was enrolled, BVCT called NO-GO on ATHENA — from Drug MoA and the protocol alone.
On , BVCT locked No. 1525 on ATHENA while the registry still listed the trial as not yet recruiting. The first participant was enrolled (registry, actual) 8 days later; further development was discontinued after a data monitoring committee futility finding on . Recorded commercial call: FAILURE (commercially insufficient additive clinical benefit to SoC with inferior-to-semaglutide efficacy in KCCQ-CSS and 6MWD). The dashboard classifies the readout as True Negative (TN).
The public record on the lock date, the call and the registry history are set out in the ATHENA case study.
The biomarker pointed up. The mechanism pointed down.
Three readouts, as the mechanism said.
First readout · ZEUS · ASCVD + CKD + inflammation
On , Novo Nordisk announced that ZEUS — a 6,385-patient cardiovascular outcomes trial of ziltivekimab in ASCVD, CKD and inflammation — missed its primary endpoint. Three-point MACE came in at a hazard ratio of 0.99 vs. placebo (95% CI 0.88–1.11). Ziltivekimab engaged its target precisely as designed: free IL-6 fell, hsCRP fell, yet none of it converted into cardiovascular benefit. Novo Nordisk said it will book a non-cash impairment charge in Q3 2026.
Outcome: higher rate of serious infections versus placebo. No difference in all-cause mortality.
BVCT sealed a NO-GO call on — 859 days before the readout — and sent it to Novo Nordisk R&D twice: on , and again on , sixteen weeks before ZEUS reported; the first sending came 547 days before the readout.
Quantified to Novo: “the effect size of ziltivekimab as an add-on [therapy] in MACE would have an HR > 1.1 compared with semaglutide as an add-on [therapy]” · email to Novo Nordisk R&D, .
True Negative (TN) · confirmed · #1524. Target engaged: IL-6 and hsCRP fell — no MACE benefit.
Then HERMES and ATHENA · heart failure + inflammation
Data monitoring committee: stopped for futility on , 897 days after the lock (locked , sent to Novo Nordisk ). True Negative (TN) · #1526 · #1525 · HERMES n=4,874 (registry, actual) · ATHENA n=673 (registry).
Three strata, one verdict. BVCT read the mechanism, not the biomarker.
BVCT identified the biological root cause.
BVCT’s identified mechanistic root cause, sent to Novo Nordisk R&D on , was this: ziltivekimab’s weak efficacy follows from the mixed clinical impact of the IL-6 inhibition mechanism itself, which carries both pro-MACE and anti-MACE clinical impacts that are highly sensitive to the precise clinical setting of ASCVD + CKD + inflammation patients.
Novo Nordisk’s announcement describes precisely that dissociation. Ziltivekimab demonstrated target engagement and inhibition of the IL-6 pathway, as reflected by expected reductions in free IL-6 and high-sensitivity C-reactive protein (Novo Nordisk) — and this did not translate into major adverse cardiovascular events (MACE) risk reduction versus placebo (Novo Nordisk).
The mechanistic rationale behind the ziltivekimab NO-GO was sent to Novo Nordisk before the readout — the IL-6 root cause on . It was generated prior-data-independently: no Novo Nordisk data and no patient-level data from any trial. Nothing in these calls depended on privileged access — only on simulating the mechanism based on public information.
The molecule did what it was designed to do by hitting the target and improving the biomarker. The patient did not benefit. In our view, that biomarker-efficacy discordance is only visible with BVCT’s reliable causal simulations of the underlying mechanisms in proprietary virtual patients, without requiring real patients.
One pipeline. Three assets.
Called before the readout, 2.4 years earlier.
The record · ziltivekimab · CagriSema · monlunabant
Called on Drug MoA and the publicly registered protocols alone. Days = lock to the sponsor’s readout or the discontinuation. All therapeutic areas: 674 of 705 prospective calls confirmed by readout.
Audit each call on the dashboard: data.bioinvestgpt.com. Outcomes: Novo Nordisk announcements and financial reports, dated in the timeline below.