Arcus Biosciences TIGIT/PD-1 Pipeline — HyperCore KM Decomposition
Arcus Biosciences
HyperCore KM decomposition of the TIGIT + PD-1 + adenosine-axis compound attribution driving the ARC-7 (NSCLC), ARC-10 (PD-1-naïve NSCLC), and ARC-9 (mCRPC) Phase 2/3 inflection points — responder/non-responder slot attribution for the next-trial PD-1-combination label-expansion conversation.
Substrate: publicly disclosed ARC-7 / ARC-10 / ARC-9 OS & PFS datasets · HyperCore 320-classifier stack
TIGIT × PD-1 co-blocking (domvanalimab + zimberelimab) and the adenosine-axis (etrumadenant + quemliclustat) constitute the highest-weight classifier co-attribution in the HyperCore stack, ahead of any single-axis read — auditable, classifier-weighted, keyed to label expansion.
Scope the Arcus Biosciences pilot →Audit-trail parity. The comparable KEYNOTE-789 sample — same HyperCore decomposition, same three-artifact ship set, same narrative shape — is readable end-to-end at /keynote-789-report. Use the KEYNOTE-789 read as the prior sample to walk an Arcus Biosciences reviewer through before scoping the pilot engagement.
Every Arcus Biosciences Engagement Ships Three Artifacts
Each Trial Intelligence Report ships in three formats so the sponsor can read however they read. Mirrors the three-ship model used on every prior engagement so an Arcus Biosciences reviewer can audit any prior deliverable — including the ARC-7 / ARC-10 / ARC-9 OS & PFS reads — in the same shape.
| Artifact | Format | What It Contains |
|---|---|---|
| HTML | 4-page printable | Numbered pages, inline SVG KM chart, 12-row classifier attribution table, decision narrative. Designed for first-pass Tier 1 sponsor review. |
| CSV | Auditable raw data | KM curves (24 monthly rows per arm) plus classifier-attribution table with weights, direction, classifier IDs (I50, N17, E11, A41, D64 — and additional TIGIT/PD-1 adenosine-axis classifiers), and notes. Importable into R, Python, or Excel for re-analysis. |
| Markdown | Version-controlled narrative | Page-by-page decision narrative with editorial hooks, decision-implications section for next-trial label-expansion framing, and a closing paid-pilot CTA. Diff-able across report revisions. |
All three formats ship on the same day. The CSV and markdown are auditable companions to the HTML — reviewers who want to re-derive the chart or check the classifier weights against their own read can do so without leaving the engagement scope.
The CTA routes into the existing Trial Intelligence Report lead form — your request lands in lead_captures, and we respond within one business day with a scoping read on which tier (single-trial brief, comparative brief, or engagement-scope pilot) fits the publicly disclosed dataset.
Scope the Arcus Biosciences pilot →What happens after you click through. You land on /trial-intelligence-report, the existing lead form posts to /api/leads, and the row lands in lead_captures for the audit trail. No NDA required to scope — NDA only if you want Tier 3 with an in-person working session.