Registered forecast

FDA first-line approval of sunvozertinib precedes the NCCN first-line listing

This is the record as published. It was written down before anyone knew the answer, and it has not been touched since; the parts held only in the ledger are named below.

Guideline listing (NCCN, ESMO): offered on judgment, not a track record — calibration not yet demonstrated.

The estimate

The number, and what it has to beat

A number on its own is unfalsifiable. These are the alternatives it was registered against, each with the reasoning that produced it, frozen at the same cutoff.

Resolution

The rule that settles the forecast

The rule is written so a reader with no view on the drug can apply it.

Two dates are read, each under its sibling's rule as amended by workspace/reviews/2026-09-05-ff-001-ff-002-falsification.md §3d, and compared at day precision, the finest either record carries. D_FDA: the submission_status_date (YYYYMMDD) of the earliest submission in the openFDA Drugs@FDA record for application number NDA219839 — queried as https://api.fda.gov/drug/drugsfda.json?search=application_number:%22NDA219839%22 — whose submission_type is SUPPL (or, if the live literal differs, is not ORIG), whose submission_status is AP, and whose approved labeling, read from Section 1 (INDICATIONS AND USAGE) of the label PDF linked in that submission's application_docs (type Label) or from the Drugs@FDA web record at https://www.accessdata.fda.gov/scripts/cder/daf/ , covers the first-line / initial-treatment setting for EGFR exon 20 insertion mutation-positive locally advanced or metastatic NSCLC. If openFDA has not refreshed by the reading date, the Action Date on the Drugs@FDA web record governs; if the two disagree, the date on the approval letter governs. A first-line approval under a new NDA counts, read from that NDA's ORIG row. Exclusions as in FF-002: later-line, maintenance, non-NSCLC, or conversion of the existing accelerated approval do not count; a narrower first-line population does. D_NCCN: the version date printed on the earliest version of the NCCN NSCLC guideline, dated on or before 2027-07-31, that names sunvozertinib or ZEGFROVY in the initial-systemic-therapy recommendation set for EGFR exon 20 insertion mutation-positive metastatic non-squamous NSCLC, at any category (1, 2A, 2B or 3), whether or not marked preferred, on the algorithm page for initial systemic therapy or in the corresponding systemic-therapy table; footnotes attached to that first-line options block count, footnotes on testing, discussion text and 'Principles of' pages do not; a listing confined to subsequent-line therapy does not count. The version date is the date NCCN prints beside the version number on the guideline's cover; if a version prints a number but its date appears only in its Updates section, that date governs. NCCN distributes only its current version, so the adjudicator reads captured copies of every NSCLC version issued inside the window (the repository is to capture each with its date and digest as it appears — an operational task, not a rule change). Bracket rule when the earliest listing version was not captured: let L be the version date of the latest version known not to carry the listing and U the version date of the earliest version known to carry it; if the exact D_NCCN cannot be established, resolve YES if D_FDA is on or before L, NO if D_FDA is on or after U, and VOID with the reason 'earliest listing version date not recoverable' if D_FDA falls strictly between L and U. Resolution: YES if and only if both D_FDA and D_NCCN exist on or before 2027-07-31 and D_FDA is strictly earlier than D_NCCN. NO if D_NCCN is earlier than or equal to D_FDA — a same-day tie resolves NO, is recorded as a tie, and is not broken by appeal to the hypothesised mechanism — or if either event has not occurred by 2027-07-31. NCCN versions dated after 2027-07-31 and FDA actions dated after 2027-07-31 are ignored. Adjudication floor: read no earlier than 2027-08-31 and no later than 2027-10-31, at the same sitting as FF-001 and FF-002. Access: openFDA and Drugs@FDA are public and unauthenticated; NCCN requires free registration. No clinical judgement about the drug is needed. VOID, with the reason recorded, if a first-line recommendation set for exon 20 insertion disease is distinguishable in no version issued inside the window, or if neither openFDA nor the Drugs@FDA web record publishes per-supplement action dates at the reading date; VOID as ill-posed if the earliest listing version is dated on or before the availability cutoff.

Source read: FDA action date from openFDA Drugs@FDA for NDA 219839; NCCN listing date from the version date of the earliest captured NCCN NSCLC version carrying the listing; compared at day precision

No independent adjudicator has been named on this record yet. Until one is, resolution is a stated rule rather than a committed arrangement.

What this record will show after 2027-07-31

The finding

Occurred, did not occur, or ambiguous — as the rule above reads it.

Who read it, and what

An adjudicator with no stake in the forecast, reading the source named above, on a date recorded here.

The number against its baselines

45% beside 3 baselines, each scored by the same rule, so a reader can see which of them the record should have listened to.

Frozen covariatesThe facts about the evidence and the authority's process that were frozen with the forecast at its cutoff

The state of the record at the cutoff

Recorded as they stood on 2026-09-22, including the 2 fields whose honest value was unknown. A covariate table with no unknowns in it has usually been tidied.

review pathway

FDA leg: Breakthrough Therapy Designation for the first-line setting, reported only by sponsor releases (SRC-001 and SRC-006: 'in this setting'; SRC-007: 'for this treatment-naive patient population'); the original NDA 219839 was reviewed under PRIORITY (SRC-003, ORIG-1, AP 20250702); the supplement's designation, priority or standard, is in no archived file. A separate April 2024 designation in the platinum-pretreated setting is reported secondarily (SRC-004). NCCN leg: no statutory clock; the panel's continuous-update process, at a cadence the archive bounds at six or more versions in the first seven months of 2026 (SRC-006). The forecast object is the ordering of the two.

submission type

FDA leg: supplemental New Drug Application, efficacy supplement for a new line of therapy, per sponsor releases only (submitted, SRC-006; accepted, SRC-001); no supplement row in openFDA as of 2026-09-02 (SRC-003). NCCN leg: none — NCCN does not publish manufacturer submission requests.

evidence class

randomized, open-label, multinational phase 3 (WU-KONG28, NCT05668988) versus carboplatin-pemetrexed: 'phase 3, international' per the NEJM abstract (SRC-002); 'open-label' per the sponsor release (SRC-005) and The ASCO Post (SRC-008); enrolled December 2022 to May 2025 (SRC-008); 15 countries per SRC-005 and 16 per SRC-007; presented as ASCO late-breaking abstract LBA8500 and published in NEJM on 2026-05-29 (SRC-002, SRC-005).

pivotal n

324 randomized 1:1 (163 sunvozertinib, 161 chemotherapy) per the NEJM abstract (SRC-002)

endpoint type

primary: progression-free survival by blinded independent central review, 10.3 vs 7.5 months, HR 0.65 (95% CI 0.50 to 0.85), P<0.001 (SRC-002; the sponsor release gives P=0.0008, SRC-005); 12-month PFS 46.1% vs 26.7%; objective response 58.9% vs 31.1%; overall survival immature at 38.9% maturity (SRC-002) with 101 of 112 (90.2%) chemotherapy patients crossed over (SRC-008); grade 3 or higher adverse events 75.5% vs 56.7% (SRC-002); serious adverse events 45.4% vs 26.7% (SRC-008). Data cutoff read as 2026-01-16 from 'As of January 16, 2026, 324 patients were randomized' (SRC-005), which the release states as a randomization-count date; the NEJM abstract gives no cutoff date.

approved competitors in class

No archived file names an approved first-line regimen for this population, and no archived file states the current NCCN first-line recommendation set for it. What the archive supports: the sponsor's 'no oral targeted therapies have been approved for the first-line treatment' and 'first-line treatment ... largely relies on chemotherapy-based regimens' (SRC-007); the Cancer Discovery news item names amivantamab with lazertinib for atypical EGFR mutations and nothing for exon 20 insertions first-line (SRC-009). FF-001 records amivantamab with carboplatin/pemetrexed as FDA-approved first-line and widely reported as the NCCN category 1 preferred option, from secondary sources archived nowhere in this repo. Load-bearing on both legs if right — a preferred regimen already in place is the main argument for a slower listing (FF-001) and for a standard review clock (FF-002) — and unverified by the archive.

months since last guideline revision

unknown in months. The archive fixes a lower bound on version count, not a date: NCCN NSCLC Version 6.2026 existed by July 2026 (SRC-006 footnote 1, 'Accessed July 2026'; repeated in SRC-001 dated 2026-09-01). Six or more versions in seven months is a cadence of roughly one a month; the date of the latest version at the cutoff, and whether any version after 6.2026 exists, are not in the archive.

panel cycle position

unknown — see months_since_last_guideline_revision. At the cadence the archive implies, ten or more NSCLC versions fall inside the eleven-month window, so procedural opportunity for the NCCN leg after any in-window FDA action is close to certain; the position of the window in the cycle is not known.

sponsor request status

FDA leg: supplemental NDA submitted on or before 2026-07-14 (SRC-006) and accepted on or before 2026-09-01 (SRC-001); receipt date, filing date, designation and goal date undisclosed; press release only. NCCN leg: unknown — NCCN does not publish manufacturer submission requests.

odac referral

none announced in any archived file as of the availability cutoff; the advisory-committee calendar is not archived (absence claim)

sponsor state change

AstraZeneca entered an exclusive global licence with Dizal on 2026-07-14 ($600m upfront, up to $900m in milestones; SRC-006) and announced its completion on 2026-09-01, four days before this cutoff, with a US launch of the second-line indication planned for the fourth quarter of 2026 (SRC-001) — a change in the commercial capacity behind both regulatory and guideline engagement. The application holder of record in openFDA is still DIZAL as of 2026-09-02 (SRC-003).