Forecast registry · FF-001
FF-001 — registration-ready.
A defined transition forecast for an NCCN guideline listing, frozen at a cutoff and scored against reference baselines. The number stays held until the registry admits it.
Registration status
Fully specified. Not yet registered.
Every field on this page is frozen: the proposition, the cutoff, the resolution rule, the reference baselines, and the archived sources. Registration adds two things, and both are out of the drafter's hands: two named adjudicators, and a minted durable identifier. Until that happens, the probability is held off the public surface.
Forecast specification
The fields that define this proposition
These fields make the proposition resolvable and scoreable. The verification hash is computed from the frozen fields on this page; the durable identifier is minted at registration.
nc-a913480a6697 How to verify Reference class: agents with a positive randomized phase 3 in a molecularly defined NSCLC subset appearing in an NCCN NSCLC line-of-therapy recommendation within 12 months of publication. Recent precedent in this guideline (alectinib adjuvant post-ALINA, amivantamab first-line post-PAPILLON, sevabertinib in ERBB2) is consistently fast. Estimate from precedent, not an enumerated count — the enumerated corpus is the object of the external-transition-forecast-v1 study and does not yet exist. Recorded as the honest state of this baseline at the cutoff.
A model that knows only NCCN's revision cadence. The NSCLC guideline is revised several times a year, so the probability that at least one revision opportunity falls inside a 12-month window is approximately 1.0. Having no evidence input, such a model cannot discriminate between pending candidates and defaults to the unconditional per-candidate incorporation rate, which is well below the phase-3-conditioned rate above. The output is the product of opportunity and that unconditional rate, recorded as the model's value at the cutoff.
A thoracic-oncology analyst reading the same public record on 2026-07-31 — NEJM-published positive phase 3, Breakthrough Therapy Designation, first-line sNDA submitted, a $600m upfront global license — would call listing highly likely and treat FDA review timing as the only serious question. Estimated by reconstructing that read, not by surveying analysts.
Resolves YES if any NCCN NSCLC guideline version dated on or before 2027-07-31 names sunvozertinib (or ZEGFROVY) anywhere 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) and whether or not it is marked preferred. The adjudicator reads the NSCLC algorithm page for initial systemic therapy in EGFR exon 20 insertion mutation-positive disease and the corresponding systemic therapy tables, in the latest version dated on or before 2027-07-31. Resolves NO if no such version names it there. Listing confined to subsequent-line or post-progression therapy does NOT count. Access requires free NCCN registration; no clinical judgement about the drug is needed to apply this rule.
Upstream states at freeze
Rung 1–4 states locked at evidence cutoff
Rungs 1–4 record the public evidence state at cutoff. Later evidence belongs to a new registration.
How the transition could happen
Mechanism and contingency
The mechanism states how the transition is expected to occur. The contingency notes state when the proposition stops being well-posed.
Hypothesis, not access — NCCN panel deliberation is closed. Primary path: the US supplemental NDA for the first-line indication, submitted on or before 2026-07-14 and carrying FDA Breakthrough Therapy Designation in this setting, is granted priority review and approved inside the window; NCCN then incorporates the new indication at its normal post-approval cadence, which for this guideline has historically been one to two versions. Secondary path: the panel lists the agent ahead of FDA action on the strength of the NEJM-published randomized phase 3, which it does occasionally for breakthrough-designated agents in molecularly defined subsets. The two paths are not independent — both are driven by the same evidence event — which is why the forecast sits below the ceiling implied by the regulatory path alone.
The proposition would be ill-posed if: NCCN restructures the EGFR exon 20 insertion pages such that 'initial systemic therapy' is no longer a distinguishable recommendation set; NCCN withdraws public version-dated distribution of the NSCLC guideline; or the agent is withdrawn from the US market, in which case no listing can occur and the proposition resolves NO on its plain terms rather than being voided. A brand-name change following the AstraZeneca license does not void it — the rule matches on the INN 'sunvozertinib' as well as the brand. Listing restricted to a narrower first-line population than the full exon20ins group still counts as YES, provided it is a first-line recommendation.
Archived sources
What the cutoff rests on
Six sources, each archived at the cutoff with a SHA-256 digest so the record can be recomputed from what was actually public then.
Dizal press release announcing WU-KONG28 phase 3 results (2026-05-29): PFS 10.3 vs 7.5 months per the release; HR 0.65, p=0.0008; confirms prior US and China accelerated approvals in relapsed/refractory disease
PubMed record for the NEJM primary publication (PMID 42212913, doi 10.1056/NEJMoa2604461, 2026-05-29), 'First-Line Sunvozertinib in NSCLC with EGFR Exon 20 Insertion Mutations'. Archived via PubMed rather than the publisher: nejm.org served a bot-challenge page, not the article.
AstraZeneca press release (2026-07-14) announcing the exclusive global license from Dizal. Load-bearing: the only archived primary source confirming the first-line supplemental NDA to the US FDA and FDA Breakthrough Therapy Designation in this setting. Both facts drive the probability.
Dizal press release: China NMPA accepted the first-line NDA with Priority Review
ASCO Post coverage of the WU-KONG28 first-line results (June 2026) — independent secondary reporting of the same readout
PubMed record for the Cancer Discovery commentary 'Rare EGFRm NSCLC: More First-Line Options Emerge' (PMID 42224429, 2026-07-01) — contemporaneous field read of the first-line landscape in this subset
Every forecast we register
What comes with the specification
Scoring only means something if the forecast is sealed off from hindsight and from your internal context. Four things hold that line, and they apply to every forecast we register.
When it registers
One email when FF-001 is admitted to the registry.
We'll email the named adjudicators, the minted identifier, and the released probability. No other mail.
What this is for
Registered before the outcome, on the public record.
FF-001 is the first record heading for registration under the forecasting program. When it is admitted, the probability goes on the record before the deadline — a dated basis to prepare documentation and set review priorities while the authority's process is still open.
Prefer the format without the specifics? The example brief walks the same anatomy with an illustrative coverage decision.