Provenance Record

Clear, well-structured News piece (634 words, within the 400-1200 News range). Sectioned as Overview / What We Know / What We Don't Know / Analysis. Appropriate technical depth on the in vivo vs ex vivo distinction without overstating; explicitly separates Phase 3 results from the earlier Phase 1/2 long-term follow-up. Properly framed as a follow-up to the April preview rather than re-reporting it.

All 3 source snapshots read from disk and verified (all fetched live, status 200, no archive fallback). source-0.html.gz (globenewswire, Intellia press release) confirms verbatim: 87% reduction (p<0.0001) in mean monthly attacks, weeks 5 to 28 efficacy window, 62% attack-free/therapy-free vs 11% placebo, secondary-endpoint table header 'Lonvo-z Arm (N=52) Placebo Arm (N=28)', 89% reduction in on-demand-treatment attacks and 91% in moderate/severe attacks (both p<0.0001), KLKB1 inactivation lowering kallikrein with a single dose, no serious adverse events in lonvo-z arm, rolling BLA initiated in April with FDA, anticipated approval/US launch H1 2027, and the John Leonard quote 'These are the first Phase 3 results to deliver on the much-heralded promise of in vivo CRISPR gene editing.' source-1.html.gz (news-medical.net) confirms the side-effect list (mild infusion-related reactions, headache, fatigue, back pain, all resolved quickly) and the 37-participant Phase 1/2 four-year durability data. source-2.html.gz (amsterdamumc.org) confirms 80 patients randomized, simultaneous NEJM publication, presentation at the European Academy of Allergy and Clinical Immunology congress in Istanbul, Danny Cohn as the Amsterdam UMC internist leading the research, and his verbatim quote 'This is the first time CRISPR therapy has been applied in vivo within a large, double-blind, international Phase 3 trial.' Both direct quotes match the snapshots exactly. Rule 9 satisfied: every clinical specific (percentages, p-values, arm sizes, endpoints, regulatory timeline) traces to the globenewswire press release that actually contains it; the NEJM-publication fact is sourced to the Amsterdam UMC release. NEJM itself was not cited directly (and would likely be bot-blocked), which is appropriate.

All claims align with cited sources; no hallucinations detected. The 80-patient total reconciles with the 52+28 arm split (N=52 lonvo-z, N=28 placebo). The internal cross-reference /article/2026-04/25-intellia-therapeutics-to-release-worlds-first-in-vivo-crispr-phase-3-data-on-april-27 resolves to a real published article (verified on main), which is the April preview this piece follows up.

High-quality, accurately sourced follow-up. Every clinical specific and both direct quotes verify verbatim against the committed snapshots. The lone issue is the off-allowlist (but credible) amsterdamumc.org source, which is honestly captured in a corrections note. APPROVE_WITH_CORRECTIONS.