This is the positive control. Every citation chain in this audit is strong. The forensic audit at entry would have said proceed with high confidence — and the outcome validated it. Krystal Biotech entered the portfolio when the NEJM Phase 3 data was published and exited near FDA approval. The science was not the risk. The only risk at entry was regulatory timing. That risk resolved in four months.
This audit is included not to find what went wrong, but to show what a clean chain looks like — and to answer the question: When the forensic audit says proceed, does the outcome follow? In this case, yes.
Topical HSV-1–based gene therapy delivering functional copies of the COL7A1 gene directly to DEB wounds will restore collagen type VII (COL7) protein expression, reconstitute anchoring fibril formation at the dermal-epidermal junction, and produce durable wound closure in patients with dystrophic epidermolysis bullosa.
| # | Paper / Source | What It Says | What Was Claimed | Gap | Confidence |
|---|---|---|---|---|---|
| 1 | Christiano et al., Nature Genetics, 1994; Hovnanian et al., Nat Genet, 1994 | DEB is caused by loss-of-function mutations in COL7A1. COL7A1 encodes collagen type VII, the major structural component of anchoring fibrils at the dermal-epidermal junction. Absence of functional COL7 → anchoring fibril loss → skin fragility and blistering. | B-VEC addresses the root genetic cause of DEB. | None. Causative genetic link is textbook and reproduced across hundreds of independent publications over three decades. | HIGH |
| 2 | Woodley et al., J Invest Dermatol, 2003; Spirito et al., J Invest Dermatol, 2003 | In DEB mouse models, delivery of COL7A1 via retroviral vector restores COL7 protein expression at the dermal-epidermal junction and improves skin integrity. Anchoring fibril structure reconstituted on electron microscopy. | Gene delivery of COL7A1 can restore function in DEB tissue. | None. Multiple independent research groups replicated functional rescue in preclinical models. The gap between mouse and human skin is acknowledged but COL7 biology is highly conserved. | HIGH |
| 3 | Krystal Biotech preclinical data, JCI Insight, 2017; Gurevich et al. | HSV-1–based vector (not retroviral) delivers COL7A1 to skin with high efficiency. Topical application achieves transduction of keratinocytes and fibroblasts. Vector designed to be non-replicating in normal tissue. Redosable — no long-term integration risk of retrovirals. | B-VEC's HSV-1 delivery platform is suitable for topical, repeated application to DEB wounds. | None. Topical HSV-1 delivery is mechanistically distinct from IV gene therapy systemic risks. The non-replicating design addresses the key safety concern for redosable use. | HIGH |
| 4 | Gurevich et al., Nature Medicine, 2022 (GEM-3 Phase 3 results, published NEJM Dec 2022) | In 31 DEB patients (paired wound design, each patient own control): complete wound closure at 6 months in 67.4% of B-VEC wounds vs 21.6% placebo (p=0.002). Complete closure at 3 months: 70.6% vs 19.7% (p=0.0005). COL7 expression confirmed by immunofluorescence in B-VEC–treated skin. | B-VEC produces clinically meaningful, statistically significant wound closure superior to placebo in DEB. | None. Phase 3 data with paired-wound design (each patient as own control) is methodologically strong. n=31 is small but trial size reflects orphan disease pragmatics. FDA had already reviewed Phase 2 data without requiring advisory committee. | HIGH |
| 5 | FDA BLA acceptance August 2022; PDUFA date extended January 2023 (manufacturing amendment) | FDA accepted the BLA for priority review. January 2023 extension was triggered by manufacturing process information — not a clinical efficacy or safety concern. No Advisory Committee required. No REMS required. | Regulatory path is clear; extension does not signal efficacy or safety problems. | None. Manufacturing amendments are common at BLA stage. FDA language at extension confirmed no efficacy or safety issues raised. | HIGH |
This is a rare clean chain. Every link from causative gene → protein function → animal rescue → clinical trial → regulatory review holds at HIGH confidence. The mechanism is not extrapolated — it is the textbook genetic rescue of a known monogenic disease. The Phase 3 used a paired-wound design that makes each patient their own control, eliminating the confounders that weaken most DEB trials. The FDA extension was administrative, not clinical.
The one honest caveat: n=31 in Phase 3. This is the practical ceiling for an ultra-rare disease. The FDA recognized this — orphan review, no advisory committee, no REMS. The science at entry was as clean as clinical-stage biotech gets.
The entire citation chain. COL7A1 causation → COL7 function → anchoring fibril reconstitution → wound closure. Established across primary biology, preclinical models, Phase 2, and Phase 3. No inflation, no substitution, no missing validation. The Phase 3 primary endpoint was met with p=0.002. COL7 expression was histologically confirmed in treated skin. FDA review did not raise clinical questions.
Long-term durability of repeated HSV-1 vector dosing. Phase 3 measured wound closure at 3 and 6 months. Durability beyond 6 months was supported by Phase 1/2 extension data (up to 2 years in limited patients) but not Phase 3 confirmed. Not a thesis-killer — rare disease with no alternative — but a known open question at entry.
Nothing. There are no claims in this chain that the evidence fails to support at entry.
A forensic audit at Q1 2023 entry would have found nothing to restructure. The science chain is clean from primary biology to Phase 3 clinical evidence. The only open item — long-term durability — is acknowledged, not hidden, and does not affect the approval thesis. The PDUFA date is known. The manufacturing extension is administrative. This is a pre-approval event trade with validated science backing it. Enter at conviction size, not a hedged position.
B-VEC approved as Vyjuvek May 19, 2023. First redosable gene therapy. European Commission approval followed. NicHealth exited Q2 2023 — consistent with a pre-approval event structure. The trade was structurally clean: enter on validated Phase 3 data, exit on regulatory catalyst. No science risk remained post-approval; commercial execution became the thesis, which is outside NicHealth's apparent hold thesis for the position.
What this audit demonstrates for the demo: When the forensic chain is clean, the audit confirms conviction rather than introducing doubt. Lisa entered a clean chain and exited on the catalyst. The audit would have said: this is the science working exactly as advertised.
All 5 chain links rated HIGH — confirmed: FDA approved B-VEC as Vyjuvek on May 19, 2023. COL7A1 causation, COL7 protein rescue, anchoring fibril reconstitution, Phase 3 wound closure data (67.4% vs 21.6%), and the manufacturing-only PDUFA extension — every link held exactly as the audit assessed.
Kill conditions — none triggered: No CRL. No HSV-1 safety signal. No competitor approval. The risk window closed cleanly at approval.
Decision validated: "Proceed at conviction size" was correct. NicHealth entered on validated Phase 3 data and exited on the approval catalyst — the textbook pre-approval event trade. The forensic audit at Q1 2023 would have increased conviction rather than reducing it.
Audit accuracy score: Perfect positive control. This is the case that proves the methodology's calibration from the other direction — when the chain is genuinely clean, the audit confirms it clearly, and the investor is not talked out of a correct position by false caution.