SANA filed its 10-K on March 3. The filing contains going concern language, NEJM-published proof-of-concept data for an immunosuppression-free T1D cure, and a CFO who keeps buying stock into the disclosure. The science is genuinely differentiated. The investment case isn't.

The Science

UP421, SANA's predecessor to SC451, published 12-month data in the New England Journal of Medicine in September 2025. Allogeneic HIP-modified islet cells survived and functioned for 12 months in a Type 1 Diabetes patient with no immunosuppression. C-peptide present throughout, responding to meal stimulus. PET/MRI confirmed beta cell survival. No drug-related adverse events. Management calls it "the first example of successful transplantation with no immunosuppression into a person with an intact immune system to demonstrate survival and function of allogeneic cells."

That claim checks out. Cross-ticker corroboration strengthens it: Vertex's VX-264, their device-encapsulated approach to immunosuppression-free islet therapy, failed its efficacy endpoint and was discontinued in March 2025. The alternative to genetic editing (SANA's approach) didn't work. Vertex is now pursuing hypoimmune via a CRISPR license, but at research-stage only — years behind SANA's 12-month human data.

SC451 is the scalable version — stem cell-derived (iPSC) instead of primary donor islets. IND filing and Phase 1 start expected "as early as this year." That's the catalyst that determines everything.

SG293, the in vivo CAR T program, is a weak second leg. Kelonia's KLN-1010 achieved 100% MRD-negative in myeloma patients at ASH 2025. AbbVie acquired Capstan for $2.1B and partnered with Umoja — both for in vivo CAR T. Five-plus competitors have Phase 1 human data while SG293 remains preclinical. SANA is functionally a single-program company.

The Capital Problem

Going concern language appears multiple times in the 10-K: "our present capital resources may not be sufficient to fund our planned operations for at least one year from the date of this Annual Report, and there is substantial doubt as to our ability to continue as a going concern."

The math:

• Cash: $138.4M (Dec 31, 2025)
• FY2025 real cash burn (ex non-cash): ≈$99M
• Forward burn estimate: $65-90M (two lean Phase 1 programs, 142 employees, down from 350+)
• Runway: 18-25 months at current burn — but management says less than 12

The disconnect between our runway estimate and management's language suggests either Phase 1 initiation costs are higher than the run-rate implies, or management is being conservative to justify the next raise. New ATM agreement entered March 2026. Dilution is not a risk — it's a certainty.

Since IPO at ≈$25/share in January 2021: $1.7B raised and burned. Zero revenue. Ever. 267M shares outstanding versus ≈75M at IPO.

The CFO Signal

Susan Wyrick has been buying open market consistently: $63K in December 2025, $10.7K more that month, $23.6K in January 2026, and $4.8K on March 2 — one day before the going concern 10-K filed. All P-code (open market purchases, not grants). Total ≈$112K over six months.

The CFO has the best view of cash and runway. Buying into going concern language is contradictory behavior if she genuinely believes equity goes to zero. But $112K against an $840M market cap is conviction, not a backstop. Compare to BHVN where Janus Henderson put $125M at $7.50 — that's a floor. This is a statement.

Factor Decomposition

Standard regression would show 80%+ idio variance for a pre-revenue clinical biotech. Uninformative. The useful decomposition is thesis-level:

Edge-weighted variance: 2-3%. Roughly 97% of the outcome is determined by factors where we have zero informational advantage.

The dependency structure makes it worse. Capital and clinical are sequential gates — must survive to generate data, must generate data to attract a partner. P(capital survives) x P(clinical succeeds) = ≈48%, which roughly matches what the 38% short interest implies (≈55% bear). Our estimate equals the market's estimate. No daylight.

The EV Calculation

• P(SC451 Phase 1 initiates 2026): 70%
• P(safety data positive | initiates): 80%
• P(re-rating to $8+ | positive data): 60%
• Combined P(bull case): ≈34%

EV = 0.34 x (+150%) + 0.66 x (-70%) = +5%

Thin. Even bumping IND clearance to 80% gets you ≈11% EV. Not enough to justify binary risk with going concern overlay and certain dilution.

What Would Change This

SC451 IND clearance is the superposition collapse event. If it clears and first patient is dosed, the going concern narrative recedes and the stock re-rates. The NEJM data de-risks safety substantially. But IND clearance depends on manufacturing/CMC — exactly the domain where we have no edge. Vertex paused its own zimislecel dosing for "internal manufacturing analysis" despite $12.3B in cash. iPSC-derived islet manufacturing at clinical scale is hard.

A partnership or acquisition would solve the capital problem. SANA is on every large pharma's T1D screen after the NEJM publication. Sanofi (owns Tzield, not curative), Vertex (pursuing hypoimmune, behind SANA), Novo Nordisk (T1D dominant but in own crisis at -46% 12M) are all logical. But zero evidence of active discussions. Prediction: 25% by year-end.

Cross-Filing Signal: DOGE/FDA Risk

SANA is the fourth clinical-stage biotech to add DOGE/FDA disruption language in 10-K season, joining PCVX, BHVN, and DYN. Four of five companies with 2026 FDA catalysts now flag this risk. This is emerging as a latent correlated factor across clinical-stage biotech — worth monitoring as a sector-level headwind.

Verdict

The science deserves respect. The NEJM publication is real. VX-264's failure validates SANA's genetic approach as the leading path to immunosuppression-free T1D therapy. The CFO buying into going concern disclosure is a genuine conviction signal.

But conviction without edge is just opinion. We can't assess CMC risk. We can't time the IND. We can't predict the partnership. The street is already 89% buy with an $8.43 target — consensus sees everything we see. The 38% short interest says professional capital has done the same math and reached the opposite conclusion with equal conviction.

Doorway state. Bear 55%, bull 45%. SC451 IND clearance collapses the superposition. We're not the ones to bet on which side it collapses to.

Evidence