Thesis

Savara Inc. is a single-asset orphan disease biotech with one drug (MOLBREEVI, inhaled molgramostim for autoimmune pulmonary alveolar proteinosis) and one catalyst: FDA PDUFA date approximately August 2026. The stock at $5.25 implies ≈47.5% probability of approval. Primary-source evidence — the 10-K filed March 13, RTW Investments' $75M post-RTF commitment, published base rates on RTF resubmissions, and BTD approval cohort statistics — supports 65%. That's a 17.5 point probability edge on a binary with +81% upside (approval) vs -67% downside (CRL).

The question is not whether SVRA is a good company. It's whether 47.5% is the right number. The evidence says no.

The Setup

Autoimmune PAP is an ultra-orphan lung disease — roughly 2,000-2,350 diagnosed patients in the US, 6-7 per million. Patients currently get whole lung lavage, an invasive procedure that is not curative. There are zero approved therapies in the US or EU.

SVRA's BLA was first submitted in March 2025. FDA refused to file it in May 2025 — not on efficacy or safety grounds, but on CMC: the manufacturing transfer from clinical-stage GEMA Biotech to commercial-scale Fujifilm Diosynth wasn't adequately documented. Company ran a Type A meeting, aligned with FDA on a resolution path, completed three PPQ validation batches at Fujifilm, and resubmitted in December 2025. FDA accepted the resubmission and granted Priority Review in February 2026.

The clinical data is not in dispute. IMPALA-2, the pivotal Phase 3 trial (164 patients, published in the New England Journal of Medicine in August 2025), met its primary endpoint (DLCO at Week 24) and secondary endpoints with statistical significance. 97% of patients completed the 48-week double-blind period. Zero drug-related discontinuations. 100% of completers enrolled in the open-label extension. For a rare disease trial, this is as clean as it gets.

Why 65%, Not 47.5%

Four independent lines of evidence converge:

1. RTW Investments committed $75M contingent on approval — after the RTF.

RTW is a $3B+ biotech-specialist hedge fund, not a spray-and-pray operation. Their last comparable deal — $75M to Avadel in March 2023 for LUMRYZ — was a clean win: drug approved May 2023, royalties collected, Avadel prepaid $60.2M to terminate upon acquisition. RTW's portfolio: 54 core positions at 1.7x MOIC, 18 exits at 2.7x MOIC.

The SVRA deal was signed October 29, 2025 — after the RTF, after the securities class action filing. RTW had full visibility into the CMC failure and litigation risk, conducted diligence on the manufacturing resolution, and still committed $75M. They're not hedging. They're declaring the manufacturing problem is fixed.

This is the strongest evidence item in the file. RTW doesn't have to make this bet. The fact that they did, at that specific moment, with that specific structure (contingent on approval), is a sophisticated investor pricing P(approval) well above 50%.

2. RTF resubmission base rate: 71.7%.

A PMC study (2008-2017) found that of 53 resubmitted RTF'd applications, 71.7% were eventually approved. CMC was the single most commonly cited RTF deficiency (19.4% of all refusal reasons). SVRA's RTF was CMC-only — no additional efficacy studies requested, no safety concerns raised. This is likely a more favorable subset than the 71.7% overall rate.

3. BTD approval rate for filed BLAs is substantially above the 53% headline.

FDA has granted 634 Breakthrough Therapy Designations. 336 (53%) have received approval. But that 53% includes products still in Phase 2 trials that haven't submitted applications yet. For the subset with filed BLAs and positive Phase 3 data — SVRA's cohort — the rate is much higher. We found no examples of orphan drug rejections with BTD + clean Phase 3 + filed BLA + no safety concerns.

4. The designation stack is the strongest possible.

Breakthrough Therapy + Fast Track + Orphan Drug (7-year US exclusivity, 10-year EU) + Priority Review. In a disease with zero approved therapies. FDA rarely assembles this combination without high confidence in the science. And they granted Priority Review on the resubmission — a formal statement that the package is complete and review-ready.

The Bear Case Is Real but Narrow

The bear case is not "the drug doesn't work." It's "Fujifilm's manufacturing facility might have quality issues."

Fujifilm Diosynth has received Form 483s at Morrisville NC (April 2021, five significant observations including endotoxin failures and environmental monitoring gaps) and College Station TX (twice in 2023). No warning letters issued, but recurring quality observations. The pre-approval inspection (PAI), expected April-June 2026, is the single largest remaining risk. If FDA inspects the facility manufacturing MOLBREEVI and finds issues, it could trigger a Complete Response Letter regardless of clinical data quality.

We don't know which specific Fujifilm facility makes MOLBREEVI drug substance. We don't know if FDA has scheduled the PAI. These are the two highest-priority open gaps.

The structural bear case — single-asset, no revenue, no pipeline — is real but it's the bet, not an objection to the bet. You know this going in. CRL = $1.75, thesis dead, exit immediately. That's why you size for survival, not for EV maximization.

Other bear signals: 19% share dilution in 2025 alone ($140M offering at $4.20 in October), G&A surged 68% to $42.1M for pre-commercial buildup (stranded cost if rejected), and the RTW royalty toll (7-9.5% of US Net Sales capped at $187.5M) clips commercial upside.

The Math

Market-implied probability, backed out from scenarios:

Conditional value if approved (probability-weighted bull+base):
  V_approve = (30/65 × $11) + (35/65 × $7.50) = $9.12

Value if rejected:
  V_reject = $1.75

Market-implied P(approval):
  $5.25 = P × $9.12 + (1-P) × $1.75
  P = 47.5%

Probability sensitivity:

You make money as long as P(approval) > 47.5%. Even at 55% — well below the 71.7% RTF resubmission base rate — you're up 10.5%. To lose money, you need to believe RTW made a $75M mistake, the base rates don't apply, BTD + clean Phase 3 = less than a coin flip, and all 8 analysts (unanimous buy, mean target $10.81) are wrong. That's a hard case to make.

Factor Decomposition

Statistical (iev regress, 250d):
  Idio:  82.2%   ← well above 75% target
  XBI:   18.1%   ← biotech sector
  MTUM:   5.8%   ← momentum
  SPY:   -6.0%   ← negative market beta

82% idiosyncratic. This stock moves on its own. The negative SPY beta (β=-1.10) means SVRA actually diversifies a long equity book — the FDA binary outcome is independent of market direction.

Economic decomposition:

Edge is concentrated in one factor. Don't pretend to have insight in the others.

Entry: Two-Stage Binary

This is not a single binary. It's two sequential catalysts:

Stage 1: Fujifilm PAI (April-June 2026). If clean, P(approval) jumps to 80%+, stock likely re-rates to $7-8. If issues found, P(approval) drops to 30-40%, stock falls to $3-4.

Stage 2: PDUFA (~August 2026). Terminal binary. Approve = $9-11. CRL = $1.50-2.00.

The staged structure favors a staged entry. Enter small now to capture PAI optionality. Scale up materially if the PAI comes back clean — P(approval) jumps to 80%+, position is de-risked. If the PAI has issues, cut early on a small position before the full downside is realized. Staged entry cuts max drawdown roughly in half versus going all-in ahead of the PAI, while preserving most of the upside. It matches position size to information flow.

Management conviction signals support near-term entry: 5-year HQ lease signed March 10, 2026 ($1.78M aggregate rent, Yardley PA), three days before the 10-K filing and five months before PDUFA. Companies don't commit to multi-year commercial infrastructure when they expect a CRL.

What I Don't Know

1. Which Fujifilm facility makes MOLBREEVI drug substance — 483 risk depends on location
2. Whether FDA has scheduled the PAI — expected April-June but unconfirmed
3. Exact PDUFA date — Priority Review from February implies August, need the 8-K
4. RTW's actual diligence depth on Fujifilm manufacturing — we're trusting their bet, not their work product
5. Peak sales pricing strategy — orphan drug pricing is typically $250-500K/year, but SVRA hasn't disclosed

Conclusion

SVRA is a probability mispricing on a binary FDA catalyst. The market prices 47.5% approval. Four independent evidence streams — RTW's burned-boat commitment, RTF resubmission base rates, BTD cohort statistics, and bulletproof clinical data — support 65%. The breakeven is 47.5%, which means you can be materially wrong on probability and still make money.

The risk is narrow but existential: Fujifilm PAI inspection. This is the one variable we can't model from the outside. RTW presumably diligenced it and committed $75M. That's the best external validation available, but it's still someone else's homework.

The staged entry — small now, scale after a clean PAI — respects what you know and what you don't. The probability edge is robust. The timing is right: RSI 38.6, down 10% in a month, pre-PAI accumulation window. The question reduces to: do you trust RTW's manufacturing diligence more than Fujifilm's 483 history?

LR = 1.6. Market underpricing approval probability by ≈17 points. Evidence quality is high (SEC filings, NEJM, PMC base rate study, RTW track record). Not extreme because the remaining bear case (Fujifilm PAI) is genuine and opaque.

Evidence