Elicio Therapeutics filed its 10-K on March 12, 2026. The filing confirms what the stock price has been sniffing out: the AMPLIFY-7P Phase 2 readout is imminent (H1 2026), the pre-readout signals are unusually strong, and the company is running on fumes.

This is a one-factor thesis. Everything — financing, partnership, governance, competitive position — is conditional on whether the Phase 2 DFS data is positive. Factor decomposition confirms: 79.1% idiosyncratic variance, all signal is company-specific, and the five causal drivers collapse into one question: does this vaccine work in pancreatic cancer?

The Setup

ELTX is a clinical-stage oncology biotech with no revenue, no approved products, and $233.7M in accumulated deficit. The AMP platform chemically links lipid tails to antigens, routing them specifically to lymph nodes for T cell education. The lead program, ELI-002 7P, is an off-the-shelf cancer vaccine targeting 7 KRAS mutations present in 88% of pancreatic ductal adenocarcinoma (PDAC) and 25% of all solid tumors.

The Phase 2 AMPLIFY-7P trial is randomized 2:1 (ELI-002 7P vs observation), with disease-free survival as the primary endpoint. Enrollment completed December 2024. Final analysis is event-driven, anticipated H1 2026.

Market cap is ≈$224M. Two analysts cover it (HC Wainwright, Rodman & Renshaw). This is undercovered micro-cap territory.

The Pre-Readout Signal Stack

Most Phase 2 biotech binaries are coin flips dressed up with press releases. This one has a signal stack I haven't seen on a micro-cap vaccine play:

IDMC August 2025: Continue without modification. The independent data monitoring committee reviewed unblinded efficacy data and recommended the trial continue to final analysis. This is a hard gate — they could have stopped for futility or safety. They didn't. An external committee saw who lived and who died and said: keep going.

November 2025: Fewer events than projected. The company disclosed that fewer disease progressions and deaths had occurred than their statistical model predicted. In a 2:1 randomized trial, two-thirds of patients are on the treatment arm. Fewer total events with more patients on drug is directionally favorable. The company explicitly stated this "may reflect a favorable impact on DFS."

Phase 2 immunogenicity: 99% response rate, 145x fold-change. 89 of 90 evaluable patients mounted mKRAS-specific T cell responses. Average 145x expansion over baseline. 80% exceeded the 9.5x threshold associated with clinical activity in Phase 1. This is among the strongest immunogenicity data ever published for a therapeutic cancer vaccine — and this is Phase 2 data at scale (n=90), not a Phase 1 poster with 12 patients.

Phase 1 Nature Medicine: HR 0.12 on relapse, p<0.0001. Published August 2025. Above-median T cell responders had 88% reduction in relapse risk and 77% reduction in death risk. Peer-reviewed in a top journal. Small (n=25) but the biomarker-survival correlation is the strongest signal in this thesis — if the Phase 2 immunogenicity data (99% response) correlates with survival as Phase 1 suggests, the DFS data should be positive.

MSK/Lustgarten IIT. Memorial Sloan Kettering and the Lustgarten Foundation independently funded a Phase 1 trial combining ELI-002 7P with chemo and checkpoint inhibitor in neoadjuvant PDAC. Third parties don't risk reputational capital on dead science. Expected to start H1 2026.

Antigen spreading in 87% of patients. T cells trained against mKRAS are also attacking other tumor neoantigens. Mechanistically important — makes immune escape harder. 90% showed both CD4+ and CD8+ responses (vs 33% in Phase 1).

Any one of these signals is soft. The stack together is unusual. The Bayesian math: starting from a 25% Phase 2 oncology base rate, sequential updates through IDMC pass (LR 2.0), event count (LR 1.5), and biomarker correlation (LR 1.5) bring the posterior to approximately 45%. That's where we land.

The Competition Vacuum

BioNTech's personalized neoantigen vaccine platform (autogene cevumeran/BNT122) has failed or stalled in 3 of 4 indications: first-line melanoma (failed 2024), adjuvant colorectal cancer (crossed futility boundary Q3 2025), bladder cancer (paused for safety event July 2025). Their PDAC trial won't read out until December 2029. Moderna/Merck's mRNA-4157 is pursuing melanoma and NSCLC — not PDAC.

ELTX has the adjuvant PDAC cancer vaccine space to itself. This matters because the off-the-shelf vs personalized distinction is structural, not cosmetic: ELTX's approach is cheaper to manufacture, faster to deploy, and showing 99% response rate (n=90) vs BioNTech's 50% (n=16). If personalized neoantigen vaccines are failing because of manufacturing complexity and patient selection, ELTX's simpler approach may actually be the answer — or at least the first one to prove it.

The Bear Case Is Modal

The bear case isn't a footnote — it's the most probable outcome at 45%.

Going concern is real. Both management and auditor Baker Tilly issued "substantial doubt" language. Cash of $18.6M funds operations only into Q3 2026. This is 4-6 months from filing. Not boilerplate.

62% dilution in one year. Share count went from 11.0M to 17.8M in 2025, funded by ATM ($16.2M), GKCC convertible note conversion (3.5M shares), January offering ($9.2M), and warrant exercises. The $40M ATM program and $200M shelf registration remain live. More dilution is coming regardless of Phase 2 outcome.

GKCC governance structure. Board member Chudnovsky controls GKCC, which owns 43.1% of outstanding stock AND holds a $10M first-lien secured note collateralized by all company assets. In the bull case, this is an aligned insider with skin in the game. In the bear case, the secured creditor is the controlling shareholder — minority shareholders get wiped while GKCC takes the IP. The $20M note-to-equity conversion in March 2025 shows commitment to equity upside, but the remaining secured note is a governance bomb in a downside scenario.

Cancer vaccine base rate: 8.3%. Only one therapeutic cancer vaccine (Provenge) has ever been FDA approved. Phase 2 to approval for cancer vaccines runs ≈8.3% historically. The ELTX pre-readout signals justify above-base-rate conviction, but the graveyard is real. Immunogenicity does not equal efficacy — MAGE-A3 and TG4010 proved that.

Cash and catalyst are racing each other. Data expected H1 2026. Cash runs out Q3 2026. If the readout slips, the company negotiates financing from desperation. If data is ambiguous (not clearly positive, not clearly negative), the going concern overhang prevents clean financing. The only good path is data so strong that pharma comes running.

Factor Decomposition

This is a one-factor thesis masquerading as a multi-factor setup.

Statistical decomposition confirms: beta 1.68 to SPY, 79.1% idiosyncratic variance (passes the >75% Paleologo threshold), XBI down -2.3% while ELTX up +45.2% — the move is entirely company-specific, not biotech sector rotation.

Factors 2 through 5 are all conditional on Factor 1. There is no diversification within this thesis. Positive DFS data solves financing, attracts partners, and validates the competitive position. Negative DFS data makes every other factor irrelevant. You cannot hedge one leg and keep the others.

Entry and Forward EV

The options market is pricing an ATM straddle implying +/-49% move — almost exactly our scenario spread. Call positioning is 4.3x puts on the May expiration. IV at 160-360% is above the 52-week range (39-114%) at every expiration. The market knows the readout is coming.

Backing out implied probabilities: if the market's EV equals the current price (≈$12.20), the market is pricing roughly 25% bull / 25% base / 50% bear. Our edge is approximately 10 percentage points of probability shifted from bear to bull, driven by the pre-readout signal synthesis. That 10pp shift on a $37.50 bull-bear spread equals ≈$3.75/share of edge — real but not overwhelming.

At $12.20 entry, risk/reward is 2.9:1. At a pullback to $9-10 (max pain gravity), risk/reward improves to 4.0-4.8:1 with the same signal stack. The stock has already moved +45% in a month — we'd be buying momentum, not panic.

Sizing: 1-2% GMV maximum. This is a catalyst bet sized for surviving the -80%. At 1% position, max loss is 0.8% of portfolio. Survivable. The bimodal return distribution (either +228% or -80%) makes expected-value sizing dangerous — proportional rule must be tempered by the binary structure.

Open Questions

Five gaps remain in the worldview:

1. Event count (HIGH): How close is AMPLIFY-7P to triggering final analysis? "Fewer than projected" tells us direction but not proximity. Weeks vs months.
2. Insider purchases (HIGH): Any open-market buying by officers/directors? Only compensation awards detected. Open-market buys would be a strong independent signal.
3. GKCC covenant terms (MEDIUM): What triggers default on the $10M secured note? Could be material in bear scenarios.
4. Crossover rate (MEDIUM): How many observation-arm patients crossed over after relapse? High crossover complicates ITT analysis.
5. Partnership discussions (MEDIUM): Any active BD signals? A pre-data deal would be transformative.

Conclusion

ELTX is a positive-EV catalyst bet with moderate edge. The pre-readout signal stack is the strongest I've seen on a micro-cap cancer vaccine play — IDMC pass, fewer events, 99% immunogenicity (n=90), Nature Medicine peer review, competition vacuum. Each signal alone is soft. Together they justify 45% probability against an 8-25% base rate.

But the bear case is modal. Going concern is real. Dilution is structural. The governance structure is a bomb in the wrong scenario. And you're buying after a +45% month, not on the floor.

The edge is in synthesis — combining soft signals that no one else has bothered to combine because only two analysts cover this stock. The alpha is in pre-positioning ahead of the readout, not in waiting for the 8-K (by which point the gap is instant). This is retail edge territory: small, undercovered, imminent catalyst, asymmetric payoff.

Size for survival. 1-2% GMV. If you can't afford to be wrong, you can't afford to be in it.

Evidence