Your Cancer Drug Has a Five-Day Shelf Life. Curium Just Bid $7 Billion on Who Controls the Atoms.

On Friday, May 22, 2026, Bloomberg reported that Curium Pharma had made a takeover offer valuing Lantheus Holdings at about $7 billion. Two of the largest pure-play radiopharmaceutical companies in the world, roughly the same size, possibly weeks from a deal. Lantheus shares slipped about 2% in extended trading, and the market filed it under routine consolidation.

That reading misses what the bid is actually pricing.

When a $7 billion company bids $7 billion for another, the premium has nothing to do with the molecules. Lantheus already sells Pylarify, a PSMA-targeted PET imaging agent that cleared $500 million in sales by 2022, and it in-licensed late-stage Lu-177 radioligand therapies from Point Biopharma. Curium is paying for control of the imaging-plus-therapy stack, and underneath it, for the thing that actually gates the whole field: the atoms, and the clock they run on.

If you formulate radioligands, this deal is a map of where your scarce inputs really are. Almost none of them are the parts you were trained to optimize.

Three numbers that define the problem

Hold three figures together: a multi-billion-dollar consolidation fight, a product that expires in less time than a hotel reservation, and a raw material so scarce its annual world supply is measured in single-digit curies. That is radioligand therapy. The science is real, the clinical data are good, and the binding constraint sits somewhere else entirely.

The clock starts the moment you make it

Lutetium-177 has a physical half-life of 6.647 days. That is not a storage guideline you can pad with a safety factor. It is physics, and roughly a tenth of the radioactivity is gone every day, whether the vial is on a manufacturing line, in a courier's van, or sitting in a clinic waiting for an infusion slot. The active ingredient degrades toward uselessness from the instant it exists.

The FDA label for Pluvicto puts the shelf life at 120 hours, five days from calibration. In practice that leaves a finished dose about a five-day window to travel from the production site to the patient's arm. Miss it and the dose is simply gone: no salvage, no relabeling, no second window.

Now set the normal realities of pharmaceutical manufacturing against a five-day clock. Quality control release, aseptic fill-finish, container-closure checks, cold-chain logistics, patient scheduling. A conventional drug program runs those steps in series with comfortable margin. Here they all have to collapse into the same five days, in order, with no slack. You cannot run a full conventional QC cycle on a product that loses 10% of its potency while you wait for the assay, so radiopharma runs on rapid and parametric release because nothing else fits inside the window.

CNBC, reporting on Novartis's Pluvicto operation, described a team of roughly 30 to 40 people running distribution around the clock, with doses scheduled to the minute and written off as unsalvageable if they arrive late. That is what formulation looks like when your product is racing its own half-life. It is a relay against decay, run every single day, rather than a batch you make and ship.

The demand-supply math is just as unforgiving in the other direction. In 2023, Pluvicto demand outran supply, and Novartis had to pause new-patient starts and delay doses until added manufacturing capacity came online (FiercePharma, 2023). The chemistry never failed. You cannot warehouse a five-day product, and you cannot conjure isotope on a clinical timescale.

The atom is scarcer than the molecule

Lutetium-177 is the workhorse. The frontier is actinium-225, and that is where the constraint graduates from logistics headache to existential supply problem.

Ac-225 is an alpha emitter with a half-life of about 9.9 days. Alpha particles deposit enormous energy over a very short range, which makes them attractive for hitting solid tumors hard while sparing nearby tissue. The pharmacology is compelling and the supply is almost nonexistent.

Global clinical-grade Ac-225 supply has historically run on the order of a few curies a year, much of it from legacy thorium stock at Oak Ridge National Laboratory (BioSpace; NIDC National Isotope Development Center). A few curies, for every actinium program on Earth, combined.

A registrational cancer trial, halted not by a safety signal or a data problem, but because the world ran out of an element.

Source: BioSpace, on Bristol Myers Squibb's RayzeBio pausing a Phase 3 trial over an Ac-225 shortage, 2024

This ceiling is not theoretical. In 2024, Bristol Myers Squibb's RayzeBio paused a Phase 3 trial specifically because of an actinium-225 shortage. That is the rate limiter for the entire alpha-emitter wave, and perfect chemistry does nothing to move it.

New production capacity is coming. Companies and labs including PanTera, Niowave, NorthStar, BWXT, Ionetix, Nusano, and TRIUMF are working to expand actinium and other medical-isotope output (NucNet; BioSpace). But isotopes come from reactors, cyclotrons, and accelerators, and that infrastructure does not scale on a drug-development calendar. New production reactors run roughly 8 to 12 years from design to commissioning. Your IND-to-approval timeline does not get to wait a decade for an upstream reactor.

Half-life governs the whole field, not just the new therapeutics. The isotope clock below shows why nuclear medicine is just-in-time by force of physics.

Radioligand therapy pushes that fragility into the highest-stakes corner of oncology.

Why this breaks normal pharma logic

Look at how completely this inverts the assumptions a formulation program is usually built on. You have no safety stock, no warehouse, and no option to make a large batch, hold it, and ship against demand. Each radioligand dose is patient-specific, time-stamped, and unsalvageable if it misses its window. The inventory buffer that absorbs every other supply hiccup in pharma does not exist here.

So where do the scarce inputs actually live? Three places.

• Isotope allocation: Whether you can get Lu-177, and especially Ac-225, in the quantity and on the schedule your program needs.
• Fill-finish slots: Aseptic fill-finish capacity that can be booked against isotope delivery and patient scheduling at the same time, to the day.
• Precedent: The approved-product landscape and regulatory record that tell you which excipient systems, specifications, and stability frameworks will actually clear review, so you are not improvising on a product that punishes every reformulation.

Getting the chemistry right is necessary and nowhere near sufficient. The binding constraints sit upstream of the bench and downstream of it, and every one of them is on the clock.

The money is following the atoms, not the chemistry

Watch where the capital has gone. Every major move in this space has been a move to secure isotope supply and the platform around it.

The players and the deals

Aktis Oncology's IPO, on January 8, 2026, was the first biotech IPO of the year and was upsized, with Lilly anchoring around $100 million. Aktis's pipeline includes an actinium-225 therapeutic candidate (its [225Ac]Ac-AKY-2519 program against B7-H3 cleared its IND in early 2026), a clean illustration of the pivot from Lu-177 beta therapy toward Ac-225 alpha therapy that the whole sector is making.

Drugmakers are also locking in long-dated isotope-supply agreements: AstraZeneca with Niowave on a 10-year term, Bayer with a roster of suppliers, precisely because supply security gates the pipeline and chemistry does not. When buyers tie up a raw material on decade-long contracts and pay billions to own the platforms that control it, they are naming the scarce asset for you.

The formulation and CMC problem, named precisely

If you are the scientist who actually has to file this, the abstract "logistics problem" resolves into a stack of concrete, hard CMC challenges, each of them sharpened by the decay clock.

• Radiolysis: The product self-degrades. Radiation from the isotope drives radiolytic breakdown that erodes radiochemical purity over the dose’s short life. Your choice of radioprotectant and quencher (ascorbate and gentisate systems, for example), your buffer, and your concentration limits are core formulation levers, not finishing touches. They decide whether the dose still meets spec when it reaches the patient.
• Radiochemical purity at administration: You must specify purity at release and demonstrate it is maintained all the way to administration, despite ongoing radiolysis the entire time. The specification has to hold across a moving target.
• Container-closure under radiation: The primary container sits in a continuous radiation flux for the product’s whole life. Extractables and leachables behavior and closure integrity have to be qualified under that radiation stress, not just under standard storage conditions.
• Stability as designed-in decay: The relevant biotech and ATMP-adjacent framework is ICH Q5C for stability, Q6B for specifications, Q2(R2) and Q14 for analytical method validation and development, with Q5A(R2) and Q13 where they apply. Stability here works backwards from the usual case: the study has to account for radioactive decay as a designed-in degradation pathway rather than one you are trying to prevent.
• Fill-finish under a decay clock: Aseptic fill-finish must be slotted against isotope delivery and patient scheduling at once. A missed slot wastes a scarce, expensive, decaying input. That is exactly why approved-product CMC precedent and a fast, defensible CMC dossier are worth disproportionately more in radiopharma than almost anywhere else in drug development.

So here is the part worth dwelling on. Your isotope is gated by physics and reactor capacity, and effort will not fix it. Your fill-finish window is gated by GMP capacity, and you can only partly fix that. The research, regulatory-precedent, and CMC-framework work sits on the critical path in front of every batch decision, it is the one bottleneck fully within your control, and it is the one that quietly eats weeks.

Where DeepC fits

DeepC will not source you an atom or book you a fill-finish slot. It does not manage isotope supply or cold-chain logistics, and any tool that claimed to would be selling you something it cannot deliver. What it does is take the research-and-CMC fraction of your timeline off the critical path, so the only things left gating your program are the atom and the clock, the two constraints that are genuinely physical.

Two of its specialist agents do this work directly.

Biologics Research Agent: know the field and the rules before you commit isotope

Before you allocate a scarce, perishable input to a program, you need a clear-eyed picture of what is already approved, who sponsors it, on what regulatory basis, and what the dossier will have to satisfy. The Biologics Research Agent assembles that picture from primary registries rather than from training-data recall.

• Approved-product landscape: A filterable view across the FDA Purple Book, EMA authorizations, and the FDA Office of Therapeutic Products cell- and gene-therapy register, the right neighborhood for charting advanced therapeutic modalities and the pathways they cleared on.
• Pipeline and competitive intelligence: Pulled from Open Targets by target, modality, development stage, sponsor, and indication, so you know who else is in Phase II or III against your target before you commit.
• Regulatory framework: Surfaced through a modality-filtered search across ICH Q5A-E, Q6B, and Q13, the exact specification, viral-safety, and continuous-manufacturing guidance your CMC dossier will have to meet.
• A fixed structure: Approved-product landscape, then pipeline and competitive intelligence, then structural and sequence references, then regulatory framework, then open questions. Every claim links back to the underlying registry or publication.

That is the "understand the landscape and the rules" deliverable, produced in the time it takes to read it instead of the weeks it takes a team to assemble it by hand.

Formulation & CMC Agent: the dossier under a decay clock

This is the agent built for the formulation-under-time-pressure problem. It produces structured CMC memos (BF01, BF02, and so on) that stream into the side panel as it works, and it maps cleanly onto the radiopharma CMC challenges above.

• Excipient rationale: Cross-referenced against the unified IID and PharmaExcipients surface for maximum-permissible levels per route. This is how you justify your radioprotectant, stabilizer, and buffer choices within precedented ranges, the core radiolysis-control levers, with the precedent to back them.
• Precedent extraction from approved-product labels: Via DailyMed (BLA to SETID to SmPC section 6.1), pulling the actual excipient systems approved products use. When rework on a decaying input is this costly, knowing the precedent before you formulate is worth more than it is anywhere else.
• Reports that follow the ICH framework verbatim: Q5C for stability, Q6B for specifications, Q2(R2) and Q14 for analytical validation and development, plus Q5A(R2) and Q13 where relevant. That is the stability-and-specifications backbone a five-day product needs, including the framing of decay as a designed-in degradation pathway under Q5C.
• CMC red flags with severity ratings: In a fixed memo structure: overview, then excipient rationale table, then regulatory framework, then CMC red flags, then precedent products, then recommended next steps. Container-closure-under-radiation risk, radiochemical-purity-at-administration risk, and fill-finish-timing risk get surfaced here, early, before any of them costs you a batch.

The CMC scope spans LNP-mRNA, ADCs, bispecifics, CAR-T, and AAV, so the radioligand work draws on the same excipient, precedent, and ICH machinery the agent already runs, not a brittle one-off path.

The molecule was never the bottleneck you could not fix, and neither was the paperwork. DeepC fixes the paperwork bottleneck (the landscape, the precedent, the Q5C and Q6B framework, the red-flag identification) so that the only thing left standing between your program and the patient is the atom and the clock.