Single-cell stochastic dosimetry for Ac-225 / Th-227 / At-211 targeted alpha therapy — short-range, high-LET, hit-or-miss biology
Every alpha decay is a two-body momentum-conservation problem played out at the femtometer scale. When Actinium-225 emits an alpha particle (a helium-4 nucleus) with 5.830 MeV of kinetic energy, the recoiling daughter nucleus — Francium-221 — is not a passive bystander. Because momentum must balance, the light, fast alpha particle carries away ~98% of the released energy while the heavy daughter recoils with tens of keV, an amount that dwarfs any chemical bond holding it to a chelator or targeting antibody.
In the rest frame of the parent nucleus, total momentum before decay is zero, so it must remain zero after decay: the alpha particle and daughter nucleus fly apart in exactly opposite directions with equal and opposite momentum.
p_alpha = p_daughter
Because kinetic energy E = p²/2m, and the daughter (Fr-221, mass ~221) is roughly 55 times heavier than the alpha particle (mass 4), the energy partitions inversely with mass:
E_daughter / E_alpha ≈ m_alpha / m_daughter ≈ 4/221 ≈ 1.8%
For a 5.830 MeV Ac-225 decay, this puts the alpha particle at roughly 5.725 MeV and the Fr-221 daughter recoil at roughly 105 keV. That 105 keV recoil energy, concentrated in a single nucleus over a path of only a few nanometers, is what makes daughter-nuclide redistribution an unavoidable physical consequence of every single decay — not a rare failure mode.
Targeted alpha therapy constructs (e.g., Ac-225-PSMA-617, Ac-225-DOTATATE) hold the actinium radiometal in a macrocyclic chelator (DOTA) coordinated by 8–9 bonds, each worth a few electron-volts of binding energy — a few eV total for the whole cage in the most generous estimate, versus ~100,000 eV of recoil kinetic energy delivered to the daughter in a few femtoseconds.
This 20,000-to-1 energy mismatch means essentially every alpha decay event mechanically ejects the daughter nuclide from its chelator, regardless of how well the chelate chemistry was engineered. This is fundamentally different from beta-emitter or gamma-emitter radiopharmaceuticals, where recoil energies are orders of magnitude lower and the chelate typically survives decay intact.
The practical consequence carried through the rest of this simulation: once Ac-225 decays inside or near a tumor cell, none of its four subsequent daughter decays (Fr-221, At-217, Bi-213, Po-213) can be assumed to still be chemically attached to the original targeting vector.
A single therapeutic Ac-225-PSMA-617 administration of a few MBq corresponds to on the order of 10^11–10^12 decaying atoms — meaning the recoil-driven daughter release happens billions of times per injection, not as a rare edge case.
A typical mammalian cell is 10–20 μm across. Alpha particles from Ac-225's decay chain travel only 50–90 μm in tissue — a handful of cell diameters — before stopping completely. Along that short path they deposit energy with extraordinary density: linear energy transfer (LET) around 80–120 keV/μm, roughly 400–500 times higher than the LET of therapeutic beta particles. This combination of short range and dense ionization is the physical basis of targeted alpha therapy's potency and its fundamentally different dosimetry.
A beta particle (electron) loses energy to matter mostly through single, glancing Coulomb interactions spread over a long, tortuous path — its LET is low (~0.2 keV/μm in water) because most of its track consists of nearly-empty space between rare ionization events. An electron can traverse an entire cell, or many cells, while depositing only a small, diffuse dose.
An alpha particle is roughly 7,000 times more massive and carries a double positive charge. It interacts far more frequently with the electron clouds of surrounding atoms, losing energy in a nearly continuous, densely-packed column of ionizations spaced only 2–5 nanometers apart along its track. Over its 50–90 μm range it produces hundreds of thousands of ionization events concentrated in a cylinder roughly the width of the DNA double helix.
Sparse, low-LET ionization from beta or gamma radiation predominantly produces single-strand DNA breaks and base damage, most of which are efficiently repaired by base-excision and single-strand-break repair pathways — cell survival curves are broad and dose-rate dependent.
High-LET alpha tracks instead produce clustered, complex damage: multiple ionizations within a few nanometers of each other frequently sever both strands of the DNA duplex close together, generating clustered double-strand breaks (DSBs) that overwhelm non-homologous end joining and homologous recombination repair. This is the molecular reason a single alpha traversal through a cell nucleus is often sufficient to be lethal, whereas beta or gamma radiation typically requires a much larger cumulative dose.
Relative biological effectiveness (RBE) quantifies this: to achieve the same biological effect as 1 Gy of alpha radiation typically requires 5–20 Gy of low-LET beta/gamma radiation, depending on cell type and endpoint (RBE for cell killing is often cited near the high end, 10–20, for clonogenic survival assays).
Conventional radiopharmaceutical dosimetry (MIRD formalism) computes a mean absorbed dose in Gray, averaged over an entire organ or tumor mass. That approach works reasonably well for beta and gamma emitters, whose particles cross many cells and smooth out spatial heterogeneity. For alpha emitters, the picture breaks down completely: because each alpha track is short and each cell either is or is not traversed, the relevant biological quantity is not the mean dose but the discrete, stochastic distribution of the number of hits per cell — governed by Poisson statistics.
Because each cell nucleus is traversed by a small, discrete, random number of alpha tracks, the probability of a cell receiving exactly n hits given a mean hit rate λ (which depends on activity concentration, exposure time, cell/target geometry, and decays per unit volume) follows the Poisson distribution:
P(n; λ) = λⁿ · e^(−λ) / n!
Even when the organ-averaged absorbed dose corresponds to a mean of, say, λ = 1.5 hits per cell, this does NOT mean every cell receives 1.5 hits — that is physically impossible for a discrete particle count. Instead:
P(0 hits) = e^(−1.5) ≈ 22% of cells are completely missed P(1 hit) = 1.5·e^(−1.5) ≈ 33% of cells take exactly one traversal P(≥2 hits) ≈ 45% of cells take two or more traversals
The missed cells (P(0)) may survive entirely untouched even though the "average dose" delivered would, under conventional dosimetry, predict substantial cell kill.
Because even a single alpha traversal often delivers enough clustered DNA damage to kill a cell (unlike beta/gamma radiation, where cell killing requires an accumulated dose from many track crossings), the key clinical question is not "what is the mean dose to this tumor?" but "what fraction of tumor cells receive at least one hit, and what fraction receive zero?"
This reframes therapy optimization: increasing administered activity primarily works by raising λ enough that the Poisson zero-hit fraction e^(−λ) approaches zero across the tumor cell population — i.e., saturating hit coverage — rather than by delivering incrementally more dose to cells that are already lethally hit.
Microdosimetric models (e.g., the ICRU/ICRP microdosimetric formalism, single-cell Monte Carlo track-structure simulations such as Geant4-DNA) are increasingly used alongside or instead of MIRD mean-dose calculations for alpha-emitter dosimetry, particularly for small or heterogeneous tumor deposits and micrometastases where cell-scale statistics dominate outcome.
A tumor micrometastasis with heterogeneous PSMA expression can have large patches of completely unhit (surviving) cells even at activities that would appear, by mean-dose calculations alone, to deliver a "curative" dose — a key reason for observed heterogeneity in alpha-RLT treatment response.
Ac-225 is prized as an in-vivo "nanogenerator": a single injected Ac-225 atom produces four subsequent alpha or beta decays as it cascades to stable Bi-209, in principle multiplying cytotoxic hits per targeting event. But as established in Stage 1, every decay in the chain recoils its daughter free of the chelator. Over the minutes-to-hours timescale of the chain, daughters can diffuse or be transported out of the tumor cell and into surrounding tissue — most consequentially the salivary glands and kidneys, which is the leading dose-limiting toxicity of Ac-225-PSMA therapy in the clinic.
Ac-225 (t½ = 9.9 days, α, 5.830 MeV) → Fr-221 (t½ = 4.9 min, α, 6.334 MeV) → At-217 (t½ = 32.3 ms, α, 7.067 MeV) → Bi-213 (t½ = 45.6 min, branches: 97.8% β⁻ to Po-213, 2.2% α to Tl-209) → Po-213 (t½ = 4.2 μs, α, 8.375 MeV) → Pb-209 (t½ = 3.2 h, β⁻) → Bi-209 (stable).
Each of the four alpha decays and two beta decays in this chain deposits additional energy and, critically, generates another independent recoil event capable of relocating the daughter nucleus. In total, one Ac-225 decay can in principle produce up to four separate alpha "hits" within or near the same tissue volume — but only if every daughter stays put, which physics says it will not.
Once recoiled free of the PSMA-617 chelate, daughter nuclides behave as free radiometal cations subject to normal biodistribution: Francium and Astatine behave chemically like alkali metals and halogens respectively; Bismuth and Polonium have their own renal and reticuloendothelial handling; Lead-209 behaves like circulating lead, with affinity for bone and kidney.
Because much of this redistribution happens on the minutes-to-hours timescale of the intermediate daughters (Bi-213 at 45.6 min is long enough for meaningful vascular transport before its own decay), free daughters can leave the tumor microenvironment and accumulate in organs with no PSMA expression at all — most importantly the salivary glands (which show substantial physiologic uptake through PSMA-independent mechanisms as well) and the proximal renal tubules.
This daughter-recoil redistribution is now recognized as a distinct, emerging challenge in alpha-emitter dosimetry, separate from the primary targeting biodistribution of the parent radiopharmaceutical — and an active area of chelator and nanocarrier engineering (e.g., macrocyclic cages, liposomal encapsulation, "in-cell" chelation strategies) aimed at retaining daughters closer to their site of production.
Xerostomia (severe salivary gland dysfunction) is the most common and often dose-limiting toxicity reported in Ac-225-PSMA-617 clinical series (Kratochwil et al., J Nucl Med 2016 and subsequent reports), attributed in significant part to both direct salivary PSMA expression and to daughter-nuclide redistribution to the glands over repeated treatment cycles.
Despite the biological brutality of alpha-particle hits, targeted alpha therapy has produced some of the most striking clinical responses seen in molecularly targeted radiotherapy — precisely because the same short range and Poisson statistics that make dosimetry complicated also create a favorable therapeutic window. Cells with high PSMA expression accumulate enough activity to make at least one hit near-certain (and one hit is often lethal), while PSMA-negative marrow stem cells — protected by low target expression — receive vanishingly few hits.
Christoph Kratochwil and colleagues at Heidelberg first reported Ac-225-PSMA-617 in patients with metastatic castration-resistant prostate cancer who had progressed despite Lu-177-PSMA-617 beta therapy. Using microdose activities on the order of 100 kBq/kg per cycle (roughly 1,000-fold lower activity than typical beta-emitter Lu-177 doses, because alpha particles are so much more potent per decay), patients who were refractory to beta-particle therapy nonetheless showed marked PSA declines and, in several cases, apparent complete biochemical and molecular imaging responses.
These results — published initially as case reports and subsequently in larger retrospective series — established Ac-225-PSMA-617 as a viable salvage option and motivated the ongoing prospective trials now underway for earlier lines of therapy.
The therapeutic window in targeted alpha therapy emerges from the interaction of three factors modeled throughout this simulation:
1. Target expression differential: PSMA is markedly overexpressed on prostate tumor cell membranes (often 100- to 1,000-fold over normal tissue) but essentially absent on hematopoietic stem and progenitor cells in bone marrow — so marrow cells rarely bind enough radioligand to be traversed at all.
2. Short range confines collateral damage: because the alpha range (50–90 μm) is only a few cell diameters, tumor cells that DO bind radioligand irradiate only their immediate neighbors — not distant marrow niches — unlike longer-range beta emitters which can deliver meaningful crossfire dose to marrow from activity bound elsewhere.
3. Single-hit lethality flips the Poisson math in the tumor's disfavor: because even 1–2 hits are often sufficient to kill a PSMA-high tumor cell (high-LET clustered DNA damage overwhelms repair), the same short-range, low-total-activity administration that leaves marrow essentially untouched (near-zero λ) can still drive tumor λ high enough that the Poisson zero-hit fraction e^(−λ) collapses toward zero.
The net effect: a therapy that is, hit-for-hit, far more destructive than beta or gamma radiation can nonetheless achieve a favorable — sometimes superior — therapeutic index, provided target expression is sufficiently differential and daughter redistribution (Stage 4) is managed.
The central engineering challenge for the field is now less about "does alpha-particle biology work" — the clinical PSA responses and imaging data say yes — and more about controlling daughter-nuclide redistribution and off-target salivary/renal toxicity well enough to move Ac-225-PSMA-617 from salvage therapy into earlier, potentially curative treatment lines.