Transarterial microsphere dosimetry — MAA lung-shunt mapping, hepatic delivery, MIRD partition modeling & post-treatment verification
Before a single Y-90 microsphere is ever infused, every radioembolization patient undergoes a rehearsal: a hepatic angiogram followed by injection of technetium-99m macroaggregated albumin (MAA) particles through the same catheter position planned for treatment. MAA particles (10–90 µm) are similar enough in size to Y-90 microspheres to approximate their flow distribution, letting the team quantify the single most dangerous failure mode of TARE — arteriovenous shunting of radioactive particles into the pulmonary circulation.
Hepatocellular carcinoma and hepatic metastases are hypervascular — they recruit dense, chaotic arterial networks to sustain rapid growth. Some of these tumors, particularly large or centrally located HCCs, develop direct arteriovenous shunts connecting the hepatic arterial supply to the hepatic or systemic venous system, which drains to the right heart and then the pulmonary capillary bed.
Any particle small enough to pass through these shunts — whether Tc-99m-MAA today or a Y-90 microsphere on treatment day — will lodge in the first capillary bed it reaches downstream: the lungs. Because the lungs have essentially zero tolerance for the kind of concentrated beta radiation microspheres deliver (unlike the liver, which can regenerate around focal radiation injury), even a modest shunt fraction can translate into a clinically significant pulmonary radiation dose if left uncorrected.
The MAA scan is therefore not a formality — it is the primary safety gate of the entire procedure. A measured lung shunt fraction (LSF) above roughly 20% is an absolute contraindication to treatment with standard dosing; between 10–20% requires a proportional reduction in administered activity; below 10% is considered low risk.
The MAA scan also identifies extrahepatic shunting to the stomach, duodenum, gallbladder, or pancreas via variant hepatic artery anatomy — a second major safety concern, since non-target embolization of the GI tract can cause radiation-induced ulceration.
After MAA injection, a gamma camera acquires planar anterior and posterior images of the chest and abdomen (or SPECT/CT for more precise volumetric quantification). The lung shunt fraction is calculated as:
LSF (%) = [Lung counts / (Lung counts + Liver counts)] × 100
Geometric mean counts (√(anterior×posterior)) correct for attenuation depth differences between anterior and posterior projections. SPECT/CT increasingly supersedes planar imaging because it separates true lung uptake from soft-tissue and chest-wall scatter, avoiding overestimation.
Once LSF is known, the administered activity for the actual Y-90 treatment is reduced according to a standard formula so that the predicted lung dose stays below the single-administration and cumulative safety ceilings — the calculation explored quantitatively in Stage 4 of this simulation.
On treatment day, an interventional radiologist advances a microcatheter — often just 1.5–2.7 French — through the femoral or radial artery, into the aorta, celiac trunk, and finally superselectively into the segmental or subsegmental hepatic artery branch feeding the tumor. Y-90 microspheres are then infused slowly under continuous fluoroscopic roadmap guidance, exploiting a simple physiological fact: tumors get almost all their blood supply from the hepatic artery, while normal liver parenchyma draws roughly 75% of its supply from the portal vein instead.
Two FDA-approved / CE-marked microsphere products dominate clinical practice, and they differ fundamentally in how activity is packaged:
• Resin microspheres (SIR-Spheres, Sirtex): each sphere carries a low activity (~50 Bq), so a typical 3 GBq treatment requires 40–80 million spheres. The large sphere count achieves better spatial coverage of the tumor vascular bed but also means more emboli reach normal parenchyma, generally limiting resin dosing to a whole-liver-tolerance-based model.
• Glass microspheres (TheraSphere, Boston Scientific/BTG): each sphere carries far higher activity (~2,500 Bq), so the same total activity is delivered in only 1–8 million spheres. Fewer particles means less embolic effect (glass microspheres are considered minimally embolic), permitting much higher tumor-absorbed doses — the basis of "radiation segmentectomy," where doses of 400–1,000+ Gy are delivered to small, well-defined tumor volumes with acceptable normal-tissue sparing.
Both platforms rely on the same underlying physics: Y-90 is a pure high-energy beta emitter (Emax 2.28 MeV, mean 0.9367 MeV) with a physical half-life of 64.1 hours and a mean tissue penetration of only ~2.5 mm (max ~11 mm) — short enough that virtually all dose is deposited locally, near where each microsphere lodges.
Because tissue penetration is only millimeters, the spatial pattern of microsphere deposition IS the dose map — unlike external beam radiotherapy, there is no way to "smear" the dose evenly after the fact. Catheter position and injection technique directly determine outcome.
Normal hepatocytes receive roughly 70–80% of their blood supply from the low-pressure portal vein and only 20–30% from the hepatic artery. Malignant hepatic tumors — whether primary HCC or metastatic deposits — undergo neoangiogenesis and lose portal venous supply almost entirely, becoming nearly 90–100% arterially fed.
Because microspheres are injected into the arterial tree and are too large (20–60 µm) to pass through capillaries without embolizing, this vascular asymmetry becomes a delivery advantage: the same injected activity preferentially lodges in tumor arterioles and precapillary vessels, achieving tumor-to-normal-liver activity concentration ratios (T/N) that commonly range from 2:1 to more than 10:1 in favorable, hypervascular tumors.
Superselective catheter positioning compounds this effect further — placing the catheter tip as close as possible to the tumor-feeding branch (segmental or subsegmental, rather than lobar) reduces reflux of microspheres into adjacent normal segments and is the technical basis of radiation segmentectomy and radiation lobectomy techniques.
Knowing how much activity was infused is not the same as knowing how much radiation dose the tumor and liver actually absorbed. The Medical Internal Radiation Dose (MIRD) committee formalism, adapted into a two- or three-compartment "partition model" for radioembolization, converts injected Y-90 activity and the measured tumor-to-normal (T/N) uptake ratio into compartment-specific absorbed doses — the calculation engine used clinically to plan and verify every TARE treatment.
The classic MIRD local-deposition dose equation for Y-90, assuming all beta energy is absorbed within the tissue compartment where the microsphere lodges (a reasonable approximation given the ~2.5 mm mean path length), is:
D (Gy) = 49.67 × A (GBq) / M (kg)
where A is the activity residing in a compartment of mass M. The partition model splits the total non-lung-shunted activity (A_liver) between tumor and normal liver compartments using the measured T/N uptake ratio (R) and each compartment's estimated mass or volume fraction (f_tumor):
A_tumor = A_liver × (R × f_tumor) / (R × f_tumor + (1 − f_tumor)) A_normal = A_liver − A_tumor
D_tumor = 49.67 × A_tumor / M_tumor D_normal = 49.67 × A_normal / M_normal
Tumor and normal liver volumes are typically contoured from contrast CT or MRI, and the T/N ratio is estimated from the pretreatment MAA SPECT/CT (or occasionally from Y-90 PET after a small test dose). The higher the T/N ratio, the more of the injected activity is mathematically credited to the tumor compartment — and the higher the achievable tumor dose for a given normal-liver dose constraint.
This simulation's "Tumor:Normal Perfusion Ratio" slider directly represents R in the partition equation above — dragging it higher concentrates more of the fixed administered activity into the (much smaller) tumor compartment, driving tumor dose up steeply while normal liver dose falls.
Three broad dosimetric strategies are used clinically, differing mainly in spatial granularity:
• Whole-liver (partition/BSA) dosing: treats the liver as one or two compartments (tumor + normal), historically also using body-surface-area (BSA) empirical formulas for resin microspheres. Simple but averages over tumor heterogeneity.
• Voxel-based (personalized) dosimetry: uses the 3D SPECT/CT or PET/CT activity distribution directly, calculating dose voxel-by-voxel via local deposition or dose-point-kernel convolution. Captures cold and hot spots invisible to compartment averaging, and is increasingly standard of care, especially for glass microspheres and radiation segmentectomy planning.
• Boosted / ablative dosimetry (radiation segmentectomy): deliberately targets very high focal tumor doses (>200 Gy, often 400–1,000+ Gy) in a small liver volume using superselective glass microsphere infusion, aiming for a tumoricidal, near-ablative effect analogous to thermal ablation — while keeping the treated normal liver volume small enough that even a high local dose is tolerated.
Across all approaches, the guiding constraint remains the same: keep mean normal (non-tumorous) liver dose below roughly 70 Gy to avoid radioembolization-induced liver disease (REILD), while pushing tumor dose as high as the T/N ratio and anatomy allow.
Every Y-90 treatment plan is ultimately bounded by a single non-negotiable safety constraint: the predicted radiation dose to the lungs. Because the measured lung shunt fraction directly determines what fraction of the administered activity will bypass the liver entirely and lodge in pulmonary capillaries, this single number can force a reduction in total administered activity — even when tumor dosimetry alone would justify treating with more.
Once the lung shunt fraction (LSF) is measured from the pretreatment MAA scan, predicted lung dose is calculated the same way as tumor and liver dose — activity in the compartment divided by compartment mass, scaled by the Y-90 dose constant:
D_lung (Gy) = 49.67 × [A_total × (LSF / 100)] / M_lung
where M_lung is typically assumed to be ~1.0 kg (combined mass of both lungs, a standard reference value; actual patient lung mass can be individualized from CT if needed). Because this equation is linear in administered activity, the maximum activity that can safely be given is capped directly by the LSF:
A_max (from lung constraint) = (30 Gy × M_lung) / (49.67 × LSF/100)
When this activity ceiling is lower than what would otherwise be given to satisfy tumor dosimetry goals, the lung constraint wins — treatment proceeds at reduced activity, is split into multiple staged sessions, or (if LSF exceeds roughly 20%) is abandoned in favor of an alternative locoregional or systemic therapy.
Radiation pneumonitis is a rare but serious complication when lung dose limits are exceeded, typically presenting weeks to months after treatment with cough, dyspnea, and characteristic imaging changes. The 30 Gy single-session / 50 Gy cumulative limits were extrapolated conservatively from decades of external-beam whole-lung irradiation tolerance data, since prospective Y-90-specific pneumonitis dose-response data remain limited.
Many patients with bilobar or multifocal disease undergo more than one Y-90 treatment session, whether staged lobar treatments separated by 4–8 weeks to allow contralateral liver hypertrophy, or repeat treatments for tumor recurrence. Because lung dose is cumulative and lung tissue does not meaningfully regenerate the way liver parenchyma can, every treatment planning session must account for lung dose already delivered in prior sessions, not just the dose from the treatment being planned.
Institutions maintain a running cumulative lung dose record for each patient, and this cumulative value — not just the single-session dose — is checked against the 50 Gy lifetime ceiling before authorizing any subsequent treatment. This is directly analogous to cumulative dose tracking in conventional external-beam radiotherapy re-irradiation planning.
Treatment does not end when the last microsphere leaves the catheter. Post-treatment imaging — either bremsstrahlung SPECT/CT or, increasingly, Y-90 PET/CT — captures where the microspheres actually went, allowing the care team to confirm the planned dose distribution was achieved, identify any unexpected extrahepatic deposition, and generate a verified voxel dosimetry map that feeds directly into outcome prediction and, if needed, retreatment planning.
Y-90 is not a conventional gamma- or positron-emitting isotope, so imaging it directly requires exploiting secondary physical phenomena:
• Bremsstrahlung SPECT/CT: as Y-90's high-energy beta particles decelerate while traveling through tissue, they emit a continuous spectrum of "braking radiation" (bremsstrahlung) photons. A gamma camera can detect these photons and reconstruct a 3D SPECT image, though image quality is inherently limited by the continuous (non-monoenergetic) photon spectrum and significant scatter, making bremsstrahlung SPECT more qualitative than precisely quantitative.
• Y-90 PET/CT: although more than 99.99% of Y-90 decays are pure beta emission, a very small fraction — approximately 32 in every 1 million decays — proceeds via internal pair production, generating a positron that subsequently annihilates and produces the back-to-back 511 keV photons that PET detectors are built to capture. Despite this vanishingly small branching ratio, modern PET/CT scanners have enough sensitivity to reconstruct high-quality, quantitatively accurate 3D activity maps, and Y-90 PET has become the preferred modality at high-volume centers wherever available.
Because Y-90 PET provides substantially better spatial resolution and quantitative accuracy than bremsstrahlung SPECT, it enables true voxel-level absorbed dose maps — allowing clinicians to retrospectively verify whether the tumor actually received the >200 Gy ablative target, not just estimate it from the pretreatment MAA surrogate.
Post-treatment quantitative imaging serves several concrete clinical purposes beyond simple documentation:
• Dose-response correlation: comparing the verified tumor absorbed dose to subsequent imaging response (mRECIST, RECIST 1.1) has established clear dose-response relationships, particularly for HCC — tumors receiving >200 Gy (glass microspheres) show significantly higher complete response rates than those receiving lower doses, refining future dose-targets.
• Detecting non-target deposition: confirms whether any microspheres reached extrahepatic sites (stomach, duodenum, gallbladder, lung) despite pretreatment MAA screening — anatomy and flow dynamics on treatment day are not always identical to the planning angiogram.
• Retreatment planning: for patients requiring a second Y-90 session for residual or recurrent disease, verified cumulative liver and lung doses from the prior treatment are essential inputs to ensure the new plan does not exceed cumulative safety limits.
• Quality assurance and outcome research: aggregated verified dosimetry across many patients continues to refine the dose thresholds (70 Gy normal liver, >200 Gy ablative tumor target, 30/50 Gy lung limits) that the entire field relies on — dosimetry improves iteratively as more verified data accumulates.