Organ-level absorbed dose simulator for [177Lu]Lu-PSMA-617 radioligand therapy — MIRD residence-time dosimetry from serial SPECT/CT
[177Lu]Lu-PSMA-617 pairs a small-molecule urea-based PSMA ligand with the beta-emitting radionuclide lutetium-177. After intravenous infusion, the ligand binds prostate-specific membrane antigen — a cell-surface glutamate carboxypeptidase overexpressed 100- to 1000-fold on prostate cancer cells relative to normal tissue — and is internalized, delivering ionizing radiation directly to tumor cells while sparing most healthy tissue. The same PSMA expression pattern in renal proximal tubules and secretory glands, however, creates unavoidable off-target absorbed dose that ultimately limits how much activity can be given.
Prostate-specific membrane antigen (PSMA, also known as glutamate carboxypeptidase II or folate hydrolase 1) is a type II transmembrane glycoprotein expressed at low levels in normal prostate epithelium, proximal renal tubules, salivary and lacrimal glands, and duodenal mucosa. In prostate adenocarcinoma, PSMA expression is upregulated 100- to 1000-fold and increases further with androgen deprivation therapy and castration resistance — the exact clinical population targeted by radioligand therapy.
PSMA-617 is a low-molecular-weight urea-based inhibitor (Glu-urea-Lys motif) linked through a naphthyl-chelator spacer to DOTA, which chelates the therapeutic radiometal. This design gives PSMA-617 rapid tumor uptake, high tumor-to-background contrast within hours, and predominantly renal clearance of the unbound fraction — properties optimized over several generations of PSMA ligands (from the earlier PSMA-11 imaging agent used with Ga-68 for PET, to I-131-MIP-1095, to the current Lu-177/Ac-225-PSMA-617 and PSMA-I&T therapeutic agents).
Once bound, the ligand-PSMA complex is internalized by receptor-mediated endocytosis, trapping the radiometal intracellularly for an extended residence time — this is what makes targeted radioligand therapy dosimetrically favorable compared to a freely diffusible tracer: activity concentrates and stays concentrated in PSMA-expressing cells while background tissue clears the compound within hours via the kidneys.
Lu-177 was chosen deliberately over higher-energy beta emitters: its short mean tissue range of 0.67 mm (versus several millimeters for Y-90) concentrates dose within small tumor deposits and limits crossfire damage to adjacent normal parenchyma, while its concurrent low-abundance 208 keV and 113 keV gamma emissions are ideal for simultaneous quantitative SPECT imaging — the same radionuclide that treats the patient is used to image and dose the therapy, a "theranostic" pairing enabled by the same molecule being labeled with Ga-68 or F-18 for pre-therapy PET selection.
Absorbed dose cannot be calculated from a single scan. Because different organs clear Lu-177-PSMA at different rates, dosimetry requires sampling the activity in each organ at multiple timepoints after infusion and fitting a time-activity curve (TAC) — typically a bi-exponential rise-then-decay — that captures both the early uptake phase and the terminal clearance half-life unique to that tissue.
Converting a raw SPECT projection dataset into an absolute activity value (MBq) inside a volume-of-interest requires several corrections applied during iterative reconstruction (typically OSEM — ordered-subset expectation maximization):
• Attenuation correction: using the co-registered CT for tissue-density-based photon attenuation maps, since 208 keV and 113 keV photons are substantially attenuated by soft tissue and bone. • Scatter correction: triple-energy-window or Monte Carlo-based scatter estimation, since Compton-scattered photons misplace counts and would otherwise overestimate deep-organ activity. • Collimator-detector response modeling: correcting for the finite spatial resolution of the medium-energy collimator used for Lu-177's intermediate-energy photons. • System calibration factor: a phantom-derived counts-to-MBq conversion factor, validated against a dose calibrator, that converts reconstructed voxel counts into absolute becquerels.
Organs and lesions are then segmented — kidneys and salivary glands by CT-based contouring, tumor lesions by SPECT-avid volume thresholding — and the total activity within each volume-of-interest is recorded at each timepoint, producing one point per organ per timepoint on the time-activity curve.
The resulting four-timepoint TAC is fit to a mono- or bi-exponential function: an initial uptake phase (minutes to a few hours, often not sampled directly and instead extrapolated from the injected activity) followed by a clearance phase whose effective half-life (combining biological clearance and physical decay) determines how long activity — and therefore radiation dose — continues to accumulate in that organ.
Because a full 4-timepoint imaging protocol is logistically demanding (patients must return at 24h, 48h and again at 168h/day 7), several groups have validated abbreviated single- or two-timepoint dosimetry protocols using population-based clearance curves — trading some individual precision for practicality, since imaging every cycle of every patient at 4 timepoints is rarely feasible outside dosimetry-focused trials like TheraP.
The Medical Internal Radiation Dose (MIRD) formalism, developed by the Society of Nuclear Medicine since the 1960s, is the standard method for converting a measured biodistribution into an absorbed radiation dose. It separates the calculation into two independent pieces: a patient-specific time-integrated activity term describing how much activity was present and for how long, and a physics-only S-value term describing how efficiently that activity deposits energy in a target organ.
The MIRD absorbed dose to a target organ rT from a source organ rS is:
D(rT) = Ã(rS) × S(rT ← rS)
where Ã(rS), the time-integrated activity coefficient (TIAC, historically called the "residence time" when normalized to administered activity), is the area under the organ's time-activity curve:
Ã(rS) = ∫₀^∞ A(rS, t) dt
In practice this integral is computed by trapezoidal numerical integration across the measured 4h/24h/48h/168h datapoints, followed by analytic extrapolation of the terminal exponential tail out to infinity (using the fitted clearance half-life) — since activity does not simply stop at the last scan, it continues decaying (and depositing dose) for days afterward. Missing the terminal tail is one of the most common sources of dosimetry error.
S(rT ← rS), the S-value, is a purely physical quantity: the absorbed dose per unit cumulated activity in the source region, tabulated from Monte Carlo radiation transport simulations through a reference anatomical phantom (or, increasingly, patient-specific CT-based voxel S-values). For a compact organ like the kidney receiving nearly all its dose from activity within itself (a "self-dose" geometry, since Lu-177 beta particles travel under 2 mm), S(kidney←kidney) dominates and cross-organ terms are small.
Software packages such as OLINDA/EXM (Organ Level INternal Dose Assessment) and SAAM II implement this pipeline end-to-end: TAC curve-fitting, TIAC integration, S-value lookup or voxel-kernel convolution, and final organ-dose reporting in Gy — turning a set of SPECT/CT activity measurements into the clinically actionable number that drives dose-limiting toxicity decisions.
Published Lu-177-PSMA-617 dosimetry cohorts converge on strikingly consistent numbers despite different centers and protocols: kidney doses of roughly 0.6–0.9 Gy per GBq administered, salivary gland doses of roughly 0.5–1.3 Gy per GBq, and — the therapeutically important contrast — tumor lesion doses that are far higher and far more variable, commonly 1 to over 100 Gy per GBq depending on lesion PSMA expression, size, and blood flow.
Tumor response is not what limits how much [177Lu]Lu-PSMA-617 a patient can receive — normal-organ toxicity is. The kidneys, which reabsorb the PSMA ligand's unbound fraction in the proximal tubules, and the bone marrow, irradiated by circulating blood activity and marrow PSMA expression, are the two dose-limiting organs that oncologists monitor cycle to cycle to decide whether to continue, reduce, or stop therapy.
Because radioligand therapy delivers dose at a continuously decreasing rate (protracted, low-dose-rate irradiation) rather than the fractionated high-dose-rate pattern of external-beam radiotherapy, absorbed dose in Gray alone is not directly comparable to the external-beam tolerance doses nephrologists and radiation oncologists are used to. The biological effective dose (BED) model corrects for this using the linear-quadratic formalism:
BED = D × [1 + (D/n)/(α/β) × (dose-rate correction factor)]
For kidneys, historically modeled as a late-responding tissue with a low α/β ratio (~2.6 Gy, consistent with external-beam renal tolerance studies from Emami et al. and subsequent QUANTEC updates), the classic renal tolerance limit translates to a BED ceiling of approximately 23 Gy for the whole course of therapy — the number oncologists track cumulatively across every cycle a patient receives. Some groups argue this externally-derived threshold is conservative for the very different dose-rate and spatial pattern of radioligand therapy, and a minority of centers use a higher working threshold (~28 Gy) — the exact number remains an area of active clinical debate.
Bone marrow dosimetry is handled differently: because marrow cannot be directly imaged with the same precision as a solid organ, blood-based surrogate dosimetry is standard — sampling venous blood activity at the same imaging timepoints and using a marrow-to-blood activity concentration ratio (assumed ≈1 for red marrow, since Lu-177-PSMA has minimal direct marrow uptake in patients without diffuse bone marrow infiltration) to estimate cumulative marrow dose. The empirical ceiling of roughly 2 Gy cumulative red marrow dose is drawn from external-beam and prior radionuclide-therapy experience with reversible versus irreversible marrow suppression.
In practice, dose-limiting toxicity from [177Lu]Lu-PSMA-617 is uncommon at standard fixed dosing: in the pivotal VISION trial, grade ≥3 renal adverse events occurred in under 3% of patients, and grade ≥3 hematologic toxicity (anemia, thrombocytopenia) in roughly 10–13% — reflecting that fixed 7.4 GBq q6-week dosing keeps most patients well within both the kidney BED and marrow dose ceilings, though patients with impaired baseline renal function or extensive bone marrow tumor infiltration are at meaningfully higher individual risk.
Two landmark trials defined how [177Lu]Lu-PSMA-617 is dosed today. VISION (Sartor et al., NEJM 2021) established efficacy and safety using a simple fixed-activity regimen — 7.4 GBq every 6 weeks for up to 6 cycles — while TheraP (Hofman et al., Lancet 2021) and subsequent dosimetry-guided protocols explored adjusting each cycle's activity to an individual patient's measured kidney dose, aiming to safely push more dose into high-avidity, favorable-clearance patients while protecting those with compromised renal reserve.
A personalized dosing strategy uses the absorbed dose measured after each cycle — primarily the kidney dose, secondarily marrow dose — as feedback for planning the next cycle's administered activity:
1. Baseline / Cycle 1: administer a standard starting activity (e.g., 7.4 GBq) alongside full 4-timepoint SPECT/CT dosimetry. 2. Post-cycle dose calculation: compute cycle-1 kidney and marrow absorbed dose via the MIRD pipeline (Stage 3), and running cumulative BED against the ~23 Gy kidney ceiling. 3. Adjustment: if projected cumulative kidney BED across a full 6-cycle course would exceed the safety ceiling, reduce subsequent cycle activity (or extend the interval between cycles); if the patient shows rapid renal clearance and low kidney dose per GBq, activity can be maintained or, in some protocols, escalated to increase tumor-absorbed dose. 4. Iterate: repeat dosimetry (abbreviated 1–2 timepoint protocols are often used for cycles 2–6 to reduce patient burden) after each subsequent cycle, refining the plan. 5. Stop criteria: therapy is discontinued at radiographic/biochemical progression, unacceptable toxicity, or on reaching the cumulative kidney/marrow dose ceiling — whichever comes first — typically after 4–6 cycles.
This approach shifts radioligand therapy from a one-size-fits-all drug regimen toward genuine internal radiotherapy treatment planning, conceptually parallel to how external-beam radiotherapy plans are individualized to organ-at-risk dose-volume constraints — the key difference being that here the "plan" is executed one infusion at a time, with the previous cycle's measured outcome directly informing the next.
The TheraP trial randomized patients to Lu-177-PSMA-617 with dosimetry-based individualized activity versus cabazitaxel chemotherapy in metastatic castration-resistant prostate cancer, and required a baseline PSMA PET/CT plus FDG PET/CT to confirm PSMA-avid, FDG-discordant-negative disease before enrollment — establishing that combining PET-based patient selection with dosimetry-guided dosing improves the therapeutic index of radioligand therapy beyond what fixed dosing alone can achieve.