European Nuclear Medicine Guide
The approved treatment regimen for [177Lu]Lu-PSMA-617 in non-compromised patients is 7.4 GBq (200 mCi) per cycle at 6-week (± 1 week) intervals for a maximum of 6 cycles. However, as there is significant inter-patient variability in the resulting dose exposure, higher treatment activities or more treatment cycles may be possible for several patients under an individualized dosimetry concept. Therefore, the evaluation of the radiation dose absorbed by organs at risk and tumour lesions following ¹⁷⁷Lu-based treatment is an important aspect to be considered in radioligand therapy. The procedural guidelines for absorbed dose calculation are outlined in a publication by the European Association of Nuclear Medicine (EANM) [20].
Lutetium-177 (¹⁷⁷Lu) is a radionuclide that undergoes β⁻decay with a half-life of 6.6 days and the emission of electrons with an average kinetic energy of 147 keV and a maximum energy of 497 keV, corresponding to penetration ranges of approximately 0.28 mm and 1.8 mm, respectively. Additionally, the decay of ¹⁷⁷Lu produces gamma radiations at 112.9 keV (~6%) and 208.4 keV (~11%), which enable multi-time point imaging of drug biodistribution via planar scintigraphy or SPECT/CT. Small-molecule PSMA ligands, such as PSMA-617 and PSMAI&T, have been radiolabelled with lutetium-177 for the treatment of metastatic prostate adenocarcinoma. Assessing the radiation dose delivered to tumours and surrounding healthy tissues is crucial for understanding the radiobiological effects, evaluating the risk of adverse events, refining treatment indication, and defining patient-specific optimized RLT protocols (administered activity per cycle, number of cycles, time intervals between cycles). The reported radiation dose estimates for [¹⁷⁷Lu]Lu-PSMA vary significantly due to the influence of various technical factors, such as image reconstruction settings, acquisition protocols, segmentation techniques, dose calculation models, lesion dimensions, and the application of partial-volume correction.
In [¹⁷⁷Lu]Lu-PSMA therapy, the main organs at risk are the salivary and lacrimal glands, as well as the bone marrow. The kidneys and liver are also considered at-risk organs. While radiobiological reactions in parotid gland and bone marrow can occur both early and late, kidney toxicity is generally delayed.
The primary factor influencing radiation dose to haematopoietic stem cells is the radiopharmaceutical's circulation within the bone marrow. Additional determinants of haematological toxicity include the extent of bone metastases and any prior myelotoxic chemotherapy or bone marrow irradiation. Haematological toxicity is the most frequent adverse event of ¹⁷⁷Lu therapy, with severe cases (Grade 3–4, typically thrombocytopenia) reported in about 10% of treated patients [21]. A recent systematic review and meta-analysis estimated a red-marrow absorbed dose of 0.03 Gy/GBq for both PSMA-617 and PSMA-I&T [22]. In analogy with ¹³¹I therapy, a bone-marrow absorbed dose of 2 Gy is generally considered the threshold for severe haematological toxicity, though further validation is needed for ¹⁷⁷Lu-based treatments [23].
Dosimetry studies have shown that radiation from the rest of the body contributes minimally to kidney dose—about 2% of the total—making self-irradiation the primary factor. Reported kidney absorbed doses for [¹⁷⁷Lu]Lu-PSMA therapy range from 0.4 to 0.8 Gy/GBq [21]. A systematic review and meta-analysis estimated average absorbed doses of 0.6 Gy/GBq for PSMA-617 and 0.7 Gy/GBq for PSMA-I&T [22]. The variability of reported data may be related to differences in patient characteristics, such as renal function and tumour burden, as well as methodological factors like imaging time points and dosimetry techniques. Additionally, there can be a significant intra-patient variability across therapy cycles [24,25]. Overall, clinically significant nephrotoxicity is uncommon with cumulative activities used in current practice [26,27]. This suggests that kidneys may tolerate higher absorbed doses than previously assumed, possibly due to non-uniform radiation exposure and relatively low dose rates.
The liver is generally not considered an organ at risk, but caution is warranted in patients with high liver tumour burden or prior hepatotoxic therapy in [¹⁷⁷Lu]Lu-PSMA therapy. It should be monitored in the case of concomitant treatments. Radiation-induced liver disease, which can develop weeks after exposure, presents as veno-occlusive disease affecting the central lobule and the small branches of the hepatic veins.
The highest accumulation of PSMA in normal tissues is in the salivary and lacrimal glands [24,25]. A key challenge in dosimetry for these organs is accurately estimating their mass due to their small size. In the salivary glands, radiation exposure can lead to xerostomia (reduced saliva production), a known side effect of [¹⁷⁷Lu]Lu-PSMA therapy. However, the threshold dose for salivary gland toxicity remains undefined. Data from EBRT suggest a low risk of toxicity when the mean absorbed dose to both parotid glands is below 10 Gy, with a proposed limit of 20 Gy [28]. Protective strategies, such as folic polyglutamate tablets or ice pack cooling, have been explored but remain investigational [29].
The reported salivary gland absorbed doses range from 0.5 to 1.9 Gy/GBq [20]. A recent meta-analysis estimated mean absorbed doses of 0.84 Gy/GBq and 0.74 Gy/GBq for the parotid and submandibular glands during PSMA-617 therapy, and 0.43 Gy/GBq and 0.64 Gy/GBq, respectively, for PSMA-I&T therapy [22].
Lacrimal glands are considered potential dose-limiting organs in [¹⁷⁷Lu]Lu-PSMA therapy [30], although no significant concerns regarding xerophthalmia have been raised. Based on EBRT, a dose constraint of 25 Gy for lacrimal glands has been suggested [31]. Reported absorbed doses for lacrimal glands range from 0.4 to 3.8 Gy/GBq [20], with pooled estimates of 1.58 Gy/GBq for PSMA-617 and 2.83 Gy/GBq for PSMA-I&T [22].
The average radiation dose delivered to tumour lesions during [177Lu]Lu-PSMA therapy ranges from approximately 1 to 8 Gy/GBq [20]. A recent systematic review and meta-analysis reported mean absorbed doses of 3.57 Gy/GBq and 4.19 Gy/GBq for bone and soft tissue metastases treated with PSMA-617, and 4.10 Gy/GBq and 2.94 Gy/GBq, respectively, for PSMA-I&T therapy [22]. Patients who responded to therapy (PSA responders) were reported to exhibit a significantly higher median absorbed dose (14 Gy) compared to non-responders (<10 Gy) when averaging the dose across all metastases [32].
While the same absorbed dose can result from different degrees of uptake and excretion rates, the rate at which the dose is delivered (absorbed dose rate) is different and influences the biological response and the effectiveness of cellular repair processes. [177Lu]Lu-PSMA therapy generally delivers lower absorbed dose rates than most radiotherapy techniques. Another key factor is treatment fractionation, as seen in [177Lu]Lu-PSMA therapy, which affects tissue recovery and repair. Furthermore, radiation exposure is often non-uniform in these therapies due to the short electron range of 177Lu, impacting the overall biological response.
A recent meta-analysis of dose estimates of [177Lu]Lu-PSMA radiopharmaceuticals conducted on 29 studies and 535 patients found that there was no significant difference in absorbed doses between [177Lu]Lu-PSMA-I&T and [177Lu]Lu-PSMA-617, despite a possible trend toward a higher kidney dose with [177Lu]Lu-PSMA-I&T and toward a higher tumour lesion dose with [177Lu]Lu-PSMA-617. Considering the methodological heterogeneity among the included studies, it remains unclear whether the higher tumour-to-kidney dose of PSMA-617 significantly affects clinical outcomes like progression-free or overall survival, and further standardization efforts are needed.
Ideally dosimetry should be performed after each administration of 177Lu-PSMA, since there is considerable intra-patient variability in the absorbed dose across treatment cycles. At least, an initial dose estimate should be obtained during the first treatment cycle. The calculation of the absorbed dose for organs at risk and tumour lesions in RLT requires measuring the activity distribution over time within the regions of interest and applying conversion factors (or Monte Carlo simulations) to estimate the absorbed dose based on cumulative activity. Several imaging protocols can be used to evaluate the activity distribution within the patient’s body, including planar, hybrid planar + SPECT, and 3D SPECT approaches. Planar imaging protocols involve acquiring sequential 2D whole-body images; while this method is fast and easy to implement, it has limitations in distinguishing overlapping structures. Hybrid imaging protocols combine planar imaging with a SPECT/CT scan at one selected time point: SPECT/CT imaging allows for activity quantification, while serial planar imaging allows determination of temporal activity changes. Finally, 3D SPECT/CT imaging protocols use multiple SPECT/CT scans at different time points to provide a fully detailed 3D activity distribution and are considered the gold standard. The imaging acquisition time points should be optimized based on the pharmacokinetics of radiopharmaceutical retention and excretion. Most studies on lutetium-based radiopharmaceuticals include at least three acquisitions: the first one within 1–4 hours post-administration, the second one 18–24 hours post-administration, and the last one 5–7 days post-administration. Additional scans, such as at 90–96 hours, could be optimal for kidney dosimetry, while even later time points could improve the accuracy of lesion time-activity curves. Finally, recent studies suggest that absorbed dose calculations can be performed using a single SPECT/CT scan when pharmacokinetics have been previously characterized in a patient population and the effective half-life in the target organ shows minimal inter-patient variability [33,34]. This method could be implemented during the first treatment cycle to facilitate the adoption of a dosimetric approach in clinical practice; alternatively, it may be applied in subsequent treatment cycles after a full dosimetric study has been conducted during the first cycle to obtain patient-specific biokinetic data.
SPECT has now become the preferred technique for dosimetry applications. Guidelines for quantitative SPECT/CT are outlined in MIRD pamphlet 23 [35] and the EANM/MIRD recommendations for quantitative 177Lu-SPECT [36]. The key practical steps for dosimetry in 177Lu-labelled compounds are detailed in the following sections.
Measurement of Therapeutic Activity
The exact activity administered to the patient can be determined by measuring the activity in the vial both before and after the administration of the radiopharmaceutical, using a proper activity calibrator [20].
Absolute Calibration of the SPECT/CT System
To determine the calibration factor for a SPECT/CT system—used to convert count rate (cps) into absolute activity (MBq)—a radioactive sample with a known activity level is prepared and imaged. The calibration factor is obtained by calculating the ratio of the measured count rate to the known activity. The same acquisition and reconstruction parameters should be used as those applied in patient imaging, and periodic evaluation of this factor is recommended to ensure system stability. One of the most widely adopted and accurate calibration methods involves the use of a large 3D phantom, similar to the calibration technique used for PET systems.
Image Acquisition and Reconstruction
It is recommended to use the same SPECT/CT system for the entirety of the dosimetry study. For lutetium-177 imaging, medium-energy collimators are recommended, with an energy window of 15–20% centered around the photopeak at 208 keV. Although lutetium-177 also has a photopeak at 113 keV, only the 208 keV peak is typically used for quantitative imaging with NaI(Tl) detectors, as it is less affected by scatter effects compared to the 113 keV peak. For systems using CZT crystals, the 208 keV peak may fall outside the acquisition energy range, so the 113 keV peak is used instead. The number of projections and the scan time per projection should be selected based on the patient's activity level, the system's sensitivity, the matrix size being used, and the noise propagation of the tomographic reconstruction. Typically, between 60 and 120 projection angles and 30–40 seconds per view are recommended, with the duration potentially reduced for early acquisitions. The auto-contour mode and a 128x128 or higher acquisition matrix should be used, with a zoom factor of 1. For image reconstruction, iterative algorithms such as ordered subset expectation maximization (OSEM) are recommended.
Partial Volume Effect Correction
The partial volume effect is due to the limited spatial resolution of SPECT systems and should be taken into account in dosimetry studies involving lutetium-177. This effect introduces an error in activity quantification that increases as the size of the lesion decreases. The most commonly used method to correct for this effect is the recovery coefficient, which involves calculating a corrective factor to apply to the measured activity from the tomographic images.
Image Segmentation
In order to perform dosimetric calculations, the volumes of interest (VOI) should be defined on a reference scan using manual, semi-automatic, or automatic tools (fixed or adaptive thresholding methods, gradient-based surface adaption, or convolutional neural networks). Subsequently, images taken at different time points are registered to the reference scan, with the segmentations being propagated across all images. Alternatively, segmentation can be performed on each individual acquired image. For organs at risk, VOI delineation based on CT information is preferable, and adjustments to VOI delineation should account for possible organ motion between the SPECT and CT acquisitions. For lesions, the use of contrast-enhanced CT images co-registered with SPECT images may be preferred for delineating VOIs.
Kinetic study of time-integrated activity
Given the activity in the source regions at various time points, the time-activity curve can be derived. To calculate the time-integrated activity, the data must be fitted to a theoretical curve, which is assumed to represent the biokinetics of the radiopharmaceutical. Typically, mono- or bi-exponential curves are used. The accuracy of the fit depends on how closely the actual activity follows the assumed pattern and on the temporal sampling. Ideally, a minimum of three data points should be used to define an exponential phase. If a precise theoretical model is unavailable, or if only a few time points are available, the trapezoidal method can be used [20].
Uncertainty evaluation
Each step of the dosimetry process is affected by uncertainty. The EANM has provided guidelines for calculating the uncertainty of the absorbed dose based on error propagation [37]. Performing uncertainty analysis helps to reduce errors and facilitates comparison between data collected across different centres.
To assess the absorbed dose in the haematopoietic bone marrow, it is necessary to calculate the time-integrated activity in the marrow itself as well as in the rest of the body. The absorbed dose in the marrow, due to the self-irradiation component, can be evaluated using either an image-based method or a blood-based method. The image-based method involves measuring the activity concentration in a bone region where the percentage of haematopoietic marrow is known, such as the L2–L4 vertebrae. The blood-based method involves collecting peripheral blood samples at various time points post- administration, which are then counted. Sampling time points should be selected to adequately represent both the early and later phases (e.g., 5 min, 60 min, 90 min, 120 min, 4 hours, 24 hours, and a later time point). Generally, a bi-exponential function provides a good representation of the time-activity curve (TIA). To derive the TIA for the whole body, measurements must be taken at various time points post-administration. The first measurement should be performed with a full bladder to represent the total activity in the patient's body at time zero. Planar imaging with a gamma camera (in both AP/PA projections to calculate the geometric mean of the counts) or an external probe can be used.