Background
Over the recent years, radionuclide therapy using Lu-177-Octreotate and Lu-177-PSMA-617 evolved as a promising approach for the treatment of metastasised and inoperable neuroendocrine tumours (NET) and metastasised, castration-resistant prostate cancer (mCRPC), respectively [
1‐
3]. The red or active bone marrow (BM) represents a main organ at risk in radionuclide therapy [
4‐
8]
. Bone marrow toxicity is particularly of concern in Lu-177-PSMA-617 therapy, as patients suffering from mCRPC often present with a high burden of bone metastases. The latter may cause pronounced activity accumulations in close proximity to the regions which potentially bear active marrow. At these locations, especially the
γ-component of the Lu-177 decay might lead to a significant photon cross-irradiation of the bone marrow [
9]. However, for Lu-177-Octreotate therapy, the bone marrow is also considered as an organ at risk, as patients with progressive cancer disease usually already underwent several pre-therapies such as external radiotherapy or chemotherapy [
3,
4]. These pre-therapies may have interfered with the haematological function of the bone marrow. Thus, bone marrow dosimetry is highly recommended in these patients for risk reduction of marrow toxicities and, at the same time, an as high as possible tumour absorbed dose [
10].
The total bone marrow absorbed dose is composed of different contributions originating from various activity source regions: (1) the bone marrow self-absorbed dose including the active bone marrow cells, the extracellular fluid, and the blood cells; (2) activity accumulations in the remaining skeleton composed of compact bone or fatty tissue (yellow or inactive marrow); (3) the cross-absorbed dose by major organs or tumours; and (4) the cross-irradiation coming from the remainder of the body (ROB; whole body minus specific or unspecific accumulations in the other source regions) [
11]. Each absorbed dose component requires a dedicated measurement procedure to derive its respective time-activity curve (TAC) and the source-specific time-integrated activity. The cumulated actvity-to-absorbed-dose conversion is usually performed via pre-calculated and standardised organ-level
S values [
11].
The appropriate data collection to accurately quantify the various possible source regions is challenging and leads to both a high clinical workload and long patient examination times, if bone marrow dosimetry shall be routinely performed in the clinic. For Lu-177-Octreotate or Lu-177-PSMA-617 therapy, the bone marrow absorbed dose from the major accumulating organs (
DBM ← organs), the ROB (
DBM ← ROB), and the blood (
DBM ← blood) can be determined from sequential quantitative SPECT images, sequential quantitative whole-body planar images, and multiple blood samples, respectively, in combination with the corresponding
S values [
8,
9,
11‐
13]. However, despite the high metastatic load which might be observed for NET and mCRPC patients, it is challenging to explicitly consider the bone marrow absorbed dose from activity accumulations in the tumours (
DBM ← tumours) via standardised and pre-calculated tumour-to-bone marrow
S values, as the latter intrinsically cannot consider the large inter-patient variability of the shape, size, and distribution of all lesions [
14].
Our institutional protocol determines the absorbed dose contribution from the ROB via sequential whole-body planar images [
11], which are acquired at three time points at 24, 48, and 72 h post-injection. In addition, we decided to derive organ (e.g. kidneys) and tumour absorbed doses from sequential quantitative SPECT measurements for improved organ and tumour dosimetry [
15‐
18]. However, full whole-body quantitative Lu-177 SPECT is still not commonly used in the clinic, implicating the need of consecutive planar and SPECT imaging at each time point to obtain both reliable bone marrow absorbed doses from the ROB and reliable organ or tumour absorbed doses [
19]. Particularly, the increased examination time in case of consecutive SPECT and whole-body planar imaging leads to an increased clinical workload and patient discomfort, as patients with progressive cancer disease may suffer from a bad health condition. Thus, the aim of this work was to derive a time-efficient, patient-friendly, and simplified bone marrow dosimetry protocol for clinical routine. Therefore, we investigated the possibility to reduce the number of image acquisitions from three whole-body planar and three quantitative SPECT scans (reference protocol (RP)) to a single whole-body planar acquisition while maintaining the institution’s usual sequential quantitative SPECT protocol (hybrid protocol (HP)). Further, we investigated the effect of this image reduction on the bone marrow absorbed dose from the ROB and on the total bone marrow dose (
DBM ← total), to prove whether the proposed hybrid protocol provides comparable absorbed dose estimates for both Lu-177-Octreotate and Lu-177-PSMA-617 therapy. For the determination of the total bone marrow absorbed dose, the energy depositions in the bone marrow due to activity accumulations in the ROB, blood, major organs, and tumours were considered. Furthermore, we determined the best-suited time point for this single whole-body planar image acquisition with respect to the time points available in our institutional protocol. All absorbed dose calculations are based on the organ-level
S values (e.g. whole ROB to bone marrow) [
11].
Discussion
Although all bone marrow absorbed dose estimates are well below the typically applied critical threshold of 2 Gy [
5] and no severe marrow toxicities have been observed for all investigated patients, bone marrow dosimetry is still a matter of interest. This is particularly true regarding the maximum absorbed dose that can be applied for patients with progressive cancer disease, who already underwent several pre-therapies. The absorbed dose estimates determined in this study are in good agreement with the findings of previous studies for both therapies [
5,
7,
8,
31].
According to the current clinical standard, an uncertainty of at least 10–20% has to be expected for the derived activity and absorbed dose values in case of quantitative Lu-177 SPECT imaging, and even greater values might be expected for planar imaging [
15‐
17,
32,
33]. Thus, the results presented in this study suggest that the application of a hybrid SPECT planar dosimetry approach based on late whole-body planar images allows for bone marrow dosimetry which is sufficiently reliable and applicable in clinical routine. In the case of Lu-177-Octreotate therapy of patients bearing NET and with regard to our institutional measurement protocol, the best time point for whole-body planar imaging was found to be approximately at 72 h p. i., with maximum deviations of the total bone marrow absorbed dose of 5% compared to the reference protocol. In patients with mCRPC receiving Lu-177-PSMA-617 therapy, the whole-body planar imaging time points 48 and 72 h p. i. provided comparable total bone marrow absorbed dose estimates with similar maximum differences of 6% to the reference-protocol-based full sequential whole-body planar approach. If five to ten Lu-177-PSMA-617 or Lu-177-Octreotate therapies are offered per week, the reduction of whole-body planar scans from three to one results in a reduction of examination time of 3.5 to 7 h per week. Simultaneously, the application of the proposed hybrid imaging protocol does not lead to an increased workload for the absorbed dose calculations.
The magnitude of deviations depends on the differences in the abdominal and whole-body washout and the positioning of the base point used for scaling of the mono-exponential pseudo-whole-body TAC. Analysis of the patient-specific reference-protocol-based and hybrid-protocol-based TAC parameters revealed that the use of a prolonged SPECT-based effective half-life is compensated by a lower y-axis intercept, if a later base point is selected. The use of a base point later than 72 h p. i. still has to be investigated; however, such a time point was unfortunately not available in our institutional measurement protocol. As expected, the deviations between the reference and hybrid protocol were larger for the bone marrow absorbed dose from the ROB compared with the total bone marrow absorbed dose, as the median ROB contribution to the total absorbed dose was found to be only 34% for Lu-177-Octreotate therapy and 45% for Lu-177-PSMA-617 therapy.
The appropriate whole-body planar imaging time point may have to be determined separately for each type of therapy. The degree of the deviations between abdominal and whole-body effective decay constants is driven by the disease- or therapy-specific retention in the organs and tumours and the corresponding typical tumour distribution. The mCRPC patients included in this study typically showed a larger tumour load compared with the NET patients, which was additionally strongly varying over the whole patient body. For most of the mCRPC patients (except P9) included in this study, the main metastatatic load was located in the torso, and consequently, the abdominal effective half-life was larger compared with the whole-body effective half-life. By contrast, patient P9 suffered from a strongly accumulating metastasis in the hip, leading to a comparatively larger whole-body effective half-life. The larger variability in the whole-body tumour distribution for mCRPC patients causes the observed larger spread in the differences between abdominal and whole-body effective half-lives. Consequently, a high tumour load outside the SPECT field of view might lead to an increased uncertainty of the proposed hybrid protocol, and this effect should be further investigated. As it was the case for most of the mCRPC patients, the investigated NET cases mainly presented with metastases in the torso, which lead to an increased retention of the radiopharmaceutical in the abdomen. However, due to the lower tumour load, the inter-patient variability in the abdominal and whole-body effective half-lives was reduced for the NET patients under study.
The change from one-bed abdominal SPECT imaging to the imaging of two or more beds could principally improve the proposed hybrid protocol for bone marrow dosimetry, as an enlarged acquisition area will lead to a more realistic estimate of the whole-body effective half-life. Furthermore, the introduction of fast multi-bed SPECT imaging in the clinical routine would be beneficial for a robust tumour and organ dosimetry over a larger part of the patient body [
15‐
18]. Attempts to introduce fast whole-body SPECT imaging into the clinic already exist [
34]. However, the effect of a reduction of scan time on absorbed dose estimates for Lu-177 therapy still has to be evaluated.
The accuracy of dosimetry based on standardised organ-level
S values is limited, as such
S values are inherently not capable to fully consider the patient-specific full 3D functional and anatomical characteristics. The latter fact remains true, even if a scaling of the
S values to the specific anatomical conditions is applied [
6,
14,
35‐
37]. For Lu-177, the ROB cross-absorbed dose of the bone marrow is mainly driven by the long-range photon component, which is more sensitive to the anatomy than the locally deposited beta absorbed dose. In a previous study based on Monte Carlo simulations, deviations of the order of up to 100% were observed, if photon cross-absorbed doses were calculated based on standardised
S values [
38]. Furthermore,
S values are determined based on the assumption of homogeneous activity accumulation. However, the activity accumulation in the ROB with the inclusion of tumours is highly heterogeneous with the degree of heterogeneity being caused by both tumour load and distribution. With regard to both aspects the limited consideration of the patient-specific functional and anatomical characteristics, the reliablity of the proposed hybrid protocol can be well accepted in the framework of organ-level
S values. Moreover, it should also be noted that the exact bone marrow distribution of each patient is a priori unknown due to the heterogeneous micro-structure of the bone marrow and its pathologically highly variable distribution, which both lead to a highly unspecified target for bone marrow dosimetry [
23]. Particularly, for mCRPC patients with a high bone tumour load, a displacement of active bone marrow from highly metastasised to tumour-free skeletal sites is possible [
39].
Our decision to include all tumours in the ROB represents a simplified approach for clinical routine bone marrow dosimetry. On the one hand, this approach is more practical, as in case of a high bone tumour load, a manual determination of the time-integrated activity is not feasible for each tumour lesion in an acceptable time. On the other hand, even if a semi-automatic or automatic tumour segmentation is available, tumour-to-bone marrow
S values for both individual tumours and the total tumour distribution are not available, as tumours are quite variable in shape, size, and position, and the pre-calculation of all possible
S values is not possible. Thus, at this point, a more simplified approach was chosen, which considered all tumours at once within the ROB compartment. The approximation to use the
S value of the compartment in which the tumours are located to estimate the bone marrow absorbed dose from lesions has also been applied in previous studies [
5]. An alternative way, proposed by Svensson et al. for bone marrow dosimetry for Lu-177-Octreotate therapy, differentiates the activity distribution in the patient body in low- and high-activity regions (background vs. main accumulating organs and tumours) with separate
S values applied to each of both compartments [
31]. The resulting bone marrow absorbed doses correlated with the change of blood parameters and were found to be in a similar range compared to previously published results. Monte Carlo studies may help in further understanding the effect of such simplifying assumptions for bone marrow dosimetry.