Introduction
Prostate cancer accounts for 20% of new cancers diagnosed every year. With a mortality rate of 10%, it is one of the most common causes of death worldwide [
1‐
3]. Treatment options include local radiotherapy, surgery, or systemic treatments such as hormonal therapy or chemotherapy. For metastasized disease, prostate-specific membrane antigen (PSMA), a protein that is overexpressed in most prostate cancer cells [
4‐
6], can also be used as a target for radionuclide therapy. In end-stage castrate-resistant metastatic prostate cancer (mCRPC) patients, [
177Lu]Lu-PSMA-617 [
7‐
19] and/or [
225Ac]Ac-PSMA-617 [
20‐
24] showed remarkable responses with, in general, a mild toxicity profile. Therefore, [
177Lu]Lu-PSMA is now also translated to earlier stages such as to hormone-sensitive prostate cancer (HSPC) with encouraging results [
25].
At present, high tumor uptake of [
68Ga]Ga-PSMA-11, [
18F]DCFPyL, or [
18F]PSMA-1007 on positron emission tomography (PET) imaging is mandatory for PSMA radioligand therapy [
26‐
30]. In some studies, PET standardized uptake value (SUV) on [
68Ga]Ga-PSMA-PET has been shown to correlate with absorbed (radiation) dose in lesions and salivary glands in both mCRPC [
31] and mHSPC patients [
32], while other studies did not find this correlation [
9,
18,
33]. Patients selected based on PET lesion SUV show a response rate of only 40–60% [
7‐
19]. A potential improvement of patient selection has been suggested by the group of Hofman and colleagues by using both FDG-positive tumor volume and mean intensity of PSMA-avid tumor uptake [
34]. An actual dose estimation based on the pre-therapeutic [
68Ga]Ga-PSMA-PET could provide more accurate information on expected treatment response, since the calculations of the absorbed doses take into account tracer kinetics and are intrinsically corrected for factors such as partial volume effect occurring in particular for small tumors. Therefore, we hypothesize that patient selection could be improved if the pre-therapeutic [
68Ga]Ga-PSMA-PET data were used to predict absorbed doses for the subsequent [
177Lu]Lu-PSMA treatment.
In addition, pre-therapeutic evaluation of risk for organ toxicity is important in order to design a patient-specific treatment plan. It can prevent clinicians from exceeding threshold doses for radiation-related toxicity and it can potentially be used to apply higher therapeutic activities. To this end, mean SUV of organs on PET are not a suitable parameter to predict organ absorbed dose, mainly due to heterogeneity in the PET signal. However, modeling the organ absorbed doses based on the pre-therapeutic [68Ga]Ga-PSMA-PET imaging could provide a tool to assess organ toxicity after treatment.
Similar studies have been carried out using PET/CT imaging for an absorbed dose prediction after radionuclide therapy, mainly using
124I for prediction of organ-absorbed dose after
131I-therapy in thyroid cancer patients [
35‐
39]. This methodology is based on the assumption that tracer kinetics for
124I and
131I are comparable, and cumulated activity derived from multi time point
124I-PET/CT can be translated to
131I-cumulated activity, thereby predicting organ absorbed dose after therapy. It is suggested that this approach can be used to design patient-specific treatment by respecting the organ threshold dose for radiation toxicity effects [
39,
40].
To date, the use of [68Ga]Ga-PSMA-PET for an absorbed dose estimation of [177Lu]Lu-PSMA treatment has not been reported in the literature. This study aims to fill this gap by investigating the predictive value of a single time point pre-therapeutic [68Ga]Ga-PSMA-PET for absorbed dose after [177Lu]Lu-PSMA therapy in organs (kidneys, salivary glands, and liver) and tumor lesions. It relies on tissue-specific radioligand kinetics that will be derived from therapeutic imaging data with [177Lu]Lu-PSMA-SPECT, in combination with tracer uptake of a single time point pre-therapeutic [68Ga]Ga-PSMA-PET. The predicted absorbed doses were compared to actually delivered absorbed doses in therapy.
Discussion
This study evaluated the possibility to use a single time point [
68Ga]Ga-PSMA-PET scan to predict the therapeutic absorbed dose in organs at risk and lesions for a subsequent treatment with [
177Lu]Lu-PSMA therapy. Tracer kinetics is a crucial part in these predictive absorbed dose calculations, which determines the shape of the uptake time-activity curve and thus the cumulated activity. In this study, these kinetics were determined as the mean kinetics of 10 patients based on the SPECT data. This means that this approach is based on two main assumptions: firstly, the typical shape of the uptake curves for organs and lesions for the different patients are nearly identical; therefore, it is justified to use general tissue-specific kinetics in the PET prediction model. Secondly, the different tracers used in PET and SPECT imaging (PSMA-11 and PSMA-617, respectively) have similar kinetics; therefore, the kinetics found for PSMA-617 on SPECT can be used to project the expected kinetics of PSMA-11 on PET. Multiple studies investigated biodistribution and kinetics for PSMA-11 [
48‐
51] and PSMA-617 [
52] and showed indeed similar kinetic behavior [
53].
Prediction of absorbed dose for lesions showed a large variation in kinetics between patients both during the uptake phase (SD of 50%) and the excretion phase (SD of 30%), indicating that the first assumption of identical lesion kinetics between patients does not hold. Therefore, tumor lesion dosimetry using a single time [
68Ga]Ga-PSMA-PET was challenging. Earlier studies found that different tracer kinetics could be the result of different lesion types (bone versus lymph node lesions) [
54]. However, in our study no statistically significant difference in tracer kinetics between the two tissue types was found (
p = 0.84). The highly variable kinetics observed in lesions are possibly the result of heterogeneity in tumor biology. Therefore, the use of a general tracer uptake pattern for lesions will introduce relevant deviations on an individual level.
However, the proposed methodology using lesion-specific kinetics results in a rather good PET/SPECT absorbed dose ratio for lesions of 1.3 ± 0.7 (0.4–2.7), with a significant correlation (r = 0.69, p < 0.01) that was not found between SUVmax on [68Ga]Ga-PSMA-PET and absorbed dose after therapy (r = 0.16, p = 0.47). So, despite a relatively large range in PET/SPECT absorbed dose ratio, an actual absorbed dose prediction could still mean a significant improvement in patient selection compared to only using lesion SUVmax, since it provides better insight in what lesion uptake is to be expected and thus whether treatment with [177Lu]Lu-PSMA is expected to be effective.
Estimation of patient-specific tracer uptake in lesions could potentially be improved by obtaining continuous information on tracer distribution during the first hour after injection of [
68Ga]Ga-PSMA using dynamic PET imaging [
54]. Moreover, obtaining uptake information at multiple later time points could provide crucial information on late tracer kinetics, which largely determine the absorbed dose. However, due to the short half-life of
68Ga (68 min), it is not possible to follow the retention of PSMA over multiple days. The positron emitter
89Zr with a 3.27 days half-life could be an attractive alternative. The first preclinical studies with [
89Zr]Zr-PSMA-617 and [
89Zr]Zr-PSMA-I&T biodistribution showed that this resembled the distribution of [
177Lu]Lu-PSMA-617 and [
177Lu]Lu-PSMA-I&T, respectively (data not published yet). Recently the first clinical study showed that several lesions had uptake on [
89Zr]Zr-PSMA-PET, which were not detected on early time point PET using
18F-FDG or [
68Ga]Ga-PSMA [
55]. Therefore,
89Zr-labelled PSMA has the potential to improve lesion absorbed dose prediction.
The large range in PET/SPECT absorbed dose ratio found in this study can also partly be explained by difficulties in calculating SPECT absorbed dose for small structures, such as the lesions found in this patient cohort. Due to limited image resolution, count statistics, and photon scatter, determination of residence times is difficult. This means that in general, larger uncertainties in absorbed dose calculations are found in these small volumes [
32,
56].
While patient selection might be improved by combining lesion SUV on PSMA-PET with evaluation of positive tumor uptake on
18F-FDG-PET [
34], this does not provide information on risk of organ toxicity. The mHSPC patient cohort for this study, acute organ toxicity, is not anticipated, since these patients tend to have a relatively good physical condition and good organ function. However, development of chronic toxicities should be prevented. In addition, presently, [
177Lu]Lu-PSMA therapy is mainly applied in mCRPC patients, which are at risk for compromised organ function and may have received prior radionuclide therapy that already deposited a radiation dose to the healthy organs. Therefore, an absorbed dose prediction based on the pre-therapeutic [
68Ga]Ga-PSMA-PET scan would provide the physician with a useful tool to manage or refrain from additional treatment cycles when there is a significant risk of organ toxicity. Our study showed that an absorbed dose prediction based on a single time point [
68Ga]Ga-PSMA-PET scan is feasible, similar to what earlier studies found for
124I-PET/CT dose prediction of
131I-therapy in thyroid cancer patients [
35‐
40]. Tissue-specific organ kinetics showed to be stable between patients, which means that uptake information at a single time point in combination with assumed tissue-specific tracer kinetics provide an effective instrument for absorbed dose prediction. Although it was shown earlier that organ tracer kinetics in mHSPC patients are very similar to those in mCRPC patients [
32], it would be advised to establish tissue-specific tracer kinetics for mCRPC patients when applying the proposed methodology in this specific patient group. Furthermore, our results are based on only 10 patients. More elaborate data of larger patient cohorts is warranted.
Initially, we found that the absorbed dose prediction based on PET for the kidneys was notably higher than the SPECT-based values: PET/SPECT absorbed dose ratio of 2.21 ± 0.46 (Fig.
3B). A possible explanation could be a difference in early phase kinetics between patients, which was the only exception found in this study that showed somewhat larger variation in tracer kinetics: 21% for the early phase kinetics up to 72 h (Table
2). In addition, there might be a difference in 1-h tracer uptake between PSMA-11 and PSMA-617. Since the PSMA tracer is cleared mainly via the kidneys, a potential faster blood and renal uptake for PSMA-11 would lead to a higher activity found in the kidney at 1 h p.i. on PET than for PSMA-617 at the same time point on SPECT. In addition, there are some differences in the coordination chemistry of [
68Ga]Ga-PSMA-11 and [
177Lu]Lu-PSMA-617; Ga
3+ forms a hexadentate binding in the HBED chelator leaving two nitrogens and Lu
3+ a octadentate binding in the DOTA chelator. In preclinical setting, it was shown that this leads to a higher kidney uptake of [
68Ga]Ga-PSMA-11 in comparison to [
111In]In-PSMA-617 [
57,
58]. This would then lead to an overestimation of the total predicted absorbed dose. A remarkable feature was that the PET/SPECT absorbed dose ratio for kidney was rather constant. After applying a scaling factor of 2.2, a PET/SPECT absorbed dose ratio of 1.01 ± 0.21 (Fig.
3B) was obtained. Thus, despite an initial overestimation of kidney absorbed dose based on [
68Ga]Ga-PSMA-PET, it can reliably be used to predict therapeutic absorbed dose for [
177Lu]Lu-PSMA after applying the scaling factor, with a maximum deviation of around 20%. For the other organs, such a deviation was not found so no scaling was performed. However, the salivary glands showed a relatively large range in PET/SPECT absorbed dose ratio for the submandibular glands (0.61–1.84) and parotid glands (0.54–1.47), respectively (Fig.
3). This indicates that, despite very comparable overall tracer kinetics in the salivary glands, the uptake at 1 h p.i. can be rather variable between patients, leading to a larger range in PET/SPECT ratio.
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.