Treatment planning in molecular radiotherapy

Machine Learning-Based Dose Prediction in [177Lu]Lu-PSMA-617 Therapy by Integrating Biomarkers and Radiomic Features from [68Ga]Ga-PSMA-11 PET/CT

PURPOSE

The study aimed to develop machine learning (ML) models for pretherapy prediction of absorbed doses (ADs) in kidneys and tumoral lesions for metastatic castration-resistant prostate cancer (mCRPC) patients undergoing [177Lu]Lu-PSMA-617 (Lu-PSMA) radioligand therapy (RLT). By leveraging radiomic features (RFs) from [68Ga]Ga-PSMA-11 (Ga-PSMA) PET/CT scans and clinical biomarkers (CBs), the approach has the potential to improve patient selection and tailor dosimetry-guided therapy.

METHODS

Twenty patients with mCRPC underwent Ga-PSMA PET/CT scans prior to the administration of an initial 6.8±0.4 GBq dose of the first Lu-PSMA RLT cycle. Post-therapy dosimetry involved sequential scintigraphy imaging at approximately 4, 48, and 72 h, along with a SPECT/CT image at around 48 h, to calculate time-integrated activity (TIA) coefficients. Monte Carlo (MC) simulations, leveraging the Geant4 application for tomographic emission (GATE) toolkit, were employed to derive ADs. The ML models were trained using pretherapy RFs from Ga-PSMA PET/CT and CBs as input, while the ADs in kidneys and lesions (n=130), determined using MC simulations from scintigraphy and SPECT imaging, served as the ground truth. Model performance was assessed through leave-one-out cross-validation (LOOCV), with evaluation metrics including R² and root mean squared error (RMSE).

RESULTS

The mean delivered ADs were 0.88 ± 0.34 Gy/GBq for kidneys and 2.36 ± 2.10 Gy/GBq for lesions. Combining CBs with the best RFs produced optimal results: the extra trees regressor (ETR) was the best ML model for predicting kidney ADs, achieving an RMSE of 0.11 Gy/GBq and an R² of 0.87. For lesion ADs, the gradient boosting regressor (GBR) performed best, with an RMSE of 1.04 Gy/GBq and an R² of 0.77.

CONCLUSION

Integrating pretherapy Ga-PSMA PET/CT RFs with CBs shows potential in predicting ADs in RLT. To personalize treatment planning and enhance patient stratification, it is crucial to validate these preliminary findings with a larger sample size and an independent cohort.

Non-linear mixed-effects modelling and population-based model selection for 131I kinetics in benign thyroid disease

PURPOSE

This study aimed to determine a mathematical model for accurately calculating time-integrated activities (TIAs) of target tissue in 131I therapy for benign thyroid disease using the population-based model selection and non-linear mixed-effects (PBMS-NLME) method.

METHODS

Biokinetic data of 131I in target tissue were collected from seventy-three patients at 2, 6, 24, 48, and 96 (N = 53) or 120 (N = 20) h after oral capsule administration with 1 MBq 131I. Based on the Akaike weight, the best sum-of-exponential function (SOEF) describing the biokinetic data was selected using PBMS-NLME modelling. Nine SOEF with three to six parameters (including the function from the European Association of Nuclear Medicine Standard Operational Procedure (EANM SOP)) were used. The fittings were repeated 1000 times with different starting values of the SOE parameters to find the optimal fit. Akaike weight was used to identify the performance of the best model from PBMS-NLME and the EANM SOP SOE function with individual fitting.

RESULTS

Based on the PBMS-NLME analysis, the SOEFλ 1 λ 2 + λ 1 - λ 3 e - λ 3 + λ phys t - e - λ 1 + λ 2 + λ phys t + a 1 e - λ 1 + λ 2 + λ phys t was selected as the function most supported by the data. The Akaike weight of the best function was approximately 100%. The best SOEF from the PBMS-NLME approach shows a better performance in describing the biokinetic data of 131I in the thyroid gland than the function from the EANM SOP with individual fitting, based on the Akaike weight.

CONCLUSIONS

The best mathematical model from the PBMS-NLME approach has one more free parameter than the EANM SOP function, which could lead to more accurate TIAs.

Accuracy and precision analyses of single-time-point dosimetry utilising physiologically-based pharmacokinetic modelling and non-linear mixed-effects modelling

PURPOSE

The aim of this study was to investigate the accuracy and precision of single-time-point (STP) dosimetry using a physiologically-based pharmacokinetic (PBPK) model with non-linear mixed-effects modelling (NLMEM).

METHODS

Biokinetic data of [111In]In-DOTA-TATE in tumours, kidneys, liver, spleen, and whole body were collected from eight patients. The imaging was performed using planar scintigraphy at 2, 4, 24, 48, and 72 h after injection. Serum activity concentration was quantified at 5 and 15 min; 0.5, 1, 2, and 4 h; and 1, 2, and 3 d after injection. The PBPK model was fitted to the biokinetic data using NONMEM software version 7.5.1. Goodness-of-fit (GoF) criteria were visual inspection of the biokinetic curves, relative standard errors (RSEs) of the fitted parameters < 50%, and the absolute values of the off-diagonal elements in the correlation matrix < 0.8. All-time-point (ATP) fitting was performed, and the obtained absorbed doses (ADs) were used as reference (rADs). The leave-one-out Jackknife method was applied to calculate STP ADs (sADs). The accuracy of STP dosimetry was evaluated using the relative deviation between sADs and rADs. The time point, which resulted in the smallest root-mean-square error (RMSE), was selected as the optimal time point for STP dosimetry. The precision of the AD was calculated as ratio of AD RSE and AD values.

RESULTS

The ATP fitting was adequate based on the GoF test. STP dosimetry at 48 h after injection provided an acceptable estimation of ADs, yielding the lowest RMSE values for the kidney and tumour, calculated as (7 ± 2)% and (14 ± 4)%, respectively. The ADs in STP dosimetry showed lower precision than in ATP dosimetry. For instance, the ADs precision in ATP and STP dosimetry for kidneys in term median[min, max] were 3[3, 3]% and 6[5, 6]%, respectively. Similar results were found for the tumours where the precision of the ADs in ATP and STP dosimetry were 4[4, 5]% and 9[8, 12] %, respectively.

CONCLUSION

STP dosimetry exhibits acceptable accuracy, although it shows a decrease in precision compared to ATP fitting. Precision information is clinically relevant for developing the optimal strategies for simplified dosimetry protocols.

Personalized metronomic radiopharmaceutical therapy through injection profile optimization via physiologically based pharmacokinetic (PBPK) modeling

Each treatment cycle of radiopharmaceutical therapy (RPT) is administered as a single dose. We aimed to investigate a personalized metronomic RPT paradigm, employing multiple lower-dose administrations, to evaluate its effect on delivering radiopharmaceuticals to tumors. We developed a physiologically-based pharmacokinetic (PBPK) model applied to metastatic castration-resistant prostate cancer patients to analyze the impact of metronomic framework and various infusion durations (1-4 h) on absorbed doses (ADs) in tumors and organ-at-risk (OAR). We designed a treatment algorithm to select optimal regimens with high AD, while investigating what we term radiopharmaceutical delivery payload (RDP). This metric evaluates the efficiency of radiopharmaceutical delivery by quantifying the proportion of the administered dose that successfully reaches the target tissue. The goal is to optimize trade-offs between RDP and tumors-AD among injection profiles, amongst varying radioactivity (1-22GBq), total radiopharmaceutical mass (25-210nmol), number of injections (2-6), and time intervals (12-36 h) between injections. Our framework applied to five patients led to increased AD between 2 and 358 Gy (between 2 and 146%) higher than normally administered to patients, safeguarding OARs. Using single-dose scenarios to match ADs in metronomic approach, led to significant increase in injected activities, requiring injection of 0 to 9GBq additional activity (reducing RDP by 3-75%). Maintaining total administered radioactivity within clinically therapeutic levels, increasing frequency, time interval, and infusion duration increases tumors and OARs AD by 0.05-73%, while it decreased tumors-to-OARs AD ratios by 0.1-30%. Based on the PBPK modeling approach, metronomic RPT appears to improve efficacy (RDP) in delivered doses to tumors for a given total injected radioactivity.

Few-time-points time-integrated activity coefficients calculation using non-linear mixed-effects modeling: Proof of concept for [111In]In-DOTA-TATE in kidneys

PURPOSE

The purpose of this study is to investigate the accuracy of few-time-points (FTP) time-integrated activity coefficients (TIACs) in peptide-receptor radionuclide therapy (PRRT) using non-linear mixed-effects (NLME) modeling.

METHODS

Biokinetic data of [111In]In-DOTA-TATE in kidneys at T-1 = (2.9 ± 0.6) h, T-2 = (4.6 ± 0.4) h, T-3 = (22.8 ± 1.6) h, T-4 = (46.7 ± 1.7) h, and T-5 = (70.9 ± 1.0) h after injection were obtained from eight patients using planar imaging. The Sum-Of-Exponentials (SOE) function with four parameters was used, which was selected as the best model for the renal biokinetic data of [111In]In-DOTA-TATE. The parameters of the SOE function were fitted to the all-time-point data in the NLME framework to derive reference (rTIACs). FTP fits, which consist of all combinations of time points, are done to calculate the estimated TIACs (eTIACs). The accuracy of the FTP-NLME TIACs calculations was assessed by calculating the relative deviations (RDs) and relative root-mean-square errors (RMSEs) between the eTIACs and rTIACs.

RESULTS

The lowest (mean ± SD) of RDs for the single-, two-, three-, four-time point FTPs were (0 ± 8) % (T-4), (1 ± 6) % (T-3 and T-4), (3 ± 5) % (T-2, T-3 and T-4), and (0 ± 2) % (T-2, T-3, T-4, and T-5), respectively. The lowest RMSEs for the one-, two-, three-, and four-time point FTPs were 8 % (T-4), 6 % (T-3 and T-4), 5 % (T-2, T-3 and T-4), and 2 % (T-2, T-3, T-4, and T-5), respectively.

CONCLUSION

Our results showed that FTP-NLME in an example of [111In]In-DOTA-TATE could lead to a high accuracy of eTIAC across various time points, when incorporating time point T-4 = (46.7 ± 1.7) h.

Single-time-point dosimetry using model selection and the Bayesian fitting method: A proof of concept

PURPOSE

This study aimed to determine the effect of model selection on simplified dosimetry for the kidneys using Bayesian fitting (BF) and single-time-point (STP) imaging.

METHODS

Kidney biokinetics data of [177Lu]Lu-PSMA-617 from mHSPC were collected using SPECT/CT after injection of (3.1 ± 0.1) GBq at time points T1(2.3 ± 0.5), T2(23.8 ± 2.0), T3(47.7 ± 2.2), T4(71.8 ± 2.2), and T5(167.4 ± 1.9) h post-injection. Eleven functions with various parameterizations and a combination of shared and individual parameters were used for model selection. Model averaging of functions with an Akaike weight of >10 % was used to calculate the reference TIAC (TIACREF). STP BF method (STP-BF) was performed to determine the STP TIACs (TIACSTP-BF). The STP-BF performance was assessed by calculating the root-mean-square error (RMSE) of relative deviation between TIACSTP-BF and TIACREF. In addition, the STP-BF performance was compared to the Hänscheid Method.

RESULTS

The function [Formula: see text] with shared parameter λ2 was selected as the best function (Akaike weight of 57.91 %). STP-BF using the best function resulted in RMSEs of 20.3 %, 9.1 %, 8.4 %, 13.6 %, and 19.3 % at T1, T2, T3, T4, and T5, respectively. The RMSEs of STP-Hänscheid were 22.4 %, 14.6 %, and 21.9 % at T2, T3, and T4, respectively.

CONCLUSION

A model selection was presented to determine the fit function for calculating TIACs in STP-BF. This study shows that the STP dosimetry using BF and model selection performed better than the frequently used STP Hänscheid method.

Population-based model selection for an accurate estimation of time-integrated activity using non-linear mixed-effects modelling

PURPOSE

Personalized treatment planning in Molecular Radiotherapy (MRT) with accurately determining the absorbed dose is highly desirable. The absorbed dose is calculated based on the Time-Integrated Activity (TIA) and the dose conversion factor. A crucial unresolved issue in MRT dosimetry is which fit function to use for the TIA calculation. A data-driven population-based fitting function selection could help solve this problem. Therefore, this project aims to develop and evaluate a method for accurately determining TIAs in MRT, which performs a Population-Based Model Selection within the framework of the Non-Linear Mixed-Effects (NLME-PBMS) model.

METHODS

Biokinetic data of a radioligand for the Prostate-Specific Membrane Antigen (PSMA) for cancer treatment were used. Eleven fit functions were derived from various parameterisations of mono-, bi-, and tri-exponential functions. The functions' fixed and random effects parameters were fitted (in the NLME framework) to the biokinetic data of all patients. The goodness of fit was assumed acceptable based on the visual inspection of the fitted curves and the coefficients of variation of the fitted fixed effects. The Akaike weight, the probability that the model is the best among the whole set of considered models, was used to select the fit function most supported by the data from the set of functions with acceptable goodness of fit. NLME-PBMS Model Averaging (MA) was performed with all functions having acceptable goodness of fit. The Root-Mean-Square Error (RMSE) of the calculated TIAs from individual-based model selection (IBMS), a shared-parameter population-based model selection (SP-PBMS) reported in the literature, and the functions from NLME-PBMS method to the TIAs from MA were calculated and analysed. The NLME-PBMS (MA) model was used as the reference as this model considers all relevant functions with corresponding Akaike weights.

RESULTS

The function [Formula: see text] was selected as the function most supported by the data with an Akaike weight of (54 ± 11) %. Visual inspection of the fitted graphs and the RMSE values show that the NLME model selection method has a relatively better or equivalent performance than the IBMS or SP-PBMS methods. The RMSEs of the IBMS, SP-PBMS, and NLME-PBMS (f3a) methods are 7.4%, 8.8%, and 2.4%, respectively.

CONCLUSION

A procedure including fitting function selection in a population-based method was developed to determine the best fit function for calculating TIAs in MRT for a given radiopharmaceutical, organ and set of biokinetic data. The technique combines standard practice approaches in pharmacokinetics, i.e. an Akaike-weight-based model selection and the NLME model framework.

Uncertainty Analysis of Time-Integrated Activity Coefficient in Single-Time-Point Dosimetry Using Bayesian Fitting Method

Purpose

Calculation of the uncertainty of the individual time-integrated activity coefficient (TIACs) is desirable in molecular radiotherapy. However, the calculation of TIAC's uncertainty in single-time-point (STP) method has never been reported in the literature. This study presents a method based on the Bayesian fitting (BF) to calculate the standard deviation (SD) of individual TIACs in the STP dosimetry.

Methods

Biokinetic data of 177Lu-DOTATATE in kidneys were obtained from PMID33443063. BF methods with extended objective function, which optimize the fitting using prior knowledge of the function's parameters, were used. Reference TIACs (rTIACs) were calculated by fitting a mono-exponential function to the all-time-point data. The goodness of fit was checked based on the visual inspection and the coefficient of variations (CV) of the fitted parameters < 0.5. BF with relative (BFr) and absolute-based (BFa) variance methods were used to obtain the calculated TIACs (cTIACs) from the STP dosimetry. Performance of the STP method was obtained by calculating the relative deviation (RD) between cTIACs and rTIACs.

Results

Visual inspection showed a good fit for all patients with CV of fitted parameters less than 50%. The mean ± SD of cTIAC's %RD were 7.0 ± 25.2 for BFr and 2.6 ± 8.9 for BFa. The range of %CV of the individual cTIAC's SD for BFr and BFa methods was 36-78% and 22-33%, respectively, while the %CV of the rTIAC SD was 0.8-49%.

Conclusion

We introduce the BF method to calculate the SD of individual TIACs in STP dosimetry. The presented method might be used as an alternative method for uncertainty analysis in STP dosimetry.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13139-024-00851-8.

Quality Assurance Considerations in Radiopharmaceutical Therapy Dosimetry Using PLANETDose: An International Atomic Energy Agency Study

Implementation of radiopharmaceutical therapy dosimetry varies depending on the clinical application, dosimetry protocol, software, and ultimately the operator. Assessing clinical dosimetry accuracy and precision is therefore a challenging task. This work emphasizes some pitfalls encountered during a structured analysis, performed on a single-patient dataset consisting of SPECT/CT images by various participants using a standard protocol and clinically approved commercial software. Methods: The clinical dataset consisted of the dosimetric study of a patient administered with [177Lu]Lu-DOTATATE at Tygerberg Hospital, South Africa, as a part of International Atomic Energy Agency-coordinated research project E23005. SPECT/CT images were acquired at 5 time points postinjection. Patient and calibration images were reconstructed on a workstation, and a calibration factor of 122.6 Bq/count was derived independently and provided to the participants. A standard dosimetric protocol was defined, and PLANETDose (version 3.1.1) software was installed at 9 centers to perform the dosimetry of 3 treatment cycles. The protocol included rigid image registration, segmentation (semimanual for organs, activity threshold for tumors), and dose voxel kernel convolution of activity followed by absorbed dose (AD) rate integration to obtain the ADs. Iterations of the protocol were performed by participants individually and within collective training, the results of which were analyzed for dosimetric variability, as well as for quality assurance and error analysis. Intermediary checkpoints were developed to understand possible sources of variation and to differentiate user error from legitimate user variability. Results: Initial dosimetric results for organs (liver and kidneys) and lesions showed considerable interoperator variability. Not only was the generation of intermediate checkpoints such as total counts, volumes, and activity required, but also activity-to-count ratio, activity concentration, and AD rate-to-activity concentration ratio to determine the source of variability. Conclusion: When the same patient dataset was analyzed using the same dosimetry procedure and software, significant disparities were observed in the results despite multiple sessions of training and feedback. Variations due to human error could be minimized or avoided by performing intensive training sessions, establishing intermediate checkpoints, conducting sanity checks, and cross-validating results across physicists or with standardized datasets. This finding promotes the development of quality assurance in clinical dosimetry.

Single-time-point dosimetry using model selection and nonlinear mixed-effects modelling: a proof of concept

PURPOSE

This project aims to develop and evaluate a method for accurately determining time-integrated activities (TIAs) in single-time-point (STP) dosimetry for molecular radiotherapy. It performs a model selection (MS) within the framework of the nonlinear mixed-effects (NLME) model (MS-NLME).

METHODS

Biokinetic data of [¹¹¹In]In-DOTATATE activity in kidneys at T1 = (2.9 ± 0.6) h, T2 = (4.6 ± 0.4) h, T3 = (22.8 ± 1.6) h, T4 = (46.7 ± 1.7) h, and T5 = (70.9 ± 1.0) h post injection were obtained from eight patients using planar imaging. Eleven functions were derived from various parameterisations of mono-, bi-, and tri-exponential functions. The functions' fixed and random effects parameters were fitted simultaneously (in the NLME framework) to the biokinetic data of all patients. The Akaike weights were used to select the fit function most supported by the data. The relative deviations (RD) and the root-mean-square error (RMSE) of the calculated TIAs for the STP dosimetry at T3 = (22.8 ± 1.6) h and T4 = (46.7 ± 1.7) h p.i. were determined for all functions passing the goodness-of-fit test.

RESULTS

The function [Formula: see text] with four adjustable parameters and [Formula: see text] was selected as the function most supported by the data with an Akaike weight of (45 ± 6) %. RD and RMSE values show that the MS-NLME method performs better than functions with three or five adjustable parameters. The RMSEs of TIA_NLME-PBMS and TIA_3-parameters were 7.8% and 10.9% (for STP at T3), and 4.9% and 10.7% (for STP at T4), respectively.

CONCLUSION

An MS-NLME method was developed to determine the best fit function for calculating TIAs in STP dosimetry for a given radiopharmaceutical, organ, and patient population. The proof of concept was demonstrated for biokinetic ¹¹¹In-DOTATATE data, showing that four-parameter functions perform better than three- and five-parameter functions.

Single-time-point estimation of absorbed doses in PRRT using a non-linear mixed-effects model

INTRODUCTION

Estimation of accurate time-integrated activity coefficients (TIACs) and radiation absorbed doses (ADs) is desirable for treatment planning in peptide-receptor radionuclide therapy (PRRT). This study aimed to investigate the accuracy of a simplified dosimetry using a physiologically-based pharmacokinetic (PBPK) model, a nonlinear mixed effect (NLME) model, and single-time-point imaging to calculate the TIACs and ADs of ⁹⁰Y-DOTATATE in various organs of dosimetric interest and tumors.

MATERIALS & METHODS

Biokinetic data of ¹¹¹In-DOTATATE in tumors, kidneys, liver, spleen, and whole body were obtained from eight patients using planar scintigraphic imaging at T1 = (2.9 ± 0.6), T2 = (4.6 ± 0.4), T3 = (22.8 ± 1.6), T4 = (46.7 ± 1.7) and T5 = (70.9 ± 1.0) h post injection. Serum activity concentration was measured at 5 and 15 min; 0.5, 1, 2, and 4 h; and 1, 2, and 3 d p.i.. A published PBPK model for PRRT, NLME, and a single-time-point imaging datum at different time points were used to calculate TIACs in tumors, kidneys, liver, spleen, whole body, and serum. Relative deviations (RDs) (median [min, max]) between the calculated TIACs from single-time-point imaging were compared to the TIACs calculated from the all-time-points fit. The root mean square error (RMSE) of the difference between the computed ADs from the single-time-point imaging and reference ADs from the all-time point fittings were analyzed. A joint root mean square error RMSE_joint of the ADs was calculated with the RSME from both the tumor and kidneys to sort the time points concerning accurate results for the kidneys and tumor dosimetry. The calculations of TIACs and ADs from the single-time-point dosimetry were repeated using the sum of exponentials (SOE) approach introduced in the literature. The RDs and the RSME of the PBPK approach in our study were compared to the SOE approach.

RESULTS

Using the PBPK and NLME models and the biokinetic measurements resulted in a good fit based on visual inspection of the fitted curves and the coefficient of variation CV of the fitted parameters (<50%). T4 was identified being the time point with a relatively low median and range of TIACs RDs, i.e., 5 [1, 21]% and 2 [-15, 21]% for kidneys and tumors, respectively. T4 was found to be the time point with the lowest joint root mean square error RMSE_joint of the ADs. Based on the RD and RMSE, our results show a similar performance as the SOE and NLME model approach.

SUMMARY

In this study, we introduced a simplified calculation of TIACs/ADs using a PBPK model, an NLME model, and a single-time-point measurement. Our results suggest a single measurement might be used to calculate TIACs/ADs in the kidneys and tumors during PRRT.

Dosimetry in Radiopharmaceutical Therapy

The application of radiopharmaceutical therapy for the treatment of certain diseases is well established, and the field is expanding. New therapeutic radiopharmaceuticals have been developed in recent years, and more are in the research pipeline. Concurrently, there is growing interest in the use of internal dosimetry as a means of personalizing, and potentially optimizing, such therapy for patients. Internal dosimetry is multifaceted, and the current state of the art is discussed in this continuing education article. Topics include the context of dosimetry, internal dosimetry methods, the advantages and disadvantages of incorporating dosimetry calculations in radiopharmaceutical therapy, a description of the workflow for implementing patient-specific dosimetry, and future prospects in the field.

Application of machine learning to pretherapeutically estimate dosimetry in men with advanced prostate cancer treated with 177Lu-PSMA I&T therapy

PURPOSE

Although treatment planning and individualized dose application for emerging prostate-specific membrane antigen (PSMA)-targeted radioligand therapy (RLT) are generally recommended, it is still difficult to implement in practice at the moment. In this study, we aimed to prove the concept of pretherapeutic prediction of dosimetry based on imaging and laboratory measurements before the RLT treatment.

METHODS

Twenty-three patients with metastatic castration-resistant prostate cancer (mCRPC) treated with 177Lu-PSMA I&T RLT were included retrospectively. They had available pre-therapy 68 Ga-PSMA-HEBD-CC PET/CT and at least 3 planar and 1 SPECT/CT imaging for dosimetry. Overall, 43 cycles of 177Lu-PSMA I&T RLT were applied. Organ-based standard uptake values (SUVs) were obtained from pre-therapy PET/CT scans. Patient dosimetry was calculated for the kidney, liver, spleen, and salivary glands using Hermes Hybrid Dosimetry 4.0 from the planar and SPECT/CT images. Machine learning methods were explored for dose prediction from organ SUVs and laboratory measurements. The uncertainty of these dose predictions was compared with the population-based dosimetry estimates. Mean absolute percentage error (MAPE) was used to assess the prediction uncertainty of estimated dosimetry.

RESULTS

An optimal machine learning method achieved a dosimetry prediction MAPE of 15.8 ± 13.2% for the kidney, 29.6% ± 13.7% for the liver, 23.8% ± 13.1% for the salivary glands, and 32.1 ± 31.4% for the spleen. In contrast, the prediction based on literature population mean has significantly larger MAPE (p < 0.01), 25.5 ± 17.3% for the kidney, 139.1% ± 111.5% for the liver, 67.0 ± 58.3% for the salivary glands, and 54.1 ± 215.3% for the spleen.

CONCLUSION

The preliminary results confirmed the feasibility of pretherapeutic estimation of treatment dosimetry and its added value to empirical population-based estimation. The exploration of dose prediction may support the implementation of treatment planning for RLT.

Marker-controlled watershed with deep edge emphasis and optimized H-minima transform for automatic segmentation of densely cultivated 3D cell nuclei

BACKGROUND

The segmentation of 3D cell nuclei is essential in many tasks, such as targeted molecular radiotherapies (MRT) for metastatic tumours, toxicity screening, and the observation of proliferating cells. In recent years, one popular method for automatic segmentation of nuclei has been deep learning enhanced marker-controlled watershed transform. In this method, convolutional neural networks (CNNs) have been used to create nuclei masks and markers, and the watershed algorithm for the instance segmentation. We studied whether this method could be improved for the segmentation of densely cultivated 3D nuclei via developing multiple system configurations in which we studied the effect of edge emphasizing CNNs, and optimized H-minima transform for mask and marker generation, respectively.

RESULTS

The dataset used for training and evaluation consisted of twelve in vitro cultivated densely packed 3D human carcinoma cell spheroids imaged using a confocal microscope. With this dataset, the evaluation was performed using a cross-validation scheme. In addition, four independent datasets were used for evaluation. The datasets were resampled near isotropic for our experiments. The baseline deep learning enhanced marker-controlled watershed obtained an average of 0.69 Panoptic Quality (PQ) and 0.66 Aggregated Jaccard Index (AJI) over the twelve spheroids. Using a system configuration, which was otherwise the same but used 3D-based edge emphasizing CNNs and optimized H-minima transform, the scores increased to 0.76 and 0.77, respectively. When using the independent datasets for evaluation, the best performing system configuration was shown to outperform or equal the baseline and a set of well-known cell segmentation approaches.

CONCLUSIONS

The use of edge emphasizing U-Nets and optimized H-minima transform can improve the marker-controlled watershed transform for segmentation of densely cultivated 3D cell nuclei. A novel dataset of twelve spheroids was introduced to the public.

Bone Marrow Mesenchymal Stem Cells Mediated Radiosensitive Promoter-Combined Sodium Iodide Symporter for the Treatment of Breast Cancer

OBJECTIVES

To develop a genetically engineered bone marrow mesenchymal stem cells (BMSCs) that carries a radiotherapy gene to target triple-negative breast cancer (TNBC) and to evaluate the efficacy of radiation damage within the tumor microenvironment (TME).

METHODS

The early growth response protein 1 (Egr1)-human sodium iodide symporter (hNIS) gene was transfected into BMSCs by lentiviral transfection and the expression levels were evaluated by RT-PCR. Transwell and adipogenesis and osteogenesis assays were performed to determine the targeting properties and adipogenic and osteogenic characteristics of the transgenic stem cells. The uptake of radioiodine and the efflux characteristics of the transgenic stem cells were determined by iodine uptake experiments. 131I-SPECT imaging was used to determine the characteristics of targeting to TNBC and to quantify the iodine uptake of transgenic stem cells in vivo. The effects of 131I treatment on BMSCs were characterized using tumor growth, immune cell infiltration and tumor invasion endpoints based on immunohistochemistry and flow cytometry analysis of tumor samples.

RESULTS

BMSCs-Egr1-hNIS cells abundantly express hNIS after radiation induction and are chemotactically attracted to TNBC tumors. Iodine uptake of BMSCs-Egr1-hNIS gradually increased with increasing induction concentrations and times. When the inductive concentration of 131I was > 100 μCi/mL and lasted for 36 h, the rate of iodine uptake in cells increased. In vitro, the radioiodine quickly flowed out from cells within 20 minutes but in vivo, the rate of radioiodine loss was significantly slower and occurred over 24 hours. After 131I therapy, tumor growth was inhibited, white blood cells infiltrated into tumor site and the levels of invasion-related cytokines significantly decreased.

CONCLUSIONS

BMSCs-Egr1-hNIS mediates 131I therapy can achieve precisely targeted radiotherapy to inhibit tumor growth, promote immune cells infiltration to the tumor sites and reduce the invasiveness and metastasis characteristics of tumor cells.

A population-based method to determine the time-integrated activity in molecular radiotherapy

BACKGROUND

The calculation of time-integrated activities (TIAs) for tumours and organs is required for dosimetry in molecular radiotherapy. The accuracy of the calculated TIAs is highly dependent on the chosen fit function. Selection of an adequate function is therefore of high importance. However, model (i.e. function) selection works more accurately when more biokinetic data are available than are usually obtained in a single patient. In this retrospective analysis, we therefore developed a method for population-based model selection that can be used for the determination of individual time-integrated activities (TIAs). The method is demonstrated at an example of [¹⁷⁷Lu]Lu-PSMA-I&T kidneys biokinetics. It is based on population fitting and is specifically advantageous for cases with a low number of available biokinetic data per patient.

METHODS

Renal biokinetics of [¹⁷⁷Lu]Lu-PSMA-I&T from thirteen patients with metastatic castration-resistant prostate cancer acquired by planar imaging were used. Twenty exponential functions were derived from various parameterizations of mono- and bi-exponential functions. The parameters of the functions were fitted (with different combinations of shared and individual parameters) to the biokinetic data of all patients. The goodness of fits were assumed as acceptable based on visual inspection of the fitted curves and coefficients of variation CVs < 50%. The Akaike weight (based on the corrected Akaike Information Criterion) was used to select the fit function most supported by the data from the set of functions with acceptable goodness of fit.

RESULTS

The function [Formula: see text] with shared parameter [Formula: see text] was selected as the function most supported by the data with an Akaike weight of 97%. Parameters [Formula: see text] and [Formula: see text] were fitted individually for every patient while parameter [Formula: see text] was fitted as a shared parameter in the population yielding a value of 0.9632 ± 0.0037.

CONCLUSIONS

The presented population-based model selection allows for a higher number of parameters of investigated fit functions which leads to better fits. It also reduces the uncertainty of the obtained Akaike weights and the selected best fit function based on them. The use of the population-determined shared parameter for future patients allows the fitting of more appropriate functions also for patients for whom only a low number of individual data are available.

Performance assessment of the ALBIRA II pre-clinical SPECT S102 system for 99mTc imaging

OBJECTIVE

The performance characteristics of the SPECT sub-system S102 of the ALBIRA II PET/SPECT/CT are analyzed for the 80 mm field of view (FOV) to evaluate the potential in-vivo imaging in rats, based on measurements of the system response for the commonly used Technetium-99 m (99mTc) in small animal imaging.

METHODS

The ALBIRA II tri-modal µPET/SPECT/CT pre-clinical system (Bruker BioSpin, Ettlingen, Germany) was used. The SPECT modality is made up of two opposite gamma cameras (Version S102) with Sodium doped Cesium Iodide (CsI(Na)) single continuous crystal detectors coupled to position-sensitive photomultipliers (PSPMTs). Imaging was performed with the NEMA NU-4 image quality phantom (Data Spectrum Corporation, Durham, USA). Measurements were performed with a starting activity concentration of 4.76 MBq/mL 99mTc. An energy window of 20% at 140 keV was selected in this study. The system offers a 20 mm, 40 mm, 60 mm and an 80 mm field of view (FOV) and in this study the 80 mm FOV was used for all the acquisitions. The data were reconstructed with an ordered subset expectation maximization (OSEM) algorithm. Sensitivity, spatial resolution, count rate linearity, convergence of the algorithm and the recovery coefficients (RC) were analyzed. All analyses were performed with PMOD and MATLAB software.

RESULTS

The sensitivities measured at the center of the 80 mm FOV with the point source were 23.1 ± 0.3 cps/MBq (single pinhole SPH) and 105.6 ± 5.5 cps/MBq (multi pinhole MPH). The values for the axial, tangential and radial full width at half maximum (FWHM) were 2.51, 2.54, and 2.55 mm with SPH and 2.35, 2.44 and 2.32 mm with MPH, respectively. The corresponding RC values for the 5 mm, 4 mm, 3 mm and 2 mm rods were 0.60 ± 0.28, 0.61 ± 0.24, 0.29 ± 0.11 and 0.20 ± 0.06 with SPH and 0.56 ± 0.20, 0.50 ± 0.18, 0.38 ± 0.09 and 0.23 ± 0.06 with MPH. To obtain quantitative imaging data, the image reconstructions should be performed with 12 iterations.

CONCLUSION

The ALBIRA II preclinical SPECT sub-system S102 has a favorable sensitivity and spatial resolution for the 80 mm FOV setting for both the SPH and MPH configurations and is a valuable tool for small animal imaging.

Important pharmacokinetic parameters for individualization of 177 Lu-PSMA therapy: A global sensitivity analysis for a physiologically-based pharmacokinetic model

PURPOSE

The knowledge of the contribution of anatomical and physiological parameters to interindividual pharmacokinetic differences could potentially be used to improve individualized treatment planning for radionuclide therapy. The aim of this study was therefore to identify the physiologically based pharmacokinetic (PBPK) model parameters that determine the interindividual variability of absorbed doses (ADs) to kidneys and tumor lesions in therapy with ¹⁷⁷ Lu-labeled PSMA-targeting radioligands.

METHODS

A global sensitivity analysis (GSA) with the extended Fourier Amplitude Sensitivity Test (eFAST) algorithm was performed. The whole-body PBPK model for PSMA-targeting radioligand therapy from our previous studies was used in this study. The model parameters of interest (input of the GSA) were the organ receptor densities [R₀ ], the organ blood flows f, and the organ release rates λ. These parameters were systematically sampled NE times according to their distribution in the patient population. The corresponding pharmacokinetics were simulated and the ADs (model output) to kidneys and tumor lesions were collected. The main effect S i and total effect S Ti were calculated using the eFAST algorithm based on the variability of the model output: The main effect S i of input parameter i represents the reduction in variance of the output if the "true" value of parameter i would be known. The total effect S Ti of an input parameter i represents the proportion of variance remaining if the "true" values of all other input parameters except for i are known. The numbers of samples NE were increased up to 8193 to check the stability (i.e., convergence) of the calculated main effects S i and total effects S Ti .

RESULTS

From the simulations, the relative interindividual variability of ADs in the kidneys (coefficient of variation CV = 31%) was lower than that of ADs in the tumors (CV up to 59%). Based on the GSA, the most important parameters that determine the ADs to the kidneys were kidneys flow ( S i  = 0.36, S Ti  = 0.43) and kidneys receptor density ( S i  = 0.25, S Ti  = 0.30). Tumor receptor density was identified as the most important parameter determining the ADs to tumors ( S i and S Ti up to 0.72).

CONCLUSIONS

The results suggest that an accurate measurement of receptor density and flow before therapy could be a promising approach for developing an individualized treatment with ¹⁷⁷ Lu-labeled PSMA-targeting radioligands.

A simulation-based method to determine optimal sampling schedules for dosimetry in radioligand therapy

AIM

For dosimetry in radioligand therapy, the time-integrated activity coefficients (TIACs) for organs at risk and for tumour lesions have to be determined. The used sampling scheme affects the TIACs and therefore the calculated absorbed doses. The aim of this work was to develop a general and flexible method, which analyses numerous clinically applicable sampling schedules using true time-activity curves (TACs) of virtual patients.

METHOD

Nine virtual patients with true TACs of the tumours were created using a physiologically-based pharmacokinetic (PBPK) model and individual biokinetic data of five patients with neuroendocrine tumours and four with meningioma. ¹¹¹In-DOTATATE was used for pre-therapeutic dosimetry. In total, 15,120 sampling schemes, each consisting of 4 time points, were investigated. Gaussian noise of different levels was added to the corresponding true time-activity points. A bi-exponential function was used to fit the simulated time-activity data. For each investigated sampling schedule, 1000 replications were performed. Patient-specific and population-specific optimal sampling schedules were determined using the relative root-mean-square error (rRMSE). Furthermore, the fractions of TIACs a˜ deviating >5% (f_Δa˜>5%) and >10% (f_Δa˜>10%) from the true TIAC a˜_true were used for additional evaluations e.g. to investigate the effect of varying single time points.

RESULTS

Almost all patient-specific and all population-specific optimal sampling schedules have t₄≥96h for all noise levels. Changing the latest time point from the population-specific optimal time to e.g. 48h leads to a median increase of f_Δa˜>10% from 0.1% to 88% for the lowest investigated noise level. Using the determined population-specific optimal schedules, results in more accurate and precise results than established schedules from the literature.

CONCLUSION

A method of determining the optimal sampling schedule for dosimetry, which considers clinical working hours and measurement uncertainties, has been developed and applied. The simulation study shows that optimised sampling schedules result in high accuracy and precision of the determined TIACs.

Treatment planning algorithm for peptide receptor radionuclide therapy considering multiple tumor lesions and organs at risk

PURPOSE

Peptide receptor radionuclide therapy (PRRT) has shown promising results in the treatment of tumors with high expression of somatostatin receptors such as neuroendocrine tumors (NETs) and meningioma. However, PRRT potentially produces high renal and red marrow (RM) toxicity, the kidneys usually being the dose-limiting organ. Previously, it was shown that an improved therapeutic index can be achieved by choosing an optimal combination of injected activity and peptide molar amount. The aim of this work was to develop a clinically applicable algorithm for treatment planning in PRRT. To demonstrate the applicability and possible advantages of the algorithm thus developed, an in silico clinical trial applying the algorithm to 177 Lu-DOTATATE therapy in nine virtual patients was conducted.

METHODS

An algorithm for treatment planning in PRRT was developed, which simultaneously considers multiple tumor lesions, maximum tolerated biologically effective doses (BEDs) for multiple organs at risk (OARs) and a maximum achievable molar activity. The algorithm, subject to the abovementioned constraints, aims at maximizing the total number of killed tumor cells in the considered lesions/metastases. An in silico clinical trial was conducted with nine virtual patients. For each virtual patient, simulations increasing the molar dose of 177 Lu-DOTATATE from 2 to 2048 nmol by factors of 25 were performed. Maximum tolerated BEDs per cycle for the kidneys (10 Gy2.5 ) and for the RM (0.5 Gy15 ) were defined based on a planned total treatment of four cycles. A maximum achievable molar activity of 420 MBq/nmol was assumed. Optimal combinations of molar dose and activity were determined by applying the developed algorithm. For comparison, simulations for a typical plan with 177 Lu-DOTATATE (7.4 GBq, 265 nmol) were performed and BEDs for the OARs and for individual tumor lesions were calculated. Furthermore, to determine treatment efficacy, overall tumor control probability (oTCP) values after a four-cycle treatment were estimated for the optimal and typical plans.

RESULTS

The conducted in silico clinical trial yielded optimal molar doses and activities ranging from 24 to 512 nmol and from 6 to 30 GBq, respectively. Tumor BEDs ranged from 2 to 107 Gy10 and from 1 to 65 Gy10 for the optimal and typical plans, respectively. The estimated oTCP values showed that the optimal plans may produce adequate tumor control in six of the nine virtual patients after four cycles of 177 Lu-DOTATATE while the typical plan may be sufficient in only two virtual patients.

CONCLUSIONS

The algorithm presented can derive plans with higher tumor control than the typically delivered plan. Therefore, we propose this algorithm for clinical validation and possibly future implementation in treatment planning in molecular radiotherapy.

Treatment planning in PRRT based on simulated PET data and a PBPK model

Aim: To investigate the accuracy of treatment planning in peptide-receptor radionuclide therapy (PRRT) based on simulated PET data (using a PET noise model) and a physiologically based pharmacokinetic (PBPK) model. Methods: The parameters of a PBPK model were fitted to the biokinetic data of 15 patients. True mathematical phantoms of patients (MPPs) were the PBPK model with the fitted parameters. PET measurements after bolus injection of 150 MBq 68Ga-DOTATATE were simulated for the true MPPs. PET noise with typical noise levels was added to the data (i.e. c=0.3 [low], 3, 30 and 300 [high]). Organ activity data in the kidneys, tumour, liver and spleen were simulated at 0.5, 1 and 4 h p.i. PBPK model parameters were fitted to the simulated noisy PET data to derive the PET-predicted MPPs. Therapy was simulated assuming an infusion of 3.3 GBq of 90Y-DOTATATE over 30 min. Time-integrated activity coefficients (TIACs) of simulated therapy in tumour, kidneys, liver, spleen and remainder were calculated from both, true MPPs (true TIACs) and predicted MPPs (predicted TIACs). Variability v between true TIACs and predicted TIACs were calculated and analysed. Variability< 10 % was considered to be an accurate prediction. Results: For all noise level, variabilities for the kidneys, liver, and spleen showed an accurate prediction for TIACs, e.g. c=300: vkidney=(5 ± 2)%, vliver=(5 ± 2)%, vspleen=(4 ± 2)%. However, tumour TIAC predictions were not accurate for all noise levels, e.g. c=0.3: vtumour=(8 ± 5)%. Conclusion: PET based treatment planning with kidneys as the dose limiting organ seems possible for all reported noise levels using an adequate PBPK model and previous knowledge about the individual patient.

Prediction of time-integrated activity coefficients in PRRT using simulated dynamic PET and a pharmacokinetic model

PURPOSE

To investigate the accuracy of predicted time-integrated activity coefficients (TIACs) in peptide-receptor radionuclide therapy (PRRT) using simulated dynamic PET data and a physiologically based pharmacokinetic (PBPK) model.

METHODS

PBPK parameters were estimated using biokinetic data of 15 patients after injection of (152±15)MBq of ¹¹¹In-DTPAOC (total peptide amount (5.78±0.25)nmol). True mathematical phantoms of patients (MPPs) were the PBPK model with the estimated parameters. Dynamic PET measurements were simulated as being done after bolus injection of 150MBq ⁶⁸Ga-DOTATATE using the true MPPs. Dynamic PET scans around 35min p.i. (P₁), 4h p.i. (P₂) and the combination of P₁ and P₂ (P₃) were simulated. Each measurement was simulated with four frames of 5min each and 2 bed positions. PBPK parameters were fitted to the PET data to derive the PET-predicted MPPs. Therapy was simulated assuming an infusion of 5.1GBq of ⁹⁰Y-DOTATATE over 30min in both true and PET-predicted MPPs. TIACs of simulated therapy were calculated, true MPPs (true TIACs) and predicted MPPs (predicted TIACs) followed by the calculation of variabilities v.

RESULTS

For P₁ and P₂ the population variabilities of kidneys, liver and spleen were acceptable (v<10%). For the tumours and the remainders, the values were large (up to 25%). For P₃, population variabilities for all organs including the remainder further improved, except that of the tumour (v>10%).

CONCLUSION

Treatment planning of PRRT based on dynamic PET data seems possible for the kidneys, liver and spleen using a PBPK model and patient specific information.

European Association of Nuclear Medicine (EANM), European Federation of Organizations for Medical Physics (EFOMP), European Federation of Radiographer Societies (EFRS), European Society of Radiology (ESR), European Society for Radiotherapy and Oncology (ESTRO). Common strategic research agenda for radiation protection in medicine

Reflecting the change in funding strategies for European research projects, and the goal to jointly improve medical radiation protection through sustainable research efforts, five medical societies involved in the application of ionising radiation (European Association of Nuclear Medicine, EANM; European Federation of Organizations for Medical Physics. EFOMP; European Federation of Radiographer Societies, EFRS; European Society of Radiology, ESR; European Society for Radiotherapy and Oncology, ESTRO) have identified research areas of common interest and developed this first edition of the Common Strategic Research Agenda (SRA) for medical radiation protection. The research topics considered necessary and most urgent for effective medical care and efficient in terms of radiation protection are summarised in five main themes: 1. Measurement and quantification in the field of medical applications of ionising radiation 2. Normal tissue reactions, radiation-induced morbidity and long-term health problems 3. Optimisation of radiation exposure and harmonisation of practices 4. Justification of the use of ionising radiation in medical practice 5. Infrastructures for quality assurance The SRA is a living document; thus comments and suggestions by all stakeholders in medical radiation protection are welcome and will be dealt with by the European Alliance for Medical Radiation Protection Research (EURAMED) established by the above-mentioned societies.

MAIN MESSAGES

• Overcome the fragmentation of medical radiation protection research in Europe • Identify research areas of joint interest in the field of medical radiation protection • Improve the use of ionising radiation in medicine • Collect stakeholder feedback and seek consensus • Emphasise importance of clinical translation and evaluation of research results.

Time-integrated activity coefficient estimation for radionuclide therapy using PET and a pharmacokinetic model: A simulation study on the effect of sampling schedule and noise

PURPOSE

The aim of this study was to investigate the accuracy of PET-based treatment planning for predicting the time-integrated activity coefficients (TIACs).

METHODS

The parameters of a physiologically based pharmacokinetic (PBPK) model were fitted to the biokinetic data of 15 patients to derive assumed true parameters and were used to construct true mathematical patient phantoms (MPPs). Biokinetics of 150 MBq (68)Ga-DOTATATE-PET was simulated with different noise levels [fractional standard deviation (FSD) 10%, 1%, 0.1%, and 0.01%], and seven combinations of measurements at 30 min, 1 h, and 4 h p.i. PBPK model parameters were fitted to the simulated noisy PET data using population-based Bayesian parameters to construct predicted MPPs. Therapy simulations were performed as 30 min infusion of (90)Y-DOTATATE of 3.3 GBq in both true and predicted MPPs. Prediction accuracy was then calculated as relative variability vorgan between TIACs from both MPPs.

RESULTS

Large variability values of one time-point protocols [e.g., FSD = 1%, 240 min p.i., vkidneys = (9 ± 6)%, and vtumor = (27 ± 26)%] show inaccurate prediction. Accurate TIAC prediction of the kidneys was obtained for the case of two measurements (1 and 4 h p.i.), e.g., FSD = 1%, vkidneys = (7 ± 3)%, and vtumor = (22 ± 10)%, or three measurements, e.g., FSD = 1%, vkidneys = (7 ± 3)%, and vtumor = (22 ± 9)%.

CONCLUSIONS

(68)Ga-DOTATATE-PET measurements could possibly be used to predict the TIACs of (90)Y-DOTATATE when using a PBPK model and population-based Bayesian parameters. The two time-point measurement at 1 and 4 h p.i. with a noise up to FSD = 1% allows an accurate prediction of the TIACs in kidneys.

Radioimmunotherapy for Prostate Cancer--Current Status and Future Possibilities

Prostate cancer (PCa) is one of the most common cancers in men and is the second leading cause of cancer-related deaths in the USA. In the United States, it is the second most frequently diagnosed cancer after skin cancer, and in Europe it is number one. According to the American Cancer Society, approximately 221,000 men in the United States would be diagnosed with PCa during 2015, and approximately 28,000 would die of the disease. According to the International Agency for Research on Cancer, approximately 345,000 men were diagnosed with PCa in Europe during 2012, and despite more emphasis placed on early detection through routine screening, 72,000 men died of the disease. Hence, the need for improved therapy modalities is of utmost importance. And targeted therapies based on radiolabeled specific antibodies or peptides are a very interesting and promising alternative to increase the therapeutic efficacy and overall chance of survival of these patients. There are currently several preclinical and some clinical studies that have been conducted, or are ongoing, to investigate the therapeutic efficacy and toxicity of radioimmunotherapy (RIT) against PCa. One thing that is lacking in a lot of these published studies is the dosimetry data, which are needed to compare results between the studies and the study locations. Given the complicated tumor microenvironment and overall complexity of RIT to PCa, old and new targets and targeting strategies like combination RIT and pretargeting RIT are being improved and assessed along with various therapeutic radionuclides candidates. Given alone or in combination with other therapies, these new and improved strategies and RIT tools further enhance the clinical response to RIT drugs in PCa, making RIT for PCa an increasingly practical clinical tool.

Sensitivity Analysis of a Physiologically Based Pharmacokinetic Model Used for Treatment Planning in Peptide Receptor Radionuclide Therapy

The aim of this work was to evaluate the sensitivity of time-integrated activity coefficients (TIACs) on the erroneously chosen prior knowledge in a physiologically based pharmacokinetic (PBPK) model used for treatment planning in peptide receptor radionuclide therapy (PRRT). Parameters of the PBPK model were fitted to the biokinetic data of 15 patients after the injection of (111)In-DTPAOC. The fittings were performed using fixed parameter values taken from literature as prior knowledge (reference case, Ref). The fixed parameters were gender, physical information (e.g., body weight), dissociation rate koff, dissociation constant KD, fraction of blood flow, and spleen and liver volumes. The fittings were repeated with changed fixed parameters (Changed). The relative deviations (RDs) of TIACs calculated from Changed and Ref were analyzed for kidneys, tumor, liver, spleen, remainder, whole body, and serum. A changed koff has the largest effect on RD, the largest RD values were found for changed koff = 0.001 L/min: RDkidneys = (3 ± 3)%, RDtumor = (0.5 ± 4)%, RDliver = (6 ± 9)%, RDspleen = (5 ± 5)%, RDremainder = (2 ± 31)%, RDserum = (-4 ± 25)%, and RDwholebody = (3 ± 16)%. For other changed parameters, the maximum RDs were <1%. The calculation of organ TIACs in PRRT using the PBPK model was little affected by assigning wrong prior knowledge to the evaluated patients. The calculation of bone marrow-absorbed doses could be affected by the inaccurate TIACs of serum and remainder in the case of an inadequate koff.

Dependence of treatment planning accuracy in peptide receptor radionuclide therapy on the sampling schedule

Background

Peptide receptor radionuclide therapy (PRRT) plays an important role in the treatment of neuroendocrine tumors (NET). Pre-therapeutic dosimetry using the area under the measured time-activity curve (AUC) is important. The sampling schedule for this dosimetry determines the accuracy and reliability of the obtained AUC.

The aim of this study was to investigate the effect of reduced number of measurement points (i.e., gamma camera image acquisition or serum measurements) on treatment planning accuracy in PRRT using ¹¹¹In-labeled-diethylenetriaminopentaacetic acid-octreotide (DTPAOC; Octreoscan™).

Methods

Pre-therapeutic biokinetic data of 15 NET patients were investigated using a recently developed physiologically based pharmacokinetic (PBPK) model. Two parameter sets were determined (standard or iterative approach) and used for calculation of time-integrated activity coefficients (TIACs) for the tumor, kidneys, liver, spleen, serum, and whole body. TIACs obtained using the full data sets were used as reference. To evaluate the effect of sampling on individual treatment planning, reduced sampling schedules were generated omitting either 1, 2, 3, or 4 organ and serum measurements or all serum measurements for each patient. Relative deviations (RDs) between these and reference TIACs were calculated and used as criterion for treatment planning accuracy. An RD < 0.1 was considered acceptable.

Results

When omitting serum measurements, TIAC accuracy remained acceptable (RD < 0.1) for the standard approach. The kidney TIACs could be estimated for both approaches with acceptable RDs using two time points (t = 4 h; 2 d); tumor RDs were <0.3. The iterative approach reduced the range of RD, but did not further reduce the number of needed measurement points (i.e., to achieve an RD <0.1). For both approaches RDs for liver, spleen and whole body were larger than 0.1. However, in the clinical setting these RDs are less relevant as liver and spleen are not organs at risk due to the low absorbed doses.

Conclusions

When using a priori information of a PBPK model structure combined with Bayesian information about PBPK model parameter distribution, the administered activity could be determined with acceptable accuracy using only two time points (4 h, 2 d) and thus allow a considerable reduction of needed data for individual dosimetry.

The role of patient-based treatment planning in peptide receptor radionuclide therapy

BACKGROUND

Accurate treatment planning is recommended in peptide-receptor radionuclide therapy (PRRT) to minimize the toxicity to organs at risk while maximizing tumor cell sterilization. The aim of this study was to quantify the effect of different degrees of individualization on the prediction accuracy of individual therapeutic biodistributions in patients with neuroendocrine tumors (NETs).

METHODS

A recently developed physiologically based pharmacokinetic (PBPK) model was fitted to the biokinetic data of 15 patients with NETs after pre-therapeutic injection of (111)In-DTPAOC. Mathematical phantom patients (MPP) were defined using the assumed true (true MPP), mean (MPP 1A) and median (MPP 1B) parameter values of the patient group. Alterations of the degree of individualization were introduced to both mean and median patients by including patient-specific information as a priori knowledge: physical parameters and hematocrit (MPP 2A/2B). Successively, measurable individual biokinetic parameters were added: tumor volume V tu (MPP 3A/3B), glomerular filtration rate GFR (MPP 4A/4B), and tumor perfusion f tu (MPP 5A/5B). Furthermore, parameters of MPP 5A/5B and a simulated (68)Ga-DOTATATE PET measurement 60 min p.i. were used together with the population values used as Bayesian parameters (MPP 6A/6B). Therapeutic biodistributions were simulated assuming an infusion of (90)Y-DOTATATE (3.3 GBq) over 30 min to all MPPs. Time-integrated activity coefficients were predicted for all MPPs and compared to the true MPPs for each patient in tumor, kidneys, spleen, liver, remainder, and whole body to obtain the relative differences RD.

RESULTS

The large RD values of MPP 1A [RDtumor = (625 ± 1266)%, RDkidneys = (11 ± 38)%], and MPP 1B [RDtumor = (197 ± 505)%, RDkidneys = (11 ± 39)%] demonstrate that individual treatment planning is needed due to large physiological differences between patients. Although addition of individual patient parameters reduced the deviations considerably [MPP 5A: RDtumor = (-2 ± 27)% and RDkidneys = (16 ± 43)%; MPP 5B: RDtumor = (2 ± 28)% and RDkidneys = (7 ± 40)%] errors were still large. For the kidneys, prediction accuracy was considerably improved by including the PET measurement [MPP 6A/MPP 6B: RDtumor = (-2 ± 22)% and RDkidneys = (-0.1 ± 0.5)%].

CONCLUSION

Individualized treatment planning is needed in the investigated patient group. The use of a PBPK model and the inclusion of patient specific data, e.g., weight, tumor volume, and glomerular filtration rate, do not suffice to predict the therapeutic biodistribution. Integrating all available a priori information in the PBPK model and using additionally PET data measured at one time point for tumor, kidneys, spleen, and liver could possibly be sufficient to perform an individualized treatment planning.

SPECT- and PET-based patient-tailored treatment in neuroendocrine tumors: a comprehensive multidisciplinary team approach

The overexpression of somatostatin receptors on the tumor cell surface of neuroendocrine tumors (NETs) detected by multimodal functional imaging modalities such as SPECT and PET tracers constitutes a therapeutic option using targeting radiolabeled compounds. We will introduce the theranostic concept in general, explain in more detail its development in NETs, and discuss available SPECT and PET tracers regarding their potential for diagnostic imaging, visualization of target expression, and treatment tailoring. Moreover, we will discuss the currently available peptide receptor radionuclide therapy principles and compare them to previously published studies. Finally, we will discuss which new concepts will most likely influence the theranostic treatment approach in NETs in the future.

Pre-therapeutic dosimetry of normal organs and tissues of (177)Lu-PSMA-617 prostate-specific membrane antigen (PSMA) inhibitor in patients with castration-resistant prostate cancer

PURPOSE

(177)Lu-617-prostate-specific membrane antigen (PSMA) ligand seems to be a promising tracer for radionuclide therapy of progressive prostate cancer. However, there are no published data regarding the radiation dose given to the normal tissues. The aim of the present study was to estimate the pretreatment radiation doses in patients who will undergo radiometabolic therapy using a tracer amount of (177)Lu-labeled PSMA ligand.

METHODS

The study included seven patients with progressive prostate cancer with a mean age of 63.9 ± 3.9 years. All patients had prior PSMA positron emission tomography (PET) imaging and had intense tracer uptake at the lesions. The injected (177)Lu-PSMA-617 activity ranged from 185 to 210 MBq with a mean of 192.6 ± 11.0 MBq. To evaluate bone marrow absorbed dose 2-cc blood samples were withdrawn in short variable times (3, 15, 30, 60, and 180 min and 24, 48, and 120 h) after injection. Whole-body images were obtained at 4, 24, 48, and 120 h post-injection (p.i.). The geometric mean of anterior and posterior counts was determined through region of interest (ROI) analysis. Attenuation correction was applied using PSMA PET/CT images. The OLINDA/EXM dosimetry program was used for curve fitting, residence time calculation, and absorbed dose calculations.

RESULTS

The calculated radiation-absorbed doses for each organ showed substantial variation. The highest radiation estimated doses were calculated for parotid glands and kidneys. Calculated radiation-absorbed doses per megabecquerel were 1.17 ± 0.31 mGy for parotid glands and 0.88 ± 0.40 mGy for kidneys. The radiation dose given to the bone marrow was significantly lower than those of kidney and parotid glands (p < 0.05). The calculated radiation dose to bone marrow was 0.03 ± 0.01 mGy/MBq.

CONCLUSION

Our first results suggested that (177)Lu-PSMA-617 therapy seems to be a safe method. The dose-limiting organ seems to be the parotid glands rather than kidneys and bone marrow. The lesion radiation doses are within acceptable ranges; however, there is a substantial individual variance so patient dosimetry seems to be mandatory.

Population-Based Modeling Improves Treatment Planning Before (90)Y-Labeled Anti-CD66 Antibody Radioimmunotherapy

For treatment planning in radioimmunotherapy (RIT), the accurate estimation of time-integrated activity coefficients (TIACs) is essential. To estimate the TIACs in RIT using (90)Y-labeled anti-CD66 antibodies, physiologically based pharmacokinetic (PBPK) models are advantageous. Further optimization in predicting therapeutic TIACs may be achieved by including population-specific parameters. Therefore, the aims of this work were (1) to estimate population parameters and (2) to show the effect of these parameters on prediction accuracy of therapeutic biodistributions. To estimate population values, a PBPK model was fitted to pretherapeutic (gamma camera and serum) and therapeutic (serum) measurements simultaneously using the standard two-stage (STS) and iterated two-stage (ITS) algorithms. Including the estimated population values as Bayesian information, the model parameters of each patient were fitted to pretherapeutic data only (simulating therapeutic TIACs). To validate the prediction accuracy of the therapeutic serum curve, the simulated and fitted TIACs were compared. Prediction accuracy expressed as relative deviation (RD) improved from RD=8%±16% to RD=0%±10% for STS and ITS, respectively. The authors demonstrated a method to estimate and apply population values for RIT using a PBPK model and population fitting. For (90)Y-labeled anti-CD66 antibodies, the prediction accuracy was substantially improved.

Patient dosimetry in nuclear medicine

In diagnostic nuclear medicine, the biokinetics of the radiopharmaceutical (actually of the radionuclide) is determined for a number of representative patients. At therapy, it is essential to determine the patient's individual biokinetics of the radiopharmaceutical in order to calculate the absorbed doses to critical normal organs/tissues and to the target volume(s) with high accuracy. For the diagnostic situations, there is still a lack of quantitative determinations of the organ/tissue contents of radiopharmaceuticals and their variation with time. Planar gamma camera imaging using the conjugate view technique combined with a limited number of SPECT/CT images is the main method for such studies. In a similar way, PET/CT is used for 3D image-based internal dosimetry for PET substances. The transition from stylised reference phantoms to voxel phantoms will lead to improved dose estimates for diagnostic procedures. Examples of dose coefficients and effective doses for diagnostic substances are given. For the therapeutic situation, a pre-therapeutic low activity administration is used for quantitative measurements of organ/tissue distribution data by a gamma camera or a SPECT- or PET-unit. Together with CT and/or MR images this will be the base for individual dose calculations using Monte Carlo technique. Treatments based on administered activity should only be used if biological variations between patients are small or if a pre-therapeutic activity administration is impossible.

DNA damage in blood lymphocytes in patients after (177)Lu peptide receptor radionuclide therapy

Purpose

The aim of the study was to investigate DNA double strand break (DSB) formation and its correlation with the absorbed dose to the blood lymphocytes of patients undergoing their first peptide receptor radionuclide therapy (PRRT) with ¹⁷⁷Lu-labelled DOTATATE/DOTATOC.

Methods

The study group comprised 16 patients receiving their first PRRT. At least six peripheral blood samples were obtained before, and between 0.5 h and 48 h after radionuclide administration. From the time–activity curves of the blood and the whole body, residence times for blood self-irradiation and whole-body irradiation were determined. Peripheral blood lymphocytes were isolated, fixed with ethanol and subjected to immunofluorescence staining for colocalizing γ-H2AX/53BP1 DSB-marking foci. The average number of DSB foci per cell per patient sample was determined as a function of the absorbed dose to the blood and compared with an in vitro calibration curve established in our laboratory with ¹³¹I and ¹⁷⁷Lu.

Results

The average number of radiation-induced foci (RIF) per cell increased over the first 5 h after radionuclide administration and decreased thereafter. A linear fit from 0 to 5 h as a function of the absorbed dose to the blood agreed with our in vitro calibration curve. At later time-points the number of RIF decreased, indicating progression of DNA repair.

Conclusion

Measurements of RIF and the absorbed dose to the blood after systemic administration of ¹⁷⁷Lu may be used to obtain data on the individual dose–response relationships in vivo. Individual patient data were characterized by a linear dose-dependent increase and an exponential decay function describing repair.

Electronic supplementary material

The online version of this article (doi:10.1007/s00259-015-3083-9) contains supplementary material, which is available to authorized users.

Monte Carlo Calculation of Radioimmunotherapy with (90)Y-, (177)Lu-, (131)I-, (124)I-, and (188)Re-Nanoobjects: Choice of the Best Radionuclide for Solid Tumour Treatment by Using TCP and NTCP Concepts

Radioimmunotherapy has shown that the use of monoclonal antibodies combined with a radioisotope like ¹³¹I or ⁹⁰Y still remains ineffective for solid and radioresistant tumour treatment. Previous simulations have revealed that an increase in the number of ⁹⁰Y labelled to each antibody or nanoobject could be a solution to improve treatment output. It now seems important to assess the treatment output and toxicity when radionuclides such as ⁹⁰Y, ¹⁷⁷Lu, ¹³¹I, ¹²⁴I, and ¹⁸⁸Re are used. Tumour control probability (TCP) and normal tissue complication probability (NTCP) curves versus the number of radionuclides per nanoobject were computed with MCNPX to evaluate treatment efficacy for solid tumours and to predict the incidence of surrounding side effects. Analyses were carried out for two solid tumour sizes of 0.5 and 1.0 cm radius and for nanoobject (i.e., a radiolabelled antibody) distributed uniformly or nonuniformly throughout a solid tumour (e.g., Non-small-cell-lung cancer (NSCLC)). ⁹⁰Y and ¹⁸⁸Re are the best candidates for solid tumour treatment when only one radionuclide is coupled to one carrier. Furthermore, regardless of the radionuclide properties, high values of TCP can be reached without toxicity if the number of radionuclides per nanoobject increases.

The NUKDOS software for treatment planning in molecular radiotherapy

The aim of this work was the development of a software tool for treatment planning prior to molecular radiotherapy, which comprises all functionality to objectively determine the activity to administer and the pertaining absorbed doses (including the corresponding error) based on a series of gamma camera images and one SPECT/CT or probe data. NUKDOS was developed in MATLAB. The workflow is based on the MIRD formalism For determination of the tissue or organ pharmacokinetics, gamma camera images as well as probe, urine, serum and blood activity data can be processed. To estimate the time-integrated activity coefficients (TIAC), sums of exponentials are fitted to the time activity data and integrated analytically. To obtain the TIAC on the voxel level, the voxel activity distribution from the quantitative 3D SPECT/CT (or PET/CT) is used for scaling and weighting the TIAC derived from the 2D organ data. The voxel S-values are automatically calculated based on the voxel-size of the image and the therapeutic nuclide ((90)Y, (131)I or (177)Lu). The absorbed dose coefficients are computed by convolution of the voxel TIAC and the voxel S-values. The activity to administer and the pertaining absorbed doses are determined by entering the absorbed dose for the organ at risk. The overall error of the calculated absorbed doses is determined by Gaussian error propagation. NUKDOS was tested for the operation systems Windows(®) 7 (64 Bit) and 8 (64 Bit). The results of each working step were compared to commercially available (SAAMII, OLINDA/EXM) and in-house (UlmDOS) software. The application of the software is demonstrated using examples form peptide receptor radionuclide therapy (PRRT) and from radioiodine therapy of benign thyroid diseases. For the example from PRRT, the calculated activity to administer differed by 4% comparing NUKDOS and the final result using UlmDos, SAAMII and OLINDA/EXM sequentially. The absorbed dose for the spleen and tumour differed by 7% and 8%, respectively. The results from the example from radioiodine therapy of benign thyroid diseases and the example given in the latest corresponding SOP were identical. The implemented, objective methods facilitate accurate and reproducible results. The software is freely available.