Impact of PET and MRI threshold-based tumor volume segmentation on patient-specific targeted radionuclide therapy dosimetry using CLR1404

Theranostics and Patient-Specific Dosimetry

Radiopharmaceutical therapy (RPT) is an invigorated form of cancer therapy that systemically delivers targeted radioactive drugs to cancer cells. Theranostics is a type of RPT that utilizes imaging, either of the RPT drug directly or a companion diagnostic, to inform whether a patient will benefit from the treatment. Given the ability to image the drug onboard theranostic treatments also lends itself readily to patient-specific dosimetry, which is a physics-based process that determines the overall absorbed dose burden to healthy organs and tissues and tumors in patients. While companion diagnostics identify who will benefit from RPT treatments, dosimetry determines how much activity these beneficiaries can receive to maximize therapeutic efficacy. Clinical data is starting to accrue suggesting tremendous benefits when dosimetry is performed for RPT patients. RPT dosimetry, which was once performed by florid and often inaccurate workflows, can now be performed more efficiently and accurately with FDA-cleared dosimetry software. Therefore, there is no better time for the field of oncology to adopt this form of personalize medicine to improve outcomes for cancer patients.

Automated lung tumor delineation on positron emission tomography/computed tomography via a hybrid regional network

BACKGROUND

Multimodality positron emission tomography/computed tomography (PET/CT) imaging combines the anatomical information of CT with the functional information of PET. In the diagnosis and treatment of many cancers, such as non-small cell lung cancer (NSCLC), PET/CT imaging allows more accurate delineation of tumor or involved lymph nodes for radiation planning.

PURPOSE

In this paper, we propose a hybrid regional network method of automatically segmenting lung tumors from PET/CT images.

METHODS

The hybrid regional network architecture synthesizes the functional and anatomical information from the two image modalities, whereas the mask regional convolutional neural network (R-CNN) and scoring fine-tune the regional location and quality of the output segmentation. This model consists of five major subnetworks, that is, a dual feature representation network (DFRN), a regional proposal network (RPN), a specific tumor-wise R-CNN, a mask-Net, and a score head. Given a PET/CT image as inputs, the DFRN extracts feature maps from the PET and CT images. Then, the RPN and R-CNN work together to localize lung tumors and reduce the image size and feature map size by removing irrelevant regions. The mask-Net is used to segment tumor within a volume-of-interest (VOI) with a score head evaluating the segmentation performed by the mask-Net. Finally, the segmented tumor within the VOI was mapped back to the volumetric coordinate system based on the location information derived via the RPN and R-CNN. We trained, validated, and tested the proposed neural network using 100 PET/CT images of patients with NSCLC. A fivefold cross-validation study was performed. The segmentation was evaluated with two indicators: (1) multiple metrics, including the Dice similarity coefficient, Jacard, 95th percentile Hausdorff distance, mean surface distance (MSD), residual mean square distance, and center-of-mass distance; (2) Bland-Altman analysis and volumetric Pearson correlation analysis.

RESULTS

In fivefold cross-validation, this method achieved Dice and MSD of 0.84 ± 0.15 and 1.38 ± 2.2 mm, respectively. A new PET/CT can be segmented in 1 s by this model. External validation on The Cancer Imaging Archive dataset (63 PET/CT images) indicates that the proposed model has superior performance compared to other methods.

CONCLUSION

The proposed method shows great promise to automatically delineate NSCLC tumors on PET/CT images, thereby allowing for a more streamlined clinical workflow that is faster and reduces physician effort.

Radiotherapy Target Volume Definition in Newly Diagnosed High-Grade Glioma Using 18F-FET PET Imaging and Multiparametric MRI: An Inter Observer Agreement Study

Background: The aim of this prospective monocentric study was to assess the inter-observer agreement for tumor volume delineations by multiparametric MRI and 18-F-FET-PET/CT in newly diagnosed, untreated high-grade glioma (HGG) patients. Methods: Thirty patients HGG underwent O-(2-[18F]-fluoroethyl)-l-tyrosine(18F-FET) positron emission tomography (PET), and multiparametric MRI with computation of rCBV map and K2 map. Three nuclear physicians and three radiologists with different levels of experience delineated the 18-F-FET-PET/CT and 6 MRI sequences, respectively. Spatial similarity (Dice and Jaccard: DSC and JSC) and overlap (Overlap: OV) coefficients were calculated between the readers for each sequence. Results: DSC, JSC, and OV were high for 18F-FET PET/CT, T1-GD, and T2-FLAIR (>0.67). The Spearman correlation coefficient between readers was ≥0.6 for these sequences. Cross-comparison of similarity and overlap parameters showed significant differences for DSC and JSC between 18F-FET PET/CT and T2-FLAIR and for JSC between 18F-FET PET/CT and T1-GD with higher values for 18F-FET PET/CT. No significant difference was found between T1-GD and T2-FLAIR. rCBV, K2, b1000, and ADC showed correlation coefficients between readers <0.6. Conclusion: The interobserver agreements for tumor volume delineations were high for 18-F-FET-PET/CT, T1-GD, and T2-FLAIR. The DWI (b1000, ADC), rCBV, and K2-based sequences, as performed, did not seem sufficiently reproducible to be used in daily practice.

Generative Adversarial Network (GAN) for Automatic Reconstruction of the 3D Spine Structure by Using Simulated Bi-Planar X-ray Images

In this study, we modified the previously proposed X2CT-GAN to build a 2Dto3D-GAN of the spine. This study also incorporated the radiologist’s perspective in the adjustment of input signals to prove the feasibility of the automatic production of three-dimensional (3D) structures of the spine from simulated bi-planar two-dimensional (2D) X-ray images. Data from 1012 computed tomography (CT) studies of 984 patients were retrospectively collected. We tested this model under different dataset sizes (333, 666, and 1012) with different bone signal conditions to observe the training performance. A 10-fold cross-validation and five metrics—Dice similarity coefficient (DSC) value, Jaccard similarity coefficient (JSC), overlap volume (OV), and structural similarity index (SSIM)—were applied for model evaluation. The optimal mean values for DSC, JSC, OV, SSIM_anteroposterior (AP), and SSIM_Lateral (Lat) were 0.8192, 0.6984, 0.8624, 0.9261, and 0.9242, respectively. There was a significant improvement in the training performance under empirically enhanced bone signal conditions and with increasing training dataset sizes. These results demonstrate the potential of the clinical implantation of GAN for automatic production of 3D spine images from 2D images. This prototype model can serve as a foundation in future studies applying transfer learning for the development of advanced medical diagnostic techniques.

Interrelation between Spectral Online Monitoring and Postoperative T1-Weighted MRI in Interstitial Photodynamic Therapy of Malignant Gliomas

Simple Summary

Treatment monitoring is highly important for the delivery and control of brain tumor therapy. For interstitial photodynamic therapy (iPDT), an intraoperative spectral online monitoring (SOM) setup was established in former studies to monitor photosensitizer fluorescence and treatment light transmission during therapy. In this work, data from patients treated with iPDT as the initial treatment for newly diagnosed glioblastoma (n = 11) were retrospectively analyzed. Observed changes in treatment light transmission were assessed, and changes in optical tissue absorption were calculated out of these. In addition, magnetic resonance imaging (MRI) data were recorded within 48 h after therapy and showed intrinsic T1 hyperintensity in the treated area in non-contrast-enhanced T1-weighted sequences. A 3D co-registration of intrinsic T1 hyperintensity lesions and the light transmission zones between cylindrical diffuser fiber pairs showed that reduction in treatment light transmission corresponding to increased light absorption had a spatial correlation with post-therapeutic intrinsic T1 hyperintensity (p ≤ 0.003).

Abstract

In a former study, interstitial photodynamic therapy (iPDT) was performed on patients suffering from newly diagnosed glioblastoma (n = 11; 8/3 male/female; median age: 68, range: 40–76). The procedure includes the application of 5-ALA to selectively metabolize protoporphyrin IX (PpIX) in tumor cells and illumination utilizing interstitially positioned optical cylindrical diffuser fibers (CDF) (2–10 CDFs, 2–3 cm diffusor length, 200 mW/cm, 635 nm, 60 min irradiation). Intraoperative spectral online monitoring (SOM) was employed to monitor treatment light transmission and PpIX fluorescence during iPDT. MRI was used for treatment planning and outcome assessment. Case-dependent observations included intraoperative reduction of treatment light transmission and local intrinsic T1 hyperintensity in non-contrast-enhanced T1-weighted MRI acquired within one day after iPDT. Intrinsic T1 hyperintensity was observed and found to be associated with the treatment volume, which indicates the presence of methemoglobin, possibly induced by iPDT. Based on SOM data, the optical absorption coefficient and its change during iPDT were estimated for the target tissue volumes interjacent between evaluable CDF-pairs at the treatment wavelength of 635 nm. By spatial comparison and statistical analysis, it was found that observed increases of the absorption coefficient during iPDT were larger in or near regions of intrinsic T1 hyperintensity (p = 0.003). In cases where PpIX-fluorescence was undetectable before iPDT, the increase in optical absorption and intrinsic T1 hyperintensity tended to be less. The observations are consistent with in vitro experiments and indicate PDT-induced deoxygenation of hemoglobin and methemoglobin formation. Further investigations are needed to provide more data on the time course of the observed changes, thus paving the way for optimized iPDT irradiation protocols.

Low-Dose Radiation Potentiates the Propagation of Anti-Tumor Immunity against Melanoma Tumor in the Brain after In Situ Vaccination at a Tumor outside the Brain

Brain metastases develop in over 60% of advanced melanoma patients and negatively impact quality of life and prognosis. In a murine melanoma model, we previously showed that an in situ vaccination (ISV) regimen, combining radiation treatment and intratumoral (IT) injection of immunocytokine (IC: anti-GD2 antibody fused to IL2), along with the immune checkpoint inhibitor anti-CTLA-4, robustly eliminates peripheral flank tumors but only has modest effects on co-occurring intracranial tumors. In this study, we investigated the ability of low-dose radiation to the brain to potentiate anti-tumor immunity against a brain tumor when combined with ISV + anti-CTLA-4. B78 (GD2+, immunologically "cold") melanoma tumor cells were implanted into the flank and the right striatum of the brain in C57BL/6 mice. Flank tumors (50-150 mm3) were treated following a previously optimized ISV regimen [radiation (12 Gy × 1, treatment day 1), IT-IC (50 µg daily, treatment days 6-10), and anti-CTLA-4 (100 µg, treatment days 3, 6, 9)]. Mice that additionally received whole-brain radiation treatment (WBRT, 4 Gy × 1) on day 15 demonstrated significantly increased survival compared to animals that received ISV + anti-CTLA-4 alone, WBRT alone or no treatment (control) (P < 0.001, log-rank test). Timing of WBRT was critical, as WBRT administration on day 1 did not significantly enhance survival compared to ISV + anti-CTLA-4, suggesting that the effect of WBRT on survival might be mediated through immune modulation and not just direct tumor cell cytotoxicity. Modest increases in T cells (CD8+ and CD4+) and monocytes/macrophages (F4/80+) but no changes in FOXP3+ regulatory T cells (Tregs), were observed in brain melanoma tumors with addition of WBRT (on day 15) to ISV + anti-CTLA-4. Cytokine multiplex immunoassay revealed distinct changes in both intracranial melanoma and contralateral normal brain with addition of WBRT (day 15) to ISV + anti-CTLA-4, with notable significant changes in pro-inflammatory (e.g., IFNγ, TNFα and LIX/CXCL5) and suppressive (e.g., IL10, IL13) cytokines as well as chemokines (e.g., IP-10/CXCL10 and MIG/CXCL9). We tested the ability of the alkylphosphocholine analog, NM600, to deliver immunomodulatory radiation to melanoma brain tumors as a targeted radionuclide therapy (TRT). Yttrium-86 (86Y) chelated to NM600 was delivered intravenously by tail vein to mice harboring flank and brain melanoma tumors, and PET imaging demonstrated specific accumulation up to 72 h at each tumor site (∼12:1 brain tumor/brain and ∼8:1 flank tumor/muscle). When NM600 was chelated to therapeutic β-particle-emitting 90Y and administered on treatment day 13, T-cell infiltration and cytokine profiles were altered in melanoma brain tumor, like that observed for WBRT. Overall, our results demonstrate that addition of low-dose radiation, timed appropriately with ISV administration to tumors outside the brain, significantly increases survival in animals co-harboring melanoma brain tumors. This observation has potentially important translational implications as a treatment strategy for increasing the response of tumors in the brain to systemically administered immunotherapies.

Evaluation of a patient-specific algorithm for predicting distribution for convection-enhanced drug delivery into the brainstem of patients with diffuse intrinsic pontine glioma

OBJECTIVE

With increasing use of convection-enhanced delivery (CED) of drugs, the need for software that can predict infusion distribution has grown. In the context of a phase I clinical trial for pediatric diffuse intrinsic pontine glioma (DIPG), CED was used to administer an anti-B7H3 radiolabeled monoclonal antibody, iodine-124-labeled omburtamab. In this study, the authors retrospectively evaluated a software algorithm (iPlan Flow) for the estimation of infusate distribution based on the planned catheter trajectory, infusion parameters, and patient-specific MRI. The actual infusate distribution, as determined on MRI and PET imaging, was compared to the distribution estimated by the software algorithm. Similarity metrics were used to quantify the agreement between predicted and actual distributions.

METHODS

Ten pediatric patients treated at the same dose level in the NCT01502917 trial conducted at Memorial Sloan Kettering Cancer Center were considered for this retrospective analysis. T2-weighted MRI in combination with PET imaging was used to determine the distribution of infusate in this study. The software algorithm was applied for the generation of estimated fluid distribution maps. Similarity measures included object volumes, intersection volume, union volume, Dice coefficient, volume difference, and the center and average surface distances. Acceptable similarity was defined as a simulated distribution volume (Vd Sim) object that had a Dice coefficient higher than or equal to 0.7, a false-negative rate (FNR) lower than 50%, and a positive predictive value (PPV) higher than 50% compared to the actual Vd (Vd PET).

RESULTS

Data for 10 patients with a mean infusion volume of 4.29 ml (range 3.84-4.48 ml) were available for software evaluation. The mean Vd Sim found to be covered by the actual PET distribution (PPV) was 77% ± 8%. The mean percentage of PET volume found to be outside the simulated volume (FNR) was 34% ± 10%. The mean Dice coefficient was 0.7 ± 0.05. In 8 out of 10 patients, the simulation algorithm fulfilled the combined acceptance criteria for similarity.

CONCLUSIONS

iPlan Flow software can be useful to support planning of trajectories that produce intraparenchymal convection. The simulation algorithm is able to model the likely infusate distribution for a CED treatment in DIPG patients. The combination of trajectory planning guidelines and infusion simulation in the software can be used prospectively to optimize personalized CED treatment.

Radiotherapy target volume definition in newly diagnosed high grade glioma using 18F-FET PET imaging and multiparametric perfusion MRI: A prospective study (IMAGG)

PURPOSE

The aim of this study was to prospectively investigate tumor volume delineation by amino acid PET and multiparametric perfusion magnetic resonance imaging (MRI) in patients with newly diagnosed, untreated high grade glioma (HGG).

MATERIALS AND METHODS

Thirty patients with histologically confirmed HGG underwent O-(2-[¹⁸F]-fluoroethyl)-l-tyrosine (¹⁸F-FET) positron emission tomography (PET), conventional Magnetic Resonance Imaging (MRI) as contrast-enhanced (CE) and fluid-attenuated inversion recovery (FLAIR) and multiparametric MRI as relative cerebral blood volume (rCBV) and permeability estimation map (K2). Areas of MRI volumes were semi-automatically segmented. The percentage overlap volumes, Dice and Jaccard spatial similarity coefficients (OV, DSC, JSC) were calculated.

RESULTS

The ¹⁸F-FET tumor volume was significantly larger than the CE volume (median 43.5 mL (2.5-124.9) vs. 23.8 mL (1.4-80.3), p = 0.005). The OV between ¹⁸F-FET uptake and CE volume was low (median OV 0.59 (0.10-1)), as well as spatial similarity (median DSC 0.52 (0.07-0.78); median JSC 0.35 (0.03-0.64)). Twenty-five patients demonstrated both rCBV and CE on MRI: The median rCBV tumor volume was significantly smaller than the median CE volume (p < 0.001). The OV was high (median 0.83 (0.54-1)), but the spatial similarity was low (median DSC 0.45 (0.04-0.83); median JSC 0.29 (0.07-0.71)). Twenty-eight patients demonstrated both K2 and CE on MRI. The median K2 tumor volume was not significantly larger than the median CE volume. The OV was high (median OV 0.90 (0.61-1)), and the spatial similarity was moderate (median DSC 0.75 (0.01-0.83); median JSC 0.60 (0.11-0.89)).

CONCLUSION

We demonstrated that multiparametric perfusion MRI volumes (rCBV, K2) were highly correlated with CE T1 gadolinium volumes whereas ¹⁸F-FET PET provided complementary information, suggesting that the metabolically active tumor volume in patients with newly diagnosed untreated HGG is critically underestimated by contrast enhanced MRI. ¹⁸F-FET PET imaging may help to improve target volume delineation accuracy for radiotherapy planning.

Simultaneous FET-PET and contrast-enhanced MRI based on hybrid PET/MR improves delineation of tumor spatial biodistribution in gliomas: a biopsy validation study

Purpose

Glioma treatment planning requires precise tumor delineation, which is typically performed with contrast-enhanced (CE) MRI. However, CE MRI fails to reflect the entire extent of glioma. O-(2-¹⁸F-fluoroethyl)-L-tyrosine (¹⁸F-FET) PET may detect tumor volumes missed by CE MRI. We investigated the clinical value of simultaneous FET-PET and CE MRI in delineating tumor extent before treatment planning. Guided stereotactic biopsy was used to validate the findings.

Methods

Conventional MRI and ¹⁸F-FET PET were performed simultaneously on a hybrid PET/MR in 33 patients with histopathologically confirmed glioma. Tumor volumes were quantified using a tumor-to-brain ratio ≥ 1.6 (V_PET) and a visual threshold (V_CE). We visually assessed abnormal areas on FLAIR images and calculated Dice’s coefficient (DSC), overlap volume (OV), discrepancy-PET, and discrepancy-CE. Additionally, several stereotactic biopsy samples were taken from “matched” or “mismatched” FET-PET and CE MRI regions.

Results

Among 31 patients (93.94%), FET-PET delineated significantly larger tumor volumes than CE MRI (77.84 ± 51.74 cm³ vs. 34.59 ± 27.07 cm³, P < 0.05). Of the 21 biopsy samples obtained from regions with increased FET uptake, all were histopathologically confirmed as glioma tissue or tumor infiltration, whereas only 13 showed enhancement on CE MRI. Among all patients, the spatial similarity between V_PET and V_CE was low (average DSC 0.56 ± 0.22), while the overlap was high (average OV 0.95 ± 0.08). The discrepancy-CE and discrepancy-PET were lower than 10% in 28 and 0 patients, respectively. Eleven patients showed V_PET partially beyond abnormal signal areas on FLAIR images.

Conclusion

The metabolically active biodistribution of gliomas delineated with FET-PET significantly exceeds tumor volume on CE MRI, and histopathology confirms these findings. Our preliminary results indicate that combining the anatomic and molecular information obtained from conventional MRI and FET-PET would reveal a more accurate glioma extent, which is critical for individualized treatment planning.

Electronic supplementary material

The online version of this article (10.1007/s00259-019-04656-2) contains supplementary material, which is available to authorized users.

Pretreatment CLR 124 Positron Emission Tomography Accurately Predicts CLR 131 Three-Dimensional Dosimetry in a Triple-Negative Breast Cancer Patient

INTRODUCTION

CLR1404 is a theranostic molecular agent that can be radiolabeled with ¹²⁴I (CLR 124) for positron emission tomography (PET) imaging, or ¹³¹I (CLR 131) for single-photon emission computed tomography (SPECT) imaging and targeted radionuclide therapy. This pilot study evaluated a pretreatment dosimetry methodology in a triple-negative breast cancer patient who was uniquely enrolled in both a CLR 124 PET imaging clinical trial and a CLR 131 therapeutic dose escalation clinical trial.

MATERIALS AND METHODS

Three-dimensional PET/CT images were acquired at 1, 3, 24, 48, and 120 h postinjection of 178 MBq CLR 124. One month later, pretherapy 2D whole-body planar images were acquired at 0.25, 5, 24, 48, and 144 h postinjection of 370 MBq CLR 131. Following the therapeutic administration of 1990 MBq CLR 131, 3D SPECT/CT images were acquired at 74, 147, 334, and 505 h postinjection. The therapeutic CLR 131 voxel-level absorbed dose was estimated from PET (RAPID PET) and SPECT (RAPID SPECT) images using a Geant4-based Monte Carlo dosimetry platform called RAPID (Radiopharmaceutical Assessment Platform for Internal Dosimetry), and region of interest (ROI) mean doses were also estimated using the OLINDA/EXM software based on PET (OLINDA PET), SPECT (OLINDA SPECT), and planar (OLINDA planar) images.

RESULTS

The RAPID PET and OLINDA PET tracer-predicted ROI mean doses correlated well (m ≥ 0.631, R² ≥ 0.694, p ≤ 0.01) with both the RAPID SPECT and OLINDA SPECT therapeutic mean doses. The 2D planar images did not have any significant correlations. The ROI mean doses differed by -4% to -43% between RAPID and OLINDA/EXM, and by -19% to 29% between PET and SPECT. The 3D dose distributions and dose volume histograms calculated with RAPID were similar for the PET/CT and SPECT/CT.

CONCLUSIONS

This pilot study demonstrated that CLR 124 pretreatment PET images can be used to predict CLR 131 3D therapeutic dosimetry better than CLR 131 2D planar images. In addition, unlike OLINDA/EXM, Monte Carlo dosimetry methods were capable of accurately predicting dose heterogeneity, which is important for predicting dose-response relationships and clinical outcomes.

FET PET reveals considerable spatial differences in tumour burden compared to conventional MRI in newly diagnosed glioblastoma

PURPOSE

Areas of contrast enhancement (CE) on MRI are usually the target for resection or radiotherapy target volume definition in glioblastomas. However, the solid tumour mass may extend beyond areas of CE. Amino acid PET can detect parts of the tumour that show no CE. We systematically investigated tumour volumes delineated by amino acid PET and MRI in patients with newly diagnosed, untreated glioblastoma.

METHODS

Preoperatively, 50 patients with neuropathologically confirmed glioblastoma underwent O-(2-[¹⁸F]-fluoroethyl)-L-tyrosine (FET) PET, and fluid-attenuated inversion recovery (FLAIR) and contrast-enhanced MRI. Areas of CE were manually segmented. FET PET tumour volumes were segmented using a tumour-to-brain ratio of ≥1.6. The percentage overlap volumes, and Dice and Jaccard spatial similarity coefficients (DSC, JSC) were calculated. FLAIR images were evaluated visually.

RESULTS

In 43 patients (86%), the FET tumour volume was significantly larger than the CE volume (21.5 ± 14.3 mL vs. 9.4 ± 11.3 mL; P < 0.001). Forty patients (80%) showed both increased uptake of FET and CE. In these 40 patients, the spatial similarity between FET uptake and CE was low (mean DSC 0.39 ± 0.21, mean JSC 0.26 ± 0.16). Ten patients (20%) showed no CE, and one of these patients showed no FET uptake. In five patients (10%), increased FET uptake was present outside areas of FLAIR hyperintensity.

CONCLUSION

Our results show that the metabolically active tumour volume delineated by FET PET is significantly larger than tumour volume delineated by CE. Furthermore, the results strongly suggest that the information derived from both imaging modalities should be integrated into the management of patients with newly diagnosed glioblastoma.

The Relevance of Dosimetry in Precision Medicine

The aim of this review is to provide an overview of the most recent technologic developments in state-of-the-art equipment and tools for dosimetry in radionuclide therapies. This includes, but is not restricted to, calibration methods for imaging systems. In addition, a summary of new developments that consider the influence of small-scale dosimetry and of biologic effects on radionuclide therapies is given. Finally, the current limitations of patient-specific dosimetry such as bone-marrow dosimetry or dosimetry of α-emitters are discussed.