Joost C. Haeck

Facilitating Tumor Functional Assessment by Spatially Relating 3D Tumor Histology and In Vivo MRI: Image Registration Approach

Background Magnetic resonance imaging (MRI), together with histology, is widely used to diagnose and to monitor treatment in oncology. Spatial correspondence between these modalities provides information about the ability of MRI to characterize cancerous tissue. However, registration is complicated by deformations during pathological processing, and differences in scale and information content. Methodology/Principal Findings This study proposes a methodology for establishing an accurate 3D relation between histological sections and high resolution in vivo MRI tumor data. The key features of the methodology are: 1) standardized acquisition and processing, 2) use of an intermediate ex vivo MRI, 3) use of a reference cutting plane, 4) dense histological sampling, 5) elastic registration, and 6) use of complete 3D data sets. Five rat pancreatic tumors imaged by T2*-w MRI were used to evaluate the proposed methodology. The registration accuracy was assessed by root mean squared (RMS) distances between manually annotated landmark points in both modalities. After elastic registration the average RMS distance decreased from 1.4 to 0.7 mm. The intermediate ex vivo MRI and the reference cutting plane shared by all three 3D images (in vivo MRI, ex vivo MRI, and 3D histology data) were found to be crucial for the accurate co-registration between the 3D histological data set and in vivo MRI. The MR intensity in necrotic regions, as manually annotated in 3D histology, was significantly different from other histologically confirmed regions (i.e., viable and hemorrhagic). However, the viable and the hemorrhagic regions showed a large overlap in T2*-w MRI signal intensity. Conclusions The established 3D correspondence between tumor histology and in vivo MRI enables extraction of MRI characteristics for histologically confirmed regions. The proposed methodology allows the creation of a tumor database of spatially registered multi-spectral MR images and multi-stained 3D histology.

Can DCE-MRI explain the heterogeneity in radiopeptide uptake imaged by SPECT in a pancreatic neuroendocrine tumor model?

Although efficient delivery and distribution of treatment agents over the whole tumor is essential for successful tumor treatment, the distribution of most of these agents cannot be visualized. However, with single-photon emission computed tomography (SPECT), both delivery and uptake of radiolabeled peptides can be visualized in a neuroendocrine tumor model overexpressing somatostatin receptors. A heterogeneous peptide uptake is often observed in these tumors. We hypothesized that peptide distribution in the tumor is spatially related to tumor perfusion, vessel density and permeability, as imaged and quantified by DCE-MRI in a neuroendocrine tumor model. Four subcutaneous CA20948 tumor-bearing Lewis rats were injected with the somatostatin-analog 111In-DTPA-Octreotide (50 MBq). SPECT-CT and MRI scans were acquired and MRI was spatially registered to SPECT-CT. DCE-MRI was analyzed using semi-quantitative and quantitative methods. Correlation between SPECT and DCE-MRI was investigated with 1) Spearman’s rank correlation coefficient; 2) SPECT uptake values grouped into deciles with corresponding median DCE-MRI parametric values and vice versa; and 3) linear regression analysis for median parameter values in combined datasets. In all tumors, areas with low peptide uptake correlated with low perfusion/density/ /permeability for all DCE-MRI-derived parameters. Combining all datasets, highest linear regression was found between peptide uptake and semi-quantitative parameters (R2>0.7). The average correlation coefficient between SPECT and DCE-MRI-derived parameters ranged from 0.52-0.56 (p<0.05) for parameters primarily associated with exchange between blood and extracellular extravascular space. For these parameters a linear relation with peptide uptake was observed. In conclusion, the ‘exchange-related’ DCE-MRI-derived parameters seemed to predict peptide uptake better than the ‘contrast amount- related’ parameters. Consequently, fast and efficient diffusion through the vessel wall into tissue is an important factor for peptide delivery. DCE-MRI helps to elucidate the relation between vascular characteristics, peptide delivery and treatment efficacy, and may form a basis to predict targeting efficiency.

Multi-modal image registration: matching MRI with histology

Spatial correspondence between histology and multi sequence MRI can provide information about the capabilities of non-invasive imaging to characterize cancerous tissue. However, shrinkage and deformation occurring during the excision of the tumor and the histological processing complicate the co registration of MR images with histological sections. This work proposes a methodology to establish a detailed 3D relation between histology sections and in vivo MRI tumor data. The key features of the methodology are a very dense histological sampling (up to 100 histology slices per tumor), mutual information based non-rigid B-spline registration, the utilization of the whole 3D data sets, and the exploitation of an intermediate ex vivo MRI. In this proof of concept paper, the methodology was applied to one tumor. We found that, after registration, the visual alignment of tumor borders and internal structures was fairly accurate. Utilizing the intermediate ex vivo MRI, it was possible to account for changes caused by the excision of the tumor: we observed a tumor expansion of 20%. Also the effects of fixation, dehydration and histological sectioning could be determined: 26% shrinkage of the tumor was found. The annotation of viable tissue, performed in histology and transformed to the in vivo MRI, matched clearly with high intensity regions in MRI. With this methodology, histological annotation can be directly related to the corresponding in vivo MRI. This is a vital step for the evaluation of the feasibility of multi-spectral MRI to depict histological groundtruth.

Computed tomography and magnetic resonance imaging.

Imaging in Oncology is rapidly moving from the detection and size measurement of a lesion to the quantitative assessment of metabolic processes and cellular and molecular interactions. Increasing insights into cancer as a complex disease with involvement of the tumor stroma in tumor pathobiological processes have made it clear that for successful control of cancer, treatment strategies should not only be directed at the tumor cells but also targeted at the tumor microenvironment. This requires understanding of the complex molecular and cellular interactions in cancer tissue. Recent developments in imaging technology have increased the possibility to image various pathobiological processes in cancer development and response to treatment. For computed tomography (CT) and magnetic resonance imaging (MRI) various improvements in hardware, software, and imaging probes have lifted these modalities from classical anatomical imaging techniques to techniques suitable to image and quantify various physiological processes and molecular and cellular interactions. Next to a more general overview of possible imaging targets in oncology this chapter provides an overview of the various developments in CT and MRI technology and some specific applications.

Magnetic resonance detection of CD34+ cells from umbilical cord blood using a 19F label

Impaired homing and delayed recovery upon hematopoietic stem cell transplantation (HSCT) with hematopoietic stem cells (HSC) derived from umbilical cord blood (UCB) is a major problem. Tracking transplanted cells in vivo will be helpful to detect impaired homing at an early stage and allows early interventions to improve engraftment and outcome after transplantation. In this study, we show sufficient intracellular labeling of UCB-derived CD34+ cells, with 19F-containing PLGA nanoparticles which were detectable with both flow cytometry and magnetic resonance spectroscopy (MRS). In addition, labeled CD34+ cells maintain their capacity to proliferate and differentiate, which is pivotal for successful engraftment after transplantation in vivo. These results set the stage for in vivo tracking experiments, through which the homing efficiency of transplanted cells can be studied.

Optimized time-resolved imaging of contrast kinetics (TRICKS) in dynamic contrast-enhanced MRI after peptide receptor radionuclide therapy in small animal tumor models

Anti-tumor efficacy of targeted peptide-receptor radionuclide therapy (PRRT) relies on several factors, including functional tumor vasculature. Little is known about the effect of PRRT on tumor vasculature. With dynamic contrast-enhanced (DCE-) MRI, functional vasculature is imaged and quantified using contrast agents. In small animals DCE-MRI is a challenging application. We optimized a clinical sequence for fast hemodynamic acquisitions, time-resolved imaging of contrast kinetics (TRICKS), to obtain DCE-MRI images at both high spatial and high temporal resolution in mice and rats. Using TRICKS, functional vasculature was measured prior to PRRT and longitudinally to investigate the effect of treatment on tumor vascular characteristics. Nude mice bearing H69 tumor xenografts and rats bearing syngeneic CA20948 tumors were used to study perfusion following PRRT administration with (177) lutetium octreotate. Both semi-quantitative and quantitative parameters were calculated. Treatment efficacy was measured by tumor-size reduction. Optimized TRICKS enabled MRI at 0.032 mm(3) voxel size with a temporal resolution of less than 5 s and large volume coverage, a substantial improvement over routine pre-clinical DCE-MRI studies. Tumor response to therapy was reflected in changes in tumor perfusion/permeability parameters. The H69 tumor model showed pronounced changes in DCE-derived parameters following PRRT. The rat CA20948 tumor model showed more heterogeneity in both treatment outcome and perfusion parameters. TRICKS enabled the acquisition of DCE-MRI at both high temporal resolution (Tres ) and spatial resolutions relevant for small animal tumor models. With the high Tres enabled by TRICKS, accurate pharmacokinetic data modeling was feasible. DCE-MRI parameters revealed changes over time and showed a clear relationship between tumor size and Ktrans .

Developing a tool for the validation of quantitative DCE-MRI

Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) is becoming an indispensable tool to non-invasively study tumor characteristics. However, many different DCE-analysis methods are currently being used. To compare and validate different methods, histology is the gold standard. For this purpose, exact co-localization between histology and MRI images is a prerequisite. In this study a methodology is developed to validate DCE-data with histology with an emphasis on correct registration of DCE-MRI and histological data. A pancreatic tumor was grown in a rat model. The tumor was dissected after MR imaging, embedded in paraffin, and cut into thin slices. These slices were stained with haematoxylin and eosin, digitized and stacked in a 3D volume. Next, the 3D histology was registered to ex-vivo SWI-weighted MR images, which in turn were registered to in-vivo SWI and DCE images to achieve correct co-localization. Semi-quantitative and quantitative parameters were calculated. Preliminary results suggest that both pharmacokinetic and heuristic DCE-parameters can discriminate between vital and non-vital tumor regions. The developed method offers the basis for an accurate spatial correlation between DCE-MRI derived parametric maps and histology, and facilitates the evaluation of different DCE-MRI analysis methods.

SSTR-mediated Breast Cancer Imaging: Is There A Role For Radiolabeled SSTR Antagonists?

Recent studies have shown enhanced tumor targeting by novel somatostatin receptor (SSTR) antagonists compared with clinically widely used agonists. However, these results have been obtained mostly in neuroendocrine tumors, and only limited data are available for cancer types with lower SSTR expression, including breast cancer (BC). To date, two studies have reported higher binding of the antagonist than the agonist in BC, but in both studies only a limited number of cases were evaluated. In this preclinical study, we further investigated whether the application of an SSTR antagonist can improve SSTR-mediated BC imaging in a large panel of BC specimens. We also generated an in vivo BC mouse model and performed SPECT/MRI and biodistribution studies. Methods: Binding of 111In-DOTA-Tyr3-octreotate (SSTR agonist) and 111In-DOTA-JR11 (SSTR antagonist) to 40 human BC specimens was compared using in vitro autoradiography. SSTR2 immunostaining was performed to confirm SSTR2 expression of the tumor cells. Furthermore, binding of the radiolabeled SSTR agonist and antagonist was analyzed in tissue material from 6 patient-derived xenografts. One patient-derived xenograft, the estrogen receptor–positive model T126, was chosen to generate in vivo mouse models containing orthotopic breast tumors for in vivo SPECT/MRI and biodistribution studies after injection with 177Lu-DOTA-Tyr3-octreotate or 177Lu-DOTA-JR11. Results: 111In-DOTA-JR11 binding to human BC tissue was significantly higher than 111In-DOTA-Tyr3-octreotate binding (P < 0.001). The median ratio of antagonist binding versus agonist binding was 3.39 (interquartile range, 2–5). SSTR2 immunostaining confirmed SSTR2 expression on the tumor cells. SPECT/MRI of the mouse model found better tumor visualization with the antagonist. This result was in line with the significantly higher tumor uptake of the radiolabeled antagonist than of the agonist as measured in biodistribution studies 285 min after radiotracer injection (percentage injected dose per gram of tissue: 1.92 ± 0.43 vs. 0.90 ± 0.17; P = 0.002). Conclusion: SSTR antagonists are promising candidates for BC imaging.

Abstract A5: Establishment of a novel spontaneous prostate cancer liver metastasis model

Objective: Prostate cancer (PCa) preferentially spreads to bone with additional metastasis to regional lymph nodes, lung and liver in a subset of patients. Understanding the determining factors of this metastatic pattern and to elucidate the involved pathways and interplay between epithelial and the stromal niche, we established a patient-derived xenograft (PDX) model showing spontaneous metastatic spread to all relevant organs and metastatic lesions in liver. Methods: PDX of PCa were obtained from NMRI tumor-bearing mice and 3D cell cultures were established by collagenase treatment o/n in prostate growth medium (PGM) containing 2% fetal calf serum. 3D cultures of PDX PC339 were established (PC339C) that could be passaged and stored in liquid nitrogen. PC339C was transfected with a lentiviral construct containing luc2-GFP fused protein (M21) and upon FACS-sorting the PC339C-M21 was retrieved. PC339C-M21 cells were inoculated subcutaneously in NSG mice and when tumors were established mice underwent tumorectomy (700-1000 mm3) to extend the life span of the animal and allow metastastic outgrowth. Mice were monitored weekly for metastatic outgrowth using IVIS, microCT and MRI. At sacrifice, M21-positive organs, including liver, lymph nodes, lung and bone, were sampled. Part of the tissue was used for culturing, part was fresh frozen for RNA and protein isolation to allow qPCR and western blot analyses, and part was formalin-fixed and paraffin-embedded for histology and immunohistochemistry. Results: PC339C-M21 tumors were established in NSG mice and after tumorectomy a spontaneous liver metastasis was established. Culturing of this tumor resulted in the PC339C-M21-L cell line. This cell line was inoculated subcutaneously in NSG mice to result in tumor formation and spontaneous liver metastasis in all mice. Tumorectomy was performed at 25-30 days after inoculation. Luciferase-positive signals in liver were detected by IVIS as early as 7 days after tumorectomy with MRI showing first visible lesions at 21 days and microCT at 28 days. 3D reconstructed microCT images and MRI-detected lesions confirmed the signals and location as seen by IVIS. When inoculated orthotopically, liver metastatic lesions became visible by IVIS at 40 days, which was not different from the subcutaneous tumors. However, these mice could not be followed longer due to the orthotopic tumor burden. On top of the PC339C-M21-L cell line, culturing of luciferase-positive organs resulted in cell lines from lymph node (PC339C-M21-L-LN), lung (PC339C-M21-L-LU), and bone (PC339C-M21-L-B) together constituting a cell line panel of metastatic PCa. This cell line panel is being further characterized in vitro and in vivo for their unique properties and metastatic potential. Conclusion: A new spontaneous PCa liver metastasis xenograft model, PCL339-M21-L, was generated. This model was extended to yield a panel of metastatic sublines from different organ origin. Together, this platform may serve as a tool to study the biology of spontaneous metastatic outgrowth in liver and other organs. Furthermore, we demonstrated the feasibility of non-invasive imaging techniques to follow metastatic spread and outgrowth allowing to monitor responses to anti-cancer therapies. Citation Format: Hanneke J.A.A. van Zoggel, Sigrun E. Erkens-Schulze, Corrina M.A. De Ridder, Wilma Teubel, Hamza Saleem, Stefan J. Roobol, Sander A.H. Hoeben, Yanto R. Ridwan, Joost C. Haeck, Wytske M. van Weerden{Authors}. Establishment of a novel spontaneous prostate cancer liver metastasis model. [abstract]. In: Proceedings of the AACR Special Conference: Patient-Derived Cancer Models: Present and Future Applications from Basic Science to the Clinic; Feb 11-14, 2016; New Orleans, LA. Philadelphia (PA): AACR; Clin Cancer Res 2016;22(16_Suppl):Abstract nr B21.

Quantification of DCE-MRI: A validation of three techniques with 3D-histology

Three different DCE-MRI quantification methods: model-free-based, compartment-model-based and principal component analysis, are compared by evaluating parameter maps for histological defined volumes of vital and non-vital tumor tissue. To obtain an accurate spatial correspondence between histology and DCE-MRI, a two-step registration process was used involving dense histological sampling, a reference plane and an intermediate ex vivo MRI. Results show that the model-free parameter washout and the second principal component score can adequately separate vital from non-vital tumor tissue, with an accuracy of respectively 99.2% and 99.7%. The other model-free parameters and the compartment-model-based Ktrans show some overlap in values between vital and non-vital tissue. The first, third and fourth pc-score have limited discriminative power.