Cancer Center Amsterdam

1.1 Artificial Intelligence (AI) for improved molecular imaging
People involved: Sandeep SV Golla, Maqsood Yaqub, Ronald Boellaard
(PhD) Students: Bart de Vries, Hanna Saadani, Ben Zwezerijnen, Vacancy

Molecular imaging with Positron Emission Tomography (PET) plays an important role for diagnosis and staging, treatment response prediction and monitoring in oncology. At present, PET image interpretation is mainly based on visual reads resulting in observer variability. Within the theme Imaging, several artificial intelligence driven image processing and classification projects are being undertaken to improve image interpretation and quantification. For example, at present a convolutional network has been trained to enhance the quality of images with poor image quality (high noise levels), such as seen with ⁸⁹Zr-labelled tracers, providing images with improved signal to noise ratio and with much less loss of signal as compared to conventional image filters.

1.2 Deep learning for improving MRI: RIMs, U-nets and unsupervised and physics-informed networks
People involved: Oliver J Gurney-Champion, Matthan Caan, Jaap Stoker, Hanneke van Laarhoven, Aart J Nederveen

We are implementing deep learning techniques for improving MRI acquisition speed and image quality, and automating workflows. We are using recurrent inference machines that reconstruct MR images from highly undersampled data, allowing substantially shorter scan times. Furthermore, we are using unsupervised physics-informed neural networks for modelling of quantitative MRI data, including IVIM and DCE. Such networks result in considerably nicer parameter maps (less noisy, sharper) then conventional approaches. Finally, U-nets enable us to automate contouring regions of interest, and can be used for denoising of images, too.

1.3 Quantitative MRI
People involved: Oliver J Gurney-Champion, Hanneke W.M. van Laarhoven, Jaap Stoker, Aart J Nederveen
PhD students: Nienke Wassenaar, Sophie van Schelt

We are able to characterize tumor microenvironment, including perfusion, diffusion and hypoxia, using various Magnetic Resonance Imaging (MRI), including 7T 31P MRS. Ultimately, we would like to use these biomarkers to predict which treatments will be most effective for individual cancer patients. Our MRI methods consist of dynamic contrast‐enhanced (DCE), intravoxel incoherent motion (IVIM), MR elastography (ASAP project) and relaxometry derived MRI parameters. At the Spinoza Centre (AMC campus) a 7T MRI is installed and is currently being upgraded with a borecoil for 31P MR spectroscopy. Phosphorus spectroscopy provides a multitude of information on cancer metabolism. We will investigate its added value in treatment prediction and the unravelling of tumor pathophysiology.

1.4 PET radiotracer validation and kinetic modelling (small molecules)
People involved: Ronald Boellaard, Sandeep SV Golla, Maqsood Yaqub, Willemien Menke, Josee Zijlstra, Yvonne Jauw, Guus van Dongen, Idris Bahce, Gem kramer, Daniela Oprea, Ben Zwezerijnen
PhD Students: Thomas Koopman, Roland Martens, Ramsha Iqbal

Positron emission tomography (PET) enables the in vivo measurement of functional and molecular processes. The main advantages of PET is its high sensitivity and the possibility to measure tracer uptake quantitatively. A tracer amount of a compound labelled with a positron emitting radionuclide can be measured using PET. By following the tracer uptake over time in combination with tracer kinetic modelling, specific (quantitative) information on molecular pathways, receptor density or binding or pathophysiological processes is obtained. Kinetic analysis is an essential part of the validation of new radiotracers. At present, several studies on the kinetic behavior of novel radiotracers for oncology are ongoing.

2.1 PET imaging of living cells
People involved: Danielle Vugts, Guus van Dongen, Willemien Menke, Bert Windhorst. Postdoc: Natascha Stergiou

The aim of this study is to develop a generic technique for stable labeling of immune-cells of interest using Zr-89 for PET imaging. Current methods suffer from release of Zr-89 from the cells due to the used radiolabeling methodology (loading of cells with Zr-oxine), therefore we propose to use membrane labeling, which is more promising to obtain stable 89Zr-labeled cells.

2.2 Bringing biological diagnostics and therapeutics behind the blood-brain-barrier

People involved: Wissam Beaino, Danielle Vugts, Guus van Dongen
Biologicals are targeting a specific receptor and therefore highly promising as drugs for example in cancer. Their application in the brain is hampered by poor BBB penetration. In this project we are investigating which methods can enhance brain uptake of diagnostic and therapeutic biological tracers.

2.3 Improving immuno-PET tools: evaluation of DFO*, the next generation chelator for 89Zr-immuno-PET
People involved: Danielle Vugts, Guus van Dongen, Harry Hendrikse, Willemien Menke-van der Houven van Oordt, Ben Zwezerijnen, Otto Hoekstra

In the preclinical setting we have observed that DFO*, an extended version of DFO, the gold standard in the clinic for immuno-PET shows increased bone uptake as well as liver uptake. The next step is bringing this new chelator to the clinic. Clinical study using trastuzumab is in development.

2.4 Targeting and Imaging glutamine metabolism in Chronic lymphocytic leukemia
People involved: Bert Windhorst, Eric Eldering, Josée Zijlstra

CLL has a highly active mitochondrial metabolism and OXPHOS and it appears glutamine is a key player. Therefore, it is highly significant that a novel mitochondrial glutamine transporter was discovered recently in pancreatic cancer cells and we found it is also highly expressed in CLL. New modes of cancer imaging in patients make use of glutamine tracers instead of the usual 18F-labeled glucose analog [18F]FDG, to visualize malignancy and therapy responses. The major aim of this research line is to develop glutamine-based metabolic tracers to image CLL in vivo.

2.5 Imaging and treatment of activated fibroblasts in oncology and fibrosis People involved: Bert Windhorst, Geert Kazemier, Rutger-Jan Swijnenburg, Gem Kramer, Geert d’Haens, Danielle Vugts, Guus van Dongen, Ronald Boellaard, Harry Hendrikse, Ben Zwezerijnen, Pieter Raijmakers

Fibroblast activation protein (Fap) has been identified as a new highly promising target for cancer detection and possibly therapy. In this project 18F-, 68Ga and 177Lu- labeled Fap tracers will be evaluated in different indications (pancreas and IDB secured funding) in the clinic. AmsterdamUMC, location VUmc, provides possibilities to make other 68Ga-labeled tracers as well.

3.1 Radiomics and machine learning for response prediction in Diffuse Large B-Cell Lymphoma patients with FDG PET/CT studies
Project leaders: R. Boellaard, J.M. Zijlstra, H.C.W. de Vet, O.S. Hoekstra
PhD students/Research assistant: J. Eertink, C. Burggraaf, S. Wiegers, Ben Zwezerijnen

PETRA consortium, headquarters at AmsterdamUMC

Up to one third of diffuse large B-cell lymphoma (DLBCL) patients experience relapse or fail to achieve complete remission during first-line treatment. To improve outcome, early identification of patients at risk of treatment failure is of paramount importance. Identification of poor prognosis patients might be further improved by radiomics. Radiomics analysis of scans provides quantitative features of tumor characteristics as intensity, shape, volume, texture and intra-and inter-lesion heterogeneity. The aim of this research line is to identify and validate prognostic radiomics features (both at patient- and lesional level) from baseline ¹⁸F-FDG PET/CT scans in DLBCL, and examine their performance in addition to currently used prognostic markers.

4.1 Molecular imaging to understand mechanisms and effectivity of immune therapy
People involved: Willemien Menke-van der Houven van Oordt, Idris Bahce, Daniela Oprea, Ben Zwezerijen, Marc Huisman, Ronald Boellaard. Bert Windhorst, Danielle Vugts, Guus van Dongen
PhD students: Iris Miedema, Hanneke Pouw

Within this theme several trials are ongoing/being developed using mAbs, nanobodies or small molecules to image targets for immune therapy such as PDL1, PD1, CTLA4 as well as tracers to track T cell (CD8 nanobody, AraG), ao within the large multicenter consortium ImmuneImage. Investigations focus on (quantification of) changes during immune therapy and the relation to the tumor micro environment using tissue multiplex IHC, RNA seq and DNA mutation profiles.

4.2 Biodistribution and PKPD modelling to understand and quantify specific drug uptake and targeting
People involved: Marc Huisman, Willemien Menke-van der Houven van Oordt, Yvonne Jauw , Idris Bahce, Daniela Oprea, Ben Zwezerijen, Ronald Boellaard.
PhD students: Iris Miedema, Hanneke Pouw, Jessica Wijngaarden

In collaboration with pharma labeled mAbs, bispecific/trispecific antibodies, nanomolecules and others are being investigated in early clinical trials to understand their biodistribution and tumor targeting on one hand, and on the other hand to model saturation of target binding and quantification of drug delivery.

5.1 Towards implementation in the guidelines: FES PET for diagnostics and therapy prediction in patients with ER+ breast cancer
People involved: Willemien Menke-van der Houven van Oordt, Daniela Oprea/Otto Hoekstra, Maqsood Yaqub, Ronald Boellaard, Bert Windhorst. PhD student: Ramsha Iqbal

Within this research theme, several questions are being asked to understand the added benefit of the novel ER-targeting PET tracer 16α-[18F]-fluoro-17β-estradiol or FES for diagnostics and staging, response prediction and therapy monitoring for patients with ER+ breast cancer. Optimization of visualization and methods for reliable quantification are being investigated.

5.2 [18F]DCFPyL/[18F]PSMA-1007 PET imaging to guide radioactive [177Lu]Lutetium Prostate-Specific Membrane Antigen I&T therapy
People involved: Philip de Witt Hamer, Daniela Oprea, Frederik Barkhof, Anna Bruynzeel, Guus van Dongen, Elsmarieke van de Giessen, Harry, Hendrikse, Mariette Labots, Myra van Linde, Frank Lagerwaard, Bastiaan Moraal, Peter Vandertop, Peter van de Ven, Pieter Wesseling, Bert Windhorst, Tom Wurdinger, Maqsood Yaqub

This comprises a proof of concept study for the theranostic approach targeting the Prostate- Specific Membrane Antigen (PSMA) in patients with glioblastoma. In this early clinical phase 0/I/II study, we aim to demonstrate (a) that [18F]DCFPyL/[18F]PSMA-1007 PET scans can be used as an imaging biomarker for PSMA expression in glioblastoma, and (b) whether radionuclide therapy with the ß-emitter [177Lu]Lu- PSMA-I&T can be effective and has tolerable toxicity in newly-diagnosed glioblastoma. A collaboration between Cancer Center Amsterdam, Imaging Center Amsterdam, Netherlands Cancer Institute/Antoni van Leeuwenhoek and Radboud UMC.

6.1 Prostate cancer imaging
People involved: Daniela Oprea, Otto Hoekstra, Ronald Boellaard, Maqsood Yaqub, Bert Windhorst, Jens Voortman, Andre Vis, Matthijs Cysouw, Bernard Jansen, Harry Hendrikse, Jeroen van Moorselaar.
PhD students: Dennie Meijer, Yves Bodar, Wietske Luining.

This research line focusses on (diagnostic) prostate imaging using various (new) radiotracers, such as 18F-FDHT and 18F-DCFPyL PSMA, and use of innovative image analysis methods (radiomics and machine learning). The research line aims at improving early detection and characterization of prostate cancer, treatment response prediction and studies the association between the baseline and post-treatment imaging variables, serum biomarkers, and the tissue biomarkers with response and disease progression. An international collaboration between Amsterdam VUmc/ Memorial sloan Kettering/ Monash Victoria, Australia is part of this effort.

6.2 Targeted PET Imaging of Pancreas Carcinoma
People involved: Daniela Oprea, Rutger-Jan Swijnenburg, Ronald Boellaard, Thomas Poels, Geert Kazemier, Joanne Verheij, Johanna Wilmink, Pieter Wesseling, Bert Windhorst, Danielle Vugts, Floris van Velden (LUMC), Alexander Vahrmeijer (LUMC), Lioe-Fee de Geus (LUMC).
PhD student: Floris Vuijk

Currently used conventional imaging modalities such as CT, MRI, EUS for diagnosis & pre-operative staging of pancreatic cancer have low specificity. The role of molecular imaging with ¹⁸F -FDG PET is suboptimal in PC, since inflammation of the pancreas cannot be distinguished from cancer, and mucinous neoplasms of the pancreas may often show no tracer uptake. Within this research line we will explore and clinically implement new tracers, such as ¹⁸F -PSMA and [¹⁸F]FP-R₀1-MG-F2, as an Integrin αvβ6-Targeted radiotracer, to improve lesion detection, staging and patient stratification

7.1 PICTURE: picturing predictions for patients with brain tumors
People involved: Philip de Witt Hamer, Frederik Barkhof, Roelant Eijgelaar, Henk-Jan Mutsaerts, Hugo Vrenken. PhD students: Ivar Kommers, Maisa van Genderen, Alexandros Ferles.

The aim of this study is to improve neurosurgical and radiotherapy decisions in patients with glioblastoma to prolong their survival and decrease complications and functional decline. A well-structured dataset is available of 1427 patients with glioblastoma, who have had state-of-the-art treatment in international expertise centers (the PICTURE consortium, http://www.pictureproject.nl). All preoperative tumors, postoperative residual tumors and tumors at progression have been segmented by trained raters. All MRIs have been aligned to a standard brain atlas for comparison of brain locations between patients. The main focus for the coming years is to predict biopsy decisions, tumor growth speed, location of tumor recurrence and patient survival time using deep learning models with both patient and treatment characteristics, as well as MRI information in a public-private partnership with Active Collective.

7.2 Frontier
People involved: Niels Verburg, Frederik Barkhof, Ronald Boellaard, Maqsood Yaqub, Otto Hoekstra, Pieter Wesseling, Philip de Witt Hamer

The aim of this project is to define the standard for image-guided surgery and radiotherapy in patients with a diffuse glioma. A well-structured dataset is available of 174 multiregion samples in 20 patients with a diffuse glioma. Multiregion stereotactic biopsies were co-localized with MRI including perfusion- and diffusion-weighted imaging and MR spectroscopy, and FET PET. The dataset will be expanded to more patients with newly-diagnosed and progressive glioma.

8.1 Development of new optical devices
People: Ton van Leeuwen, Marloes Groot, Johannes de Boer, Maurice Aalders, Dirk Faber, Dick Sterenborg, Martijn de Bruin, Anouk Post, Imran Avci,

The interactions of light with tissue create the contrast observed in in images obtained with optical techniques. In this research line, we continuously investigate better ways to describe light-tissue interactions in order to improve contrast between healthy and pathological tissues and use that knowledge to develop new methods using novel techniques as Optical Coherence Tomography, (single fiber) reflectance spectroscopy, Raman and fluorescence spectroscopy, non-linear microscopic and hyperspectral imaging, using new quantitative algorithms and using new technologies like Integrated photonics and MEMS.

8.2 Clinical validation of novel devices and algorithms
People: Martijn de Bruin, Ton van Leeuwen, Peter Bonta, Jouke Annema, Dirk Faber, Jorg Oddens, Theo de Reijke, Mark van Berge Henegouwen, Suzanne Gisbertz, Roel Hompes, Pieter Tanis, Patricia Zondervan, Otto van Delden, Guido Kamphuis, Joyce Baard, Joseph Liaso

This research line aims to advance the current diagnostic accuracy of various urological, gastro and pulmonary tumors to the next level by combining novel multimodal endoscopic techniques with novel confocal endomicroscopy (CLE), optical coherence tomography, fluorescence imaging and multi fiber spectroscopic methods, potentially supported by a real-time artificial intelligence (AI) based feedback platform to aid point of care diagnosis. This research line aims to develop a more accurate, reliable, less operator dependent real-time diagnosis which accelerates point of care treatment decision-making and patient stratification.

8.3 Development of new contrast methods and biomarkers
People: Rienk Nieuwland, Edwin van der Pol, Annemieke van Dam, Maurice Aalders, Martijn de Bruin, Ton van Leeuwen

All body fluids may contain important biomarkers which can be labeled and subsequently imaged. In this research line we focus on the use of extracellular vesicles in blood and urine as potential biomarkers e.g. for disease screening and monitoring treatment and biomarkers present in saliva and fingerprints. In our research, we will develop improved and unique antibodies and contrast agents.

8.4 Precision Surgery for Hepato-Pancreato-Biliary Cancers (PrecisionHPB)
People involved: Rutger-Jan Swijnenburg, Geert Kazemier, Daniela Oprea, Bert Windhorst, Ronald Boellaard, Arantza Farina, Joanne Verheij, Alexander Vahrmeijer, Sven Mieog, Lioe-Fee de Geus Oei, Koos Burggraaf, Rob Valentijn

The PrecisionHPB Surgery Group (PrecisionHPB) focusses on more precise, image-guided, surgical treatment of HPB cancers. “Precision” can be explained in two ways: 1) Oncological Precision: Tumor biology-specific and 2) Surgical Precision: Image-guided, minimally-invasive robotic surgery.

Examples of studies that are currently conducted are: The MIMIC trial: a multicenter prospective cohort study on the effect of ICG-fluorescence on radical resection rates in minimally-invasive colorectal liver metastases surgery. Development and validation of an Integrin avb6 targeting Near Infrared Fluorescent (IRDye800) tracer for pancreatico-biliary cancers. Besides optical techniques, avb6, PSMA and FAPI targeting PET studies, described in earlier paragraphs, will play a more important role in the future.

9.1 Multiparametric ultrasound for detection and treatment of cancer
People involved: Hessel Wijkstra, Massimo Mischi (TUe), Ruud van Sloun (TUe), Simona Turco (TUe), Martijn Smeenge, ChristopheMannaerts, Mark Bloemendaal (Angiogenesis Analytics)enAuke Jager.

Together with the Signal Processing Systems (SPS) group of the TU/e we are developing and validating multi parametric ultrasound techniques for the detection of cancer. The focus at this moment is Contrast Ultrasound Dispersion Imaging (CUDI) for the detection of prostate and kidney cancer. Artificial Intelligence is used to optimally combine the different utrasound techniques. Together with the startup company Agiogenesis Analystics we will validate the techniques is a large multi center study.