In this project, we aim to understand PET/CT-based imaging features and their molecular and clinical correlates pre/post therapy in gastroesophageal junction (GEJ) adenocarcinoma. In the DKTK-funded MEMORI trial we evaluated PET-directed chemo- (CTX) or salvage radiochemotherapy (CRT) and assembled sequential high-quality tumor tissue at PET/CT imaging time points pre/post therapy. The fully recruited trial, met its primary endpoint of improved negative surgical margins (R0 rate) upon salvage intensified CRT not responding to standard neoadjuvant CTX determined by PET response 14 days after CTX initiation. Salvage CRT also led to a highly increased rate of pathologic complete remissions (pCR) suggesting high local activity of CRT. However, an important result from the trial was distant recurrence in a subgroup of patients despite high R0 resection and pCR rates. Thus, current standard-of care PET/CT or histological diagnostics do not identify high- risk patients with bad outcome and their distinguishing tumor features remain unknown.

We here aim to investigate molecular subtypes and tumor metabolism patterns in the MEMORI patient cohort. We will focus on understanding therapy-induced molecular dynamics. We will correlate these findings with PET-CT derived metabolic and anatomical imaging data using standard (SUVmax/mean) and advanced (radiomics) techniques leveraging the infrastructure provided by the Joint Imaging Platform (JIP) to perform an integrated analysis of clinical, molecular and imaging data. The available clinical and multimodal imaging data combined with available tumor tissue before, during and after neoadjuvant therapy are a unique opportunity to address fundamental questions of imaging and tissue- based features in tumor metabolism, heterogeneity and therapy response. 

In healthy cells, a number of different processes ensure that damage or mutations to DNA are repaired efficiently. In cancer cells of various tumor entities, however, genes whose products are involved in one of these repair processes can also suffer somatic mutation. These cells then lose some of their DNA repair ability.

If it is the homologous recombination repair system that is affected in this way, the cells are particularly sensitive to a group of drugs called Poly(ADP-ribose) polymerase inhibitors (PARP inhibitors). PARP inhibitors prevent cancer cells from being able to repair themselves, e.g. as a result of DNA damage caused by chemotherapy treatments, and are already used to treat a number of different types of cancer. One such form of acquired sensitivity to a particular group of drugs is called synthetic lethality. 

However, we do not know all the genes or mutations involved in synthetic lethality where PARP inhibitors are concerned. Other biological processes besides mutations, such as hypermethylation of a gene promoter involved in homologous recombination, may contribute to their inactivation. This means that it is not possible using current technical methods, including high-throughput technologies, to systematically identify all the causes of loss of homologous recombination function. On the other hand, these methods can identify a large number of ‘passenger’ mutations that arise following the repair defect. We have developed an integrated genomic biomarker that can detect the fingerprint of this repair defect on the genome of tumor cells with the help of pattern recognition, thereby increasing precision levels when detecting DNA repair defects. Subsequently, we designed a clinical trial in which patients who have tested positive with this biomarker, are treated with a synergistic combination of a PARP inhibitor and matched drug.

In future, as well as conducting a broad, in-depth genomic characterization of the samples from these treated patients, we hope to (i) gain a better understanding of the biological process of DNA repair via homologous recombination, (ii) improve the precision of our biomarker, (iii) improve predictions of the effectiveness of synthetic lethality drugs in DNA repair defect cases, (iv) investigate resistance development mechanisms to these drugs and prevent them occurring in patients, and (v) find new synergistic therapies.​

Glioblastomas are particularly aggressive, and generally incurable, brain tumors. The CAR2BRAIN trial treats patients with an HER2-positive glioblastoma who have suffered a relapse. The research team uses genetically modified natural killer cells (NK cells) with a chimeric antigen receptor (CAR) that enables them to detect the HER2 antigen and selectively attack glioblastoma cells. Since the HER2 antigen is often formed by glioblastoma cells but is not detected in healthy brain tissue, it makes a good target for cellular immunotherapy.

The NK cells are injected into the edge of the surgical cavity during the resection operation. The idea is that they will attack any remaining tumor cells directly and also activate the patient’s own immune system against them. In the trial’s expansion cohort, CAR NK cells will also be repeatedly administered into the resection cavity via a reservoir and catheter. The DKTK’s ‘upgrade project’ will enable a detailed characterization of the immune architecture changes triggered by the CAR NK cell therapy in the blood, reservoir fluid and tumor tissue. In addition, the immune response produced by the CAR NK therapy alone will be compared with the result of a combined therapy that also involves immune checkpoint blockade.

Image-based subtyping of locally advanced pancreatic cancer

Pancreatic ductal adenocarcinoma (PDAC) has a very poor prognosis because the tumor is often already at an advanced stage when first diagnosed and because of resistance to conventional chemotherapy treatments. The genome, transcriptome and proteome of these tumors are highly heterogeneous. This heterogeneity is reflected in a complex tissue architecture with different molecular phenotypes: the classic phenotype with a somewhat better response rate, and the particularly aggressive quasi-mesenchymal phenotype.

There are currently no reliable biomarkers to differentiate between these subtypes in routine clinical practice. This again is partly because of the high degree of heterogeneity, which makes it harder to ensure representative sampling and draw histopathological and molecular conclusions. Medical imaging provides information on total tumor volume. Recent developments in the field of machine learning (e.g. radiomics) are showing promising results in terms of non-invasive tumor characterization and risk stratification based on medical imaging. Translational development and testing of these kinds of prognostic models require a standardized data matrix.

The NEOLAP phase II clinical trial (NCT02125136) is one of the biggest randomized prospective trials that aims to evaluate intensified neoadjuvant chemotherapy in locally advanced PDAC. The image datasets, tumor samples and clinical information collected prospectively for the NEOLAP trial offer a unique opportunity to develop and test the image-based algorithms mentioned above.   

Pediatric low-grade gliomas (pLGGs), including pilocytic astrocytomas (PAs), are the most common childhood brain tumors. While the overall survival rate is very good, progression-free survival, especially in subtotal resection cases, is low. Chemotherapy is currently the standard treatment, but it is associated with secondary damage. Moreover, tumor-associated morbidity is particularly high in cases that cannot be fully resected and do not respond well to conservative therapy. Despite the very good overall survival rate, therefore, there is a need to establish new therapy approaches to significantly improve the quality of life of patients with this chronic disease. The researchers working on the DKTK’s Next Gen LOGGIC project aim to generate preclinical data in collaboration with the LOGGIC pLGG trial that will form the basis for next-generation clinical trials. Working with the Berlin, Düsseldorf, Freiburg and Heidelberg sites, they are investigating the signaling pathways, the proteome and the response to novel low-molecular substances in pLGG models and primary tumors. Integration with the molecular and clinical data from LOGGIC Core and the ongoing clinical trial for pLGGs (LOGGIC trial) ensures the translational value of the preclinical data.

Anti-angiogenic agents are a key element of drug therapy for patients with metastatic colorectal cancer (mCRC). Despite intensive efforts over recent years, there is still a lack of biomarkers that can predict whether these medicines will be effective. Complex interactions between the tumor cells and the surrounding stroma have made it harder to identify suitable biomarkers.

The aim of the RAMTAS study-related research project is to identify a molecular signature in patients with mCRC that can predict a response to anti-angiogenic therapy. To this end, researchers are analyzing tumor and blood samples from a phase II clinical trial (RAMTAS, NCT03520946) that was held at Arbeitsgemeinschaft Internistische Onkologie (AIO) that investigates the use of a monoclonal antibody against vascular endothelial growth factor receptor 2 (VEGFR-2). This receptor plays a key role in tumor neoangiogenesis and is an attractive target structure for modern molecular drugs.

There is a particular focus on the analysis of somatic gene mutations, gene expression profiles and post-translational protein modifications. In addition, IT-assisted machine learning and data mining make it possible to increase our understanding of the complex interactions between the individual profiles and to explore and validate the prognostic value regarding the effectiveness of the anti-angiogenic agent.