Tobias Sjöblom's projects on mutations causing colorectal cancer

Whole genome and transcriptome sequencing of colorectal cancers

Luis Nunes, Klara Hammarström, Nicole Yacoub, Erik Osterman, Viktor Ljungström, Per-Henrik Edqvist, Chatarina Larsson, Emma Lundin

The aim of this project is to increase knowledge of colorectal cancer tumorigenesis through detailed characterization of the somatic genomic and transcriptomic landscapes of colorectal tumours from Swedish patients. This study will constitute the largest single whole genome sequencing effort in colorectal cancer to date and is possible through collaboration between the U-CAN colorectal cancer investigators at Uppsala University, Umeå University, and KTH.
 

Role of Ephrin receptors in metastasis

Snehangshu Kundu, Luis Nunes, Lucy Mathot

Ephrin receptors have a role in tumour growth, invasiveness, and angiogenesis and have previously been associated with metastasis. We recently demonstrated a link between inactivating EPHB1 mutations and metastasis in colorectal cancer (Mathot, Ljungström et al, Cancer Res 2017). To provide better insight into Ephrin receptor signalling in cancer and metastasis we aim to determine the phenotypes of the most frequently mutated positions in Ephrin receptors. To this end, mutational data from 10 000 TCGA tumours from 33 different cancers have been collated and ten hotspot mutations chosen for further functional analyses in cell models.
 

Ras pathway mutations in colorectal cancer

Snehangshu Kundu, Joakim Ekström

Genes in the Ras pathway are frequently subject to somatic mutations in colorectal cancers and we have recently linked several cancer genes with low mutation frequency to the Ras pathway using forward genetic tools (Kundu et al., Genome Med 10(1):2, 2018). The focus of the current project is on the phenotypes of prevalent mutations in the Ras pathway oncogenes KRAS and BRAF. For this we have produced isogenic cell models of frequent somatic mutations by genome editing. These cell models have undergone transcriptome sequencing, LC-MS global proteomics and LC-MS metabolomics. Integrative analyses will provide better understanding of commonalities and differences between common KRAS and BRAF mutations at the transcriptomic, proteomic and metabolomics levels.
 

Comprehensive discovery and validation of plasma biomarkers

Joakim Ekström, Klara Hammarström, Natallia Rameika

good diagnostic plasma biomarker should be specific for the cancer type and be able to accurately identify a patient at an early disease stage. However, most cancer types lack FDA-approved biomarkers for early detection, and the literature lacks strategies for how biomarker studies should be designed to maximize the chances of success. In this project, we have developed and validated a new study design framework for biomarker discovery based on regulatory authority requirements and novel statistical approaches. Now, the aim is discovery and validation of new diagnostic plasma biomarkers for colorectal, lung and obsterics/gynaecological cancers by applying the statistical framework to data generated from U-CAN patient samples within the project.
 

Software for somatic mutation analysis in molecular pathology

Ivaylo Stoimenov, Tom Adlerteg, Marina Rashyna

The somatic mutations of a tumour govern the selection of modern cancer drugs that the patient will be treated with. Therefore, tumour DNA is sequenced and analysed for mutations in known cancer genes as part of the routine diagnostic workflow. To improve the precision of mutational diagnostics, we are creating an integrated analysis pipeline optimized for cancer sequence data interpretation. The current pipeline contains aligner, caller, trimmer and quality control steps. A start-up company, Oncodia AB, was formed in 2018 to launch a CE/IVD compliant software for clinical diagnostics.
 

Exploiting loss of heterozygosity for a novel anti-cancer therapy

Ivaylo Stoimenov, Snehangshu Kundu, Natallia Rameika

Successful anti-cancer therapies selectively kill cancer cells, while normal tissues are spared. We have developed a concept based on frequent loss-of-function genetic variants naturally occurring in the human population and the cancer specific phenomenon loss of heterozygosity (LOH). We therefore identified human enzymes with variant amino acids in their active sites as result of SNPs and ranked the putative targets according to the prevalence of SNPs and LOH in common human cancers. We then constructed and validated colorectal cancer cell model systems for the top candidates in several genetic backgrounds. For the top candidate, a metabolic enzyme, we estimate that >3 % of patients with colorectal cancer could benefit from a tailored drug therapy, which translates to >35 000 cases worldwide per year. Subsequent drug discovery efforts uncovered a compound with 3-fold increased cytotoxicity in cells lacking the top candidate gene in vitro and in vivo (Ivaylo Stoimenov, Veronica Rendo et al., in peer review). Currently, cell models are established to further study also a second candidate gene.
 

Image-analysis based evaluation for medical imaging

Ali Teymur Kahraman, Tomas Fröding, Dimitris Toumpanakis

The aim of the project is to develop software that will be used clinically to support fast and accurate screening for pulmonary emboli in CT images. For this, we are developing methods for automatic interpretation of CT pulmonary angiography volume images to be able to segment structures of the thorax cavity in image stacks and detect emboli in segmented pulmonary arteries. The collaborative project is part of AIDA (Analytic Imaging Diagnostic Arena), a national arena at Linköping University for AI in medical diagnostics.