gms | German Medical Science

Introduction: Over the last decade, oncology has undergone significant changes in how cancer patients are treated. Instead of a generic approach, there is now a greater emphasis on precision medicine tailored to individual genomic variations [1].

Translational research projects require a close connection from data collected specifically for the research target, the accessible data in healthcare repositories and relevant data in external research repositories. This requires a multi-layered infrastructure with connection to the internal documentation, a protected zone for the actual research data and a link to the available external data.

Here we report about the usage of cBioPortal [2] in the CRU5002 at the University Medical Centre Göttingen a clinical research unit focused on subtype-specific genome dynamics in pancreatic cancer.

State of the art: After comparing multiple research platforms for cancer genomic data we found that for our purposes cBioPortal was the best tool. cBioPortal is an open-access resource for storing, integrating, and visualizing genomic alterations and clinical information. It can be run locally to enable institutions to store and analyze patient data while maintaining data privacy [3]. Our work is based on advances and insights from other sites that deployed cBioPortal [4], [5].

Concept: The goal is to periodically pseudonymize and integrate clinical data from the tumor documentation system ONKOSTAR, genomic data from the human genetics department and additional data provided by individual scientists to a secured network area.

cBioPortal provides exploratory tools like data visualization, subsample selection and perform queries using customized lists of genes or sample/patient identifiers.

Additionally, cBioPortal features annotations and metadata on genes, mutations and variants that it obtains from external databases, like the COSMIC database.

This way clinical researches have a timely access to an integrated data set about their cancer patients, derived mouse models, cell lines, organoid, their biomarkers and the relevant mutation data.

Implementation: In order to implement this pipeline in CRU5002, we created a local installation of cBioPortal. The download from our local FAIRDOM-SEEK’s REST-API, the pseudonymization and merging with the data from ONKOSTAR was implemented in python as well as the import into cBioPortal that requires a combination of tab separated data files and meta files describing the data.

Lessons learned: cBioPortal can be applied for correlative studies of defined molecular pancreatic cancer subtypes and their implications on survival by analyzing patients’ molecular characteristics and clinical data within the CRU5002 cohort. We investigated samples of patients’ primary tumors with SMAD4 deficiency or genetic alterations in the ARID1A gene.

Furthermore, data of Patient-derived Xenografts mouse models (PDX models), PDX-derived primary PDAC cells (CDX models), and organoids generated from the CRU5002 patients are integrated into cBioPortal. Currently we try to understand how the growth patterns of the CDX and PDX models reflect the donor patients’ disease course. Therefore, we associate growth characteristics of PDX tumors and the chemotherapeutic treatment responses of individual CDX lines with tumor-specific features of the donor patient.

Our study shows the potential of cBioPortal for integrating and analyzing clinical, experimental and omics data.

The authors declare that they have no competing interests.

The authors declare that a positive ethics committee vote has been obtained.