Research Projects

DECISIONS - DEcision support by ClusterIng based on SImilarity measures in precision Oncology of Neuroendocrine neoplasia and Sarcoma

Summary:

It’s a research project focused on improving how to begin treating two types of cancer: sarcomas and neuroendocrine neoplasms (NEN). It uses advanced techniques to analyze patient molecular and clinical data and find the best possible therapy for each individual.

Goal: The aim of this project is to better predict which treatments will work for each patient by looking at patterns in their clinical and molecular profiles. This approach helps in choosing the most promising medication from the start.

This research could lead to more effective treatment for patients, potentially increasing the chance of successful outcomes. It’s particularly crucial for cancer types where choosing the initial therapy is challenging, such as sarcomas and NEN.

Clinical Machine Learning approach:

The trial is subdivided into two parts: the training based on retrospective data and the prospective study. During the training phase a cohort of retrospective cases treated with selected therapy classes will be used to develop a similarity measure for therapy-response clusters based on comprehensive molecular profiling information and patients’ medical history data. The training results will be cross-validated. The prospective observational study applies the novel similarity measure to sarcoma and neuroendocrine neoplasia (NEN) patient data and has the aim to inform the physician’s decision when selecting a second-line therapy. The steps involved are: (a) recruitment of early stage sarcoma and NEN patients at all DKTK sites, (b) genomic and transcriptomic analyses through the MASTER program, (c) data analysis and prediction of response to therapy through DECISIONS, (d) molecular tumor board (MTB) discussion of the Omics data (as established in the MASTER program) and results from the similarity measure for early line patient stratification, (e) treatment as chosen by the treating physician (taking into account the additional information provided by DECISIONS) and (f) follow-up of 2 years within MASTER: documentation of response to treatment. An interim analysis will be performed after recruitment of one third of the estimated patient number and, if necessary, used to adapt and optimize the similarity measure. We will evaluate the predicted treatment response compared to observed response and the rate of non-responders in patients receiving the DECISIONS-recommended vs. a non-recommended treatment (based on physician’s choice).

Research focus:

• Focus on sarcomas and neuroendocrine tumors.
• Enhance the effectiveness of initial cancer treatments.
• Integration of genetic profiling and clinical data.
• Leverages machine learning for treatment predictions.

• Funded by: NCT - Full Proposal for Proof-of-Concept Trial Grants
• Contact person: Karsten Wendt
• Funding period: 01/2022 - 12/2025