Purpose:

Multiple Myeloma (MM) is a rare, non-curable aggressive malignancy of plasma cells in the bone marrow. Being newly diagnosed with MM, there is a high demand for improved risk classification and treatment outcome prediction to ensure best individual clinical care. Most importantly, successful treatment outcome in MM is hampered by a substantial fraction of patients not responding adequately to standard treatment. Further, improved risk stratification in MM is needed for targeted drug development and better design of medical studies.

Even though most patients develop myeloma-related bone disease (MDB) in the course of the disease, current clinical scores (ISS, R-ISS, DSS) for risk classification and treatment outcome prediction in MM mainly involve blood samples, bone marrow biopsy and patient frailty assessment. This is mainly due to the heterogeneous nature of bone disease in MM where imaging analysis is highly vulnerable to observer-bias and generally error-prone when used in prognostics. Several studies however still claim that imaging techniques such as PET/CT and MRI could add prognostic value.

At the PETRA consortium, we aim to explore imaging techniques for prognostics in MM to ultimately develop better decision tools for patients in first and second-line therapy.

Methodology:

• Radiomics and further machine-learning methods on MM imaging data to identify recurring, observer-independent patterns in PET and CT with prognostic value regarding survival and treatment response

• Development of decision tools for survival and therapy response prediction that combine image-derived patterns with clinical data

• Development of robust delineation methods for MM to account for the heterogeneity in bone disease manifestation among patients