Computational bone health phenotyping in Multiple Myeloma patients

• Applicants: Professor Dr. Sebastian Böttcher; Privatdozentin Dr. Sarah Charlotte Foreman; Professor Dr. Jan Stefan Kirschke; Professor Dr. Marc-André Weber
• Subject Area: Radiology ; Medical Physics, Biomedical Technology
• Term: since 2016
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 283653538

Multiple myeloma (MM) is the second most common malignant hematologic disease, exhibiting significant heterogeneity at both transcriptional and genetic levels, which contributes to differences in disease presentation, progression, and response to therapy. Bone disease is highly prevalent, with 70-80% of patients experiencing skeletal-related events such as fractures, bone pain, and spinal cord compression, significantly affecting their quality of life and mobility. Furthermore, the presence of bone morbidity is associated with poorer prognosis and survival outcomes, highlighting the importance of comprehensive assessment and management of bone lesions. According to the German AWMF S3 guidelines, whole-body CT is considered the standard of care for evaluating disease severity and progression, due to the effective visualization of osteolytic lesions, overall bone mineral density, and potential for stratifying the risk of an imminent fracture. Yet attaining reproducible assessments remains challenging due to the substantial heterogeneity of imaging phenotypes, including the challenge of categorizing diffuse bone marrow involvement and the assessment of numerous (frequently >100) individual lesions. In addition, this heterogeneity is associated with prognostic differences, influencing not only inter-individual but also intra-individual variations of treatment response, complicating the effective management of bone disease. Consequently, obtaining reproducible, objective, and scalable methods to assess the imaging phenotype and provide insights into therapy response at a lesion level is of great clinical interest and represents an unmet clinical need that has never been investigated in a systematic manner. To close this gap, the aim of this project is to implement a comprehensive imaging phenotype profile within automated, scalable image processing routines, for the assessment of initial disease burden and treatment response. Notably in addition to osteolytic lesions, focal lesions are also recorded for the first time by generating bone tissue maps from virtual non-calcium CT images. This enables an objective assessment of the imaging phenotype and reproducible application to all lesions in all scans acquired for a given patient. Moreover, this phenotype is augmented with quantitative radiological descriptors at a lesion level, facilitating individualized pattern recognition of the transcriptional and genetic variance within each lesion, and additionally correlated with laboratory and cytogenic parameters, providing insights into therapy response at both lesion and patient level. The innovative methods developed in this context create a novel approach to automatically record a comprehensive bone health phenotype in multiple dimensions and analyze it in large data sets. These automated, reproducible, and more comprehensive diagnostics are poised to ultimately facilitate earlier and more targeted therapy, along with more sensitive therapy monitoring.

• DFG Programme: Research Grants
• International Connection: Austria, Switzerland
• Partner Organisation: Fonds zur Förderung der wissenschaftlichen Forschung (FWF); Schweizerischer Nationalfonds (SNF)
• Cooperation Partners: Professor Dr.-Ing. Georg Langs; Professor Dr. Björn Menze