Super-Resolved Multimodal Imaging Biomarkers for Frontotemporal Dementia

PROJECT SUMMARY This proposal is responsive to NIA solicitation PAR-22-094/NOT-AG-21-036 for projects involving the development of novel approaches to diagnose and study Alzheimer's Disease and Related Dementias (ADRD) with an emphasis on the need for biomarkers for dementia types other than Alzheimer’s disease in the ADRD spectrum, including frontotemporal dementia (FTD). In this project, we will leverage our MPI team’s expertise with MR-based accurate thalamic nuclei segmentation (TNS) and PET super-resolution (SR) to develop thalamic-nuclei-based measures of atrophy, connectivity, and hypometabolism as possible FTD biomarker candidates. FTD, the second most common cause of dementia in adults under 65 years of age after AD, is clinically, genetically, and pathologically heterogeneous. There is an urgent need for antemortem biomarkers for FTD that are sensitive across its different subtypes. Thalamic atrophy is a common feature of early disease pathogenesis for all FTD subtypes. We propose to develop an integrated MR and PET imaging framework for deriving quantitative imaging measures from the thalamic nuclei and validate in secondary-use multimodal imaging data. Our dataset will include both sporadic FTD and familial C9orf72+ FTD patients. We will use a more advanced variant of our TNS approach which, as per our preliminary results from an Alzheimer’s disease cohort, leads to significantly better discrimination between healthy and impaired groups than FreeSurfer’s Bayesian segmentation method, which is one of the current state-of-the-art TNS methods. We will develop and validate an SR PET platform that uses MR-based thalamic nuclei labels as additional inputs to the model to enhance thalamic nuclei contrast. To assess the clinical utility of the thalamic-nuclei-derived multimodal biomarker set for atrophy, connectivity, and hypometabolism, we will compute receiver operating characteristic curves for FTD subtype groups vs. cognitively normal subjects. We will also characterize our multimodal biomarker by establishing a temporal ordering via event-based modeling. To assess the sensitivity of our nuclei-derived atrophy and connectivity biomarkers, we will conduct cross-sectional spatiotemporal analyses in a larger cohort with MR-only data that can reveal the age of divergence of each biomarker in a diseased vs. control group. Finally, we will conduct longitudinal analyses to predict changes in clinical dementia rating and its behavioral and language subscores from changes in atrophy and connectivity measures from the thalamic nuclei. Unlike Alzheimer’s disease, for which the ATN research framework is well-developed for biomarker-based characterization of dementia, there is a pressing need for imaging biomarkers for FTD, which this project can address. We therefore envision that our proposed research will have high clinical impact on FTD characterization and could play a role in FTD drug development efforts.