[18F]Florzolotau PET for the Differential Diagnosis of Parkinsonism in Patients with Suspected 4-Repeat Tauopathies.

The second-generation tau radioligand [18F]florzolotau is a promising biomarker for 4-repeat (4R) tauopathies such as progressive supranuclear palsy (PSP) and corticobasal degeneration (CBD), which are difficult to disentangle clinically. Prior studies evaluating the diagnostic accuracy of [18F]florzolotau PET focused on highly selected patient populations (e.g., PSP-Richardson syndrome or amyloid-β-negative corticobasal syndrome). The present study assesses the diagnostic performance of [18F]florzolotau PET in conjunction with visual reads in a real-world clinical cohort. Methods: Ninety-four consecutive patients with parkinsonism and possible 4R tauopathy undergoing [18F]florzolotau PET for differential diagnosis were enrolled and retrospectively analyzed. The interdisciplinary consensus diagnosis based on comprehensive clinical and imaging data (most notably, [18F]FDG PET) served as the reference standard. [18F]florzolotau PET was assessed visually using predefined 4R-like and Alzheimer disease (AD)-like binding patterns (on a 4-point scale). In addition, 4R-like cases were rated with respect to the cortical-subcortical gradient of 4R-like binding. The diagnostic performance was assessed by receiver operating characteristic (ROC) analyses. Results: The 4R-like pattern was more prevalent and more strongly expressed (84.3%, mean score, 2.0 ± 1.1) in patients with a consensus diagnosis of PSP/CBD (joint diagnostic group of clinically likely 4R tauopathies) than in all other groups (11.6%, 0.26 ± 0.75, P < 0.0001). An AD-like pattern was present in all patients with a consensus diagnosis of AD (100%, 2.5 ± 0.9) and at high frequency, albeit with lower magnitude, in all other patient groups (67.4%, 1.2 ± 1.1, P < 0.01). ROC analysis for the 4R-like pattern (PSP/CBD vs. all other patients) yielded an area under the ROC curve (AUC) of 0.87 (sensitivity, 84.3%; specificity, 88.4%). The diagnostic performance of [18F]florzolotau PET did not change when also considering the AD-like pattern (AUC, 0.88; logistic regression, factor AD-like pattern; P = 0.53) or excluding all cases with AD (AUC, 0.86). The presence of corticobasal syndrome in patients with 4R-like binding was strongly associated with preferentially cortical binding (AUC, 0.89). Conclusion: Based on a real-world population of patients with parkinsonism, we demonstrate that simple visual evaluation of [18F]florzolotau PET by an a priori-defined 4R-like binding pattern allows highly accurate identification of patients with a consensus diagnosis of PSP/CBD. Thus, [18F]florzolotau PET is a promising biomarker for differential diagnosis of neurodegenerative parkinsonian syndromes.