This study explored the optimal reference regions for comparison of cortical SUVRs derived from the Centiloid cortex VOI of eFBB PET and ECD SPECT. There was a strong correlation between SUVRs of eFBB PET and ECD SPECT with cerebellum-based normalization methods. The GN method showed high but relatively lower than cerebellar-based normalized methods. There was a moderate correlation with the pons and no correlation with CWM normalization methods. We also performed a voxel-wise comparison by SPM and eFBB PET showed higher uptakes in cerebral cortices, striatum, and brain stem compared to ECD SPECT. To the best of our knowledge, this was the first study to compare eFBB PET and ECD SPECT.
In our study, cerebellum-based methods demonstrated stronger correlations compared to GN. Previous studies comparing kinetic parameters and SUVRs of early-phase amyloid and perfusion PET (H2O PET) showed a moderate to strong correlation with cerebellar normalization 10,13. Meanwhile, a previous study comparing eFBB and FDG PET of patients with mild cognitive disorder and dementia showed a stronger correlation with GN methods, contrary to our results (r = 0.86 for GN and r = 0.76 for WC) 8. The cerebellum is less susceptible to hypometabolism and deposition of misfolded proteins in neurodegenerative disease until late stages 14–16. The cerebellar deposition of Aβ showed minimal impact on SUVR values of FBB PET even in advanced stages of AD 17. On the other hand, the whole brain as a reference region is vulnerable to a reduction of cerebral blood flow in neurodegenerative disease or normal aging. Global decrease of brain metabolism may also cause artifactual hyperperfusion in unaffected cerebral regions under the GN method 18. We propose to use cerebellum-based references for these reasons.
For CWM reference, there was no regional correlation between eFBB PET and ECD SPECT. This finding is contrary to the previous study, which demonstrated a moderate to a strong correlation between SUVRs of early-phase florbetapir and H2O PET (range of r: from 0.70 to 0.86) 10. The discrepancy between our results and the aforementioned results could be explained by the non-specific binding of amyloid tracer in subcortical white matter 19. Our voxel-wise analysis also showed different distribution between two tracers throughout the white matter. The SUVR values of ECD SPECT were higher than eFBB PET in periventricular white matter. On the contrary, SUVRs of eFBB was higher than ECD in some part of centrum semiovale. The central white matter is also susceptible to age-related ischemic changes. These indicate that central white matter may not be the optimal reference for comparison of eFBB PET and ECD SPECT. Our study is limited by the difference in spatial resolution of the two studies and the absence of partial volume correction of SPECT images. Further studies are needed to validate white matter as an optimal reference.
The voxel-wise comparison between eFBB PET and ECD SPECT demonstrated a discrepant uptake pattern mainly as grey-white matters contrast. The SUVRs of cerebral cortices, basal ganglia, and brain stem were higher in eFBB PET, while the SUVRs of periventricular white matter were higher in ECD SPECT. These findings suggest that the cortical grey-to-white matter contrast is better in eFBB PET. The SUVRs in the outer edge of brain cortices seemed to be higher in ECD SPECT. This might arise from the lower spatial resolution of the SPECT showing more blurred images compared to PET. Previous studies have reported that perfusion SPECT has the lower diagnostic performance to discriminate neurodegenerative diseases compared to FDG PET, which might be related to poor resolution and grey-to-white contrast20,21. The high grey-to-white contrast and high resolution of eFBB PET may be helpful to clinicians in visual assessment of perfusion images to discriminate neurodegenerative diseases.
The ICA analysis revealed more ICNs from eFBB PET compared to ECD SPECT in the same cohort. The eFBB PET may also be more appropriate for functional network analysis. The difference in the number of ICN extractions is thought to be mainly due to the difference in resolution of the two images and the difference in grey-to-white matter contrast. However, further studies comparing eFBB PET to well-known metabolic imaging (e.g. FDG PET) are needed for validation.
There are several limitations to this study. This was a retrospective study that included a small number of patients. The population included only early PD patients with mostly negative amyloid deposits. In this study, not only HC, but also the patient groups (AD or MCI) where FBB PET is the most used were omitted. However, there have been arising interests in the impact of proteinopathies on the cognitive disorder in PD patients22,23. Furthermore, the PD cognition-related pattern (PDCP), as well as PD motor-related pattern (PDRP), was elucidated based on FDG PET and perfusion studies24. These findings emphasize the clinical need for dual-phase amyloid PET for patients with PD with or at risk of cognitive disorder, which can assess perfusion and amyloid deposition at once. We did not evaluate the performance of kinetic parameters of eFBB PET. In a previous study, a kinetic parameter of amyloid PET (PiB-R1) showed a higher correlation to H2O PET compared to early-frame SUVRs of PiB PET (ePIB) on a cortical level in patients with AD. Furthermore, there was a weak positive correlation between mean cortical binding potential and ePIB, but not with PiB-R1, suggesting that the kinetic parameter is robust to amyloid burden 25. Further studies to figure the contamination of eFBB PET images by amyloid deposition are needed. For spatial normalization of SPECT, there was an error in the co-registration step to MRI images, and the built-in SPECT template in SPM was used instead. The template was derived from HMPAO SPECT, not from ECD. However, the impact of using HMPAO based template would have been small in the spatial normalization.
To conclude, eFBB PET is an imaging test suitable for cortical perfusion evaluation and provides reliable SUVR based on cerebellum references. Based on its higher grey-to-white matter contrast and resolution than perfusion SPECT, eFBB PET provides benefits to be used in clinical management for neuronal injury evaluation in neurodegenerative diseases, in addition to its cost-benefit and less radiation exposure.