The diagnosis of Parkinson's disease (PD) is mainly based on the typical clinical presentation of cardinal signs and its being well-responsive to levodopa treatment (Armstrong and Okun, 2020); however, misdiagnosis of PD is not uncommon. The most common misdiagnoses are related to parkinsonism-plus syndromes, such as multiple system atrophy (MSA), which could be classified as the parkinsonian subtype (MSA-P) with predominant parkinsonism, and the cerebellar subtype (MSA-C) with predominant cerebellar features (Koga and Dickson, 2018). Asymmetry of the cardinal features of parkinsonism and good response to levodopa, which are characteristic manifestations associated with PD, could also be seen in some patients with MSA-P (Jost et al., 2019), which makes it difficult to make the differential diagnosis on the basis of clinical criteria alone. In clinical pathological studies, only 76% of postmortem-confirmed cases were diagnosed correctly as PD (Rajput et al., 1991), while 55% of postmortem-confirmed MSA cases were misdiagnosed (Litvan et al., 1997).

Many objective methods have been tested for the differential diagnosis between PD and MSA-P, such as serum neuronal exosomes (Jiang et al., 2020), proteins in cerebrospinal fluid (Singer et al., 2020), and neuroimaging (Saeed et al., 2020). Multiple imaging modalities such as magnetic resonance imaging (MRI) and positron emission tomography (PET) have been increasingly used to investigate the morphologic and functional characteristics of MSA, PD, and other atypical parkinsonisms (Zhao et al., 2020). The second consensus statement on the diagnosis of MSA (Gilman et al., 2008) and the recent Movement Disorders Society clinical diagnostic criteria for PD (Postuma et al., 2015) have included the results of a few of these neuroimaging techniques to serve as supportive criteria or exclusion criteria for the diagnosis of diseases. Thus, cerebral MRI and PET may aid in an early, accurate, and objective diagnostic classification by highlighting the underlying neurochemical and neuroanatomical changes that underlie the spectrum of disorders.

The differential diagnosis can be assisted by the characteristic features of MSA revealed by MRI, such as putaminal hyperintensive rim and “hot cross bun” sign of the pons (Hughes et al., 2002; Poewe and Wenning, 2002). However, putaminal hyperintensive rim occurs in only 31.8% of MSA-P patients and is also present in 6.7% of PD patients (Zhao et al., 2020). The “hot cross bun” sign exhibits the highest specificity in MSA-C patients but is not found in MSA-P patients (Zhao et al., 2020). In PET studies, ¹⁸F-fluorodopa (¹⁸F-Dopa) can only differentiate healthy subjects from parkinsonism patients, but it is inadequate for differentiating between MSA-P and PD because nigrostriatal dopaminergic neurons degenerate in these disorders (Burn et al., 1994). Imaging disease-specific patterns of regional glucose metabolism with ¹⁸F-flurodeoxyglucose (¹⁸F-FDG) PET allow for a highly accurate distinction between PD and MSA, such as PD-related pattern (Matthews et al., 2018) and MSA-related pattern (Akdemir et al., 2014). Due to the low spatial resolution of FDG PET images, visual analysis is difficult to detect significant anomalies; hence, these disease-specific patterns are calculated by voxel-based statistical analysis in comparison with healthy control subjects with statistical parametric mapping (SPM) software (Ko et al., 2017). SPM is useful in the group analysis of FDG PET images but is hampered in individual diagnosis. Single-patient SPM analysis could be also used, but this statistical method is controversial because only one patient is included in one group of the data. Single SPM analysis only derives a calculated image, which may differ from the actual brain structure of the patient; thus, the result could be affected by the selection of a normal control group. Variations in PET scanners and image reconstruction algorithms have been shown to systematically shift image quality in SPM analysis (Kogan et al., 2019).

We hypothesize that there are disease-specific patterns related to PD and MSA-P in both ¹⁸F-Dopa and ¹⁸F-FDG PET, and the two kinds of imaging modalities have their own advantages and disadvantages. In the present study, we used the co-registration analysis of ¹⁸F-Dopa and ¹⁸F-FDG PET to visualize the characteristics of the images without any calculation or image distortion, which could be easily used in the individual differential diagnosis between PD and MSA-P.

This study is the first to demonstrate the diagnostic value of co-registration analysis of ¹⁸F-FDG and ¹⁸F-Dopa PET for differentiating MSA-P from PD. Segmentation analysis of the rabbit-shaped appearance of the striatum segment of PET images not only could distinguish PD and MSA-P patients from healthy controls but also further differentiated PD from MSA-P. The segmentation analysis has also been verified by the traditional regional PET uptake value analysis and the latest SPM analysis used for identifying disease-specific metabolic patterns.

The segmentation analysis showed a full rabbit-shaped appearance of the striatal segment of PET images in both ¹⁸F-Dopa and ¹⁸F-FDG PET in the HC group, indicating intact integrity of the brain dopamine system and preserved function of the basal ganglia-related pathway. In patients with PD, a defective striatal segment was shown in ¹⁸F-Dopa PET due to the degeneration of nigrostriatal dopaminergic neurons. However, the rabbit-shaped appearance of the striatal segment was intact in the ¹⁸F-Dopa PET image, indicating that the postsynaptic function of the dopaminergic neurons and related pathways were not damaged (Ko et al., 2018; Shine et al., 2019). Defective striatal segments were shown in both ¹⁸F-Dopa and ¹⁸F-FDG PET images in patients with MSA-P, which is consistent with neuropathologic features of MSA (Yoshida, 2007; Jellinger, 2014). A neuropathologic study on MSA showed that the glial cytoplasmic inclusions, neuronal nuclear inclusions, and diffuse homogeneous alpha-synuclein staining in neuronal nuclei and cytoplasm were distributed widely in lesions in the pontine nuclei, putamen, substantia nigra, and other parts of the brain, indicating that both presynaptic function (detected by ¹⁸F-Dopa PET) and postsynaptic pathway (detected by ¹⁸F-FDG PET) of dopaminergic neurons were impaired.

The segmentation analysis could be verified by regional PET uptake value analysis. A significant reduction of ¹⁸F-Dopa uptake was observed in the bilateral striatum of patients with PD and MSA-P. There was no significant difference in ¹⁸F-Dopa uptake in the bilateral striatum between patients with PD and MSA-P, which is consistent with previous studies (Burn et al., 1994; Hu et al., 2002). The ¹⁸F-Dopa and other presynaptic dopaminergic radiotracers of PET or single-photon emission computed tomography are useful to separate healthy subjects from Parkinsonism patients, but they are inadequate for differentiating between MSA-P and PD (Perju-Dumbrava et al., 2012; Kaasinen et al., 2019). The further ¹⁸F-FDG PET showed a significant decrease in the bilateral putamen of patients with MSA-P and a significant increase in ¹⁸F-FDG uptake in the bilateral striatum of patients with PD. The diametrically opposite results suggest that the pathogeneses of the two diseases are different. ¹⁸F-FDG PET measures cerebral glucose metabolism, which reflects synaptic and neuronal activity. Loss of striatal dopamine in PD leads to changes of activity in both the direct and indirect striatal pathways, which could be detected by the ¹⁸F-FDG PET. The primary pathological abnormality in PD is confined to the substantia nigra, and the degeneration of dopaminergic projection neurons from the substantia nigra to the striatum results in widespread alterations in the functional activity (Alexander et al., 1990; Parent and Hazrati, 1995) of the basal ganglia. The loss of inhibitory dopaminergic input to the striatum results in increased activity of the striatum, which is shown in striatal hypermetabolism in ¹⁸F-FDG PET in patients with PD. The metabolic patterns in MSA-P represent regional postsynaptic neuronal loss directly related to abnormal protein deposition that characterizes the neurodegenerative disease (Tripathi et al., 2013). A susceptibility-weighted imaging study showed iron deposition in the putamen of patients with MSA-P caused by iron-mediated oxidative stress, which might appear as glucose hypometabolism in ¹⁸F-FDG PET, which could differentiate MSA-P from PD (Yoon et al., 2015). We also observed a significant positive correlation of ¹⁸F-FDG uptake with ¹⁸F-Dopa uptake in the contralateral posterior putamen in MSA-P patients in our study, indicating that presynaptic function and postsynaptic pathway of dopaminergic neurons are synchronously damaged in MSA-P. In addition, regarding the similar disease duration, MSA-P patients in our study exhibited higher severity than PD patients with higher H&Y score, which is consistent with previous PET studies (Schönecker et al., 2019; Zhao et al., 2020), supporting the fact that MSA-P develops more rapidly than PD. PET studies showed that striatal uptake of ¹⁸F-Dopa significantly correlated with the H&Y stage in both patients with PD (Eshuis et al., 2006) and MSA (Taniwaki et al., 2002). However, ¹⁸F-FDG PET exhibited an opposite pattern of 18F-FDG uptake in the striatum between PD and MSA. For example, a longitudinal FDG-PET study showed that, as evidenced by PD-related pattern activity, a metabolic network characterized by hypermetabolism in the putamen increased linearly with disease progression of PD (Huang et al., 2007). In contrast, MSA-related pattern, a metabolic network characterized by hypometabolism in the putamen, has been shown to have a positive correlation with disease duration in patients with MSA (Ko et al., 2017). Collectively, these studies suggest that with the progression of the disease, the striatal metabolism in PD patients increases, whereas the striatal metabolism in MSA patients decreases. Therefore, H&Y stage scores do not affect the differential diagnosis of the two diseases in our co-registration analysis of PET images.

Our SPM analysis also confirmed the results of the above segmentation analysis and ¹⁸F-FDG uptake changes in patients with PD and MSA-P. PD patients were found to have a significant glucose hypermetabolism in the bilateral putamen, cerebellum, and occipital lobes, and a significant glucose hypometabolism in the bilateral putamen was detected in patients with MSA-P. SPM analysis can clearly differentiate between these two forms of parkinsonism: the glucose metabolism in the putamen is preserved or elevated in PD but significantly decreased in MSA-P, which is useful in the group analysis of FDG PET image. Our results are consistent with previous studies (Ma et al., 2007; Meles et al., 2020). However, the widespread implementation of such cerebral disease-related metabolic patterns in multicenter collaborations and clinical practice has been hindered by differences between PET scanners as well as reconstruction algorithms (Kogan et al., 2019). For example, hypermetabolism in the putamen was shown in the abnormal metabolic network in PD (Ma et al., 2007; Meles et al., 2020), but there was no significant change in metabolism of the putamen in the PD-related pattern analysis in several FDG PET studies (Holtbernd et al., 2015; Ko et al., 2017). The selection of a normal control group with differing ages, genders, and ethnicities could also cause potential bias in the SPM analysis (Kogan et al., 2019). It is interesting that significant glucose hypometabolism in the bilateral putamen can be found in MSA-P patients compared with healthy controls, but SPM analysis did not show reduced cerebellar metabolism. Our result is consistent with previous studies (Zhao et al., 2012, 2020), and the character of FDG PET image is consistent with the neuropathologic feature of MSA-P (predominated in basal ganglia) and MSA-C (predominated in cerebellum) (Ozawa et al., 2004).

There were several limitations in this study. First, our study sample was relatively small and longitudinal; hence, studies with a large sample size are needed in the future. Second, single-patient SPM analysis was not performed because there would have been the need to change the age-matched HC group from a large database if the patient's age changed. Finally, other types of parkinsonism-plus syndromes, such as progressive supranuclear paralysis and corticobasal degeneration, were not included in the study.

In conclusion, the co-registration analysis of ¹⁸F-FDG and ¹⁸F-Dopa PET demonstrated great diagnostic values for differentiating MSA-P from PD, which could be easily used in the individual analysis in clinics. Further studies are needed to evaluate the segmentation analysis in the differential diagnosis of other atypical Parkinsonisms, such as progressive supranuclear paralysis and corticobasal degeneration.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by local Ethical Committee of the First Affiliated Hospital of Sun Yat-sen University. The patients/participants provided their written informed consent to participate in this study.

W-bX: conceptualization, data curation and analysis, validation, and writing original draft. X-cS: data curation and analysis, validation, and writing original draft. G-hL and CY: data curation and analysis, and validation. X-sZ and ZP: project administration, resources, supervision, data analysis, review, and editing. All the authors read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.