The role of substantia nigra sonography in the differentiation of Parkinson’s disease and multiple system atrophy

The diagnosis of Parkinson’s disease (PD) mainly depends on the clinical manifestations [1], which results in the accuracy of diagnosis affected by the physicians’ experiences in a great extent. Although as high as 90% accurate diagnosis of PD could be reached in the movement disorders clinics [2, 3], it was already the best scenario. A recent clinical pathology study demonstrated that in the early stage of the disease, only 50% parkinsonian patients received an accurate diagnosis of PD [4]. The most diagnostic confusing diseases are atypical parkinsonian disorders (APD), including progressive supranuclear palsy (PSP), multiple system atrophy (MSA), corticobasal ganglionic degeneration (CBD) and dementia with Lewy bodies (DLB). Those APD share some clinical features with PD, but progress aggressively and have much poorer prognosis. The misdiagnosis not only posed psychosocial and economic burden to patients and their families, but also set up obstacles to the neuroprotective or modified therapies for both PD and APD. Therefore, it is currently eager to find a battery of tools to help diagnosis and differential diagnosis. Transcranial sonography (TCS) is a non-invasive, convenient and economic tool. Recent evidence has shown that it not only helps in the diagnosis of PD, but also helps to differentiate it from APD [5, 6]. However, most of evidence came from Caucasians [6]. Moreover, in most previous differentiation studies not only the number of APD recruited was relatively small, but those distinct entities were pooled into one single APD group for comparison [6]. Among those APD, MSA with predominant parkinsonism (MSA-P), a subtype of MSA, presents most similar clinical features with PD and could be most easily confused with non-tremor dominant (non-TD) PD, especially in the early stage [7, 8]. In our study, we collected 86 clinical probable MSA-P and 147 age and gender matched non-TD PD patients who were clinically diagnosed, and explored if TCS could help the differential diagnosis.

This cross-sectional study was performed at the Movement Disorder Clinic, Department of Neurology, Ruijin Hospital affiliated to Shanghai Jiao Tong University School of Medicine from April 2014 to September 2017.

Consistent with previous results [12, 15, 16], our present study demonstrated that more percentage of SN hyperechogenicity was detected in non-TD PD patients than in MSA-P patients. However, the sensitivity for the non-TD PD patients in our cohort was only 74.1%, which was lower than most of previous studies performed by other researchers [5, 6, 12‐22]. The majority of earlier studies were conducted in Caucasians and the sensitivity was almost more than 90%, even reaching up to 100% [5, 6, 12‐19]. Only a few studies were performed in the Asian patients, and the sensitivity varied widely (50%~ 80%) [11, 20‐28]. Ethnic differences might not be completely excluded between Caucasians and Asians, but the spectrum bias should be the more reasonable cause. It has been well known that PD is a heterogeneous disorder, and it can be divided into various phenotypes based on the onset age or prominent clinical presentations, such as early or late onset PD, TD PD or non-TD PD [29‐31]. Different patterns of SN echogenicity have been shown to be associated with different clinical subtypes [32] and different onset ages [22]. Also, gender and disease duration could affect the SN echogenic pattern [11, 20, 22, 33]. In view of above-mentioned various factors, it was not surprising that varied sensitivity was found in our own studies. Lower sensitivity in our previous studies with a larger number of PD patients could also be due to different patient’s enrollment standard [11]. In the current study, PD patients were limited to the non-TD patients, and their ages and genders were strictly matched with that of MSA-P. Finally, the possibility of misdiagnosis could not be absolutely ruled out since a fairly part of non-TD PD patients were at their early disease stages. However, similar with a recent study conducted in Japanese patients where the sensitivity of SN hyperechogenicity was around 50% [23], all the patients in our cohort were followed up for at least 2 years to confirm the diagnosis. In supporting for the minimum possibility of misdiagnosis, the differing characteristics of non-TD PD and MSA-P in our study were obvious, that is, MSA-P patients had higher H-Y stages with shorter disease duration, which suggesting more aggressive progression.

In contrast, the sensitivity of SN hyperechogenicity for MSA-P was 38.4%. Even only the marked SN hyperechogenicity was considered, the sensitivity was up to 19.8%, which was still higher than most of previous reports [5, 6, 12, 15, 16, 18, 19, 21, 23]. The majority of sensitivity of SN hyperechogenicity for MSA was reported as about 10%, except for two studies, which were 25 and 50%, respectively [12, 27]. Just as occurred in PD, the most probable reason for the discrepancy between earlier studies and ours could be the spectrum bias. In the previous studies, the number of MSA patients enrolled was relatively small and ranged from 8 to 32 [5, 6]. A small sample size could not only affect statistical power, but also influence the statistical results even with a few cases of misdiagnosis. Moreover, there are two variants of MSA, which are MSA with predominant cerebellar ataxia (MSA-C) and MSA-P. No evidence could make sure that their SN hyperechogenicity pattern should be the same. Therefore, to minimum the bias mentioned above, the MSA patients selected in our study were all MSA-P variants and were followed up for at least two years to confirm the diagnosis in the same movement disorder clinic. In addition, the sample size of MSA-P in the present study reached up to 86, which was the largest one so far.

For overall patients, the sensitivity of SN hyperechogenicity for PD was 74.1%, and the specificity was 61.6%. If SN hyperechogenicity was marked, the specificity increased to 80.2%, but the sensitivity decreased to 46.3%. With either cutoff value, the positive predictive value for the diagnosis of non-TD PD was around 80%. However, among the patients with equal to or less than 3 years disease duration, the sensitivity and specificity further declined to 69.8% and 52.2% respectively, so did PPV decrease to 66.7%. The exact reason was not clear, and the possibility of misdiagnosis could not be definitely excluded since a fairly part of patients were at their early stages. However, it might be explained by the fact that there were different patterns between non-TD PD and MSA-P with regard to the correlation of SN hyperechogenicity with disease duration. In non-TD PD, it seems that with disease course prolonged, the area of SN hyperechogenicity became larger, which was discrepant with recent studies indicating that SN hyperechogenicity was stable during disease progression [17, 32, 34, 35]. Although the correlation between SN_L and disease duration was weak, several investigations [20, 22, 33] and our previous study [36] did support this positive correlation. In contrast, a tendency of reversed correlation was found in MSA-P patients, although without statistical significance. Therefore, it could be expected that within shorter disease duration, non-TD PD patients had smaller SN echogenicity, which could lead to lower percentage of SN hyperechogenicity, whereas MSA-P patients might have larger SN echogenicity, which could lead to higher percentage of SN hyperechogenicity. Additionally, this different correlation of SN hyperechogenicity with disease duration might also support the hypothesis about the distinct underlying mechanisms in the pattern of SN echogenicity of PD and MSA. So far, the origin of SN hyperechogenecity was not completely elucidated, nonchelated forms of iron, protein bound iron and microglia activation were all supposed to be involved in the reflection of SN hyperechogenecity signals [33, 37, 38]. Tissue iron has been shown to be elevated in both PD and MSA-P [39], but the detailed iron metabolism and gliosis might be different in PD and MSA [33, 37‐39].

Limitations of this study need to be mentioned. Indeed, we could not promise absolutely accurate diagnosis of PD and MSA, since no post-mortem verification was achieved. This also was the problem for the differentiation between MSA and other APD, such as CBD and DLB. SN hyperechogenicity has been reported frequent in CBD and DLB [40, 41], therefore misdiagnosis of those disorders could also contribute to higher sensitivity of SN hyperechogenicity in MSA-P. Additionally, although non-TD PD was considered as the phenotype that was most difficult to discriminate from MSA-P, one should note that non-TD and TD PD subtypes do not necessarily represent different subtypes of PD. TD subtype was much more likely to shift to non-TD subtype during the disease progression [42].

In summary, our present study demonstrated that SN hyperechognicity could help to discriminate non-TD PD from MSA-P; however, both sensitivity and specificity were not satisfactory, particularly in the earlier stage. Combination with other auxiliary tests seemed a necessary way to help differentiation of these two entities.