 |
|
||||
| Home | Login | |||||
| About | Editorial board | Articles
|
NSI Publications
|
Search | Instructions | Online Submission | Subscribe | Videos | Etcetera | Contact |
|
||||||||||||||||||||
|
|
|
Neuromelanin-Sensitive Magnetic Resonance Imaging as a Measure for Differential Diagnosis of Essential Tremor and Parkinson's Disease
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.383826
Keywords: Essential tremor, MRI, Parkinson's disease, substantia nigra
Both the authors Tingting Xuan and Xue Fang are co.authors and contributed equally to this article. Parkinson's disease (PD) and essential tremor (ET) are the most common movement disorders in neurological diseases, especially in the elderly. The prevalence of PD in the population over 65 years is about 5% and the number is expected to double by 2030, especially in developing countries.[1] Clinically, typical symptoms of PD include progressive aggravation of bradykinesia, myotonia, resting tremor, and postural balance disturbance.[2] ET has a prevalence of 4–5% in people >65 years old, and it is characterized by a feature of symmetric 4–8 Hz action tremor that frequently affects the hands and arms.[3] Resting tremor occurs when the body is completely relaxed and does not contract, which is considered to be the most important symptom for distinguishing PD and ET in clinical work. However, ET and PD may show overlapping clinical features. For example, ~30% of ET patients have resting tremors,[4],[5] 25% of PD patients lack resting tremors,[4],[6] and even some ET patients can develop PD. The diagnosis of tremor-dominant PD and ET is still challenging, especially in the early stages of the disease. About 25% of patients with PD were misdiagnosed as ET patients in the early stage.[7] Up to date, the diagnosis of PD usually relies on medical history and motor symptoms. The misdiagnosis rate is high; thus, effective diagnostic biomarkers of PD are urgently needed. In patients with PD, pathological alteration mainly occur in the substantia nigra (SN) region of the brain. Mei et al.[7],[8] demonstrated that transcranial sonography (TCS) examination could precisely distinguish PD from other movement disorders in the clinic. However, TCS examination has limited applicability in patients with insufficient temporal windows, which is present in approximately 15–60% of the Asian population.[9] Single photon emission computed tomography (SPECT) and positron emission tomography (PET) images are reproducible for the quantitative assessment of dopamine transporter (DAT) binding in both ET and PD patients;[10],[11] however, the disadvantages of high cost, limited availability, and long scan time significantly restrict their application in clinical work. A pathogenic characteristic of PD is the selective loss of neurons containing neuromelanin in SN and locus coeruleus, with an estimated 5–15 years progression to a loss of 60%–80% of the nigral dopamine neurons before symptoms appear.[12] The level of neuromelanin in the SN appears to continuously decrease with disease progression in PD patients, and the changes can be effectively detected by NM-MRI. Studies from independent groups have reported that the NM-related signal is significantly decreased in the SN of PD brains and shows high accuracy in distinguishing PD patients from healthy individuals.[13],[14] Recently, NM-MRI has been used to define the disparity between ET and PD in two studies, and the results showed that neuromelanin level in the SN was remarkably lower in the PD group than that in the ET and control groups.[15],[16] In this study, we explore the relationship between the area and contrast-to-noise ratio (CNR) of SN to PD and ET, which will help patients with early diagnosis and early treatment to the greatest extent, and explore the difference of area and CNR of SN in PD with a different course of disease and subtypes, to validate the feasibility of NM-MRI as a biomarker to identifying PD progression and subtypes.
Participants Twenty-three patients with treated PD, 20 patients with treated ET, and 18 healthy controls were recruited at the General Hospital of Ningxia Medical University between September 2020 and October 2021. All patients were diagnosed by the International Parkinson and Movement Disorder Society (MDS) in 2015 for PD[17] or the Consensus Statement of the Movement Disorders Society on Tremor.[18] All core symptoms were scored by two experienced professionals independently during the off-period on the unified Parkinson's disease rating scale (UPDRS) score and the modified Hoehn–Yahr rating scale (H-Y). According to the grouping method of Stebbins et al.,[19] PD patients were divided into a Tremor dominant (TD) group, a postural instability/gait difficulty (PIGD) group, and a mixed Parkinson's disease (PDM) group. The healthy subjects were enrolled from patients with no history of neurological diseases. The exclusion criteria included a history of mental illness such as dementia, severe anxiety, depression, and schizophrenia, MRI contraindications, MMSE score < 24 points, and Hamilton Depression Scale (HAMD 24 item version > 20 points). Consent forms were obtained from all participants, and all the procedures in this study were approved by the Institutional Review Board of the General Hospital of Ningxia Medical University. MRI imaging acquisition All MRI scans were performed on a 48-channel 3.0T MRI scanner (SIGNA, Architect, GE, America). The MRI pulse sequences were similar to those described by Sasaki et al.[20] The imaging parameters of T1-weighted fast spin echo (FSE) included 1. repetition time/effective echo time (TR/TEeff), 2. 633 ms/10 ms; echo train length (ETL), 3, slice thickness, 2.5 mm; no intersection gap; the number of slices, 20; matrix size, 512 × 320; field of view, 220 × 220 mm2; and acquisition time, 8 mins. The sections were obtained in an oblique axis plane perpendicular to the floor of the fourth ventricle, with the coverage extending from the posterior commissure to the inferior border of the bridge. Furthermore, T1-weighted images and T2-weighted fluid-attenuated inversion recovery (FLAIR) images using conventional MRI sequences were acquired. All images were evaluated by an experienced neuroradiologist, and images with other pathological features were excluded from further assessment. Image analysis The MRI images were analyzed on a GE AWD 4.7 Workstation. High signal on T1-weighted images in SN was observed in three slices, the location and signal intensity distribution of the regions of interest (ROI) were determined on the intermediate slice of SN according to the previous measurement method[16] [Figure 1]. A radiologist with 10 years of experience who was blinded to the subjects' information processed the images of all included cases twice, with a time interval greater than 7 days. Measurements were performed for both the left and the right sides, and the averaged value was used because of no significant difference between the two sides. The ratio of the area of the high signal region of the SN to the midbrain area of the same layer was then determined.
The signal intensity (SI) of each SN site and the SI of the adjacent cerebral peduncle (CP) were measured. The CNR between SN and background signal was: (CNR) SN = (SISN-SICP)/SICP, where SISN is the average signal intensity of bilateral SN and SICP is the average signal intensity of adjacent midbrain cerebral peduncles at the same level. The area of the aperture of the ROI used the same size, respectively: CP was 10 mm2, and each sub-region of SN was 5 mm2. Each subregion of SN was divided into three sub-regions that were parallel to the largest axis of the SN as the largest longitudinal axis, which was divided into three equal parts [Figure 2]. The measurements were performed three times to compute the average value for subsequent statistical analysis. The intraclass correlation coefficient (ICC) of the measured values of ROI was 0.85–0.96.
Data analysis Statistical analyses and plotting of data were performed using SPSS 26.0 (IBM, Armonk, NY, USA) and GraphPad Prism 8.0 (GraphPad Software Inc., San Diego, CA, USA), respectively. The measurement values that conform to a normal distribution are expressed as the mean ± standard deviation (SD). Measurement data that did not conform to a normal distribution are expressed as medians (25th percentile, 75th percentile). When the measurement data conformed to the normal distribution, a t-test or one-way analysis of variance (ANOVA) was used for comparison between groups; for the measurement data that did not conform to the normal distribution, the non-parametric rank sum test was used. Spearman's correlation was used to correlate the SN area, SN to the midbrain area ratio in the same layer, SN signal intensity with H-Y staging, and UPDRS third part score. A P value less than 0.05 was considered significant. Receiver operation characteristic (ROC) curve analysis was performed to determine the discrimination capacity of the measures of SN area, the ratio of SN to midbrain area in the same layer, and SN signal intensity for separating ET and PD patients, as well as PD patients and controls. In addition, specificity and sensitivity values, positive and negative predictive values, the area under the ROC curve, and the Yoden index were calculated. Optimal cut-point values were obtained through a method that simultaneously maximizes both sensitivity and specificity.
Demographic statistics and clinical data analysis There were no significant differences in gender and age among the three groups. The median disease duration was 4 years (3–5.75 years) in the ET patients and 2 years (1–5 years) in the PD patients. In the PD patients, the severity was relatively mild, the Hoehn and Yahr (H-Y) staging score was 1.5 (1.5–2.5), and the average score of the UPDRS III was 21. The demographic and clinical details of all subjects are shown in [Table 1].
There were no significant differences in gender, age, disease duration, H-Y stage, and UPDRS-III score among the three groups [Table 2].
Analyses of the SN area and the ratio of SN to midbrain area By manual analysis of the T1-weighted images, we obtained an SN area of 27.0400 mm2 (14.8400 –35.1200 mm2) in the PD group, 40.2100 mm2 (25.3850–51.7425 mm2) in the ET group, and 41.63 mm2 (30.4550–54.7800 mm2) in the control group. Compared with the control group, the T1 high-intensity signal area of SN only showed an insignificant reduction in the ET group (P > 0.05, P = 0.573), whereas it was significantly reduced in the PD group (P = 0.003, PD vs ET; P = 0.001, PD vs. control) [Figure 3]a]. The area of the midbrain in the same layer was manually calculated and then used for determining the ratio of the SN area to the midbrain area. The median value of the ratio was 0.0512 (0.0315–0.0636) for the PD group, 0.0701 (0.0458–0.0919) for the ET group, and 0.0747 (0.0525–0.0881) for the control group. Compared with the control group, the ratio of SN area to midbrain area in the same layer in ET patients was slightly lower (P > 0.05, P = 0.884). However, in the PD group, the ratio was significantly lower than that in the ET and the control groups (P = 0.006, PD vs ET; P = 0.005, PD vs. control) [Figure 3]b.
The SN area of PD patients with H-Y stage ≤2 was (27.8812 ± 10.8441) mm2, and PD patients with H-Y stage >2 was (22.2100 ± 8.5517) mm2. There was no significant difference in SN area between the two groups (P = 0.2610, P > 0.05) [Figure 4]a; the ratio of SN area to midbrain area in PD patients with H-Y stage ≤2 was 0.0528 ± 0.0227, and the ratio between SN area and midbrain area in PD patients with H-Y stage >2 was 0.0419 ± 0.0132, and there was no significant difference in the ratio of SN area to midbrain area in the same layer between the two groups (P = 0.2820, P > 0.05) [Figure 4]b.
The SN area of TD-type patients was 26.3250 ± 10.0935 mm2, that of PDM-type patients was 29.7878 ± 11.7683 mm2, and that of PIGD-type patients was 22.6500 ± 8.9172 mm2. There was no significant difference in SN area among the three groups (P = 0.3880, P > 0.05) [Figure 5]a, the ratio of the SN area to the midbrain area in the TD type patients was 0.0444 ± 0.0148, that of PDM type patients was 0.0570 ± 0.0217, that of PIGD patients was 0.0461 ± 0.0239, and there was no significant difference in the ratio of SN area to midbrain area among the three groups (P = 0.4430, P > 0.05) [Figure 5]b.
Receiver operation characteristics curve analysis of the SN area and ratio of SN area to midbrain area in the same layer When the cut-off value (or critical value) of the SN high signal area was set to 37.1700 mm2 [[Figure 6]a1], the sensitivity and specificity to distinguish ET and PD patients were 65.0% and 87.0%, respectively. Receiver operating characteristic (ROC) analyses indicated that the area under the ROC curve (AUC) was 0.7630 and the Youden index was 0.5200 [[Figure 6]a2]. In the aspect of the ratio of SN area to midbrain area, the cut-off value to distinguish ET from PD was 0.0683 [[Figure 6]b1], and the sensitivity and specificity were found to be at 60% and 53.3%, respectively. The AUC was 0.7478 and the Youden index was 0.47 [[Figure 6]b2].
To distinguish PD patients from healthy controls, a cut-off value of 28.1350 mm2 was found for the SN area [[Figure 7]a1], and ROC analysis determined the sensitivity and specificity as 89.0% and 61.0%, respectively. In addition, the relevant AUC value and Youden index were 0.8104 and 0.5000, respectively [[Figure 7]a2]. When the cut-off value of 0.0705 was found for the ratio of SN area to midbrain area [Figure 7b1], the sensitivity and specificity were 55.6% and 87.0%, respectively, the AUC was 0.7585, and the Youden index was 0.4260 [Figure 7b2].
Compared with the control group, there was no significant difference in the SN area or ratio of SN area to the midbrain in the same layer in ET patients (P > 0.05) [Figure 8].
Analyses of the SN signal intensity Compared with the ET group, the mean CNR value of the SN and the respective CNR values of the three subregions were all weakened in the PD group, and only the CNR in the middle part was significantly different from the control group (P = 0.006). There was no significant difference in CNR value between the lateral part and the control group (P > 0.05). Compared with the control group, the mean CNR value of the SN and the respective CNR values of the three subregions in the PD group were weakened. There was no significant difference in the average CNR value of each subregion and the CNR value of the medial part compared with the control group (P > 0.05) [Table 3].
There were no significant differences in the average CNR value of each subregion and the CNR value of the medial part, middle part, and the lateral part between the two groups [Table 4].
There were no significant differences in the average CNR value of each subregion and the CNR value of the medial part, middle part, and lateral part among the three subtypes groups [Table 5].
Receiver operation characteristics curve analysis of the SN signal intensity The sensitivity of the CNR value of the middle part of the SN for differentiating PD from controls was 77.8%, the specificity was 73.9%, the AUC was 0.7802, and the Youden index was 0.5170. The sensitivity of the CNR value of the lateral part of the SN for differentiating PD from controls was 72.2%, the specificity was 78.3%, the AUC was 0.8043, and the Youden index was 0.5050 [Figure 9].
The sensitivity of the CNR value of the middle part of the SN for differentiating ET from PD was 65%, the specificity was 87%, the AUC was 0.7500, and the Youden index was 0.5200 [Figure 10].
Correlation analysis We found that the size of the SN area in PD patients was correlated with UPDRS third part score and H-Y stage; there was no correlation between the ratio of SN area to midbrain area and the above indicators [Table 6].
In this study, we assessed the area and CNR values of SN in PD, ET, and control groups on NM-MRI images. The SN area is an optimal diagnostic biomarker to distinguish ET from PD, with an AUC of 0.7630, a sensitivity of 65%, and a specificity of 87%. Similarly, the ratio of SN area to midbrain area in the same layer had an AUC of 0.7478, a sensitivity of 60%, and a specificity of 87% in diagnosis. It is worth noting that the midbrain area hardly affects the determination of the SN area. Thus, in clinical practice, PD and ET could be distinguished by rapid measurement of SN area. Our results are consistent with previous neuroimaging studies.[15],[16] Compared with the ET group, the mean CNR value of the SN and the respective CNR values of the three subregions were all weakened in the PD group, and only the CNR in the middle part was significantly different from the control group (P = 0.006). There was no significant difference in CNR value between the lateral part and the control group (P > 0.05). A major clinical challenge we are currently facing is the differential diagnosis of PD from ET, as there is considerable overlap in clinical symptoms and a high rate of misdiagnosis, whereas patients with ET have been shown to have an increased risk of developing PD in the future,[21] and it has also been suggested by some that the action tremor of PD may be a variant of ET or augmented physiological tremor.[22] Therefore, the differential diagnosis may allow patients to benefit significantly from early treatment and a good prognosis. Currently, a good history and detailed neurological examination are usually the main basis for identifying PD and ET, leading to a certain subjectivity in the assessment of the condition and the diagnosis of the disease. Early understanding of the differences in neuroimaging of PD and ET provides an objective basis for accurate patient diagnosis, identifies specific features of both diseases to reveal differences and commonalities, and enriches our understanding of the underlying pathophysiologic mechanisms to improve existing diagnostic tools and treatment options. PD is characterized by a marked neuromelanin loss, there are numerous reports to verify the diagnosis of PD by measuring SN area, volume, and signal.[23],[24] We found that the amount of neuromelanin in the SN was not significantly different between the ET patients and the healthy subjects, dysfunctions of the cerebellar-thalamocortical network are intrinsically altered in ET. This may explain our results. However, compared with the ET patients and the healthy subjects, the PD patients showed a significantly lower T1 high signal in the SN region. This difference was more apparent in accordance with known pathological changes.[23],[25] These results indicated that the neuromelanin alterations of SN in ET and PD have different patterns, which are consistent with a previous report by Reimão et al.[15],[16] Neuromelanin is a dark pigment mainly present in SN, and the signal intensity on NM-MRI is closely related to the number of dopaminergic neurons in SN. Although PD is characterized by a marked neuromelanin loss, a recent study has also shown that aging is also associated with neuromelanin loss in the healthy brain.[26],[27] In the current study, the age and gender distribution of subjects in the healthy control group was comparable to the PD group, thereby minimizing the interference of age-related neuromelanin alterations. We measured both the T1 hyperintense area and the ratio of SN area to midbrain area in the same layer to estimate the neuromelanin in SN. Both methods generated similar results, which showed high sensitivity and specificity in discriminating between ET patients and PD patients. The MRI is a safe and widely utilized imaging system, and our semi-quantitative assessment methods for NM-MRI images can be easily integrated into daily clinical practice. Taken together, our findings demonstrated the promising applicability of NM-MRI in the differential diagnosis of ET and early-stage PD. However, the correlation between neuromelanin and iron content in SN has not been determined in the current study, which needs further analysis.[27] In addition, the different pathological patterns that affect SN in ET and PD remain to be determined. Compared to TD patients, it has been proposed that patients with PIGD exhibit more severe symptoms, faster disease progression, depressive tendencies, and more severe cognitive impairment;[28] however, the neural mechanisms that differ between subtypes are unclear. Previous reports have researched different motor subtypes of PD patients in NM-MRI; however, the results differ.[24],[29],[30] Our results showed no statistical difference in SN area, SN area to midbrain area ratio, and intensity among the TD, PIGD, and PDM groups, which may be related to the small sample size. Previous pathology reported that patients in the PIGD group had more severe loss of neurons in the SN than those in the TD group.[31] Therefore, we hypothesize that the occurrence of tremors is not closely associated with nigrostriatal pathway disorders, but may be caused by cerebellar–thalamic–cortical pathway disorders and further larger sample sizes are needed to compare SN changes among motor subtypes. We speculate that the occurrence of tremors is not closely related to nigrostriatal pathway disorders, and further sample sizes are needed to compare SN changes among motor subtypes. Finally, our findings show a gradual decrease in SN area as motor symptoms worsen and progress in PD patients, showing potential value in disease monitoring. Our study has some limitations: (1) Small number of patients in each group, and a lack of pathological confirmation and autopsy data. The clinical diagnosis may be incomplete as a reference. The small sample size limits our further stratification of PD patients in clinical subtypes for further analysis and limits the accuracy of the sensitivity and specificity estimation. Therefore, further study is needed to improve this method for distinguishing PD and ET patients. The big data analysis of the SN area and the ratio of the SN area to the midbrain area of the same layer can better monitor the changes in neuromelanin. (2) Disadvantages of imaging technology: The spatial resolution was relatively low, the scan time was relatively long, and there was in-plane signal inhomogeneity. It is necessary to further optimize the sequence, shorten the time, and extend the applicability while ensuring image quality for clinical practice. (3) The SN hyperintense areas were determined by manual drawing, which is prone to observer deviation. (4) Insignificant asymmetry in the SN region: We did not find a significant asymmetry in the SN region of the PD patients, and there was no correlation between the side and the most severe clinical impact. This may be caused by the limited number of subjects in our study.
Based on our findings, we conclude that NM-MRI can improve diagnostic accuracy in PD, and can be used as a specific and sensitive potential diagnostic biomarker for PD. This indicates that NM-MRI has a high potential value in the diagnosis of PD. To differentiate the motor subtypes of Parkinson's disease, it is necessary to expand the sample size and optimize the experimental protocol for further investigation and analysis. The NM-MRI assessment methods used in this study may also apply to other conditions, namely atypical Parkinson's syndrome. Financial support and sponsorship This work was supported by the Key Research and Development Program of Ningxia (2022BEG02046,2022BEG03130). Conflicts of interest There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 6], [Figure 7], [Figure 8], [Figure 9], [Figure 10], [Figure 5]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5], [Table 6]
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||
