Keywords Head/Neck - Neurology - Physics and Technology
Introduction
Huntington’s disease (HD) is a neurodegenerative disease that affects the central nervous system and causes movement disorders, personality changes, and intellectual decline owing to expanded trinucleotide CAG sequence in the huntingtin gene located on chromosome 4 [1 ]. The mutant huntingtin gene causes degenerative changes in several brain regions, particularly in the striatum, which consists of the caudate nucleus (CN) and putamen [2 ].
Transcranial sonography (TCS) can detect echogenicity changes in various brain structures and can be used in the diagnosis of various neurodegenerative diseases such as Parkinson’s disease, multiple system atrophy, progressive supranuclear palsy, and Wilson’s disease [3 ]
[4 ]
[5 ]
[6 ]
[7 ]
[8 ]
[9 ]. The low cost and noninvasiveness of TCS allow the use of this technique for serial examinations during the disease course [10 ]. However, to date, only a few studies have evaluated the changes in the echogenicity of basal ganglia and brainstem structures in HD patients using qualitative (presence of pathology assessed as Yes/No) or semi-quantitative brain structure assessment, which is limited by the dependence on the image quality of the bone window and experience of the sonographer [11 ]
[12 ]
[13 ]
[14 ]
[15 ].
Fusion imaging combining TCS with other imaging methods such as MR imaging or computed tomography using Virtual Navigator and digitized image analysis improves the evaluation of visually detectable structures on TCS images and can be used for effective assessment of other brain structures, such as the insula, which are not easily discernable on TCS, as has been reported in other neurodegenerative diseases [16 ]
[17 ]. The present study aimed to compare the echogenicity of the CN, substantia nigra (SN), lentiform nucleus (LN), insula, and brainstem raphe (BR) between HD patients and healthy controls by TCS-MR fusion imaging using digitized image analysis, and to evaluate the accuracy of this technique in HD diagnosis.
Materials and Methods
Study population and clinical evaluation
All consecutive patients with genetically confirmed HD treated in the university hospital between January 2014 and December 2021 were selected for this study. The inclusion criteria for HD patients were (a) adult onset of HD in the early or advanced stage of the disease according to International Guidelines for the Treatment of Huntington’s Disease [18 ], (b) genetically confirmed HD, and (c) signed informed consent. The exclusion criteria for HD patients were (a) presence of other neurological or psychiatric disease, excluding HD, (b) contraindication to MR examination, such as cardiac implantable electronic device, implantable neurostimulation systems, claustrophobia, or gravidity; (c) intracranial pathology on brain MR images unrelated to HD, and (d) insufficient temporal bone window in TCS examination.
For the healthy controls (control group), age-, and gender-matched healthy volunteers from the Hospital Healthy Volunteer Registry were selected. The inclusion criteria for healthy volunteers were (a) willingness to participate and (b) signed informed consent. The exclusion criteria for healthy volunteers were (a) any symptoms of HD, (b) family history of HD, (c) any permanent medication, (d) any known neurological or psychiatric disease, (e) contraindication to MR examination, (f) intracranial pathology on brain MR images, or (g) insufficient temporal bone window in TCS examination.
The basic demographic data and medical history (age, sex, age of onset of motor symptoms, and depression) were collected from all HD patients at admission. Patients with current symptoms or previous history of depression were considered to suffer from depression. As part of genetic examination, the number of CAG triplets was determined in the HD patients. Physical and neurological examinations, including Unified Huntington’s Disease Rating Scale (UHDRS) evaluation of HD symptoms, were performed by a trained and certified investigator. The study was conducted in accordance with the ethical standards of our Institutional Ethics Committee (Approval No. 238/2013 and No. 111/14) and the 1975 Declaration of Helsinki (as revised in 1983 and 2008). All subjects provided signed informed consent.
Structural magnetic resonance
MR images were acquired using a 1.5 T Magnetom Avanto scanner (Siemens, Erlangen, Germany) or a 1.5 T whole-body Achieva scanner (Philip, Eindhoven, The Netherlands). The MR protocol included the following routine clinical spin echo images: T1-weighted (spatial resolution = 1 × 1 × 5 mm3 , echo time [TE] = 8.7 ms, repetition time [TR] = 500 ms), T2-weighted (spatial resolution = 0.56 × 0.56 × 5 mm3 , TE = 87 ms, TR = 4990 ms), and T2 fluid-attenuated inversion recovery (spatial resolution = 1 × 1.1 × 5 mm3 , TE = 81 ms, TR = 7000 ms). These sequences covered the entire brain and were used to evaluate brain abnormalities and generate anatomic images for fusion with TCS.
TCS
All patients were examined using the ultrasound scanner Esaote MyLab Twice (Esaote, Genoa, Italy) with a transcranial phased array probe at 1–5 MHz within 4 weeks after brain MR examination by an experienced neurosonographer. The device settings were established according to the recommendations by Walter and Školoudík [7 ]. The ultrasound system was equipped with virtual navigation and sensor mount, which allowed real-time image fusion of TCS and MR images. The approach was similar to that employed in previous studies on patients with Wilson’s and Parkinson's disease [17 ]. In brief, the Virtual Navigator procedures were implemented using an electromagnetic tracking system composed of a transmitter and a small receiver mounted on the ultrasound probe. The transmitter position, considered as the basis of the reference system, was fixed through a support, and the receiver provided the position and orientation of the ultrasound probe relative to the transmitter. If necessary, the image could be manually adjusted to overlap the brain structures of the MR with the TCS image. A proper head support was also used to keep the subject’s head as steady as possible. The patients were examined in a supine position on a horizontal bed with a tilt of 0°.
Two transverse sections were used to evaluate the intracranial structures. The butterfly-shaped structure of the BR and SN area in the transverse mesencephalic section was assessed. The CN, LN, and insula were evaluated in the transverse thalamic section. If a sufficient bone window was present, bilateral examination was performed. MR images were uploaded through Virtual Navigator, and eight facial markers were used for fusion imaging and visualization of brain structures. For each transverse TCS image, a corresponding MR image was acquired ([Fig. 1 ]).
Fig. 1 TCS-MR fusion image in thalamic plane. Normal echogenicity of caudate nucleus (horizontal arrow), lenticular nucleus (vertical arrow), and insula (stars) in healthy control 1a . Increased echogenicity of caudate nucleus (horizontal arrow), lenticular nucleus (vertical arrow), and insula (stars) in patient with Huntington’s disease 1b .
All images were saved in the Digital Imaging and Communications in Medicine format. Measurements were performed using a digital analysis called the B-mode assist program. The images were loaded into the application and cropped to a specific area window for each structure as follows: CN (area: 140 mm2 ), SN (area: 50 mm2 ), LN (area: 200 mm2 ), BR (area: 40 mm2 ), and insula (area: 196 mm2 ). [5 ]
[16 ]
[17 ]
[19 ]
[20 ]
[21 ] The elliptic region of interest was manually placed in the echogenic region of each structure. The echogenicity index was calculated as the total sum of areas (area under the curve) for each image. The higher of the two measured values for each structure (except the BR) was used for analysis.
Statistics
The Shapiro-Wilk test was used for normality assay. Data with normal distribution are reported as mean ± standard deviation, and for the CN and LN, the variables were logarithmized based on lognormal distribution hypothesis. The Mann-Whitney U-test was employed for comparisons between HD patients and healthy controls, and Spearman Rho analysis was used for correlations analysis. Receiver operating characteristic (ROC) curves were plotted, and sensitivity/specificity was calculated for determined cutoff points of the CN, LN, BR, and insula echogenicity [22 ]. Based on the cutoff point, the percentile value of the healthy population was calculated, and statistical evaluations were performed using Stata 17.0 software (StataCorp, College Station, TX, USA).
Results
Out of the 46 subjects selected for the study (22 HD patients and 24 healthy controls), two subjects (one HD patient and one healthy control) were excluded from the analysis due to an insufficient bone window. A total of 21 HD patients and 23 healthy controls underwent brain MR examination followed by TCS-MR fusion imaging examination of predefined brain structures. Demographic and clinical data are in [Table 1 ]. In total, 76.8% of HD patients reported current depression symptoms or previous history of depression.
Table 1 Demographics and clinical data.
HD patients
Healthy controls
P-value
CAG – trinucleotide repeats (in normal population: < 26 CAG repeats); HD – Huntington’s disease; UHDRS – Unified Huntington's Disease Rating Scale; SD – standard deviation
No. of subjects
21
23
NA
Age; mean ± SD; years
48.4 ± 17.4
38.0 ± 10.3
0.053
Female sex; n (%)
13 (61.9)
15 (65.2)
0.857
Number of CAG, mean ± SD
43.5 ± 6.4
NA
NA
UHDRS score; mean ± SD; points
49.0 ± 13.8
NA
NA
Duration of symptoms; mean ± SD; years
4.0 ± 4.9
NA
NA
Depression; n (%)
16 (76.8)
0 (0.0)
<0.001
The echogenicity of the CN, LN, and insula significantly differed between the HD patients and healthy controls ([Table 2 ]). The mean echogenicity indices for the CN (67.0 ± 22.6 vs. 37.9 ± 7.6), LN (110.7 ± 23.6 vs. 59.7 ± 11.1), and insula (121.7 ± 39.1 vs. 70.8 ± 23.0) were higher in the HD patients than in the control group (p < 0.001 in all cases). In contrast, BR echogenicity was lower in the HD patients than in the control group (24.8 ±5.3 vs. 30.1 ± 5.0, p < 0.001). There were no significant changes in the SN echogenicity index between the HD patients and healthy controls (30.8 ± 5.4 vs. 30.7 ± 5.5, p < 0.805), with five (23.8%) HD patients presenting an SN hyperechogenicity.
Table 2 Results of measurement of echogenicity indices on transcranial sonography.
TCS results
HD patients
Healthy controls
P-value
HD – Huntington’s disease; CN – caudate nucleus; LN – lentiform nucleus; SN – substantia nigra; BR – brainstem raphe; SD – standard deviation
CN echogenicity index; mean ± SD
67.00±22.56
37.94±7.64
<0.001
LN echogenicity index; mean ± SD
110.67±23.59
59.68±11.07
<0.001
SN echogenicity index; mean ± SD
30.79±5.35
30.70±5.45
0.805
Insula echogenicity index; mean ± SD
121.71±39.11
70.82±22.96
<0.001
BR echogenicity index; mean ± SD
24.82±5.30
30.10±5.34
<0.001
The cutoff values for the echogenicity indices for the CN (47.37), LN (81.97), insula (94.50), and BR (26.96) were calculated, which corresponded to the 95th , 99th , 90th , and 20th percentiles of the healthy population, respectively. The sensitivity, specificity, and area under the ROC curve (AUC) for the CN, LN, insula, and BR are presented in [Table 3 ] and [Fig. 2 ]. The echogenicity of the CN, LN, insula, SN, and BR did not correlate with age, gender, brain atrophy, UHDRS score, and duration of the disease (all Spearman correlation coefficients r < 0.5) (Supplementary Table S1) .
Table 3 Sensitivity and specificity of echogenicity indices on transcranial sonography.
Cutoff value
Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
AUC
AUC – area under the ROC curve; BR – brainstem raphe; CN – caudate nucleus; LN – lentiform nucleus; PPV – positive predictive value; NPV – negative predictive value; SN – substantia nigra
CN
47.37
85.71
95.65
94.74
88.00
90.91
LN
81.97
90.48
100
100
92.00
95.45
Insula
94.50
76.19
91.30
88.89
80.77
84.09
BR
26.96
76.19
86.96
84.21
80.00
81.82
Fig. 2 Receiver operating characteristic (ROC) of caudate nucleus (CN), lentiform nucleus (LN), insula, substantia nigra (SN), and brainstem raphe (BR) echogenicity of Huntington’s disease patients. Values above the reference curve represent hyperechogenicity in a selected brain structure, except for BR which was reverted to 1/BR in the ROC regression.
The differences in the measured width of the third ventricle by TCS and MRI were less than 1.0 mm in all patients. However, the measurement was not performed blindly because the width of the third ventricle was measured in the fusion images.
Discussion
By employing digital image analysis of TCS-MR fusion images, the present study observed significantly higher echogenicity of the CN, LN, and insula, and lower echogenicity of the BR in HD patients than in the healthy population. The CN and LN showed hyperechogenicity in 86% and 90% of HD patients, respectively, but there was no significant correlation between the clinical status of HD patients and echogenicity of the CN and LN. Similar findings for CN echogenicity (80%) and LN echogenicity (53.3%) have also been reported in a previous study. However, only hypokinetic HD patients were enrolled in that study [13 ]. It must be noted that most of the previous studies used only a semi-quantitative visual assessment scale for the evaluation area of the hyperechogenic CN and LN, and reported a lower incidence of CN hyperechogenicity in the range of 13–18% and a lower rate of LN hyperechogenicity in the range of 6–16% [11 ]
[12 ]
[14 ].
The reason for increased hyperechogenicity in the CN and LN region remains unknown, and only one study found a correlation between CN echogenicity and signal changes in the CN on MR T2-weighted images [10 ]. Furthermore, previous studies suggested an association between signal changes on MR images and an increase in the amount of iron or myelin level due to a decrease in neuron density [23 ]
[24 ]. MR studies had confirmed an increased amount of iron accumulation in the LN region [24 ], which may be one of the reasons for the changes in the echogenicity of the LN. Further investigations are needed to clarify the pathological chances in the area. The same pathological processes could have an effect on the change in the CN echogenicity on TCS.
The present study is the first to measure the insula echogenicity index in HD patients. A value of 94.5 was established as a cutoff point for insula hyperechogenicity in HD patients with a sensitivity and specificity of 76% and 91%, respectively. Higher hyperechogenicity of the insula has also been reported in other neurodegenerative diseases, such as Wilson’s disease [17 ]. In the present study, the clinical status of HD patients did not correlate with the echogenicity changes in the insula. However, other studies investigating insula degeneration have found associations with executive dysfunction and apathy [25 ]
[26 ]. Changes in the insula have been reported to occur at the beginning of the disease in the form of microglia activation, which is associated with neuronal loss [27 ], and further investigations are required to understand the alterations in the echogenicity in the insula region and the effects on HD patients.
In the present study, SN echogenicity in the HD patients did not significantly differ from that in the control group, and there was no correlation between the clinical status of patients and SN echogenicity. Nevertheless, 24% of the HD patients had a hyperechogenic SN. In an early study, a hyperechogenic SN was found in 26% of HD patients, with significant correlation between the number of CAG repeats and the severity of the clinical status [11 ]. However, a later study did not confirm correlation between SN hyperechogenicity and disease status, and the percentage of hyperechogenic SNs in HD patients was 41% [12 ]. Furthermore, a higher frequency of SN hyperechogenicity was noted among those with the juvenile form of HD (100%), whereas SN hyperechogenicity was detected in only 29% of those with adult HD. In addition, SN hyperechogenicity was correlated with bradykinesia subscore in UHDRS, thereby suggesting that the changes in SN echogenicity may be a marker of bradykinetic and rigid motor phenotype of HD [14 ]. In a recent study, a higher hyperechogenicity of the SN was visualized in 66% of HD patients, and was correlated with the general motor status measured by UHDRS-TMS [15 ]. These variations in the percentages of SN hyperechogenicity in these studies might be attributed to different motor phenotypes of HD.
Although previous studies have observed echogenicity changes in the raphe nuclei in HD patients, these studies used only a semi-quantitative visual assessment scale for echogenicity evaluation. In the present study, by employing digitized image analysis, hypoechogenicity in the raphe nuclei was noted in 76% of HD patients. A total of 12 out of 16 HD patients with decreased BR echogenicity reported current or previous depression symptoms. These results are in agreement with the findings of previous studies in which BR hypoechogenicity was observed in 49–67% of HD patients [15 ]. In the study by Krogias et al., 72% and 63% of patients exhibiting BR hypoechogenicity had current symptoms of depression and a history of depressive episodes, respectively, whereas all patients without a history of depressive episodes showed normal BR echogenicity. The cause of changes in BR echogenicity in HD patients is still not completely clear. One possible explanation is a reduction in echogenicity due to cellular and fiber tracts causing a density reduction of nuclei raphe. Dysfunction of the serotoninergic system in HD patients has already been demonstrated in earlier studies, and may be of significant importance because of its association with depression and depressive states [28 ]
[29 ].
Limitations
The present study has several limitations. First, the number of included patients was based on the sample size calculation, and the relatively low prevalence of HD limits the number of patients in particular TCS studies. Nevertheless, the number of examined patients allows calculation of the cutoff values and determination of the sensitivity and specificity of individual brain regions. Second, TCS as well as clinical examinations were performed by only one expert, and hence, inter-investigator variability could not be evaluated. Third, as only symptomatic patients were included in this study, the results could not be readily extrapolated to prodromal HD. Fourth, detailed analysis between patients’ symptoms (including UHDRS subscores and symptoms of depression) and TCS findings were not performed. Fifth, only HD patients with neurological symptoms were enrolled in the study. However, a study with HD patients in prodromal stage is planned. Finally, both neurosonographers and the neurologist were not blinded to the patients’ symptoms.
Conclusion
The results of the present study confirm that the echogenicity indices for the CN, LN, insula, and BR evaluated by TCS-MR fusion images using digital image analysis are sensitive and specific markers of HD. Nevertheless, further studies are needed to clarify the causes of changes in echogenicity in the given brain regions.
