Subregional Pattern of Striatal Dopamine Transporter Loss on ¹⁸F FP-CIT Positron Emission Tomography in Patients With Pure Akinesia With Gait Freezing

Question  Is the subregional pattern of striatal dopamine transporter loss in patients with pure akinesia with gait freezing similar to that in patients with progressive supranuclear palsy?

Findings  In this case-control study using F-18-fluorinated-N-3-fluoropropyl-2β-carboxymethoxy-3β-(4-iodophenyl)-nortropane (¹⁸F FP-CIT) positron emission tomography, the preferential subregional pattern of striatal dopamine transporter loss in patients with pure akinesia with gait freezing was similar to that in patients with progressive supranuclear palsy, which results from early dopamine transporter loss in the caudate nuclei, but not to that in patients with Parkinson disease.

Meaning  These results suggest a similar distribution of regional neuronal loss in the substantia nigra pars compacta between pure akinesia with gait freezing and progressive supranuclear palsy.

Importance  Pure akinesia with gait freezing (PAGF) is a clinical syndrome characterized by freezing of gait, handwriting, and speech without abnormal eye movement or cognitive impairment. Several studies have suggested that PAGF may be a variant of progressive supranuclear palsy (PSP). However, the characteristics of striatal dopamine transporter loss in PAGF are unknown.

Objective  To investigate the subregional pattern of striatal dopamine transporter loss in patients with PAGF in comparison with patients with PSP and those with Parkinson disease (PD).

Design, Setting, and Participants  This retrospective case-control study included 15 patients with PAGF, 27 with PD, 20 with PSP, and 11 healthy controls who underwent F-18-fluorinated-N-3-fluoropropyl-2β-carboxymethoxy-3β-(4-iodophenyl)-nortropane (¹⁸F FP-CIT) positron emission tomography between September 1, 2008, and July 31, 2014. The positron emission tomographic images were analyzed with 12 striatal subregional and 1 occipital volume-of-interest templates. The specific to nonspecific binding ratio (SNBR) and intersubregional ratio (ISR) in patients with PAGF were compared with those in patients with PD and those with PSP.

Main Outcomes and Measures  Comparisons of SNBRs of striatal subregions and ISR among patients with PAGF, PD, and PSP and healthy controls.

Results  The mean (SD) SNBRs (1.4 [0.7]) of the whole striatum in the 15 patients with PAGF (mean [SD] age, 71.4 [6.6] years; 7 men and 8 women) were similar to those in the 20 patients (mean [SD] age, 70.6 [4.5] years; 11 men and 9 women) with PSP (1.5 [0.5]) but significantly lower than those in the 27 patients (mean [SD] age, 67.7 [5.3] years; 10 men and 17 women) with PD (3.0 [1.3]). The mean (SD) SNBRs of the caudate nuclei in patients with PAGF (1.3 [0.9]) were significantly lower than those in patients with PD (3.5 [1.5]; P < .001) but slightly higher than those in patients with PSP (1.2 [0.5]). The mean [SD] anterior caudate to ventral striatum ISRs in patients with PAGF (0.5 [0.3]) were similar to those in patients with PSP (0.4 [0.2]) but not in patients with PD (1.0 [0.2]). The mean (SD) posterior to anterior putamen ISRs in patients with PAGF (0.4 [0.2]) were similar to those in patients with PD (0.5 [0.2]) and those with PSP (0.4 [0.2]).

Conclusions and Relevance  On ¹⁸F FP-CIT positron emission tomography, patients with PAGF show a pattern of preferential dopaminergic loss similar to that seen in patients with PSP. These results suggest a similar distribution of regional neuronal loss in the substantia nigra pars compacta between PAGF and PSP. This finding may be one of the pathophysiological results suggesting that PAGF is a phenotypic variant of PSP.

Pure akinesia with gait freezing (PAGF) is a clinical syndrome characterized by freezing of gait, handwriting, and speech. At presentation, however, patients with PAGF show no rigidity, tremor, cognitive impairment, or abnormal eye movement and do not respond to levodopa, in contrast with patients with other parkinsonian disorders.¹

In 1974, Imai and Narabayashi² described 2 patients with pure akinesia who had no abnormal eye movement or cognitive impairment. Riley et al³ reported that 5 patients with stereotyped pure akinesia developed vertical gaze palsy later in life. Based on the results in previous studies, Williams et al¹ proposed clinical diagnostic criteria under the name pure akinesia with gait freezing and reported that PAGF and progressive supranuclear palsy (PSP) have a common tau pathologic process and suggested that PAGF is a subtype of PSP. However, isolated gait freezing can also be present in Parkinson disease (PD), dementia with Lewy bodies, and Binswanger disease.⁴ Thus, there has been a debate about whether PAGF is a subtype of PSP or a distinct disease that overlaps pathophysiologically with PSP.

Degenerative loss of specific dopaminergic neurons in the nigrostriatal pathway occurs in patients with PD and atypical parkinsonian disorders.⁵ In a previous study using F-18-fluorinated-N-3-fluoropropyl-2β-carboxymethoxy-3β-(4-iodophenyl)-nortropane (¹⁸F FP-CIT) positron emission tomography (PET), patients with PD and atypical parkinsonisms showed significantly decreased striatal dopamine transporter (DAT) loss with distinct patterns of subregional DAT loss.⁶ This result suggests that the different pattern of preferential subregional DAT loss may reflect the preferential site and pathophysiologic features of the disease in the midbrain.

Few studies using dopaminergic imaging^7,8 reported decreased striatal uptake in patients with PAGF. However, these studies did not evaluate the subregional pattern of striatal dopaminergic loss and simply compared the results with those of healthy controls. Therefore, we investigated the subregional pattern of striatal DAT loss in patients with PAGF and whether it is similar to that in patients with PSP or PD by using ¹⁸F FP-CIT PET.

Between September 1, 2008, and July 31, 2014, we enrolled 15 patients with PAGF (mean [SD] age, 71.4 [6.6] years; 7 men and 8 women) in this study. The diagnosis of PAGF was based on the proposed clinical diagnostic criteria.¹ We also enrolled 20 patients with clinically probable PSP (mean [SD] age, 70.6 [4.5] years; 11 men and 9 women) diagnosed based on the National Institute of Neurological Disorder and Stroke and the Society for Progressive Supranuclear Palsy clinical criteria⁹ and 27 patients with PD (mean [SD] age, 67.7 [5.3] years; 10 men and 17 women) diagnosed based on the UK Parkinson’s Disease Society Brain Bank Clinical Diagnostic Criteria.¹⁰ The patients’ Hoehn-Yahr stages and Unified Parkinson’s Disease Rating Scale, Part III, scores immediately before PET scan and after discontinuation of antiparkinsonian drugs for at least 12 hours were available in the medical records. All patients were assessed by neurologists in the movement department (S.J.C. and H.K.P.). Healthy controls were selected for the purpose of age matching from the data pool maintained at Asan Medical Center, University of Ulsan College of Medicine, of individuals with normal ¹⁸F FP-CIT results. None of the controls had a neurologic deficit, and all had normal findings on magnetic resonance imaging of the brain. The Asan Medical Center Institutional Review Board approved this study and waived the need to obtain consent from patients with PAGF, PSP, and PD. All healthy control participants gave written informed consent.

¹⁸F FP-CIT was synthesized as previously described.¹¹ All patients stopped taking antiparkinsonian drugs at least 12 hours before undergoing PET. Two patients with PAGF, 2 with PSP, and 4 with PD were taking selective serotonin reuptake inhibitors; they did not discontinue these medications before undergoing PET. All ¹⁸F FP-CIT PET scans were performed with a Biograph Truepoint 40 PET/CT camera (Siemens) 3 hours after intravenous injection of 185 MBq of ¹⁸F FP-CIT. Positron emission tomographic images were acquired for 10 minutes in 3-dimensional mode immediately after the brain computed tomographic scan for attenuation correction and image fusion. Computed tomographic scanning was performed at 120 kVp and 228 mAs and with a slice thickness of 1.5 mm. ¹⁸F FP-CIT PET images were reconstructed with a TrueX algorithm and an all-pass filter using a 336 × 336 matrix.

Image processing was performed with Statistical Parametric Mapping (Wellcome Department of Imaging Neuroscience, Institute of Neurology, University College London) within MATLAB R2013a for Windows (The MathWorks Inc) and MRIcro, version 1.40 (Chris Rorden, Columbia, SC; http://www.mccauslandcenter.sc.edu/crnl/). All reconstructed PET images were spatially normalized to Talairach space by use of a standard ¹⁸F FP-CIT PET template, which was made in-house, as described previously.^6,12-14 Images were reoriented so that the striatum contralateral to the symptomatic side was on the left side. If there was no laterality, the anatomic left became the left side of the standard space.

Quantitative analyses were based on 12 volume-of-interest (VOI) templates of bilateral striatal subregions (ventral striatum, anterior caudate, posterior caudate, anterior putamen, posterior putamen, and ventral putamen) and 1 template of the occipital subregion. The anterior commissure coronal plane was divided between the anterior caudate and posterior caudate and between the anterior putamen and posterior putamen. The dividing boundary between the posterior putamen and the ventral putamen was the anterior-posterior commissure transaxial plane.^6,12-14

The automatically normalized VOI template was adjusted manually by one of us (J.S.O.) under the supervision of a nuclear medicine physician with 20 years of experience (J.S.K.), using our in-house VOI editing software, called ANIQUE (Asan Medical Center Nuclear Medicine Image Quantification Toolkit of Excellence).¹⁴

The level of activity in each VOI was calculated. The specific to nonspecific binding ratio (SNBR) was defined as follows: (mean standardized uptake value of the striatal subregional VOI − mean standardized uptake value of the occipital VOI)/mean standardized uptake value of the occipital VOI, considering occipital uptake to be nonspecific binding. The normalized SNBR (%BR) was defined as follows: (SNBR for patient/SNBR for healthy control) × 100. The intersubregional ratio (ISR) was defined as the ratio of the SNBR of 1 striatal subregion to that of another striatal subregion. The ISRs for the anteroposterior gradient (posterior putamen to anterior putamen and posterior caudate to anterior caudate), the ventrodorsal gradient (anterior caudate to ventral striatum, anterior putamen to ventral striatum, and posterior putamen to ventral putamen), and the asymmetry (left to right) of the SNBR for each of the 6 striatal subregions were calculated.

We used Welch analysis of variance for the comparison of continuous variables among groups. The Dunnett T3 test was applied to a post hoc analysis of between-group comparisons. Comparisons according to sex were performed using the χ² test. We used SPSS for Windows, version 18.0 (SPSS Inc), for statistical analyses, and P < .05 was considered statistically significant. Data for the study variables were expressed as the mean (SD).

The clinical characteristics of the individuals enrolled in this study are summarized in Table 1. There were no significant differences in age, sex, or duration of symptoms among the study groups. Those in the PAGF group had a significantly lower mean (SD) Hoehn-Yahr score than did patients with PSP (2.7 [0.7] vs 3.7 [1.1]; P < .05) but had a higher score than did patients with PD (2.0 [1.0]; P < .05). Patients with PAGF had lower mean (SD) Unified Parkinson’s Disease Rating Scale, Part III, scores than did patients with PSP (20.2 [9.8] vs 29.9 [11.5]; P < .05). All 15 patients with PAGF showed freezing of gait at the time of ¹⁸F FP-CIT PET imaging. In the PD group, 6 of 27 patients (22%) had the symptom of freezing of gait. Among patients with PD, no statistically significant difference was found between those with freezing of gait and those without freezing of gait in mean (SD) age (67.7 [4.6] vs 67.7 [5.5] years; P = .97), duration of symptoms (5.4 [6.7] vs 4.2 [4.3] years; P = .61), Hoehn-Yahr score (2.3 [1.8] vs 2.0 [0.7]; P = .62), or Unified Parkinson’s Disease Rating Scale, Part III, score (25.3 [21.3] vs 18.4 [10.2]; P = .47). In 8 of 20 patients with PSP (40%), freezing of gait was present. During the follow-up period (mean [SD], 8.9 [2.0] years), 8 of 15 patients with PAGF developed PSP-like symptoms (1 patient with vertical gaze palsy, 3 patients with axial rigidity, 3 patients with dysarthria, and 1 patient with cognitive impairment).

The mean (SD) SNBR values for the whole striatum in the PAGF group (1.4 [0.7]; more affected side, 1.3 [0.6]; less affected side, 1.5 [0.7]) were similar to those in the PSP group (1.5 [0.5]; more affected side, 1.3 [0.5]; less affected side, 1.6 [0.6]) but significantly lower than those in the PD group (3.0 [1.3]; more affected side, 2.9 [1.3]; less affected side, 3.1 [1.3]; P < .001) (Table 2), as were the %BR values (Figure 1). In all subregions except the ventral striatum, the PAGF group showed significantly lower SNBR and %BR values than the PD group, especially in the caudate nuclei. The mean (SD) SNBRs of the caudate nuclei in patients with PAGF (1.3 [0.9]) were significantly lower than those in patients with PD (3.5 [1.5]; P < .001) but similar to those in patients with PSP (1.2 [0.5]). The SNBR and %BR values of caudate nuclei in the PAGF group were significantly lower than those in the PD group with or without freezing of gait (eTable 1 in the Supplement). No subregion showed significantly different SNBR or %BR values between the PAGF and PSP groups. The representative images of the PAGF, PSP, and PD groups and healthy controls are shown in Figure 2.

The ISRs of the SNBRs for the anteroposterior and ventrodorsal gradients in the PAGF, PSP, and PD groups and healthy controls are listed in eTable 2 in the Supplement. Of ISRs showing an anteroposterior gradient, the mean (SD) ISRs for the posterior putamen to anterior putamen in the PAGF (0.4 [0.2]; more affected side, 0.4 [0.1]; less affected side, 0.4 [0.2]), PSP (0.4 [0.2]; more affected side, 0.4 [0.2]; less affected side, 0.5 [0.2]), and PD (0.5 [0.2]; more affected side, 0.5 [0.2]; less affected side, 0.5 [0.1]) groups were significantly lower than those in the healthy controls (0.9 [0.1]; P < .001). There was no significant difference in the ISR values for the posterior putamen to anterior putamen among the disease groups (Figure 3A). However, the mean (SD) ISRs for the anterior caudate to ventral striatum suggesting a ventrodorsal gradient in the PAGF group (0.5 [0.3]; more affected side, 0.5 [0.3]; less affected side, 0.5 [0.2]) were not different from those in the PSP group (0.4 [0.2]; more affected side, 0.3 [0.1]; less affected side, 0.5 [0.2]) but were significantly lower than those in the PD group (1.0 [0.2]; more affected side, 0.9 [0.2]; less affected side, 1.0 [0.2]; P < .001) (Figure 3B). The ISRs for the anterior caudate to ventral striatum in the PAGF group (0.5 [0.3]; more affected side, 0.5 [0.3]; less affected side, 0.5 [0.2]) were significantly lower than those in the PD group without freezing of gait (1.0 [0.2]; more affected side, 0.9 [0.2]; less affected side, 1.0 [0.2]; P < .001), as well as those in the PD group with freezing of gait (1.0 [0.2]; more affected side, 0.9 [0.2]; less affected side, 1.0 [0.2]; P < .05) (eTable 3 in the Supplement). The changes in the ISRs for the anterior caudate to ventral striatum according to duration of symptoms in the PAGF, PSP, and PD groups are shown in the eFigure in the Supplement. Regarding asymmetry, the mean (SD) ISRs of the anterior caudate (0.7 [0.2]; P < .05 vs healthy controls) and posterior caudate (0.7 [0.3]; P < .05 vs healthy controls) in the PSP group and the anterior putamen (0.7 [0.2]; P < .001 vs healthy controls), posterior putamen (0.7 [0.2]; P < .001 vs healthy controls), and ventral putamen (0.8 [0.2]; P < .05 vs healthy controls) in the PD groups showed asymmetry. No subregion showed significant asymmetry in the PAGF group.

In our study, use of ¹⁸F FP-CIT PET in patients with PAGF revealed a significant loss of DAT in the whole striatum. This finding is suggestive of presynaptic dopaminergic neuronal degeneration in PAGF and consistent with previous studies using dopaminergic imaging^7,8 that reported diffuse striatal dopaminergic loss. In addition, the preferential subregional pattern of striatal DAT loss in patients with PAGF was similar to that in patients with PSP, which results from early DAT loss in the caudate nuclei, but not similar to that in patients with PD.

This early caudate involvement could have implications for the nigrostriatal pathway, the dopaminergic pathway that connects the striatum with the substantia nigra in the midbrain. Patients with PD showed more severe involvement in the posterior putamen than the anterior putamen, indicating an anteroposterior gradient in preferential DAT loss. However, the relative sparing of the caudate nuclei resulted in a preserved ISR for the anterior caudate to ventral striatum. In a previous study, PD tended to involve the ventrolateral part of the pars compacta of the substantia nigra, resulting in DAT loss in the posterior part of the putamen.⁵ In contrast, our results showed extensive DAT loss in the caudate nuclei and putamen in patients with PSP. In these patients, there was no predilection for the ventrolateral tier of the substantia nigra but a tendency for wide involvement of both ventral and dorsal tiers. These pathologic characteristics result in extensive DAT loss seen on ¹⁸F FP-CIT PET images of the nigrostriatal pathway.¹⁵ The ventral striatum, which is mainly projected by the ventral tegmental area, was relatively spared compared with the putamen or caudate nuclei.

The similar pattern of DAT loss in patients with PAGF and those with PSP might suggest a similar distribution of underlying dopaminergic neuronal loss in the substantia nigra. Ahmed et al¹⁶ reported that the 8 cases of pathologically confirmed PSP whose pathologic findings were also consistent with pallido-nigro-luysian atrophy (PNLA) have many clinical features similar to those of PAGF. Their pathologic findings showed extensive neuronal loss in the substantia nigra, which was similar to that in PSP, although the neuronal loss in the medial portion was more severe in the cases of pathologically combined PSP and PNLA. Other studies also reported severe neuronal loss in the substantia nigra in patients with PAGF^17,18 and those wth PNLA,^19-21 even though the regional pathologic changes within the substantia nigra were not described. The association between PAGF and PNLA remains unclear.

On the other hand, in a functional imaging study with ¹⁸F fluorodeoxyglucose PET, patients with PAGF showed decreased glucose metabolism in the midbrain, which was similar to that seen in patients with PSP, but less cortical hypometabolism than in those with PSP.⁸ Progressive supranuclear palsy is characterized by hypometabolism of the midline frontal cortex and midbrain, seen on ¹⁸F fluorodeoxyglucose PET images.²² In the ventral tiers of the substantia nigra, dopaminergic neurons project exclusively to the putamen. Dopaminergic neurons in the dorsal tier of the substantia nigra extend to limbic and cortical targets, forming the mesolimbic and mesocortical pathways, as well as to the striatum. The mesocortical pathway is associated mainly with the midline frontal cortex.²³ There is no known tendency for tau pathologic involvement in the midline frontal cortex. Therefore, the results might imply less severe dopaminergic neuronal loss in the dorsal tier of the substantia nigra in patients with PAGF than in those with PSP. However, dopaminergic pathways other than the nigrostriatal pathway are beyond the scope of this study. Further studies are needed to elucidate the pattern of neuronal loss in the substantia nigra in patients with PAGF.

Pathologically, PAGF shares many characteristics with PSP. In PAGF, severe neuronal loss in the globus pallidus, substantia nigra, and subthalamic nucleus were described; these areas are also vulnerable to the pathologic conditions of PSP.²⁴ Neurofibrillary tangle pathologic findings in the substantia nigra, subthalamic nucleus, globus pallidus, and dentate nucleus were also observed in patients with PAGF and those with PSP.^25,26 In addition, a similarity in the tau pathology distribution of PAGF and PSP has been noted.²⁷ In both diseases, the substantia nigra and subthalamic nucleus were consistently the most severely affected structures, although the severity of the tau pathology was milder in patients with PAGF than in those with PSP. Regionally, tau pathology was significantly less severe in patients with PAGF than in those with PSP in the parietal cortex, pontine nucleus, dentate nucleus, and cerebellar white matter.

The clinical symptoms of PAGF have often been associated with those of PSP. Since Imai and Narabayashi² first described the syndrome of pure akinesia, Riley and coworkers³ subsequently reported that patients with pure akinesia eventually developed PSP-like symptoms, such as falls, axial rigidity, and vertical gaze palsy, later in their disease course. Imai and coworkers²⁸ also described patients with pure akinesia who developed vertical gaze palsy later in life. Pure akinesia with gait freezing and PSP are not responsive to levodopa. The syndrome of PAGF has been known by different terms, such as gait ignition failure²⁹ or primary progressive freezing gait.³⁰ Likewise, patients who were diagnosed as having primary progressive freezing gait developed symptoms of vertical gaze palsy and postural instability resembling PSP.^4,31

Each part of the striatum has been associated with certain parkinsonian features. The posterior putamen has been thought to be associated with main motor symptoms, including tremor, rigidity, and akinesia, except for bradykinesia, which is thought to be mediated by a cortical mechanism. Apathy, anxiety, and depression are associated with the limbic part of the striatum—the ventral striatum. Cognitive dysfunction has been associated with dopaminergic loss in the caudate nuclei.³² Cognitive decline tends to develop later in patients with PAGF than in those with PSP. However, DAT loss in the caudate nuclei in patients with PAGF did not differ significantly from that in patients with PSP, although those with PAGF showed a milder degree of DAT loss in the caudate nuclei.

Freezing of gait is not only a characteristic clinical feature in patients with PAGF or PSP but also common in those with PD. It is known to occur more often in patients with advanced PD. In our study, the prevalence of gait freezing in the PD group is consistent with a previous study of a large group of patients with PD,³³ considering that the patients with PD in our study were, on average, in a mild stage of disease. Nevertheless, our results showed that the decreased ISRs for the anterior caudate to ventral striatum were a distinct feature of the PAGF group and not the PD group, regardless of the presence of freezing of gait. Freezing of gait is thought to be associated with locomotor circuit dysfunction. The locomotor circuit consists of the frontal cortical region, basal ganglia, and midbrain locomotor region. The basal ganglia, facilitated by dopamine, normally inhibits the mesencephalic locomotor region.³⁴ However, it is not yet known which part of the basal ganglia is responsible for locomotor circuit dysfunction. Other distinct clinical features of PAGF compared with PSP, such as absence of vertical gaze palsy or falls in the early stage of disease, could also not be explained by nigrostriatal dopaminergic neuron loss alone. Vertical eye movement is associated with the rostral interstitial nucleus of the medial longitudinal fasciculus, which contains the vertical gaze center.³⁵ Early falls and postural instability in PSP are associated with reduced activation of the thalamus.³⁶

Our study had several limitations. First, the retrospective nature of the study was a limiting factor, even though a substantial number of patients with PAGF were included. Second, pathologic confirmation of diagnosis was not available. Although all clinical diagnoses were based on strict diagnostic criteria, the possibility of misdiagnosis exists.³⁷ Third, partial volume effects could lower the count of the small striatal subregions. However, the high-spatial resolution of the PET scanner can minimize this issue, and the subregional pattern of DAT loss differed significantly among the disease groups, suggesting that this characteristic is not likely to affect the results. Finally, misregistration of template-based VOI analysis could have occurred. Therefore, we manually adjusted each striatal subregion separately after automatic normalization of the VOI templates.

Pure akinesia with gait freezing and PSP represent a similar subregional pattern of preferential dopaminergic loss seen on ¹⁸F FP-CIT PET. The similarity of the pattern of DAT loss between PAGF and PSP suggest a similar distribution of underlying regional neuronal loss in the substantia nigra pars compacta. This finding may be one of the pathophysiological results suggesting that PAGF is a phenotypic variant of PSP.

Corresponding Author: Jae Seung Kim, MD, PhD, Department of Nuclear Medicine, Asan Medical Center, University of Ulsan College of Medicine, 88 Olympic-ro 43-gil, Songpa-gu, Seoul 05505, Korea (jaeskim@amc.seoul.kr).

Accepted for Publication: June 20, 2016.

Published Online: October 10, 2016. doi:10.1001/jamaneurol.2016.3243

Author Contributions: Dr Kim had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Han, M. Oh, J. S. Oh, Kim.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Han, M. Oh, J. S. Oh, Lee, Kim.

Critical revision of the manuscript for important intellectual content: Han, J. S. Oh, S. J. Oh, Chung, Park, Kim.

Statistical analysis: Han, J. S. Oh.

Obtained funding: Kim.

Administrative, technical, or material support: Lee, Chung, Kim.

Study supervision: M. Oh, S. J. Oh, Chung, Park, Kim.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported by grant HI14C2768 from the Korea Health Technology Research and Development Project through the Korea Health Industry Development Institute, funded by the Ministry of Health & Welfare, Republic of Korea.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.