Abstract
Background
The differential diagnosis of progressive supranuclear palsy (PSP) and Lewy body disorders, which include Parkinson disease and dementia with Lewy bodies, is often challenging due to the overlapping symptoms.
Purpose
To develop a diagnostic tool based on diffusion tensor imaging (DTI) to distinguish between PSP and Lewy body disorders at the individual-subject level.
Materials and Methods
In this retrospective study, skeletonized DTI metrics were extracted from two independent data sets: the discovery cohort from the Swedish BioFINDER study and the validation cohort from the Penn Frontotemporal Degeneration Center (data collected between 2010 and 2018). Based on previous neuroimaging studies and neuropathologic evidence, a combination of regions hypothesized to be sensitive to pathologic features of PSP were identified (ie, the superior cerebellar peduncle and frontal white matter) and fractional anisotropy (FA) was used to compute an FA score for each individual. Classification performances were assessed by using logistic regression and receiver operating characteristic analysis.
Results
In the discovery cohort, 16 patients with PSP (mean age ± standard deviation, 73 years ± 5; eight women, eight men), 34 patients with Lewy body disorders (mean age, 71 years ± 6; 14 women, 20 men), and 44 healthy control participants (mean age, 66 years ± 8; 26 women, 18 men) were evaluated. The FA score distinguished between clinical PSP and Lewy body disorders with an area under the curve of 0.97 ± 0.04, a specificity of 91% (31 of 34), and a sensitivity of 94% (15 of 16). In the validation cohort, 34 patients with PSP (69 years ± 7; 22 women, 12 men), 25 patients with Lewy body disorders (70 years ± 7; nine women, 16 men), and 32 healthy control participants (64 years ± 7; 22 women, 10 men) were evaluated. The accuracy of the FA score was confirmed (area under the curve, 0.96 ± 0.04; specificity, 96% [24 of 25]; and sensitivity, 85% [29 of 34]).
Conclusion
These cross-validated findings lay the foundation for a clinical test to distinguish progressive supranuclear palsy from Lewy body disorders.
© RSNA, 2019
Online supplemental material is available for this article.
See also the editorial by Shah in this issue.
Summary
The study provided cross-validated evidence that a normalized score derived from diffusion tensor MRI fractional anisotropy values distinguishes between individuals affected by progressive supranuclear palsy and Lewy body disorders with a high level of specificity and sensitivity.
Key Results
■ To distinguish progressive supranuclear palsy versus patients with Lewy body disorders, a fractional anisotropy (FA)–based score was superior to other MRI diffusion tensor metrics.
■ A normalized FA value distinguished between patients with progressive supranuclear palsy and patients with Lewy body disorders with specificity of 91%–96% and sensitivity of 85%–95%.
■ The diagnostic performance of the FA score was area under the curve of 0.96–0.97 for distinguishing between patients affected by progressive supranuclear palsy and patients affected by Lewy body disorders.
Introduction
The diagnosis of progressive supranuclear palsy (PSP) (1) is often challenging, especially during the earlier stages of the disease course when the clinical symptoms, particularly parkinsonian symptoms, often overlap with other conditions (2,3). Clinical examination based on symptoms can lead to misdiagnoses between PSP and α-synucleinopathies, such as idiopathic Parkinson disease (4) and dementia with Lewy bodies (5). The misdiagnosis rate can reach up to 30% between Parkinson disease and PSP (2,6). Misdiagnosis has implications for clinical care (eg, treatment responses to dopaminergic medications) and for disease-modifying tau- or α-synuclein–targeted treatment trials. From a biologic perspective, PSP is characterized by the abnormal accumulation of 4-repeat isoform tau. Neuropathologic studies have shown that one of the hallmarks of PSP is the accumulation of 4-repeat isoform tau in glial cells, as well as a strong involvement of white matter tracts (7). In contrast, Lewy body disorders, such as Parkinson disease and dementia with Lewy bodies, tend to spare supratentorial white matter, especially early in the disease process (7,8). Several diffusion tensor imaging (DTI) studies have shown pronounced differences between PSP and both healthy control participants and other patients groups (9–11) in several white matter tracts including the superior cerebellar peduncle, premotor aspect of the superior longitudinal fasciculus, the corpus callosum, and the anterior aspect of the fronto-occipital fasciculus (12,13). A growing body of evidence supports the use of diffusion MRI to improve the differential diagnosis of PSP and other neurodegenerative conditions (11,14). A study by Sajjadi and colleagues (11) revealed widespread differences in DTI that accurately distinguish 4-repeat tauopathies (including PSP) from other neurodegenerative conditions, such as Alzheimer disease and semantic variant primary progressive aphasia, at individual level.
We hypothesized that the combination of hypothesis-driven anatomic regions along with DTI analysis could provide an accurate differential diagnosis of PSP and Lewy body disorders, which is currently challenging. Specifically, we focused on the superior cerebellar peduncle and on a set of frontal white matter structures (including but not limited to premotor aspect of the superior longitudinal fasciculus and the genu of the corpus callosum). A large region of interest allows one to obtain a score that encompasses several regions that have been reported in the literature to be sensitive to pathologic features of PSP (12). At the same time, the exclusion of temporal, parietal, and occipital regions may improve the specificity of the marker because changes in diffusion metrics due to microstructural damages or neuronal death can occur in these regions for conditions such as dementia with Lewy bodies (15).
We aimed to develop and investigate a DTI-based strategy for single-subject classification between PSP and Lewy body disorders based on two independent cohorts of patients.
Materials and Methods
Participants
This retrospective study was conducted in accordance with the Declaration of Helsinki. All participants gave written informed consent and participated in an informed consent procedure. All study protocols were approved by the review boards of Lund (protocol numbers, 2008–290 and 2011–277) and the University of Pennsylvania (protocol number, 298201).
Discovery cohort.—Sixteen patients with PSP, 34 patients with Lewy body disorders (23 with Parkinson disease, 11 with Parkinson disease with dementia), and 44 neurologically healthy control participants were tested. The participants are part of the Swedish BioFINDER study (http://biofinder.se). The cohort partially overlapped (approximately 70%) with previously published studies in Parkinson disease and PSP (9,16), but the analyses in the current article are completely independent of these prior reports. Participants were recruited from 2012 to 2018 and constitute a convenience series.
Validation cohort.—Thirty-four patients with PSP, 25 patients with Lewy body disorders (18 with dementia with Lewy bodies, seven with Parkinson disease, one with dementia), and 32 neurologically healthy control participants were recruited at the Frontotemporal Degeneration Center and related clinics of the University of Pennsylvania (Philadelphia, Pa). Ten patients (six with PSP and four with Lewy body disorders) had a neuropathologic confirmation of disease. All participants in the validation cohort underwent an MRI within 2 years from the first visit. Participants were recruited from 2010 to 2018 and constitute a convenience series.
In both cohorts, board-certified neurologists (S.H., D.J.I., A.F.D., M.A.S., M.G., and O.H., all with more than 10 years of experience) diagnosed the patients according to published consensus criteria that include neurologic examination, neuropsychologic testing, and MRI (5,17–19). Patients were also followed longitudinally to confirm the diagnosis with a clinical visit every 2 years. Participants were included in the study only if a diffusion-weighted imaging sequence with a sufficient image quality on neuroradiologic revision was available. Patients with a neurologic condition such as stroke or hydrocephalus, a primary psychiatric disorder, or a medical condition causing cognitive difficulty were excluded (Fig 1). In the validation cohort, four patients who had a clinical diagnosis other than PSP at the time of death but a neuropathologic diagnosis of PSP were also included in the cohort.
Figure 1a:
Flowchart shows summary of selection process of participants in the study for (a) discovery cohort and (b) validation cohort. PSP = progressive supranuclear palsy.
Figure 1b:
Flowchart shows summary of selection process of participants in the study for (a) discovery cohort and (b) validation cohort. PSP = progressive supranuclear palsy.
MRI Protocols
Discovery cohort.—Diffusion-weighted imaging data were acquired by using a 3.0-T Skyra MRI scanner (Siemens Healthineers, Erlangen, Germany). A single-shot echo-planar imaging sequence with 99 diffusion-weighted imaging volumes was performed (repetition time, 8100 msec; echo time, 103 msec; resolution, 2.3 × 2.3 × 2.3 mm3; field of view, 294 × 294 × 120 mm3; 96 imaging volumes with b values of 250, 500, 1000, and 2750 sec/mm2 distributed over six, six, 20, and 64 directions and three volumes with b value of 0 sec/mm2; twofold parallel acceleration; and partial Fourier factor, 6/8).
Validation cohort.—A diffusion-weighted imaging sequence was performed by using a 3.0-T Trio scanner (Siemens Healthineers) as a single-shot spin-echo echo-planar imaging sequence (repetition time, 6700 msec; echo time, 85 msec; resolution, 1.9 × 1.9 × 2 mm3; field of view, 245 × 245 × 114 mm3; three repetitions of a series of 34 images; 30 directions with b value of 1000 sec/mm2 and four with b value of 0 sec/mm2; threefold parallel acceleration; and partial Fourier factor, 6/8).
Analysis Strategy
The analysis strategy described here was developed and optimized in the discovery cohort by two authors (N.S., with 8 years of experience in MRI analysis and M.N., with more than 10 years of experience in MRI research) and subsequently validated in the validation cohort. The authors were blinded to clinical diagnosis during the quality control and preprocessing stages of the MRI analysis for both cohorts. For finalizing the analysis procedure, the authors were unblinded to the data of the discovery cohort but still blinded to the data of the validation cohort. Diffusion-weighted imaging data were processed by using a combination of in-house and open-source algorithms (DTI-TK algorithm and FMRIB Software Library, version 5.0.11; Oxford, United Kingdom; including the tract-based spatial statistics pipeline; see Appendix E1 [online] for details).
To obtain scalar values from our target regions, first a mask of the frontal lobe was manually delineated in template space following anatomic landmarks (by N.S.). The frontal lobe mask was then intersected with the skeletonized white matter obtained with the tract-based spatial statistics analysis to create the frontal white matter mask. This mask included as main tracts the anterior portion of the corpus callosum, the anterior limb of the internal capsule, part of the external capsule, anterior corona radiata, superior corona radiata, anterior cingulum, anterior aspect of the superior longitudinal fasciculus, and the uncinate fasciculus. The frontal white matter mask was delineated in the discovery cohort and warped to the space of the validation cohort (Fig 2, A) (one author [O.H.] inspected the mask to verify accuracy). Second, a mask of the left and right superior cerebellar peduncle was obtained by warping the ICBM-DTI-81 white-matter labels atlas (24) to the template space, and the labels for the left and right superior cerebellar peduncles were combined and intersected with the skeleton (Fig 2, B).
Figure 2:
Image shows illustrations of two masks used to compute MRI fractional anisotropy (FA) score. A, Supratentorial mask (frontal white matter mask, in green) is overlaid onto mean FA map of discovery cohort. B, Mask of skeletonized superior cerebellar peduncle overlaid onto mean FA map of discovery cohort. Images are displayed following radiologic convention.
Fractional anisotropy score.—Median values extracted from the frontal white matter and superior cerebellar peduncle masks for each DTI metric were averaged together, resulting in a single value (per subject). This value was subsequently normalized by using a z-score procedure with the healthy control participants as the comparison group (healthy control participants average [frontal white matter and superior cerebellar peduncle masks]: mean ± standard deviation [SD] for fractional anisotropy [FA], 0.49 ± 0.03 (unit from 0 to 1); mean ± SD for mean diffusivity, 0.74 mm2/sec ± 0.04; mean ± SD for radial diffusivity, 0.53 mm2/sec ± 0.04; mean ± SD for axial diffusivity, 1.19 mm2/sec ± 0.03), thereby generating an estimate of the distance between the specific value of each participant and the values of a typical population. In the discovery cohort, the score based on FA was the metric that provided better classification performance. Therefore, we propose as main outcome an FA-based score.
Neuropathologic Examination
Autopsy confirmation was obtained for 10 patients of the validation cohort. Postmortem assessments were performed by expert neuropathologists (E.B.L. and J.Q.T., with more than 10 years of experience) by using established consensus criteria and immunohistochemistry (25).
Statistical Analysis
Demographic factors (age and sex) and clinical characteristics (disease duration and Mini-Mental State Examination [MMSE] score) were compared by using χ2 and Mann-Whitney tests (P < .05 was considered to indicate statistical significance). Group differences on the main outcome metric (FA score) were assessed with general linear analysis including age, sex, and MMSE score in the model as possible confounding variables (significance of P < .05). The classification performance of the FA score was tested with a receiver operating characteristic analysis and a logistic regression model with fivefold cross-validation and the classification variable was the diagnostic group. The cutoff score was set to z = −1.65 based on the standard normal distribution. According to the cumulative standard distribution function, at a z score of −1.65, the 95% of the distribution is above this value while the 5% is below the threshold (cumulative probability <0.05). Analyses were performed by using the packages Statsmodels (version 0.9; https://www.statsmodels.org/stable/index.html) and scikit-learn (version 0.20.3; https://scikit-learn.org) implemented in Python (version 3.6; https://www.python.org/) (open source).
Results
Demographics
Discovery cohort.—Sixteen patients with PSP (mean age, 73 years ± 5, eight women, eight men), 34 patients with Lewy body disorders (mean age, 71 years ± 6; 14 women, 20 men), and 44 healthy control participants (mean age, 66 years ± 8; 26 women, 18 men) were tested. No difference in the sex distribution across diagnostic group was found (group comparison specified as group 1–group 2; healthy control participants–Lewy body disorders, P = .18; healthy control participants–PSP, P = .74; Lewy body disorders–PSP, P = .78). Age did not differ between groups of patients (P = .08) but differed between healthy control participants and patients (healthy control participants–Lewy body disorders, P = .003; healthy control participants–PSP, P < .001). Mean disease duration did not differ between groups of patients (6 years ± 2 for PSP vs 6 years ± 5 for Lewy body disorders; P = .32).
Validation cohort.—Thirty-four patients with PSP (mean age, 69 years ± 7; 22 women, 12 men), 25 patients with Lewy body disorders (mean age, 70 years ± 7; nine women, 16 men), and 32 healthy control participants (mean age, 64 years ± 7; 22 women, 10 men) were included in the study. No difference in the sex distribution across diagnostic group was found except between Lewy body disorders and healthy control participants (healthy control participants–Lewy body disorders, P = .03; healthy control participants–PSP, P = .93; Lewy body disorders–PSP, P = .06). Age did not differ between groups of patients (P = .28) but differed between healthy control participants and patients (healthy control participants–Lewy body disorders, P = .002; healthy control participants–PSP, P = .002). Mean disease duration did not differ between groups of patients (3 years ± 2 for PSP vs 3 years ± 2 for Lewy body disorders; P = .19).
Clinical and demographic information of both cohorts are provided in Table 1.
Table 1:
Demographic Data of the Two Independent Cohorts
Discovery cohort.—For analysis purposes, the patients affected by Parkinson disease and Parkinson disease with dementia were grouped under the label Lewy body disorders. The results revealed that the FA score based on the FA accurately separated PSP from Lewy body disorders (mean FA score, −3.73 ± 1.29 for PSP vs −0.35 ± 1.20 for Lewy body disorders; mean area under the curve over five cross-validation, 0.97 ± 0.04; specificity, 91% [31 of 34]; and sensitivity, 94% [15 of 16] based on a cutoff score of −1.65 corresponding to a cumulative probability of <.05) (Fig 3). The other DTI metrics (ie, mean diffusivity, radial diffusivity, and axial diffusivity) maintained good levels of specificity but showed lower sensitivity (see Table 2 for the complete results). Therefore, FA was selected as the primary DTI metric in the subsequent analyses. A multiple linear regression analysis showed a difference in the FA score between PSP and Lewy body disorders also when adjusting for the potential confounding effects of age, sex, and MMSE score (group [Lewy body disorders–PSP]: β = −3.1; P < .001; where the FA score is on a z-score scale), despite each of these features not individually contributing to our model (age, β = −.03; P = .40; sex, β = −.01; P = .99; MMSE score, β = .09; P = .10; where age is expressed in years, sex is coded as 1 for female and 0 for male, and MMSE score is on a scale between 0 and 30). A supplementary analysis suggested that the FA score accurately distinguished PSP from healthy control participants (mean FA score: healthy control participants, 0 ± 1; area under the curve, 0.98 ± 0.03; specificity, 93% [41 of 44]; and sensitivity, 94% [15 of 16]). Similarly, a multiple linear regression analysis showed a difference in the FA score between PSP and healthy control participants also when correcting for age, sex, and MMSE score (group [healthy control participants–PSP]): β = −3.3; P < .001; age, β = −.02; P = .26; sex, β = −.09; P = .75; MMSE score, β = .06; P = .34; where the FA score is on a z-score scale, age is expressed in years, sex is coded as 1 for female and 0 for male, and MMSE score is on a scale between 0 and 30).
Figure 3:
Images show diagnostic accuracy of fractional anisotropy (FA) score in discovery cohort. Left image shows analysis of FA score in discovery cohort with distribution of values of FA score across three clinical groups. Box extends from lower to upper quartile values, with line at median. Whiskers extend from box to show range of data (matplotlib [version 3.0.3] in Python [version 3.6]). Dotted black line represents cutoff value of FA score. Right image shows receiver operating characteristic (ROC) curve analysis. Dotted red line represents chance level. Blue line is mean curve across fivefold cross validation. Each thinner line represents one of five repetitions and gray areas cover one standard deviation around mean. AUC = area under the curve, HC = healthy control participants, LB = Lewy body, PSP = progressive supranuclear palsy. *** = P < .001.
Table 2:
Classification Performance for Distinguishing between PSP and Lewy Body Disorders of the Most Commonly Used Diffusion Tensor MRI Metrics in the Discovery Cohort
Validation cohort.—The accuracy of FA score classification was validated in a separate cohort. For consistency with the analysis in the discovery cohort, the patients affected by α-synucleinopathies were grouped under the label Lewy body disorders. The FA score distinguished PSP from the Lewy Body disorders with high classification accuracy (mean FA score, −2.79 ± 1.04 for PSP vs 0.03 ± 0.82 for Lewy body disorders; area under the curve, 0.96 ± 0.04; specificity, 96% [24 of 25]; and sensitivity, 85% [29 of 34] based on the same cutoff established in the discovery cohort) (Fig 4). Also in the validation cohort, a multiple linear regression analysis showed a difference in the FA score between PSP and Lewy body disorders even when accounting for the potential confounding effect of age, sex, and MMSE score (group [Lewy body disorders–PSP]: β = −3.1; P < .001; age, β = .03; P = .12; sex, β = −.35; P = .18; MMSE score, β = .05; P = .11; where the FA score is on a z-score scale, age is expressed in years, sex is coded as 1 for female and 0 for male, and MMSE score is on a scale between 0 and 30). The PSP–healthy control participants classifier was also high performing (mean FA score for healthy control participants, 0 ± 1; area under the curve, 0.96 ± 0.03; specificity, 90% [29 of 32]; and sensitivity, 85% [29 of 34]), as well as the results of the related multiple linear regression analysis (group [healthy control participants–PSP]: β = −2.5; P < .001; age, β = .01; P = .50; sex, β = .5; P = .09; MMSE score, β = .1; P = .04; where the FA score is on a z-score scale, age is expressed in years, sex is coded as 1 for female and 0 for male, and MMSE score is on a scale between 0 and 30). In this instance, MMSE reached significance as well but an analysis of variance showed no effect of interaction between MMSE and the diagnostic groups (group [healthy control participants–PSP]: F value = 66.9; P < .001; MMSE: F value = 4.00; P = .06; Group × MMSE: F value = 0.78; P = .38) revealing that MMSE score does not differentially affect the FA score in the two groups of participants. The FA score correctly classified nine of 10 patients with pathologically confirmed PSP. The four participants with only a neuropathologic diagnosis of PSP had the following values of FA score: −1.90, −1.93, −2.10, and −0.95. Therefore, three of them were correctly classified by using the FA score with typical FA score values for patients with PSP within 1 SD from the mean of this diagnostic group (−2.79 ± 1.04), while one case was misclassified by using the FA score.
Figure 4:
Images show main results in validation cohort. Left image shows analysis of fractional anisotropy (FA) score in validation cohort with distribution of values of FA score across three clinical groups. Box extends from lower to upper quartile values, with line at median. Whiskers extend from box to show range of data (matplotlib [version 3.0.3] in Python [version 3.6]). Dotted black line represents cutoff value of FA score. Right image shows receiver operating characteristic (ROC) curve analysis. Dotted red line represents chance level. Blue line is mean curve across fivefold cross validation. Each thinner line represents one of five repetitions and gray areas cover one standard deviation around mean. AUC = area under the curve, HC = healthy control participants, LB = Lewy body, PSP = progressive supranuclear palsy. *** = P < .001.
See Appendix E1 (online) for a possible alternative analysis strategy showing converging results to the FA score, as well as for the results of a correlation analysis between the FA score and clinical metrics of disease severity (see Fig E1 [online]).
Discussion
The diagnosis of progressive supranuclear palsy (PSP) is often challenging due to the overlapping symptoms with other conditions, including Lewy body disorders such as Parkinson disease and dementia with Lewy bodies. Several diffusion tensor MRI studies have shown pronounced differences between PSP and both healthy control participants and other patient groups (9–11) in multiple white matter tracts (12,13). We proposed a diffusion tensor imaging (DTI)–based marker (fractional anisotropy [FA] score) that could improve the differential diagnosis of PSP and Lewy body disorders. The results showed that the FA score can distinguish between PSP and Lewy body disorders with specificity of 91% (31 of 34) and sensitivity of 94% (15 of 16) in the discovery cohort, and with specificity of 96% (24 of 25) and sensitivity of 85% (29 of 34) in the validation cohort.
One of the main strengths of our study is that the classification accuracy was demonstrated in two independent cohorts imaged with different MRI scanners with different imaging protocols. The relevance of the approach is further strengthened by the fact that the validation cohort was constrained to participants who were imaged within 2 years from the first visit, which is arguably the time window in which the inclusion of an accurate imaging marker would be most beneficial to the diagnostic process. Today, the clinical diagnosis of PSP typically occurs when distinctive signs such as the vertical supranuclear gaze palsy become evident, but these signs could emerge several years from the onset of the disease (1,26).
Other markers, such as serum levels of neurofilament light chain (27) and morphometric measures based on structural T1-weighted images (28,29), have also been proposed to have high levels of accuracy. However, different methodologies tend to capture different aspects of the degenerative process. For example, DTI is more sensitive to pathologic changes in white matter, whereas the morphologic changes reflect the atrophy of different structures. Therefore, complementary markers could be taken into consideration during the diagnostic process.
Our approach is based on DTI-related metrics, which have both benefits and drawbacks compared with more advanced diffusion MRI methods. A major benefit of using DTI is its robustness and general availability. The robustness was suggested by the successful application of the FA score in two separate cohorts. Combining data is arguably one of the biggest challenges to multicenter DTI studies due to differences in the acquisition parameters. Here we used a standardized score against a site-specific group of control participants, which helped to overcome this limitation. We thus expect the FA score to be generally applicable, particularly because DTI can be performed with most current clinical scanners. The analysis strategy is also available and based on free software. Most importantly, the output of the analysis is a single score per subject that could be easily interpreted in a clinical setting to enhance the diagnostic process.
A major drawback of a DTI-based method is that a measure such as FA is sensitive both to anisotropy (caused by the presence of elongated cell structures such as axons) and orientation dispersion (caused by fanning or crossing fibers) (30). Therefore, the biologic underpinnings of an FA reduction cannot be extrapolated from DTI, but alternative acquisition strategies have been explored (30–32). However, estimation of an exact biologic property remains generally elusive despite the use of both advanced modeling and acquisitions (33).
Our study had some limitations, which need to be taken into account when interpreting the data. The analyses are based on patients who were diagnosed with the most common variant of PSP (PSP-Richardson syndrome). The patients with PSP with neuropathologic confirmation that was not correctly classified had an atypical presentation suggesting corticobasal syndrome. However, other four participants with PSP but with a diagnosis of corticobasal syndrome at the time of the MRI scan were correctly classified by using the FA score. The demographic characteristics of our cohorts were not entirely balanced. This was a limitation of this retrospective study, although the regression models showed no relation between age and sex and the FA score when the diagnostic group was included in the model. Moreover, neuropathologic confirmation was only available for a subset of patients included in our study, which was a limitation that will be important to address in future large-scale studies when autopsy confirmation is available for all evaluated cases. Another limitation was the lack of a comparison to other disorders, such as multiple system atrophy and corticobasal degeneration, for which PSP can be mistaken. The frontal white matter mask was partially manually delineated and this was arguably a potential limitation in terms of reproducibility in novel cohorts, although since it was only generated in the discovery cohort and warped in the space of the validation cohort, we anticipate minimal issues associated with reproducibility. The original frontal white matter mask is available on request from the lead author (N.S.).
Early accurate diagnosis is increasingly important as clinical trials of pathologic-specific therapeutics for progressive supranuclear palsy and Lewy body disorders are growing in number. This study based on two independent cohorts with cross validation demonstrated the fractional anisotropy score as an imaging marker that was shown to be robust across two centers and reliable at the single-subject level. Further studies are needed to test the accuracy of the fractional anisotropy score across subtypes of progressive supranuclear palsy, as well as to compare this disease with other neurodegenerative conditions. Moreover, analyses integrating multimodal approaches (eg, diffusion tensor imaging and cerebrospinal fluid–based markers) may achieve even greater accuracy.
APPENDIX
SUPPLEMENTAL FIGURES
Supported by the European Research Council, Swedish Research Council, Knut and Alice Wallenberg Foundation, Marianne and Marcus Wallenberg Foundation, Strategic Research Area MultiPark (Multidisciplinary Research in Parkinson’s Disease) at Lund University, Swedish Brain Foundation, Parkinson Foundation of Sweden, Parkinson Research Foundation, Skåne University Hospital Foundation, Swedish federal government under the ALF agreement, National Institutes of Health (AG043503, AG017586), and Dana Foundation. The Wellcome Centre for Human Neuroimaging is supported by core funding from Wellcome Trust (203147/Z/16/Z).
Disclosures of Conflicts of Interest: N.S. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: has grants/grants pending with Swedish Alzheimer Association; received payment for travel/accommodations/meeting expenses unrelated to activities listed from Schörling Foundation. Other relationships: disclosed no relevant relationships. S.H. disclosed no relevant relationships. D.J.I. Activities related to the present article: institution receives research grants from BrightFocus Foundation, National Institutes of Health (NIH), and Penn Institute of Aging; received payment for travel/accommodations/meeting expenses unrelated to activities listed from General Electric. Other relationships: disclosed no relevant relationships. T.R. disclosed no relevant relationships. J.A.C. disclosed no relevant relationships. A.F.D. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: is a consultant for Adamas Pharmaceuticals and Sunovion Pharmaceuticals; has grants/grants pending with Adamas Pharmaceuticals, Sunovion Pharmaceuticals, and Revance Therapeutics; receives royalties from UpToDate. Other relationships: disclosed no relevant relationships. M.A.S. disclosed no relevant relationships. E.B.L. Activities related to the present article: institution receives grant from NIH. Activities not related to the present article: is a consultant for National Disease Research Interchange; has grants/grants pending with Doris Duke Charitable Foundation and NIH; received payment for lectures including service on speakers bureaus from International Society for CNS Clinical Trials, Thomas Jefferson University, University of Pittsburg, and University of Tsukuba; received payment for travel/accommodations/meeting expenses unrelated to activities listed from Alzheimer's Association International Conference Scientific Planning Committee Meeting, American Association of Neuropathologist Education Committee Meeting, and American Board of Pathology Test Development. Other relationships: disclosed no relevant relationships. J.Q.T. disclosed no relevant relationships D.v.W. disclosed no relevant relationships. M.N. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: had research grant with Random Walk Imaging; has patents (planned, pending, or issued) with and holds stock/stock options in Random Walk Imaging. Other relationships: disclosed no relevant relationships. M.G. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: is board member of the Association for Frontotemporal Degeneration; is a consultant for Biogen; has grants/grants pending with NIH. Other relationships: disclosed no relevant relationships. P.J.N. disclosed no relevant relationships. C.T.M. Activities related to the present article: institution receives grant from NIH. Activities not related to the present article: is a consultant for Axon Advisors. Other relationships: disclosed no relevant relationships. O.H. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: is a board member and consultant for Roche. Other relationships: disclosed no relevant relationships.
Abbreviations:
- DTI
- diffusion tensor imaging
- FA
- fractional anisotropy
- MMSE
- Mini-Mental State Examination
- PSP
- progressive supranuclear palsy
- SD
- standard deviation
References
- 1. Höglinger GU, Respondek G, Stamelou M, et al . Clinical diagnosis of progressive supranuclear palsy: The movement disorder society criteria . Mov Disord 2017. ; 32 ( 6 ): 853 – 864 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Hughes AJ, Daniel SE, Kilford L, Lees AJ. . Accuracy of clinical diagnosis of idiopathic Parkinson’s disease: a clinico-pathological study of 100 cases . J Neurol Neurosurg Psychiatry 1992. ; 55 ( 3 ): 181 – 184 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Holtbernd F, Eidelberg D. . The utility of neuroimaging in the differential diagnosis of parkinsonian syndromes . Semin Neurol 2014. ; 34 ( 2 ): 202 – 209 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Postuma RB, Berg D, Stern M, et al . MDS clinical diagnostic criteria for Parkinson’s disease . Mov Disord 2015. ; 30 ( 12 ): 1591 – 1601 . [DOI] [PubMed] [Google Scholar]
- 5. McKeith IG. . Consensus guidelines for the clinical and pathologic diagnosis of dementia with Lewy bodies (DLB): report of the Consortium on DLB International Workshop . J Alzheimers Dis 2006. ; 9 ( 3 Suppl ): 417 – 423 . [DOI] [PubMed] [Google Scholar]
- 6. Poewe W, Wenning G. . The differential diagnosis of Parkinson’s disease . Eur J Neurol 2002. ; 9 ( Suppl 3 ): 23 – 30 . [DOI] [PubMed] [Google Scholar]
- 7. Zhukareva V, Joyce S, Schuck T, et al . Unexpected abundance of pathological tau in progressive supranuclear palsy white matter . Ann Neurol 2006. ; 60 ( 3 ): 335 – 345 . [DOI] [PubMed] [Google Scholar]
- 8. Halliday GM, Holton JL, Revesz T, Dickson DW. . Neuropathology underlying clinical variability in patients with synucleinopathies . Acta Neuropathol (Berl) 2011. ; 122 ( 2 ): 187 – 204 . [DOI] [PubMed] [Google Scholar]
- 9. Surova Y, Nilsson M, Lätt J, et al . Disease-specific structural changes in thalamus and dentatorubrothalamic tract in progressive supranuclear palsy . Neuroradiology 2015. ; 57 ( 11 ): 1079 – 1091 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Agosta F, Kostić VS, Galantucci S, et al . The in vivo distribution of brain tissue loss in Richardson’s syndrome and PSP-parkinsonism: a VBM-DARTEL study . Eur J Neurosci 2010. ; 32 ( 4 ): 640 – 647 . [DOI] [PubMed] [Google Scholar]
- 11. Sajjadi SA, Acosta-Cabronero J, Patterson K, Diaz-de-Grenu LZ, Williams GB, Nestor PJ. . Diffusion tensor magnetic resonance imaging for single subject diagnosis in neurodegenerative diseases . Brain 2013. ; 136 ( Pt 7 ): 2253 – 2261 . [DOI] [PubMed] [Google Scholar]
- 12. Cochrane CJ, Ebmeier KP. . Diffusion tensor imaging in parkinsonian syndromes: a systematic review and meta-analysis . Neurology 2013. ; 80 ( 9 ): 857 – 864 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Whitwell JL, Höglinger GU, Antonini A, et al . Radiological biomarkers for diagnosis in PSP: Where are we and where do we need to be? Mov Disord 2017. ; 32 ( 7 ): 955 – 971 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Talai AS, Sedlacik J, Boelmans K, Forkert ND. . Widespread diffusion changes differentiate Parkinson’s disease and progressive supranuclear palsy . Neuroimage Clin 2018. ; 20 : 1037 – 1043 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Watson R, Blamire AM, Colloby SJ, et al . Characterizing dementia with Lewy bodies by means of diffusion tensor imaging . Neurology 2012. ; 79 ( 9 ): 906 – 914 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Surova Y, Lampinen B, Nilsson M, et al . Alterations of diffusion kurtosis and neurite density measures in deep grey matter and white matter in Parkinson’s disease . PLoS One 2016. ; 11 ( 6 ): e0157755 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Emre M, Aarsland D, Brown R, et al . Clinical diagnostic criteria for dementia associated with Parkinson’s disease . Mov Disord 2007. ; 22 ( 12 ): 1689 – 1707 ; quiz 1837 . [DOI] [PubMed] [Google Scholar]
- 18. Litvan I, Agid Y, Calne D, et al . Clinical research criteria for the diagnosis of progressive supranuclear palsy (Steele-Richardson-Olszewski syndrome): report of the NINDS-SPSP international workshop . Neurology 1996. ; 47 ( 1 ): 1 – 9 . [DOI] [PubMed] [Google Scholar]
- 19. Gelb DJ, Oliver E, Gilman S. . Diagnostic criteria for Parkinson disease . Arch Neurol 1999. ; 56 ( 1 ): 33 – 39 . [DOI] [PubMed] [Google Scholar]
- 20. Nilsson M, Szczepankiewicz F, van Westen D, Hansson O. . Extrapolation-based references improve motion and eddy-current correction of high B-value DWI data: Application in Parkinson’s disease dementia . PLoS One 2015. ; 10 ( 11 ): e0141825 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Zhang H, Yushkevich P, Rueckert D, Gee J. . Unbiased white matter atlas construction using diffusion tensor images . Med Image Comput Comput Assist Interv 2007. ; 10 ( Pt 2 ): 211 – 218 . [DOI] [PubMed] [Google Scholar]
- 22. Zhang H, Yushkevich PA, Alexander DC, Gee JC. . Deformable registration of diffusion tensor MR images with explicit orientation optimization . Med Image Anal 2006. ; 10 ( 5 ): 764 – 785 . [DOI] [PubMed] [Google Scholar]
- 23. Bach M, Laun FB, Leemans A, et al . Methodological considerations on tract-based spatial statistics (TBSS) . Neuroimage 2014. ; 100 : 358 – 369 . [DOI] [PubMed] [Google Scholar]
- 24. Mori S, Oishi K, Jiang H, et al . Stereotaxic white matter atlas based on diffusion tensor imaging in an ICBM template . Neuroimage 2008. ; 40 ( 2 ): 570 – 582 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Toledo JB, Van Deerlin VM, Lee EB, et al . A platform for discovery: The University of Pennsylvania Integrated Neurodegenerative Disease Biobank . Alzheimers Dement 2014. ; 10 ( 4 ): 477 – 484.e1 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Armstrong MJ. . Progressive Supranuclear Palsy: an Update . Curr Neurol Neurosci Rep 2018. ; 18 ( 3 ): 12 . [DOI] [PubMed] [Google Scholar]
- 27. Westin CF, Knutsson H, Pasternak O, et al . Q-space trajectory imaging for multidimensional diffusion MRI of the human brain . Neuroimage 2016. ; 135 : 345 – 362 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Eriksson S, Lasic S, Topgaard D. . Isotropic diffusion weighting in PGSE NMR by magic-angle spinning of the q-vector . J Magn Reson 2013. ; 226 : 13 – 18 . [DOI] [PubMed] [Google Scholar]
- 29. Shemesh N, Özarslan E, Komlosh ME, Basser PJ, Cohen Y. . From single-pulsed field gradient to double-pulsed field gradient MR: gleaning new microstructural information and developing new forms of contrast in MRI . NMR Biomed 2010. ; 23 ( 7 ): 757 – 780 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Lampinen B, Szczepankiewicz F, Novén M, et al . Searching for the neurite density with diffusion MRI: challenges for biophysical modeling . Hum Brain Mapp . 2019. ; 40 ( 8 ); 2529 - 2545 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Hansson O, Janelidze S, Hall S, et al . Blood-based NfL: A biomarker for differential diagnosis of parkinsonian disorder . Neurology 2017. ; 88 ( 10 ): 930 – 937 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Oba H, Yagishita A, Terada H, et al . New and reliable MRI diagnosis for progressive supranuclear palsy . Neurology 2005. ; 64 ( 12 ): 2050 – 2055 . [DOI] [PubMed] [Google Scholar]
- 33. Quattrone A, Nicoletti G, Messina D, et al . MR imaging index for differentiation of progressive supranuclear palsy from Parkinson disease and the Parkinson variant of multiple system atrophy . Radiology 2008. ; 246 ( 1 ): 214 – 221 . [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.