FDG-PET patterns associated with underlying pathology in corticobasal syndrome

Participants

Twenty-nine patients with (1) a diagnosis of CBS, (2) available FDG-PET brain imaging, and (3) a postmortem neuropathologic diagnosis of CBD, AD, or PSP were included in the study. Patients were derived from a cohort of 98 patients with CBS who were enrolled at the National Institute of Neurological Disorders and Stroke (NINDS) from 2001 to 2009; 36 of those had a pathologic postmortem diagnosis, as previously described from the clinical point of view by our group.^10,11 The 69 patients with CBS not included did not fulfill the criteria above. Patients were recruited from across the United States via advertisement and had to have an established diagnosis of CBS before coming to our attention. Clinical documentation was reviewed and the diagnosis of CBS was confirmed by a behavioral neurologist with experience in movement disorders (E.M.W.) and a neuropsychologist (J.G.) via consensus following a 1-week evaluation. Initial motor impairments in all of the patients with CBS consisted of an asymmetric progressive rigid syndrome resistant to therapeutic doses of levodopa evaluated in the months preceding the referral and confirmed at the clinical examination at the NINDS, associated with evidence of cortical deficits (as described by Boeve et al.¹²; the frequency of each cortical deficit for this cohort has been previously reported by our group¹⁰). All patients underwent detailed clinical and neuropsychological examination and structural brain imaging (MRI in all but 4 patients who underwent CT scans). Presence of significant vascular disease at structural imaging represented an exclusion criterion for the study; available images were reviewed also at the time of the statistical analysis to confirm the lack of significant vascular pathology. All imaging studies were carried out at NINDS. Neuropathologic studies were carried out at the Indiana University School of Medicine by dementia neuropathology experts (B.G. and S.S.). For all patients, laterality of the motor symptoms was available on file. Motor impairment was assessed with finger tapping test. Overall cognitive impairment was assessed using the Mattis Dementia Rating Scale–2 (MDRS-2). The Test of Oral and Limb Apraxia (TOLA)¹³ and the Wechsler Adult Intelligence Scale III (WAIS-III) vocabulary score were used to assess praxis and language deficits. No quantitative sensory test was available. Patients' demographics are described in table 1.

Neuropathologic evaluation

Brains were cut along the midsagittal plane. The left hemibrain was fixed in 10% buffered formalin, whereas the right hemibrain was frozen for genetic and biochemical studies. Sections were obtained from the following regions: superior and middle frontal gyri, cingulate gyrus, superior and middle temporal gyri, amygdala, superior parietal lobule, calcarine cortex, basal ganglia at the level of the anterior commissure, thalamus at the level of the subthalamic nucleus, anterior and posterior hippocampus, cerebellar cortex, dentate nucleus, midbrain, pons, and medulla.¹⁴ Tissue was processed for hematoxylin & eosin, thioflavin S, and tau (AT8), β-amyloid, α-synuclein, and pTDP43 immunohistochemistry according to protocols published previously.¹⁴ Neuropathologic diagnoses were made using published criteria.^15–21 Neuropathologic diagnoses at the single-subject level are described in table 2. The primary pathologic diagnosis was used to divide the patients with CBS into the following subgroups: CBS-CBD, CBS-AD, and CBS-PSP.

Healthy controls (HCs)

Thirteen HC participants matched on demographics were selected to undergo the same procedures at the NIH Clinical Center (table 1). All participants were evaluated by a board-certified neurologist and found neurologically normal. None of the HC participants presented with any cognitive or motor complaint and all presented with an unremarkable medical history.

Standard protocol approvals, registrations, and patient consents

All patients gave written assent for the study. The study and consent procedure were approved by the NINDS institutional review board and by the Indiana University institutional review board. Sample and data storage and analysis including de-identification were compliant with Health Insurance Portability and Accountability Act (HIPAA) rules and the Declaration of Helsinki. HCs were compensated for their participation in the study.

FDG-PET acquisition and analysis

Participants fasted after midnight on the day of their scan and had no caffeine, alcohol, or nicotine for 24 hours before the scan. During the 25 minutes after FDG injection, patients rested with their eyes covered and ears plugged and were instructed to remain awake. The patients underwent FDG-PET scans on a GE (Cleveland, OH) Advance scanner (4.25-mm slice separation, 35 slices, axial field of view 15.3 cm, transverse field of view 55.0 cm). Forty-five minutes after injection, a 2-minute positioning scan was performed, and then a 10-minute emission scan immediately followed by an 8-minute transmission scan (used to correct the emission data for attenuation) were obtained. Acquisition was performed in 3D mode. Images were reconstructed using an ordered subset-expectation maximization algorithm. Images were then exported as DICOM files and transformed to analyze files for subsequent postprocessing.

To account for the usually asymmetric nature of CBS, scans were flipped in order to represent the hemisphere contralateral to the most affected limbs on the right side of the image.²² FDG-PET images were subjected to affine and nonlinear spatial normalization into Talairach and Tournoux space using SPM12 (Wellcome Department of Cognitive Neurology, London, UK). All the default choices of statistical parametric mapping (SPM) were followed with the exception of spatial normalization. For this study, the ¹⁵H₂O SPM-default template was replaced by a dementia-optimized brain FDG-PET template as described by Della Rosa et al.²³ The spatially normalized set of images was then smoothed with a 10-mm isotropic Gaussian filter to blur individual variations in gyral anatomy and to increase the signal-to-noise ratio. In addition, a conjunction analysis was used to identify common clusters of hypometabolism among neuropathologic subgroups. This approach tests for the conjunction of different hypotheses (each specified by a contrast) by assessing the significance of the contrasts combined and then eliminating voxels that evidenced significant differences (i.e., interactions) among the different contrasts,²⁴ thus allowing, in our case, the highlighting of areas that are hypometabolic compared to controls irrespective of the underlying pathology. A threshold of p < 0.05 (voxel-level) family-wise error (FWE)–corrected for multiple comparisons and a minimum cluster size of 100 was applied to all analyses except for the comparison between patients with CBS-PSP and controls where, given the low number of patients in the CBS-PSP group, a threshold of uncorrected p < 0.0001 was used, and for the correlations with clinical scores, where a threshold of uncorrected p < 0.001 was used. Age at the time of the PET scan and sex were used as nuisance variables.

Finally, to visually describe the pattern of hypometabolism at the single-subject level and thus to aid in the interpretation of the group analyses, we selected the following regions based on published studies^25–27: FDG-PET patterns and neuropathologic data for AD, PSP, and CBD, i.e., anterior and posterior cingulate (both as separate values and as an anterior–posterior asymmetry index), precuneus, striatum (limited to the neostriatum, i.e., putamen and caudate nuclei), and midbrain. Regions of interest (ROI) were defined using the Automated Anatomical Labeling atlas and to account for asymmetry, we selected the lower value between the left and right hemisphere mask for each ROI. All data are calculated after whole brain normalization and reported as z scores based on HC data. Given the focus on group analyses and the descriptive nature of this work, ROI data are reported only using median and range of observed values.

Data availability statement

Anonymized data will be made available to qualified investigators for collaborative research in accordance with HIPPA guidelines.

Clinical features and pathologic examinations

The primary pathologic diagnoses of enrolled patients after the neuropathologic examination were as follows: CBD (14 patients), AD (10 patients), and PSP (5 patients). Groups were matched on education and time between FDG-PET and death (overall effect education: p = 0.55, time before death: p = 0.28) while CBS-AD patients were younger (p = 0.002). Clinical data are reported in table 1. There was no difference between the CBS groups in motor impairment as assessed with finger tapping (dominant hand p = 0.91; nondominant hand p = 0.25) and all patients presented with parkinsonism. There were no differences between groups in overall cognitive performance as assessed with the MDRS-2 (total score: p = 0.20), in apraxia as assessed with the TOLA (p = 0.84), or in language ability as assessed by WAIS-III Vocabulary subtest scores (p = 0.45).

Pattern of hypometabolism in the whole CBS cohort compared to controls

Patients with CBS presented with markedly asymmetric frontoparietal hypometabolism, centered in the precentral and postcentral gyri (table 3 and figure 1). In the hemisphere contralateral to the more affected limbs, a large cluster of hypometabolism was found including the dorsolateral prefrontal gyri, the precentral and postcentral gyri, the insula, and more posteriorly the superior parietal lobule as well as a second cluster centered in the thalamus. In the hemisphere ipsilateral to the more affected limbs, a much smaller cluster was identified, with a hypometabolic focus in the precentral gyrus, extending to the superior and middle frontal gyri.

Pattern of hypometabolism correlations with clinical variables in the whole CBS cohort

In the whole CBS group, MDRS-2 performance directly correlated with metabolic levels in a frontal cluster including the superior and middle frontal gyri and the precentral gyrus located in the hemisphere contralateral to the more affected limbs (table 4 and figure 2A).

Finger tapping performance directly correlated with metabolic levels in bilateral fontal pole, anterior cingulate, basal ganglia, and orbitofrontal gyrus. Unilateral correlations were found in the insula and the ventral diencephalon, respectively, in the hemisphere contralateral and ipsilateral to the more affected limbs (table 4 and figure 2B).

Apraxia was directly associated with frontal hypometabolism in the superior and middle frontal gyri and in the supplementary motor cortex cluster, all in the hemisphere contralateral to the more affected limbs (table 4 and figure 2C).

Language deficits directly correlated with anterior insula and opercular metabolic levels in the hemisphere contralateral to the more affected limbs (table 4 and figure 2D).

Pattern of hypometabolism in patients with CBS-CBD compared to controls

Patients with CBS-CBD presented with markedly asymmetric frontoparietal hypometabolism, centered in the precentral and postcentral gyri, and in the deep gray matter (table 3 and figure 3A). In the hemisphere contralateral to the more affected limbs, a large cluster of hypometabolism was found spanning from the superior, middle, and inferior frontal gyri, including the precentral and postcentral gyri, to the superior parietal lobule and the supramarginal gyrus, as well as a second hypometabolic cluster that included the thalamus and the caudate nucleus. In the hemisphere ipsilateral to the most affected limb, a much smaller cluster centered in the precentral gyrus and extending in the superior frontal gyrus was identified, together with caudate hypometabolism. There was no material change in results after removing CBS-CBD cases with a secondary pathologic diagnosis of coexisting AD or Lewy body disease (table 2). There was no significant difference in FDG-PET metabolism between CBS-CBD patients with and without concomitant AD pathology even at the lenient threshold of p(uncorrected) < 0.005.

Pattern of hypometabolism in patients with CBS-AD compared to controls

Patients with CBS-AD presented with an asymmetric posterior hypometabolism, without deep gray matter involvement (table 3 and figure 3B). In the hemisphere contralateral to the most affected limbs, a large cluster of hypometabolism was found, including the inferior temporal gyrus and the angular gyrus, which extended to the posterior cingulate, the middle temporal gyri, and the superior parietal lobule. In the same hemisphere, a second, smaller cluster was also found, in the precentral gyrus. In the hemisphere ipsilateral to the most affected limbs, a smaller cluster was identified in the supramarginal gyrus, with a hypometabolic focus in the supramarginal gyrus and a smaller cluster in the superior parietal lobule. Of note, no cluster of hypometabolism was found in either the thalami or the basal ganglia in both hemispheres. There was no material change in results after taking out those CBS-AD cases with a secondary pathologic diagnosis of coexisting Lewy body disease (table 2). There was no significant difference in FDG-PET metabolism between patients with CBS-AD with and without concomitant Lewy body disease even at the lenient threshold of p(uncorrected) < 0.005.

Pattern of hypometabolism in patients with CBS-PSP compared to controls

Patients with CBS-PSP presented with markedly asymmetric anterior hypometabolism (table 3 and figure 3C). In the hemisphere contralateral to the most affected limbs, the precentral gyrus, the middle and anterior cingulate gyri, the middle frontal gyri, and the caudate nucleus were hypometabolic. There were no foci of hypometabolism in the other hemisphere or in the midbrain. There was no significant difference in FDG-PET metabolism between patients with CBS-PSP with and without concomitant AD pathology even at the lenient threshold of p(uncorrected) < 0.005.

Common areas of hypometabolism among the CBS-CBD, CBS-AD, and CBS-PSP groups

The conjunction analysis disclosed a cluster of hypometabolism focused in the precentral gyrus, extending in the superior frontal gyrus of the hemisphere contralateral to the most affected limbs, the only common area of hypometabolism among the CBS-CBD, CBS-AD, and CBS-PSP subgroups (table 3 and figure 3D). Visual inspection confirmed the presence of a reduction in FDG tracer retention in the precentral/postcentral gyrus region in all patients with CBS except 2 (included in the CBS-PSP and the CBS-AD groups).

Patterns of hypermetabolism in the whole CBS cohort and the CBS-CBD, CBS-AD, and CBS-PSP groups compared to controls

There were no areas of relative hypermetabolism in the whole CBS cohort and the CBS-CBD, CBS-AD, and CBS-PSP groups compared to controls at the following thresholds: p(FWE) < 0.05, p(uncorrected) < 0.005.

Regional hypometabolism differences at the single-subject level: Exploratory evidence

To further interpret the group results, median z scores and the range of observed values are reported in table 5. There was a full separation of the observed values between CBS-AD and CBS-PSP in the following ROIs: anterior cingulate, precuneus, and the striatum, as well as on the anterior–posterior cingulate gradient. Moreover, there was a full separation between CBS-AD and CBS-CBD in the basal ganglia ROI and between CBS-PSP and CBS-CBD patients on the anterior–posterior cingulate asymmetry index.