Abstract
We studied topographic distribution of tau and amyloid-β in a patient with variant Alzheimer’s disease with spastic paraparesis (VarAD) by comparing AD patients. The proband developed progressive memory impairment, dysarthria, and spastic paraparesis at age 23. Heterozygous missense mutation (L166P) was found in exon 6 of presenilin-1 gene. The proband showed prominently increased amyloid binding in striatum and cerebellum and asymmetrical tau binding in the primary sensorimotor cortex contralateral to the side more affected by spasticity. We suspect that upper motor neuron dysfunctions may be attributed to excessive abnormal tau accumulation rather than amyloid-β in the primary motor cortex.
INTRODUCTION
Presenilin-1 (PSEN1) gene mutation is the most common cause of familial Alzheimer’s disease (AD). At least 200 pathogenic mutations in PSEN1 gene have been found in more than 430 families with various clinical phenotypes [1, 2]. Mutations in PSEN1 gene may cause variant AD with spastic paraparesis (VarAD) that is characterized by early-onset dementia, spastic paraparesis, psychiatric symptoms, clumsiness, and dysarthria [2 –4].
Neuropathological studies of VarAD showed characteristic amyloid-β positive “cotton wool” plaques, which are devoid of congophilic cores and dystrophic neurites, in entire cerebral cortices and subcortical nuclei. Also, cored amyloid-β plaques are found in the cerebellar cortex [5, 6]. Cortical neurons may show hyperphosphorylated tau pathology corresponding to Braak’s stage V-VI [5]. In VarAD, a positron emission tomography (PET) study with 11C-Pittsburgh Compound B (PIB) showed highly increased binding in the striatum, cingulate, occipital, frontal, and parietal cortices [6]. In contrast to well-known distribution of the amyloid-β pathology, little is known about the topographic distribution of tau pathology.
Recent development of tau selective radiotracers enabled in vivo visualization of tau pathology [7]. We presented a very young age-onset VarAD patient associated with a rare pathogenic known mutation in PSEN1 gene. Using 18F-florbetaben (for amyloid) and 18F-AV-1451 (for tau) PET, we studied differences in topographic distribution of amyloid-β and tau between the VarAD patient and AD patients.
MATERIALS AND METHODS
Clinical characteristics and laboratory findings in the VarAD patient
A 29-year-old male patient visited our movement disorders clinic for memory dysfunction and gait disturbance. He was born normally at full term and had normal developmental milestone. He developed progressive memory disturbance at the age of 23. Three years later, he developed weakness in his left leg and within several months in the right leg. He developed progressive dysarthria at the age of 27. At his age 29, his speech was almost unintelligible, and he could barely walk without cane.
He was the first child of healthy parents. His younger brother and sister were neurologically normal. On neurological examination, he showed markedly slow tongue movement and severe dysarthria. His eye movements were normal with no gaze-evoked nystagmus. There was fine postural finger tremor when he stretched out his arms in front of him. Hand rolling and finger tapping were clumsy on both sides and worse on the left side. Limb ataxia was absent. Both legs were markedly spastic, but there was mild weakness only in the left proximal leg. Sensory functions were normal. He walked slowly with very short stride. He showed staggering when he was turning. He showed pathologically increased deep tendon reflexes in all four limbs with bilateral Hoffmann’s sign and ankle clonus. Babinski’s sign was positive on the left side.
Nerve conduction, electromyography, and sensory evoked potential studies were normal. Magnetic resonance image (MRI) study of whole spinal cord was normal, but brain MRI showed mild diffuse cerebral cortical atrophy.
Korean version of Mini-Mental State Examination (MMSE) score was 24, which was less than 0.01 percentile of the norm. Standardized neuropsychological test battery (Seoul Neuropsychological Screening Battery) showed severe dysfunction (less than 1 percentile of the norm) in all cognitive domains (verbal and visual memory, language, visuospatial, and executive functions) except attention (18 percentile of norm) [8, 9].
He had a heterozygous missense mutation (73653577 T>C, p.L166P) in exon 6 of PSEN1 gene, but other family members had no mutation in the PSEN1 gene.
PET studies
The proband underwent PET imaging studies for amyloid with 18F-florbetaben and tau with 18F-AV-1451. We also performed same PET studies in healthy controls and AD patients, and finally included 10 consecutive amyloid-negative controls (mean age 68.8±8.2 years, M: F = 4:6) and 10 consecutive amyloid-positive AD patients (mean age 71.9±8.6 years, M: F = 3: 7) for comparison (see Supplementary Table 1 for demographic characteristics).
Using Biograph mCT PET/CT scanner (Siemens Medical Solutions; Malvern, PA, USA), the 18F-florbetaben and 18F-AV1451 PET images were acquired for 20 min at 90 min after the injection of 18F-florbetaben and at 80 min after the injection of 18F-AV1451. Finally, 3D-PET images were reconstructed in 256× 256× 223 matrix with 1.591×1.591×1 mm voxel size.
In all subjects, high resolution axial T1-weighted MR images were obtained with 3D-spoiled gradient-recalled sequences (3D-SPGR sequences; repetition time = 8.3 ms, echo time = 3.3 ms, flip angle = 12°, 512×512 matrix, voxel spacing 0.43× 0.43× 1 mm) in 3.0 Tesla MR scanner (Discovery MR750, GE Medical Systems, Milwaukee, WI) for spatial normalization of PET images.
This study was approved by Institutional Review Board (IRB) of Gangnam Severance Hospital and written informed consent was obtained from all subjects.
Image processing steps
All image processing steps were done with statistical parametric mapping 8 (SPM8; Wellcome Trust Centre for Neuroimaging, London, UK) implemented in MATLAB 7.1 (MathWorks, Natick, MA). Using diffeomorphic anatomical registration using exponentiated lie algebra (DARTEL) toolbox in SPM8, MR images were corrected for inhomogeneity and spatially normalized to in-house MR template for DARTEL [10]. All PET images were coregistered to MR images and spatially normalized with the flow field normalizing gray and white matter segments of MR images. For measuring regional standardized uptake values (SUV) of PET images, we used in-house volume-of-interest (VOI) template containing 15 cortical and 4 subcortical regions.
Regional SUVR values of amyloid PET were calculated by dividing the SUV values of basis points [6], while cerebellar cortex was used as reference tissue for the regional SUVR values of tau PET. Likewise, SUVR images were created by dividing the SUV values of each reference tissue. For AD patients, mean SUVR values of each region were calculated, and averaged SUVR images were created. By using mean and standard deviation of regional SUVR values of 10 elderly healthy controls, regional Z-scores of SUVR values were calculated. Voxel-wise Z-score maps were created by using the mean and standard deviation images of controls.
RESULTS
With the cut-off Z-score 2, the proband showed increased amyloid binding in the striatum, thalamus, cerebellar cortex, and entire cerebral cortices except hippocampus and entorhinal cortex. Striatum, cerebellar cortex, occipital cortex, and precuneus were the most prominently involved regions. In tau PET study, binding values were increased in entire cerebral cortices except hippocampus. Occipital cortex was most prominently involved, and the involvement of primary sensorimotor cortex was also prominent (Table 1).
Unlike the proband, AD patients showed increased amyloid binding in all subcotical and cortical regions except hippocampus. Increase of striatal, thalamic, and cerebellar SUVR values were not as prominent as those of the proband. Primary sensorimotor cortex was more intensely involved than the proband. Moreover, the cortical binding pattern in tau PET was different. In contrast to the greatest binding in the inferior temporal cortex, primary sensorimotor cortex showed only mildly increased binding (Table 1).
Similar binding pattern was observed in voxel-based studies. In contrast to AD, the proband showed highly increased amyloid binding in the striatum and cerebellar cortex, and tau binding was highly increased in primary sensorimotor cortex. Interestingly, tau binding in the primary sensorimotor cortex was asymmetrically increased in the right side, which was contralateral to the clinically more affected side (Fig. 1).
DISCUSSION
In a patient who developed very young age-onset dementia and spastic paraparesis associated with L166P mutation in PSEN1 gene, we found amyloid-β accumulation in the striatum and cerebellar cortex and tau accumulation in the primary motor cortex.
There is wide phenotypic variation in patients with PSEN1 mutation, and about 29 PSEN1 mutations have been linked to the VarAD phenotype [2]. However, L166P mutation has been identified in only one VarAD patient, who developed generalized seizure at age 15 and subsequently memory impairment, ataxia, and spasticity over a period of 12 years very similar to our patient [11]. Other family members of our patient had no mutation in PSEN1. Although we did not check biological paternity, we suspect that the L166P mutation found in our patient might be pathogenic de novo mutation as several patients have been reported who developed very early-onset AD associated with de novo mutation in PSEN1 gene [12].
Loss of Betz cells in primary motor cortex and degeneration of corticospinal tract in the brainstem and spinal cord may be responsible for spastic paraparesis in VarAD patients [5, 13]. Because there was no visible amyloid-β and tau pathology in the corticospinal tract in the brainstem and spinal cord, spastic paraparesis was attributed to intracerebral amyloid-β and tau pathology [5].
Our VarAD patient showed asymmetric upper motor neuron signs more severely affecting the left side. He had symmetrically increased amyloid binding in the cerebral cortex, but the involvement of primary motor cortex was less prominent than in the AD. Unlike the AD patients, amyloid binding was prominently increased in the striatum and cerebellar cortex, which play role in motor control. Our patient also showed prominently increased tau binding in the primary motor cortex, which is final common pathway of motor output. Moreover, the binding was asymmetrically increased in the right primary motor cortex contralateral to the body side more severely affected by spasticity. Therefore, we suspect that upper motor neuron dysfunctions seem to be associated with excessive accumulation of abnormal tau protein in the primary motor cortex rather than amyloid-β accumulation.
