Pattern of structural and functional brain abnormalities in asymptomatic granulin mutation carriers

Alzheimer’s & Dementia - (2014) 1–10

Research Article

Pattern of structural and functional brain abnormalities in asymptomatic granulin mutation carriers Michela Pievania, Donata Paternic oa, Luisa Benussib, Giuliano Binettib, Alberto Orlandinic, Milena Cobellic, Silvia Magnaldic, Roberta Ghidonid, Giovanni B. Frisonia,e,* a

Laboratory of Epidemiology, Neuroimaging and Telemedicine, IRCCS Istituto Centro San Giovanni di Dio, Fatebenefratelli, Brescia, Italy b NeuroBioGen Lab—Memory Clinic, IRCCS Istituto Centro San Giovanni di Dio, Fatebenefratelli, Brescia, Italy c Neuroradiology and Angiology Unit, Fondazione Poliambulanza, Brescia, Italy d Proteomics Unit, IRCCS Istituto Centro San Giovanni di Dio, Fatebenefratelli, Brescia, Italy e Memory Clinic and LANVIE - Laboratory of Neuroimaging of Aging, University Hospitals and University of Geneva, Geneva, Switzerland

Abstract

Background: To investigate the patterns of brain atrophy, white matter (WM) tract changes, and functional connectivity (FC) abnormalities in asymptomatic granulin (GRN) mutation carriers. Methods: Ten cognitively normal subjects (five mutation carriers, GRN1; years to estimated disease onset: 12 6 7; five mutation noncarriers, GRN2) underwent a clinical and imaging (structural, diffusion tensor, and resting-state functional magnetic resonance imaging) assessment. Brain atrophy was measured with cortical thickness analysis, WM abnormalities with tract-based spatial statistics, and FC with independent component analysis. Results: GRN1 showed smaller cortical thickness than GRN2 in the right orbitofrontal and precentral gyrus and left rostral middle frontal gyrus. WM tracts abnormalities were limited to increased axial diffusivity in the right cingulum, superior longitudinal fasciculus, and corticospinal tract. There were no differences in FC of resting-state networks. Conclusion: Brain atrophy and WM tract abnormalities in frontal-parietal circuits can be detected at least a decade before the estimated symptom onset in asymptomatic mutation carriers. Ó 2014 The Alzheimer’s Association. All rights reserved.

Keywords:

FTLD; Progranulin; Magnetic resonance imaging; Cortical thickness; Diffusion tensor; Resting-state functional MRI

1. Introduction Frontotemporal lobar degeneration (FTLD) denotes a large group of neurodegenerative conditions characterized by frontal and temporal symptoms and atrophy. FTLD patients have a strong familial component with between 20% and 50% of all cases reporting a familial history for the disease. Three genetic mutations have been identified to date as major causes for the disease: mutations in the gene encoding the granulin protein (GRN), the microtubule associated tau protein (MAPT), and, recently, the chromosome 9 open *Corresponding author. Tel.: 139-030-3501361; Fax: 139-0303501592. E-mail address: [email protected]

reading frame 72 (C9orf72) [1]. Prevalence estimates vary according to geographic differences [1]. In northern Italy, GRN mutations are among the most common genetic causes of FTLD, whereas MAPT mutations are relatively less frequent [2–5]. All GRN mutations identified thus far cause disease through a uniform disease mechanism: the loss of functional progranulin or haploinsufficiency [6,7]. Progranulin and granulins are secreted growth factors involved in multiple biological functions, including neuronal development and synaptic maintenance. The loss of progranulin in patients carrying pathogenic GRN null mutations is thus thought to increase susceptibility to neuronal death [7]. Because FTLD mutations are inherited in an autosomal dominant manner and in GRN null mutation carriers the shortage of progranulin precedes clinical

1552-5260/$ - see front matter Ó 2014 The Alzheimer’s Association. All rights reserved. http://dx.doi.org/10.1016/j.jalz.2013.09.009

2

M. Pievani et al. / Alzheimer’s & Dementia - (2014) 1–10

symptoms [8–11], asymptomatic members of families with GRN mutations offer a unique possibility to investigate early pathologic changes and identify disease biomarkers. Recent evidence from familial Alzheimer’s disease has shown that the pathologic changes start approximately 25 years before the onset of clinical symptoms, and brain atrophy occurs approximately 15 years before clinical onset [12]. In familial FTLD, a similar pattern of early pathology is likely to occur; however, evidence to date is scarce and mostly limited to symptomatic cases. In FTLD patients carrying a GRN mutation, the main pathologic signatures on magnetic resonance imaging are an involvement of the fronto-temporo-parietal circuits, with prominent parietal and asymmetric atrophy [13–15], impaired connectivity in long-distance intrahemispheric tracts [13], and salience network disruption [16]. The involvement of the parietal regions seems to be unique to GRN mutations compared with mutations in MAPT or C9orf72 [17,18]. Which brain abnormalities occur first (atrophy, structural, or functional disconnection) and how many years before the clinical symptoms these changes occur remains to be confirmed. One previous study assessed atrophy and functional connectivity changes in asymptomatic GRN mutation carriers, reporting increased salience network connectivity with no brain atrophy [16]. In another study by the same group, early impairment of long-distance association tracts was reported [19], consistently with the white matter (WM) abnormalities observed in full-blown FTLD [20]. Although promising, these findings have been replicated by others only in part [21]. Moreover, only one study has assessed brain atrophy, functional disconnection, and WM abnormalities together in the same patients [21]. The aim of this study was to comprehensively investigate the pattern of brain atrophy, WM tract abnormalities, and functional connectivity abnormalities in asymptomatic GRN muta-

tion carriers. We hypothesized that structural and functional abnormalities might be detected in asymptomatic subjects along the circuits typically affected in symptomatic FTLD.

2. Methods 2.1. Subjects Subjects were recruited at the Istituto di Ricovero e Cura a Carattere Scientifico Centro S. Giovanni di Dio Fatebenefratelli (www.irccs-fatebenefratelli.it) from five unrelated GRN-positive pedigrees identified in northern Italy. Four families carried the GRN p.Leu271LeufsX10 mutation [22], and one carried the GRN p.Thr278SerfsX7 mutation [23]. The phenotypes of these families were behavioral-variant frontotemporal dementia (bv FTD; n 5 2), primary progressive aphasia (PPA, n 5 1), and FTD with motor neuron disease (n 5 1) for the four pedigrees with the GRN p.Leu271LeufsX10 mutation, and PPA for the family with the p.Thr278SerfsX7 mutation. Age of onset in these pedigrees was on average 61 6 4 years, ranging from 49 to 66 years. Subjects were included if they (1) were screened for the presence or absence of GRN mutations, (2) were cognitively normal, and (3) underwent magnetic resonance imaging (MRI). Ten unaffected family members fulfilled these criteria: five screened positive for a GRN mutation (GRN1, 4 with the GRN p.Leu271LeufsX10 mutation and 1 with the GRN p.Thr278SerfsX7 mutation), and the remaining five screened negative for GRN mutations and served as a control group (GRN2). Among GRN1 subjects, two p.Leu271LeufsX10 mutation carriers were from the same family (PPA phenotype), and the remaining three were from three unrelated pedigrees (1 FTD with motor neuron disease, 1 PPA, 1 bv FTD). Plasma progranulin levels were 23 6 10 ng/mL in GRN1 and 94 6 32 ng/mL in GRN2 (P , .05; Table 1).

Table 1 Demographic and cognitive features of asymptomatic progranulin mutation carriers (GRN1) and control nonmutation carriers (GRN2)

Progranulin levels (ng/mL)* Age (y) Gender (women) Education (y) Time to estimated onset (y) MMSE Rey-Osterrieth Figure Copy Rey-Osterrieth Figure Copy Recall Boston Naming Test Category verbal fluency Letter verbal fluency Token Test Rey’s Word List Immediate Recall Rey’s Word List Delayed Recall Trail Making Test Part A Trail Making Test Part B

Whole sample

GRN2 (n 5 5)

GRN1 (n 5 5)

P

Pnp

50 6 41 [11–127] 46 6 12 [28–63] 8 10 6 3 [5–13] — 29 6 1 [26–30] 31 6 6 [22–36] 15 6 6 [10–25] 28.5 6 1 [27–30] 45 6 8 [35–60] 34 6 9 [22–48] 34 6 2 [29–35] 57 6 9 [40–70] 12 6 2 [8–15] 50 6 26 [30–101] 116 6 49 [62–221]

94 6 32 46 6 15 5 964 — 29 6 1 33 6 6 16 6 5 28.5 6 2 41 6 3 31 6 6 33.5 6 3 51 6 9 12 6 3 48 6 23 116 6 63

23 6 10 45 6 10 3 11 6 3 12 6 7 29 6 2 30 6 5 14 6 7 28.5 6 1 49 6 11 38 6 12 34 6 1 62 6 7 13 6 2 53 6 33 115 6 33

.003 .88 .44 .46 — .68 .47 .71 1.00 .27 .37 .61 .15 .56 .83 .98

.03 .84 .44 .54 — 1.00 .11 .73 1.00 .29 .41 .89 .23 .63 .91 .56

NOTE. Numbers denote mean 6 SD [range], or frequency. P denotes significance on two-tailed Student’s t test for continuous variables or c2 test for dichotomous variables. Pnp, significance on two-tailed Mann-Whitney U test. MMSE, Mini-Mental State Examination. *Plasma progranulin levels were available in all mutation carriers and in three noncarriers.

M. Pievani et al. / Alzheimer’s & Dementia - (2014) 1–10

Age of onset in the affected parent of GRN1 subjects was on average 58 6 4 (range 53–63 years). Mutation carriers were on average at 12 6 7 years from the affected parent’s age of disease onset (range 218 to 21 years; Table 1). The sociodemographic and clinical features of GRN1 and GRN2 subjects are summarized in Table 1. 2.2. Neuropsychological assessment The cognitive assessment included neuropsychological tests for global cognition (Mini-Mental State Examination) [24]; memory (Rey Auditory Verbal Learning Test and Rey-Osterrieth Complex Figure Recall) [25,26]; attentionexecutive (Trail Making Test, Parts A and B) [27]; language (letter and category fluency; Token Test; Boston Naming Test) [28,29]; visuo-constructional abilities (Rey-Osterrieth Complex Figure Copy) [26]. Anxiety and depression were assessed with the State-Trait Anxiety Inventory Form Y1 and Y2 (STAI-Y1 and Y2) scale [30] and the Beck Depression Inventory (BDI) [31]. Exclusion criteria were cognitive or functional impairment, a diagnosis of dementia, a history of depression or psychosis of juvenile onset, a history or neurologic signs of major stroke, alcohol abuse, craniocerebral trauma, and heavy use of psychotropic drugs. The study protocol was approved by the local ethics committee (Comitato Etico delle Istituzioni Ospedaliere Cattoliche, CEIOC, Brescia, Italy; 34/2003, 8/ 2003, 1/2007). All participants signed an informed participation consent before inclusion in the study. 2.3. MRI acquisition Structural, diffusion tensor, and functional MRI scans were acquired on a 1.5-T GE Signa HDx system (General Electric Healthcare, Waukesha, WI, USA) at the Neuroradiology and Angiology Unit, Fondazione Poliambulanza, Brescia, Italy. The following sequences were obtained from all the subjects: (1) inversion recovery spoiled gradient echo for cortical thickness measurement (repetition time [TR] 5 11.6 milliseconds, echo time [TE] 5 5 milliseconds, inversion time 5 600 milliseconds, flip angle 5 8 , slice thickness 5 1 mm, matrix size 5 256 ! 256, field of view [FOV] 5 256 ! 256 mm); (2) gradient echo echo-planar imaging sequence for resting-state functional MRI (fMRI) analysis (TR 5 3000 milliseconds, TE 5 30 milliseconds, flip angle 5 90 , matrix size 5 64 ! 64, FOV 5 220 ! 220 mm, 40 axial slices, slice thickness 5 3 mm, number of volumes 5 200). Subjects were instructed to keep their eyes closed during resting-state scanning, not to think of anything in particular, and not to fall asleep; (3) spin echo echo-planar imaging for diffusion tensor imaging (DTI) analysis (TR 5 7760 milliseconds, TE 5 100 milliseconds, slice thickness 5 2 mm with no gap, matrix size 5 128 ! 128, FOV 5 224 ! 224 mm; acceleration factor 5 2), with five images acquired with no diffusionencoding gradients applied (b factor 5 0 s/mm2), and

3

30 diffusion-encoding gradients applied in noncollinear directions (b 5 700 s/mm2). 2.4. MRI analysis 2.4.1. Cortical thickness analysis T1-weighted MRI were analyzed with FreeSurfer version 5.0 [32] using the standard processing pipeline. The preprocessing of T1-weighted MRI included intensity variations correction, intensity normalization, affine registration to the Talairach atlas [33], skull stripping, and gray matter (GM) and WM segmentation [34,35]. The boundaries between WM and cortical GM, as well as between GM and cerebrospinal fluid (CSF), were identified, and the left and right hemispheres were separated. A WM surface was then generated for each hemisphere by tiling the outside of the WM mass for that hemisphere. This initial surface was then refined to follow the intensity gradients between the WM and GM. The WM surface was then nudged to follow the intensity gradients between the GM and CSF, obtaining the pial surface. Cortical thickness measurements were obtained by calculating the distance between those surfaces (WM and pial surface) at each of approximately 160,000 points per hemisphere across the cortical mantle [32]. To relate and compare anatomic features across subjects, the transformations mapping each brain to a common reference brain were computed. The pial surface of each subject was inflated to determine the large-scale folding patterns of the cortex and subsequently transformed into a sphere to minimize metric distortion. Each individual folding pattern was then aligned to an average folding pattern using a highresolution surface-based averaging technique. Thickness measures of each participant were mapped to the aligned and inflated surface, allowing visualization of data on a common cortical surface for all the subjects. Finally, cortical thickness was smoothed with a 10-mm full-width halfmaximum Gaussian kernel to reduce local variations in the measurements for further analysis. 2.4.2. DTI analysis Diffusion tensor MRI data were analyzed with the tractbased spatial statistics (TBSS, version 1.2; part of FSL; http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/TBSS) [36]. Each fractional anisotropy (FA) image was aligned to a target image using the following procedure: (1) a target image was selected automatically as the most representative FA image using the FMRIB Non-linear Image Registration Tool (FNIRT, part of FSL), (2) the nonlinear transformation that mapped each subject’s FA to the target image was computed using FNIRT, (3) the target image was affine-registered to the Montreal Neurological Institute 152 standard space using FLIRT (part of FSL), and (4) the same transformation was used to align each subject’s FA to the standard space. A mean FA image was then created by averaging the aligned individual FA images and thinned to create a FA skeleton representing the WM

4

M. Pievani et al. / Alzheimer’s & Dementia - (2014) 1–10

tracts common to all subjects [36]. The FA skeleton was thresholded at 0.5 to exclude voxels with low FA values, which likely include GM or CSF. Individual FA, mean diffusivity (MD), axial diffusivity (AxD), and radial diffusivity (RaD) data were projected onto this common skeleton. 2.4.3. Resting-state fMRI analysis Preprocessing of fMRI data was carried out using the Statistical Parametric Mapping (SPM5) software package (http://www.fil.ion.ucl.ac.uk/spm) running on Matlab 7.0.1 (MathWorks, Natick, MA, USA). After discarding the first five volumes of each time series for magnetic field stabilization, functional images were corrected for small head movements by realignment to the first scan of the series with a 6 rigid-body transformation. None of the participants showed a head motion .1.5 mm maximum displacement in any direction of x, y, and z, or .1.5 of rotation in any plane. Images were then spatially normalized to the SPM default echo-planar imaging template and smoothed using an isotropic Gaussian filter (6-mm full-width half-maximum). The group independent component analysis for fMRI toolbox (GIFT v1.3g; http://icatb.sourceforge.net) was then used to identify spatially independent and temporally coherent networks [37]. First, each subject’s fMRI data were reduced to a lower dimensionality by using principal component analysis. fMRI data were then concatenated and reduced to one group. The independent group components were estimated using the Infomax approach [38], and each component was represented by a spatial map and a temporal profile. The resulting maps were used to compute the individual subject components (back reconstruction). The number of independent group components was 27, a dimension determined using the minimum description length criteria [37]. The estimated spatial maps were then converted into Z scores and moved into SPM to identify the networks of interest and to exclude components related to artifacts. Visual inspection of spatial patterns and frequency spectra of the independent components allowed the removal of those clearly related to artifacts and noise and the identification of the networks of interest—according to previous literature, the default mode [16] and the salience [16] networks. Moreover, because GRN mutation carriers show significant frontal-parietal involvement [17], we assessed two additional networks: the bilateral executive control networks made up of lateral frontal-parietal nodes [39]. 2.5. Statistical analysis 2.5.1. Sociodemographic and cognitive features Differences between groups in sociodemographic and cognitive features were assessed with Student’s t test for continuous variables and paired c2 test for dichotomous variables (two-tailed). Because of the small sample size, a nonparametric test (Mann-Whitney U test, two-tailed) was used to confirm results on the t test. The estimated years to

disease onset were computed for each subject as the difference between the subjects’ age at the time of the study and the age of symptom onset in the affected parent. Parent’s age at onset was preferred to other possible alternatives (e.g., the mean age at onset in the family or the mean mutation age at onset) for consistency with previous studies [12] and to avoid the need of an arbitrary cutoff. 2.5.2. Whole brain analysis Cortical thickness differences between GRN1 and GRN2 groups were assessed using a vertex-by-vertex analysis and a two-tailed t test in FreeSurfer. A global thickness index for each hemisphere was computed by averaging thickness values from all surface points in the respective hemisphere. An asymmetry index was then computed as the right/left ratio on hemispheric cortical thickness and compared between groups. WM voxelwise changes on DTI were assessed using a permutation based inference tool for nonparametric statistical thresholding (“randomise” program, part of FSL). Between-group comparisons of MD, FA, AxD and RaD values within the skeleton were tested by using a two-sample t test (two-tailed). The number of permutations was set at 5000. The anatomic location of significant clusters was detected by using the Johns Hopkins University WM tractography atlas and the International Consortium of Brain Mapping DTI WM labels atlas (http://fsl.fmrib.ox.ac.uk/ fsl/fslwiki/Atlases). Resting-state fMRI differences were assessed by entering the GIFT spatial maps into a voxelwise two-sample t test in SPM5. Each contrast was restricted to the corresponding resting-state network as follows. First, resting-state spatial maps were extracted using a one sample t test (two-tailed, P , .01 corrected for false discovery rate, cluster threshold k 5 100 voxels). Then the binarized restingstate network masks were extracted from each spatial map using the MarsBaR toolbox [40] and used to constrain the analysis to the voxels within the mask. For all the comparisons (cortical thickness, resting-state networks, DTI changes), significance was set to P , .005 uncorrected for multiple comparisons, and cluster size was set to .50 vertices (for cortical thickness) or voxels (for DTI and resting-state analysis). 2.5.3. Correlation analysis The association between MRI abnormalities (cortical thinning, WM tract changes, resting-state fMRI abnormalities) and the years to the estimated symptoms onset age was assessed with the Pearson’s correlation coefficient (P , .05, one-tailed to test whether MRI abnormalities increased in proximity of the expected age at onset). Because the variable “estimated years to disease onset” is strongly associated with subjects’ age (in our sample, r 5 0.92, P , .001; Pearson’s correlation), significant correlations could also reflect age effects. To control for whether the associations were influenced by age, the same correlations were also assessed in mutation noncarriers. The detection of a significant correlation in carriers but not in noncarriers would then assure that the associations are explained by the disease and not by age. Correlations between MRI abnormalities, cognitive deficits, and

M. Pievani et al. / Alzheimer’s & Dementia - (2014) 1–10

progranulin levels were also assessed (P , .05, one-tailed to test whether MRI abnormalities were associated with lower cognitive scores or progranulin levels). Significant correlations on Pearson’s test were confirmed with a nonparametric test (Spearman’s coefficient, one-tailed). For this analysis, a region-of-interest approach was used, as described below. The mean cortical thickness was computed in regions defined by the Anatomical region of interest FreeSurfer tool. The mean resting-state connectivity within the networks of interest was computed by averaging the Z scores across all voxels of a given network, using the binarized network masks extracted with MarsBaR as the spatial constraints. The mean WM tract measures (MD, FA, AxD, RaD) were measured by averaging the values across all voxels of a given tract. WM tracts definition was derived from the JHU and ICBM WM atlases. Finally, to reduce the number of cognitive variables, composite scores were computed for each cognitive domain (memory, attention-executive, language, and visuospatial) by transforming the raw neuropsychological scores into Z scores (i.e., computing the difference between each subject score and the average score, divided by the standard deviation of the score) and by averaging the Z scores from multiple tests within a cognitive domain. 3. Results 3.1. Cognitive features There were no differences between GRN1 and GRN2 in age, gender, education, and cognition (Table 1). Depression and anxiety scores were similar in the two groups (BDI: 17 6 14 vs. 14 6 11 for GRN1 and GRN2, P 5 .67; STAIY1: 41 6 8 vs. 39 6 14, P 5 .84; STAI-Y2: 43 6 8 vs. 45 6 15, P 5 .84). 3.2. Patterns of brain atrophy and connectivity abnormalities Cortical thinning. GRN1 showed smaller cortical thickness than GRN2 in the right lateral orbitofrontal cortex (P , .001; percentage reduction: 20%; Fig. 1), in the left middle frontal gyrus (P , .001 in two clusters; mean percentage reduction: 16%; Fig. 1), and in the right precentral gyrus (P , .005; percentage reduction: 16%; Fig. 1). The analysis of cortical asymmetry revealed a rightward asymmetry in two GRN1 subjects (Supplementary Fig. 1). The asymmetry index did not differ significantly between GRN1 and GRN2 (0.99 and 1.01, respectively, P 5 .18). White matter changes. AxD was increased in GRN1 compared with GRN2 in the right superior longitudinal fasciculus (percentage AxD increase: 20%; Fig. 2), in the right cingulum (12% increase; Fig. 2), and in the right corticospinal tract (19% increase; Fig. 2). There were no differences in MD, FA, and RaD values (P . .005). Functional connectivity changes. There were no differences between asymptomatic GRN mutation carriers and noncarriers in functional connectivity within the major

5

resting-state networks (P . .005, data not shown). Because thickness and WM analyses showed abnormalities in the motor circuit (precentral gyrus and corticospinal tract), we analyzed retrospectively functional connectivity of the sensorimotor network [39]. This post hoc analysis revealed no differences between GRN1 and GRN2 (P . .005). 3.3. Correlation analysis Thickness of the orbitofrontal and middle frontal cortex in mutation carriers decreased as a function of years to the estimated symptoms onset (right orbitofrontal gyrus: r 5 2.88, P 5 .03; left middle frontal gyrus: r 5 2.91, P 5 .02; Fig. 3). Mutation carriers also showed a positive correlation between AxD values and the years to the estimated symptoms onset in the right superior longitudinal fasciculus (r 5 .85, P 5 .04; Fig. 3) and in the right corticospinal tract (r 5 .84, P 5 .04; Fig. 3). No significant correlation was detected between the years to the estimated symptoms onset and (1) the right precentral gyrus thickness and (2) AxD in the right cingulum (P . .05). As expected, these was no age-dependent change in MRI biomarkers in mutation noncarriers (P . .09; Fig. 3). Spearman’s correlation confirmed significance values for the orbitofrontal cortex thickness (P 5 .03) and AxD in the superior longitudinal fasciculus (P 5 .04) but not for the middle frontal cortex (P 5 .10) and the right corticospinal tract (P 5 .25). Correlations between MRI changes and cognition and progranulin levels were not significant (P ,.05) except for a positive association between visuospatial scores and thinning of the orbitofrontal cortex (r 5 .98, P 5.01, Pearson) and middle frontal cortex (r 5 .90, P 5 .05, Pearson), which were not confirmed by Spearman’s test (P  .10). No association was detected between orbitofrontal cortex thinning and depressive or anxiety symptoms in GRN1 (BDI: r 5 2.61, P 5 .27; STAI-Y1: r 5 2.11, P 5 .86; STAI-Y2: r 5 2.30, P 5 .62). 4. Discussion The main findings of our study are as follows: (1) brain atrophy and structural connectivity changes were detected in asymptomatic GRN mutation carriers at least a decade before the estimated age of symptoms onset and (2) these abnormalities targeted frontal-parietal-temporal circuits. The combined analysis of MRI, DTI, and fMRI data showed that frontal cortex thinning and abnormal connectivity in long-association tracts were detected on average 12 years before the estimated disease onset, whereas functional connectivity markers were normal. These results thus suggest that structural abnormalities might represent early markers of pathology, whereas functional connectivity changes might occur later. These findings are in line with recent evidence from familial AD showing that brain atrophy occurs decades before estimated disease onset in mutation carriers [12,41] and show that in FTLD pathologic changes might start approximately 10 to 15 years before the clinical syndrome.

6

M. Pievani et al. / Alzheimer’s & Dementia - (2014) 1–10

Fig. 1. Maps of the differences in cortical thickness between asymptomatic subjects carrying a GRN mutation (GRN1) compared with nonmutation carriers (GRN2). Maps are shown at P , .05 (in blue) for illustrative purpose. Peaks of significant differences (P , .005 uncorrected, cluster size K . 50) are shown in circles. Coordinates denote the peak voxels of each cluster in standard space. Mean values denote the average cortical thickness values (SD) in the cluster for each group. % diff denotes the percentage difference between GRN1 and GRN2. K, cluster size (number of contiguous significant vertices); L, left; R, right.

A possible clinical implication of these results is that current interventions are administered too late in the disease course to be effective because structural and functional damage is typically widespread and severe in patients [16,21,39].

Conversely, targeting the disease at a stage when the abnormalities are restricted to specific neuronal populations might prove more effective in slowing or even halting the disease progression. Moreover, the identification of

Fig. 2. Maps of the differences in WM tracts connectivity between asymptomatic subjects carrying a GRN mutation (GRN1) compared with non–mutation carriers (GRN2). Maps are shown at P , .05 (in blue) for illustrative purpose. Peaks of significant differences (P , .005 uncorrected, cluster size K . 50) are shown in circles. Coordinates denote the peak voxels of each cluster in standard space. Mean values denote the average AxD values in the cluster for each group. % diff denotes the percentage difference between GRN1 and GRN2. K, cluster size (number of contiguous significant vertices); L, left; R, right.

M. Pievani et al. / Alzheimer’s & Dementia - (2014) 1–10

7

Fig. 3. Scatter plots of the correlations between the years to the estimated symptoms onset and (1) cortical thickness changes (upper row) and (2) WM tract changes (lower row) in GRN mutation carriers (filled circles and solid lines) and noncarriers (empty circles and dotted lines). Values denote Pearson’s correlation coefficient (r) and significance (P, one-tailed). *Significance confirmed by Mann-Whitney U test (P , .05, one-tailed). L, left; R, right. SLF, superior longitudinal fasciculus.

8

M. Pievani et al. / Alzheimer’s & Dementia - (2014) 1–10

biomarkers sensitive to disease progression is important for their use in clinical trials. In our study, orbitofrontal and middle frontal cortex thinning were progressively abnormal as subjects approached the estimated age of disease onset; a similar pattern was observed at trend level for axial diffusivity in the superior longitudinal fasciculus and corticospinal tract. Conversely, no association was detected in mutation noncarriers. These correlations suggest that these biomarkers might be useful to track disease progression along the trajectory from asymptomatic to symptomatic stages. Whether these abnormalities reflect neurodegenerative or neurodevelopmental features, however, remains an open question, because we do not know whether structural changes were present during childhood or adolescence. The lack of an association between MRI changes and progranulin levels might reflect the floor effect in progranulin levels because this marker is pathologic in all mutation carries, whereas MRI markers become progressively abnormal with disease course. The pattern of MRI and DTI abnormalities observed in GRN mutation carriers is consistent with the view that these mutations target an asymmetric fronto-temporo-parietal network, starting from the orbitofrontal cortex and then spreading to the parietal and temporal cortices [17,42]. Our data agree with this view in several ways. First, brain atrophy in asymptomatic individuals mapped to the orbitofrontal and middle frontal cortex, two regions affected by neuropathology in the very mild and mild stages, respectively [43]. Second, WM tract abnormalities were detected along associative pathways connecting the orbitofrontal cortex to the parietal and temporal cortices. Indeed, the cingulum is a major tract connecting the orbitofrontal cortex with the posterior cingulate cortex and traveling along the midline [44]. The superior longitudinal fasciculus is a major tract connecting the dorsolateral prefrontal cortex to the dorsal parietal cortex and entering into the temporal lobe [44]. The involvement of these tracts is in line with previous evidence from both symptomatic and asymptomatic subjects showing long association tracts damage [13,19]. Third, we detected structural abnormalities predominantly in the right hemisphere, in line with the hypothesis of an asymmetric pattern. In this study, we also observed cortical thinning in the precentral gyrus and WM abnormalities in the corticospinal tract. These regions are part of the motor circuits and are commonly involved in motor disorders. Although GRN mutations cause disease through a common mechanism, the clinical presentations associated with these mutations are heterogeneous, ranging from cognitive to movement disorders [7]. In our cohort, we previously showed that the GRN p.Leu271LeufsX10 mutation is a major disease determinant for FTLD disorders with a motor component: the mutation was found in 31% of corticobasal syndrome, 29% of frontotemporal dementia with motor neuron disease, and 15% of behavioral-variant FTD [3]. The present finding of structural abnormalities in the motor circuit is therefore consistent with the prevalence estimates. However, because we did not

systematically assess motor symptoms in our sample, we can only speculate about the relationships between motor and MRI abnormalities. The longitudinal assessment of these subjects will help to clarify whether precentral cortex and corticospinal tract abnormalities predate the onset of motor symptoms in these families. Finally, although a motor phenotype is frequently associated with GRN mutations in Italian pedigrees, this feature is less common in other geographic and ethnic groups [45,46]. Thus, caution should be used in generalizing our motor findings to other populations. In the WM, we detected AxD increases but no change in FA, MD, and RaD values. These findings are in line with the view that AxD might be a potential early marker of pathology in the WM. Indeed, we and others have previously shown that AxD increases are among the most prominent changes associated with pathology in early Alzheimer’s disease [47,48]. These markers might therefore be more appropriate to detect small changes than traditional markers such as FA, which reflects axonal loss and might therefore occur in later (symptomatic) disease stages. The possible mechanisms underlying AxD increases are not yet clear [49], however, and future studies are required to clarify the pathologic substrate of these changes. This study is not without limitations. First, the sample size was small, and the imaging analyses did not survive multiple comparisons correction. These findings should therefore be interpreted with caution, and future studies with larger samples are required to confirm these results. Because the number of subjects available for biomarkers study is generally limited by the rare frequency of GRN mutations (w5%–10% of all FTLD cases), and this limitation is further increased for longitudinal studies, large-scale initiatives such as the Genetic Frontotemporal Dementia Initiative (http://www.coen.org/ projects.html) will likely enable future work to address these issues. Second, cases were not matched by gender, and each imaging modality used different statistical approaches, thus limiting cross-modality comparisons. Third, age-of-onset can vary considerably even within FTLD pedigrees, and hence results based on the estimated years to symptom onset should be interpreted with caution. Finally, behavioral data were limited to anxiety and depression symptoms, and scales specifically assessing behavioral symptoms (e.g., the NeuroPsychiatric Inventory, the Frontal Behavioral Inventory) were not assessed. Strengths of this study include the collection of structural MRI, DTI, and resting-state fMRI data in the same cohort and the comprehensive assessment of biomarkers of pathology (MD, FA, AxD, and RaD indexes on DTI, functional connectivity abnormalities in the major resting-state networks, and whole brain cortical thinning). In conclusion, our findings indicate that brain atrophy and structural connectivity changes might occur at least a decade before the estimated symptom onset in asymptomatic GRN mutation carriers. The identification of early pathologic changes is of paramount importance for the design of future treatments. Interventions in the early stages of the disease

M. Pievani et al. / Alzheimer’s & Dementia - (2014) 1–10

indeed hold the greatest promise for prevention of irreversible brain tissue destruction. GRN is a particularly appealing gene for drug targeting because boosting its expression may be beneficial for mutation carriers in preventing or delaying the onset of the disease. Acknowledgments This work was supported by grants 2009-2633 from the Fondazione CARIPLO; 2012-ID13 from the Associazione Fatebenefratelli per la Ricerca (AFaR); Ricerca Corrente, Italian Ministry of Health; and Monzino Foundation. Dr. G.B. Frisoni serves on the editorial boards of Lancet Neurology, Aging Clinical & Experimental Research, Alzheimer’s Disease & Associated Disorders, and Neurodegenerative Diseases and is a Section Editor for Neurobiology of Aging. He serves or has served on the advisory boards for Lilly, BMS, Bayer, Lundbeck, Elan, Astra Zeneca, Pfizer, Taurx, and Wyeth. He has received grants from Wyeth, Lilly, Lundbeck Italia, and the Alzheimer’s Association. The other authors have indicated they have nothing to disclose. RESEARCH IN CONTEXT

1. Systematic review: We searched PubMed for articles published in English between January 1990 and December 2012 with the search terms: “granulin,” “progranulin,” “imaging,” “atrophy,” and “connectivity.” Previous imaging studies have shown that granulin (GRN) mutations target fronto-parieto-temporal circuits, but how early, and in which order, these changes occur has been poorly investigated. 2. Interpretation: We detected brain atrophy and structural connectivity changes in asymptomatic GRN mutation carriers at least a decade before the estimated age of symptom onset. These brain abnormalities were consistent with the pattern involved in patients and might therefore represent early disease biomarkers. These results suggest that current interventions might be administered too late (i.e., in symptomatic stages) to be effective. 3. Future directions: The identification of early biomarkers will aid the design of future treatments. Interventions in the early disease stages could be more effective in preventing or delaying the disease onset (e.g., by increasing GRN expression).

References [1] Sieben A, Van Langenhove T, Engelborghs S, Martin JJ, Boon P, Cras P, et al. The genetics and neuropathology of frontotemporal lobar degeneration. Acta Neuropathol 2012;124:353–72.

9

[2] Signorini S, Ghidoni R, Barbiero L, Benussi L, Binetti G. Prevalence of pathogenic mutations in an Italian clinical series of patients with familial dementia. Curr Alzheimer Res 2004;1:215–8. [3] Benussi L, Ghidoni R, Pegoiani E, Moretti DV, Zanetti O, Binetti G. Progranulin Leu271LeufsX10 is one of the most common FTLD and CBS associated mutations worldwide. Neurobiol Dis 2009;33:379–85. [4] Tremolizzo L, Gelosa G, Galbussera A, Isella V, Arosio C, Bertola F, et al. Higher than expected progranulin mutation rate in a case series of Italian FTLD patients. Alzheimer Dis Assoc Disord 2009;23:301. [5] Benussi L, Ghidoni R, Binetti G. Progranulin mutations are a common cause of FTLD in northern Italy. Alzheimer Dis Assoc Disord 2010; 24:308–9. [6] Rademakers R, Rovelet-Lecrux A. Recent insights into the molecular genetics of dementia. Trends Neurosci 2009;32:451–61. [7] Ghidoni R, Paterlini A, Albertini V, Binetti G, Benussi L. Losing protein in the brain: the case of progranulin. Brain Res 2012; 1476:172–82. [8] Ghidoni R, Benussi L, Glionna M, Franzoni M, Binetti G. Low plasma progranulin levels predict progranulin mutations in frontotemporal lobar degeneration. Neurology 2008;71:1235–9. [9] Ghidoni R, Stoppani E, Rossi G, Piccoli E, Albertini V, Paterlini A, et al. Optimal plasma progranulin cutoff value for predicting null progranulin mutations in neurodegenerative diseases: a multicenter Italian study. Neurodegener Dis 2012;9:121–7. [10] Finch N, Baker M, Crook R, Swanson K, Kuntz K, Surtees R, et al. Plasma progranulin levels predict progranulin mutation status in frontotemporal dementia patients and asymptomatic family members. Brain 2009;132:583–91. [11] Sleegers K, Brouwers N, Van Damme P, Engelborghs S, Gijselinck I, van der Zee J, et al. Serum biomarker for progranulin-associated frontotemporal lobar degeneration. Ann Neurol 2009;65:603–9. [12] Bateman RJ, Xiong C, Benzinger TL, Fagan AM, Goate A, Fox NC, et al. Dominantly Inherited Alzheimer Network. Clinical and biomarker changes in dominantly inherited Alzheimer’s disease. N Engl J Med 2012;367:795–804. [13] Rohrer JD, Ridgway GR, Modat M, Ourselin S, Mead S, Fox NC, et al. Distinct profiles of brain atrophy in frontotemporal lobar degeneration caused by progranulin and tau mutations. Neuroimage 2010; 53:1070–6. [14] Bozzali M, Battistoni V, Premi E, Alberici A, Giulietti G, Archetti S, et al. Structural brain signature of FTLD driven by granulin mutation. J Alzheimers Dis 2013;33:483–94. [15] Whitwell JL, Jack CR Jr, Boeve BF, Senjem ML, Baker M, Rademakers R, et al. Voxel-based morphometry patterns of atrophy in FTLD with mutations in MAPT or PGRN. Neurology 2009; 72:813–20. [16] Borroni B, Alberici A, Cercignani M, Premi E, Serra L, Cerini C, et al. Granulin mutation drives brain damage and reorganization from preclinical to symptomatic FTLD. Neurobiol Aging 2012;33:2506–20. [17] Rohrer JD, Warren JD. Phenotypic signatures of genetic frontotemporal dementia. Curr Opin Neurol 2011;24:542–9. [18] Whitwell JL, Weigand SD, Boeve BF, Senjem ML, Gunter JL, DeJesus-Hernandez M, et al. Neuroimaging signatures of frontotemporal dementia genetics: C9ORF72, tau, progranulin and sporadics. Brain 2012;135:794–806. [19] Borroni B, Alberici A, Premi E, Archetti S, Garibotto V, Agosti C, et al. Brain magnetic resonance imaging structural changes in a pedigree of asymptomatic progranulin mutation carriers. Rejuvenation Res 2008;11:585–95. [20] Zhang Y, Schuff N, Du AT, Rosen HJ, Kramer JH, Gorno-Tempini ML, et al. White matter damage in frontotemporal dementia and Alzheimer’s disease measured by diffusion MRI. Brain 2009; 132:2579–92. [21] Dopper EG, Rombouts SA, Jiskoot LC, Heijer Td, de Graaf JR, Koning Id, et al. Structural and functional brain connectivity in presymptomatic familial frontotemporal dementia. Neurology 2013; 80:814–23.

10

M. Pievani et al. / Alzheimer’s & Dementia - (2014) 1–10

[22] Benussi L, Rademakers R, Rutherford NJ, Wojtas A, Glionna M, Paterlini A, et al. Estimating the age of the most common Italian GRN mutation: walking back to Canossa times. J Alzheimers Dis 2013;33:69–76. [23] Rossi G, Piccoli E, Benussi L, Caso F, Redaelli V, Magnani G, et al. A novel progranulin mutation causing frontotemporal lobar degeneration with heterogeneous phenotypic expression. J Alzheimers Dis 2011; 23:7–12. [24] Folstein MF, Folstein SE, McHugh PR. “Mini-mental state.” A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189–98. [25] Spinnler H, Tognoni G. Standardizzazione e taratura italiana di test neuropsicologici. Ital J Neurol Sci 1987;6:1–120. [26] Caffarra P, Vezzadini G, Dieci F, Zonato F, Venneri A. Rey-Osterrieth complex figure: normative values in an Italian population sample. Neurol Sci 2002;22:443–7. [27] Giovagnoli AR, Del Pesce M, Mascheroni S, Simoncelli M, Laiacona M, Capitani E. Trail making test: normative values from 287 normal adult controls. Ital J Neurol Sci 1996;17:305–9. [28] Novelli G, Papagno C, Capitani E, Laiacona N, Vallar G, Cappa SF. Tre test clinici di ricerca e produzione lessicale. Taratura su soggetti normali. Arch Psicol Neurol Psichiatr 1986;47:477–506. [29] Kaplan E, Goodglass H, Weintraub S. The Boston Naming Test. Philadelphia: Lea & Febiger; 1983. [30] Spielberger CD, Gorsuch RL, Lushene RE, Vagg PR, Jacobs GA. Manual for the State-Trait Anxiety Inventory. Palo Alto, CA: Consulting Psychologists Press; 1983 [Italian edition: Pedrabissi L, Santinello M. State-Trait Anxiety Inventory-Y Form (in Italian). Florence: Organizzazioni Speciali; 1989]. [31] Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuring depression. Arch Gen Psychiatry 1961;4:561–71. [32] Fischl B, Dale AM. Measuring the thickness of the human cerebral cortex from magnetic resonance images. Proc Natl Acad Sci U S A 2000;97:11050–5. [33] Talairach J, Tournoux P. Co-planar Stereotaxic Atlas of the Human Brain: 3-Dimensional Proportional System—an Approach to Cerebral Imaging. New York: Thieme Medical Publishers; 1988. [34] Fischl B, Sereno MI, Dale A. Cortical surface-based analysis. II: inflation, flattening, and a surface-based coordinate system. Neuroimage 1999;9:195–207. [35] Dale AM, Fischl B, Sereno MI. Cortical surface-based analysis. I. Segmentation and surface reconstruction. Neuroimage 1999;9:179–94.

[36] Smith SM, Jenkinson M, Johansen-Berg H, Rueckert D, Nichols TE, Mackay CE, et al. Tract-based spatial statistics: voxelwise analysis of multi-subject diffusion data. Neuroimage 2006;31:1487–505. [37] Calhoun VD, Adali T, Pearlson GD, Pekar JJ. Spatial and temporal independent component analysis of functional MRI data containing a pair of task-related waveforms. Hum Brain Mapp 2001;13:43–53. [38] Bell AJ, Sejnowski TJ. An information-maximization approach to blind separation and blind deconvolution. Neural Comput 1995;7:1129–59. [39] Pievani M, de Haan W, Wu T, Seeley WW, Frisoni GB. Functional network disruption in the degenerative dementias. Lancet Neurol 2011;10:829–43. [40] Brett M, Anton JL, Valabregue R, Poline JB. Region of interest analysis using an SPM toolbox. 8th International Conference on Functional Mapping of the Human Brain, Sendai, Japan 2002. Available on CD-ROM in Neuroimage 16(2), Abstract 497. [41] Ridha BH, Barnes J, Bartlett JW, Godbolt A, Pepple T, Rossor MN, et al. Tracking atrophy progression in familial Alzheimer’s disease: a serial MRI study. Lancet Neurol 2006;5:828–34. [42] Warren JD, Rohrer JD, Hardy J. Disintegrating brain networks: from syndromes to molecular nexopathies. Neuron 2012;73:1060–2. [43] Broe M, Hodges JR, Schofield E, Shepherd CE, Kril JJ, Halliday GM. Staging disease severity in pathologically confirmed cases of frontotemporal dementia. Neurology 2003;60:1005–11. [44] Catani M, Thiebaut de Schotten M. A diffusion tensor imaging tractography atlas for virtual in vivo dissections. Cortex 2008;44:1105–32. [45] Chen-Plotkin AS, Martinez-Lage M, Sleiman PM, Hu W, Greene R, Wood EM, et al. Genetic and clinical features of progranulin-associated frontotemporal lobar degeneration. Arch Neurol 2011;68:488–97. [46] Le Ber I, Camuzat A, Hannequin D, Pasquier F, Guedj E, RoveletLecrux A, et al. French research network on FTD/FTD-MND. Phenotype variability in progranulin mutation carriers: a clinical, neuropsychological, imaging and genetic study. Brain 2008;131:732–46. [47] Pievani M, Agosta F, Pagani E, Canu E, Sala S, Absinta M, et al. Assessment of white matter tract damage in mild cognitive impairment and Alzheimer’s disease. Hum Brain Mapp 2010;31:1862–75. [48] Acosta-Cabronero J, Williams GB, Pengas G, Nestor PJ. Absolute diffusivities define the landscape of white matter degeneration in Alzheimer’s disease. Brain 2010;133:529–39. [49] Zhang Y, Du AT, Hayasaka S, Jahng GH, Hlavin J, Zhan W, et al. Patterns of age-related water diffusion changes in human brain by concordance and discordance analysis. Neurobiol Aging 2010; 31:1991–2001.

M. Pievani et al. / Alzheimer’s & Dementia - (2014) 1–10

Supplementary Fig. 1. Asymmetry index (right/left ratio on hemispheric cortical thickness) in GRN1 and GRN2 subjects. Light gray lines denote the range of normality derived from the GRN2 group.

10.e1