Impaired glymphatic system in genetic frontotemporal dementia: a GENFI study

Abstract The glymphatic system is an emerging target in neurodegenerative disorders. Here, we investigated the activity of the glymphatic system in genetic frontotemporal dementia with a diffusion-based technique called diffusion tensor image analysis along the perivascular space. We investigated 291 subjects with symptomatic or presymptomatic frontotemporal dementia (112 with chromosome 9 open reading frame 72 [C9orf72] expansion, 119 with granulin [GRN] mutations and 60 with microtubule-associated protein tau [MAPT] mutations) and 83 non-carriers (including 50 young and 33 old non-carriers). We computed the diffusion tensor image analysis along the perivascular space index by calculating diffusivities in the x-, y- and z-axes of the plane of the lateral ventricle body. Clinical stage and blood-based markers were considered. A subset of 180 participants underwent cognitive follow-ups for a total of 640 evaluations. The diffusion tensor image analysis along the perivascular space index was lower in symptomatic frontotemporal dementia (estimated marginal mean ± standard error, 1.21 ± 0.02) than in old non-carriers (1.29 ± 0.03, P = 0.009) and presymptomatic mutation carriers (1.30 ± 0.01, P < 0.001). In mutation carriers, lower diffusion tensor image analysis along the perivascular space was associated with worse disease severity (β = −1.16, P < 0.001), and a trend towards a significant association between lower diffusion tensor image analysis along the perivascular space and higher plasma neurofilament light chain was reported (β = −0.28, P = 0.063). Analysis of longitudinal data demonstrated that worsening of disease severity was faster in patients with low diffusion tensor image analysis along the perivascular space at baseline than in those with average (P = 0.009) or high (P = 0.006) diffusion tensor image analysis along the perivascular space index. Using a non-invasive imaging approach as a proxy for glymphatic system function, we demonstrated glymphatic system abnormalities in the symptomatic stages of genetic frontotemporal dementia. Such measures of the glymphatic system may elucidate pathophysiological processes in human frontotemporal dementia and facilitate early phase trials of genetic frontotemporal dementia.


Graphical Abstract Introduction
][3] Clinical phenotypes mainly encompass behavioural-variant (bv) frontotemporal dementia and primary progressive aphasia (PPA). 1,2The pathogenic mechanisms of frontotemporal dementia are determined by intracellular accumulation of aberrant proteins, including mainly tau [frontotemporal lobar degeneration-tau (FTLD-tau)] or TAR DNA-binding protein 43 (TDP-43, FTLD-TDP). 4,5Around 30% of frontotemporal dementia is familial, most commonly caused by autosomal dominant genetic mutations within microtubule-associated protein tau (MAPT), granulin (GRN) and chromosome 9 open reading frame 72 (C9orf72) genes. 6enetic frontotemporal dementia cases represent a privileged scenario to study the earliest phases of the disease in presymptomatic individuals and assess the changes in defined neuropathological subtypes, such as FTLD-tau, associated with MAPT mutations, or FTLD-TDP, due to GRN or C9orf72 genetic variations. 70][11][12][13][14] The glymphatic pathway promotes the flow of CSF into the brain along arterial perivascular spaces and subsequently into the brain interstitium through an astrocytic aquaporin-4-dependent mechanism.The pathway then directs the CSF flow towards the venous perivascular and perineuronal spaces, ultimately clearing solutes into meningeal and cervical lymphatic drainage vessels. 15,16spite the expanding concepts on glymphatic physiology and pathology, the in vivo assessment of the glymphatic function is still hampered by the need of non-invasive imaging techniques for its quantification. 17Recent studies have demonstrated the possibility of measuring glymphatic functions using MRI. 18In particular, the diffusion tensor image analysis along the perivascular space (DTI-ALPS) technique has been utilized to estimate the efficiency of the glymphatic system by the DTI-ALPS-index, as a measure of perivascular clearance activity in the human brain. 18This index showed its significant consistency with the classical detection clearance rate calculated on glymphatic MRI after intrathecal administration of gadolinium. 19][24] However, a few questions still need to be addressed.It is not yet known whether the glymphatic dysfunction (i) is an early event in frontotemporal dementia, already detectable in presymptomatic disease stages; (ii) is associated with a specific FTLD proteinopathy or may represent a common pathway related to neurodegeneration; (iii) is a marker of disease severity; (iv) correlates with markers of neuronal dysfunction or axonal damage, such as plasma neurofilament light chain (NfL) or glial fibrillary acidic protein (GFAP) 25,26 ; and/or (v) predicts disease progression over time.
These questions prompted the current cross-sectional and longitudinal study.Our specific aim was to evaluate glymphatic system abnormalities in presymptomatic and symptomatic individuals with pathogenic mutations within the MAPT, GRN and C9orf72 genes, to provide novel insight into the pathophysiology of genetic frontotemporal dementia.

Participants
From the GENFI cohort study, individuals carrying a pathogenic frontotemporal dementia variation and their noncarrier (NC) family members were recruited from research centres across Europe and Canada (www.genfi.org.uk).Inclusion and exclusion criteria have been previously described. 27Here, we considered a consecutive sample of 374 participants recruited from 30 January 2012 to 30 January 2021.Among them, 112 were C9orf72 expansion carriers (68 presymptomatic and 44 symptomatic), 119 were GRN mutation carriers (88 presymptomatic and 31 symptomatic), 60 were MAPT mutation carriers (43 presymptomatic and 17 symptomatic), and 83 were mutation NC individuals recruited among siblings.
All participants underwent the GENFI standardized assessment at enrollment. 7During the first visit, demographic information was collected.The years to expected onset were calculated as the difference between age at assessment and mean age at onset within the family, as previously described. 28We assessed disease stage using the Clinical Dementia Rating (CDR)® Dementia Staging Instrument plus National Alzheimer's Coordinating Centre (NACC) behaviour and language domains (CDR® plus NACC FTLD), 29 hereinafter referred to as CDR-FTLD.
Mutation carriers were divided into two disease stages based on their global CDR-FTLD score: presymptomatic (0 or 0.5) and fully symptomatic (1 or more).
Among the 291 mutation carriers, 282 participants underwent CDR-FTLD at baseline.A subgroup of mutation carriers also underwent CDR-FTLD at different follow-ups (180 with at least two evaluations, 105 with at least three, 55 with at least four, 15 with at least five and 3 with at least six), for a total of 640 evaluations.
Local ethics committees approved the study at each site, and all participants provided written informed consent; the study was conducted according to the Declaration of Helsinki.

MRI acquisition
MRI protocol was common to all the GENFI sites and adapted for different scanners (see Supplementary Table 1 for the scanner list); no pre-study phantom harmonization was performed at local level.Each subject underwent a 3 T MRI at their local site, which have scanners from three different manufacturers (Philips Healthcare, GE Healthcare Life Sciences and Siemens Healthcare Diagnostic).The protocol included a volumetric T 1 -weighted (magnetizationprepared rapid gradient echo) scan and a diffusion-weighted imaging scan (consisting of with either four or five b0 images and 61 diffusion-weighted images or 64 diffusion-weighted images both with a b-value of 1000 s/mm 2 , depending on the different acquisition sites), as previously reported. 7,27

MRI preprocessing and analyses
Structural grey matter data T 1 -weighted images were processed with the voxel-based morphometry pipeline implemented in the Computational Anatomy Toolbox (CAT12 v.1742) (www.neuro.uni-jena.de/cat) for SPM12 (SPM12 v.7219) (www.fil.ion.ucl.ac.uk/spm/ software/spm12) running on MATLAB R2019b (the MathWorks, Inc., Natick, MA, USA).The voxel-based morphometry pipeline steps consist of tissue segmentation, spatial normalization to a standard Montreal National Institute template, modulation and smoothing. 30The pipeline was also used to calculate the total intracranial volumes for each subject.

Diffusion-weighted data
Diffusion-weighted imaging scans were preprocessed using Mrtrix3 and FSL software. 31Diffusion-weighted imaging scans were analysed using the FDT tool of the FMRIB Software Library (FSL, v6.0).Diffusion-weighted imaging data were denoised and corrected for eddy current-induced distortions and subject movements.These slices are replaced by non-parametric predictions by Gaussian process.Diffusion tensor measures of fractional anisotropy (FA) and mean diffusivity were calculated running DTIFIT script, which fits a diffusion tensor model at each voxel, and subsequently registered to the study template.

DTI-ALPS index computation
The DTI-ALPS method is based on the assumption that the perivascular interstitial fluid movement in the white matter at the level of the lateral ventricle body is dominant along the parallelly aligned medullary veins (x-axis), which run perpendicular to the ventricular wall; the y-axis is given by projection fibres running in the head-foot direction, mainly adjacent to the lateral ventricle, and the z-axis represents association fibres, running in the anterior-posterior direction outside the projection fibres. 18In the projection area, dominant fibres run in the z-axis direction, perpendicular to both the x-and y-axes, whereas in the association area, dominant fibres run in the y-axis direction, perpendicular to both the xand z-axes.The DTI-ALPS index is calculated from the diffusivity in each direction of the projection and association fibre regions at the lateral ventricular body level and estimates the effect of the glymphatic system impairment on the diffusivity along the perivascular space of medullary veins.In calculating the DTI-ALPS index, the FA maps of each subject were registered linearly first and nonlinearly subsequently into the high-resolution FSL_HCP1065_FA standard space image.Spherical regions of interest (ROIs) measuring 5 mm in radius were placed a priori in the projection and association areas at the level of the lateral ventricle bodies in the left and right hemispheres onto the same FA template (see Supplementary Fig. 1).The ROIs were then automatically registered to the subjects' FA map, and all subjects underwent visual inspection (by mean diffusivity and by an expert neurologist [initials: IM]) to check correct ROI placement.If ROIs were not correctly placed, due to artefacts or poor image quality, the subject was excluded from the study (n = 57, 14.3%).Finally, in line with the previous literature data, three DTI-ALPS indices were calculated using the following formula 18,32 : DTI -ALPS = mean(Dxx proj, Dxx assoc) mean(Dyy proj, Dzz assoc) .
The left and right DTI-ALPS (L-DTI-ALPS and R-DTI-ALPS) indices were calculated as a ratio of the mean of the x-axis diffusivity in the projection area (Dxx, proj) and x-axis diffusivity in the association area (Dxx, assoc) to the mean of the y-axis diffusivity in the projection area (Dyy, proj) and the z-axis diffusivity in the association area (Dzz, assoc) on each hemisphere, respectively.The bilateral mean DTI-ALPS index (hereinafter referred to as DTI-ALPS) was calculated as the average of the R-DTI-ALPS and L-DTI-ALPS.
A DTI-ALPS index close to 1.0 reflects lower diffusivity with greater glymphatic system impairment, whereas higher values indicate greater diffusivity and less glymphatic system impairment.In all the analyses, the bilateral mean DTI-ALPS index has been used as primary measure. 18

Plasma NfL and plasma GFAP quantification
In a subset of participants, at the time of clinical assessment, plasma was collected by venepuncture and centrifuged (2000 g, 10 min, at room temperature), according to GENFI protocol. 33Serum was frozen at −80°C within 3 h after collection, shipped and analysed without any previous thaw-freeze cycle.Samples were measured using the multiplex Neurology 4-Plex A kit (102153, Quanterix, Lexington, KY, USA) on the SIMOA HD-1 Analyser following the manufacturer's instructions, as previously published. 33The lower limit of detection of the assay for NfL and GFAP was 0.104 and 0.221 pg/mL, respectively.Quality control samples had a mean intra-assay and inter-assay coefficient of variation of <10%.Technicians were blinded to the genotypic and clinical status of the samples.Plasma NfL and GFAP measures were available for 284 and 235 participants, respectively.

Statistical analyses
Since symptomatic individuals are older than the presymptomatic ones and to avoid biases due to the comparisons of groups with different age, we stratified the NC group into two subgroups based on the mean age (49 ± 13 years): young NC (40 ± 6 years) and old NC (62 ± 7 years).Moreover, due to the large number of pairwise comparisons, we focused only on pre-defined comparisons of interest: young NC versus presymptomatic mutation carriers, old NC versus symptomatic mutation carriers and presymptomatic versus symptomatic within the same genetic mutation groups.
Differences among groups in the sociodemographic and clinical features were assessed using Kruskal-Wallis rank sum tests for continuous variables or tests for equality of proportions for categorical variables.
Associations between DTI-ALPS and other variables were assessed using linear models adjusted for covariates, namely age, sex and MRI site.When plasma markers were considered, site where samples were analysed was added as covariate as well.
In order to assess the association between DTI-ALPS at baseline and longitudinal change in disease severity (CDR-FTLD), baseline DTI-ALPS scores of mutation carriers were converted into Z-scores using mean and standard deviation of NC (both the young and the old ones), and mutation carriers were classified as having 'high' (Z-score > 1), 'average' (Z-score between −1 and 1) or 'low' (Z-score < −1) baseline DTI-ALPS.Subsequently, the association between DTI-ALPS at baseline and longitudinal change in disease severity was assessed using a linear mixed model with longitudinal CDR-FTLD as the dependent variable; DTI-ALPS groups (i.e.'high', 'average' and 'low'), time (years) from baseline and their interaction as independent variables; and age, sex, CDR-FTLD at baseline and site as covariates; random intercepts and slopes at the subject level were computed.
When the number of groups in the comparison was larger than two, post hoc pairwise comparisons were adjusted using the false discovery rate correction.Significance was set at P < 0.05.All statistical analyses were performed with R, version 4.3.0(The R Project for Statistical Computing, https://www.r-project.org/).

Participants
Table 1 illustrates demographic and clinical features of the 374 study participants.As expected, presymptomatic mutation carriers (mean age: 39-45 years; mean CDR-FTLD score: 0.1-0.2) were younger and had better global cognition and functioning than symptomatic mutation carriers (mean age: 59-62 years, P < 0.05; mean CDR-FTLD score: 1.7-2.0,P < 0.05).In presymptomatic mutation carriers, the expected age at onset was similar in the three genetic groups (P = 0.446).In symptomatic mutation carriers, bv frontotemporal dementia was the most prevalent clinical phenotype.

Glymphatic system according to clinical, genetic and neuropathological features
Figure 1 shows the distribution of DTI-ALPS according to disease stage and genetic group.

Figure 1 Distribution of DTI-ALPS across disaggregating by disease stage (A) or by genetic mutation and disease stage (B).
DTI-ALPS, diffusion tensor image analysis along the perivascular space.Comparisons of interest were 'young NC versus presymptomatic', 'old NC versus symptomatic' and 'presymptomatic versus symptomatic' in A (i.e. three pairwise comparisons in total) and 'young NC versus all presymptomatic groups', 'old NC versus all symptomatic groups' and 'presymptomatic versus symptomatic within the same mutation groups' within the three genetic groups in B (i.e.nine pairwise comparisons in total).Statistics values: F(3) = 6.05,P < 0.001 for A and F(7) = 3.46, P = 0.001 for B. Post hoc pairwise comparisons of interest were adjusted using the false discovery rate correction.

Associations between glymphatic system and disease severity, expected years of onset and plasma biomarkers
In all mutation carriers combined, lower DTI-ALPS was associated with worse disease severity (CDR-FTLD: β = −1.16,P < 0.001) (Fig. 2A).In presymptomatic mutation carriers, a trend of statistical significance was found between having lower DTI-ALPS and closer expected years of onset (β = −9.94,P = 0.096) (Fig. 2B).

Association between the glymphatic system status at baseline and longitudinal worsening in disease severity
All mutation carriers with low (β = 0.20, P < 0.001) or average (β = 0.05, P = 0.010) DTI-ALPS at baseline showed worsening in disease severity, while those with high DTI-ALPS remained stable over time (β = −0.01,P = 0.788) (Fig. 3).Worsening in disease severity was faster in mutation carriers with low DTI-ALPS at baseline than in those with average (P = 0.009) or high (P = 0.006) DTI-ALPS at baseline.

Discussion
The glymphatic system represents a potential novel therapeutic target for the treatment of neurodegenerative disorders, while innovative non-invasive imaging techniques can quantify glymphatic system in vivo to elucidate pathophysiology.In the present study, we reported that glymphatic system, as indirectly measured by DTI-ALPS index, is impaired in genetic frontotemporal dementia.In particular, glymphatic system was significantly impaired in symptomatic C9orf72 expansion carriers as compared with healthy controls, but the same trend was also observed in other mutation subgroups.Furthermore, glymphatic system dysfunction was not an early event and was undetectable in the presymptomatic disease stages, although it is unclear if this reflects late emergence of glymphatic deficits or limited sensitivity of non-invasive methods.
We have used the non-invasive MRI index, the DTI-ALPS, as a measure of diffusivity along the perivascular space.This we interpret as a measure of glymphatic function, at least for what concerns its perivascular component. 18][14]17,20,21 This confirms and extends previous research data, claiming that glymphatic system impairment is not disease-specific but reflects brain distress and neurodegeneration. 34In this perspective, a preserved glymphatic system might be protective against disease progression and clinical worsening, even in carriers of frontotemporal dementia-related mutations, while an altered glymphatic system might be associated with worse clinical outcomes over time.
These results support the hypothesis that brain clearance system continues to decline with disease progression, leading to increased tau or TDP-43 pathological burden, further raising the protein concentration to a level that favours aggregation.Glymphatic system dysfunction may participate in the symptom worsening of genetic frontotemporal dementia through its role in the clearance of pathogenetic proteins, autoimmune cell infiltration and promotion of inflammatory activation. 17n this view, we also reported a mild negative association between glymphatic function and expected age at onset of symptoms and between glymphatic function and markers of neurodegeneration, such as NfL, 25,26,33 although not statistically significant.
Interestingly, longitudinal assessment of participants in the GENFI cohort allowed us to demonstrate that glymphatic function may clearly predict disease progression over time, suggesting that DTI-ALPS index may be regarded as marker of progression, possibly predicting transition towards progressive stages.
These results suggest the potential of glymphatic change as a therapeutic target in frontotemporal dementia, counteracting disease progression by enhancing the brain endogenous waste clearance pathways.The evidence for specific interventions remains indirect, e.g. that sleep deprivation can suppress brain clearance via the glymphatic dysfunction [35][36][37] and consequently behavioural or pharmacological interventions that preserve night sleep could be considered.Although the study of sleep disturbances in FTD is in its infancy, there is emerging evidence that hypothalamic dysfunction, manifesting as disturbances in sleep, is an integral component of neurodegeneration in frontotemporal dementia. 38In addition to sleep, another potential molecular target could be represented by aquaporin-4, a key protein involved in the glymphatic system function, 39 already demonstrated to be impaired in FTD patients in the GENFI cohort. 40In the near future, both pharmacological strategies or nonpharmacological approaches (e.g.][43] We acknowledge that this study entails some limitations.First, the DTI-ALPS index was calculated based on the diffusion of perivascular space and may not be directly equivalent to the glymphatic function, although it has been shown to be reliable and has been widely applied in multiple diseases.In addition, in line with more recent studies on this topic, we did not use susceptibility-weighted imaging for ROI placement. 17,32Second, we assessed DTI-ALPS index of the whole brain rather than a regional function measurement, even though we tested left and right hemispheres with no substantial differences across clinical phenotypes.Third, we did not take into account possible additional confounders, such as white matter hyperintensities.Notwithstanding these limitations, we believe that by using a non-invasive diffusion-based imaging approach, as a proxy of the glymphatic system function, we have successfully demonstrated that glymphatic system is impaired in the symptomatic stages of genetic FTD, irrespective of the clinical syndrome or neuropathological features.Moreover, glymphatic system impairment was associated with disease stage and predicted its clinical progression.The glymphatic system may therefore represent a novel potential target for future treatment approaches in genetic FTD.