Synopsis

Keywords: Multiple Sclerosis, Neurofluids

We applied FAST-T2 multi-component T2 relaxometry to 200 MS patients and 66 healthy controls and found that the thalamic cerebrospinal fluid fraction (CSFF) increases with age and follows a different trajectory in MS. In 13 MS patients, we also found a strong correlation between CSFF and [¹¹C]PK11195 uptake on PET in the thalamus and putamen, suggesting a connection between glymphatic dysfunction and microglial inflammation in the MS brain.

INTRODUCTION

Multiple sclerosis (MS) is a neuroinflammatory disorder characterized by focal demyelination and axonal injury in WM and chronic progressive neurodegeneration in GM¹. Recently, impairment in the glymphatic perivascular flow of cerebrospinal fluid (CSF)² has been implicated as a potential pathological mechanism underpinning reduced waste clearance, tissue damage and clinical disability in the MS brain³. Increased CSF fraction (CSFF) on the microscopic level is a potential early indicator of glymphatic clearance dysfunction and can be mapped efficiently using advanced multi-echo T2 relaxometry sequences⁴ such as Fast Acquisition with Spiral Trajectory and adiabatic T2prep (FAST-T2)⁵. This approach can provide within each imaging voxel the fraction of highly mobile CSF occupying the perivascular space and that of the much less mobile water in the myelin sheath and the intra/extracellular space by utilizing differences in their T2 spectra⁶. We hypothesized that chronic inflammation in MS will have an adverse effect on the brain glymphatic clearance and correspondingly aimed to 1) to establish the difference in age-related CSFF trajectory between MS patients and healthy controls (HCs) over the adult lifespan, and 2) to study the relationship between MRI-derived CSFF and translocator protein (TSPO) PET, an imaging marker of microglial activation.

METHODS

Study cohorts. For the age-related CSFF change study, a total of 200 MS patients (age range 22.3-79.5y, 145 women (72.5%), 55 men (27.5%), all RRMS, median EDSS=2) and 66 HCs (age range 22.4-79.8y, 40 women (60.6%), 26 men (39.4%)) who had FAST-T2 MRI scan were included. For PET-MRI correlation study, thirteen MS patients (9 women, 4 men, age 44.5y±13.4, 8 RRMS, 3 PPMS, 2 SPMS) who had both [¹¹C]PK11195 TSPO PET and FAST-T2 MRI scans within a short interval (median 2 days) were included. Patients with active inflammatory disease as evidenced by new Gd-enhancing lesions were excluded.
MRI acquisition. The 3T MRI protocol (Siemens and GE scanners) included a 4 min FAST-T2 sequence (1x1x5 mm³ voxel size) with geometric echo spacing for multi-component T2 relaxometry⁵ in addition to the conventional T1W, T2W and FLAIR sequences. A spatially regularized three-pool non-linear least squares algorithm using the L-BFGS solver was used to compute myelin water fraction, intra/extracellular water fraction, and CSFF maps from the six-echo T2 decay data⁵. The lower and upper T2 bounds for each of the three water pools (in msec) were set to [5 20], [20 200], and [200 2000], respectively. Freesurfer v6 recon-all command was applied to T1W/T2W images to obtain brain segmentation⁷. CSFF maps were aligned to Freesurfer T1W image using FMRIB’s Linear Image Registration Tool⁸. To reduce contamination from CSF occupying the brain ventricles and the subarachnoid space due to partial voluming, the segmented brain regions of interest were first eroded by 1mm and then further eroded to be at least 1 mm in-plane and 5 mm through-plane from the boundary of these CSF-filled spaces.
[¹¹C]PK11195 PET acquisition. PET images were acquired on a lutetium oxyorthosilicate (LSO) time-of-flight PET/CT scanner (Siemens/CTI) and reconstructed using an iterative-plus-time of flight list-mode reconstruction algorithm provided by the manufacturer using OSEM methods. Tissue concentrations were estimated by reconstructing data into 22 frames for a total scan time of 60 min. Regional distribution volume ratios (DVRs) of radioligand uptake were calculated using Logan graphical model⁹. A computational supervised clustering methodology (SuperPK) was used to extract a reference curve on TSPO PET. PET images were aligned to Freesurfer T1W image using PMOD software.
Statistical analysis. Based on visual inspection of thalamic CSFF vs. age data, a linear regression model was used to regress CSFF on age (with indicator for >50y group) adjusting for sex, disease status (MS or HC), and thalamic volume (normalized to skull size) with a three-way interaction among disease status, age, and age indicator. Linear correlation was obtained between mean TSPO PET DVR and MRI CSFF values in the thalamus and putamen, two large subcortical GM regions that have been implicated in MS in recent TSPO PET studies^10,11.

RESULTS

In the linear model with CSFF as an outcome, we found a significant association between thalamic CSFF and >50y age indicator (p=0.042) and that having MS is likely to alter the relationship between thalamic CSFF and age before the age of 50 (p=0.072). The other statistically significant variables in the model were MS disease status (p=0.006) and normalized thalamic volume (p<0.001). Figure 1 shows age-related CSFF changes in MS and HC subjects, showing a steady increase of thalamic CSFF with age over the whole adult lifespan in MS patients, while the rise in CSFF only becomes noteworthy after the age of 50 in HCs. We found a strong linear correlation between TSPO PET DVR measurements and corresponding MRI-derived CSFF in the thalamus and putamen (Fig.2, R=0.827, p=1.9e-7).

DISCUSSION

We found different trends in CSFF increase with age in MS vs. HC, which underlies the adverse impact of ongoing neuroinflammation and neurodegeneration on the MS brain beyond that observed in normal aging. Our results also provided the initial evidence that the MRI-derived CSFF measures in subcortical GM correlate well with [¹¹C]PK11195 uptake by TSPO PET, suggesting a link between glial inflammation and reduced glymphatic clearance in these areas.

Acknowledgements

No acknowledgement found.