RPS 611b - Multiple sclerosis

Purpose:

Brain atrophy and the neurodegeneration underlying it starts early in multiple sclerosis (MS). However, the long-term clinical relevance of early atrophy is unknown.

Methods and materials:

132 people presenting with a clinically isolated syndrome were recruited between 1984-87 and followed-up clinically and with an MRI 1, 5, 10, 14, 20, and 30 years later. Third ventricular width, medullary width, and white matter lesions were measured at all available time points.

Results:

At 30 years, 27 had a clinically isolated syndrome (CIS), 34 relapsing-remitting (RR) MS, 26 secondary progressive (SP) MS, and 16 had died due to MS-related (MSRD) causes. Of those who were alive, the mean expanded disability status scale (EDSS) in the CIS groups was 1 (0-2), in the RRMS group 1.5 (1-2), and in the SPMS group 6 (6-6,5).

Nested mixed effect models estimated significant faster atrophy rates in SPMS (p<0.0000) and MSRD patients (p<0.0000).

Using logistic regression, a reduction of 1 mm in the medullary width from baseline to 5 years, predicted a 6-fold increase in the probability of SPMS or MSRD over the next 25 years (p<0.013).

Atrophy rates within the first 5 years after CIS predicted a broad range of cognitive and motor function-based outcome measures at 30 years.

Conclusion:

Our findings suggest that brain atrophy within 5 years of symptom onset predicts disability a decade or more later, indicating its relevannce prior to clinical progression.

Limitations:

The analysis was restricted by the image quality of archival films analysed at baseline, one, and five-year follow-up.

Ethics committee approval

Approved by the National Research Ethics Service (15/LO/0650).

Funding:

L.H. was supported by ECTRIMS/MAGNIMS and ESNR-Fellowships.

Purpose:

To assess the value of the bi-caudate ratio as an MRI marker of white matter atrophy in multiple sclerosis and ischaemic leukoencephalopathy patients and to set a cut-off value to differentiate between patients with white matter atrophy and normal subjects.

Methods and materials:

115 patients (54 male and 61 female) (mean age {+SD}, 52.5 {+ 18.2} years) diagnosed with white matter leukoencephalopathy (MS in 51 patients and ischaemic leukoencephalopathy in 64 patients) and a control group of 60 subjects (27 male and 33 female) with normal white matter signal were scanned. The BCR was compared for both patients and controls. The BCR was compared for patients with MS and ischaemic leukoencephalopathy.

Results:

BCR for the patients ranged from 0.13-0.27 mm (mean {± SD}=0.16±0.02), while for the control group it ranged from 0.05-0.13 mm (mean {± SD}=0.09±0.01). The difference between the two groups was statistically significant (P<0.001). A cut-off value of 0.13 was used to differentiate between the BCR in both patients and control groups with a sensitivity, specificity, and accuracy of 99.2%, 100%, and 99%, respectively. The difference in BCR for patients diagnosed with MS and ischaemic leukoencephalopathy was also statistically significant (P<0.001).

Conclusion:

The bi-caudate ratio represents a linear measurement of subcortical atrophy that can be useful as a surrogate marker of global supra-tentorial white matter atrophy instead of the usually performed visual, and therefore subjective, assessment. It is an easily obtained measure that can be performed without complex time consuming volumetric studies. Our findings also revealed that the BCR is higher in patients with ischaemic leukoencephalopathy than in patients with MS.

Limitations:

n/a

Ethics committee approval

n/a

Funding:

No funding was received for this work.

Purpose:

MRI is important in monitoring multiple sclerosis (MS) but conventional side-by-side reading (SBSR) of follow-up examinations for detecting new or changing T2/FLAIR lesions is time-consuming and prone to misdetection. Several recent articles presented methods for the detection of T2/FLAIR lesions based on commonly available software packages that accelerate the reading process. The aim was to compare two novel reading methods against SBSR in an everyday reading environment.

Methods and materials:

30 consecutive MS patients were included. MRI examinations were performed on the same 1.5T scanner and included a 3D-FLAIR sequence. 4 radiologists with different experience levels analysed two follow-up examinations, current and previous, respectively, on the same workstation. Automatic co-registration of the follow-up examinations along with colour coding of the current examination (CE) for the colour-coded reading (CCR) and contrast inversion of the CE for the inverted contrast reading (ICR) were performed. We analysed the specificity, sensitivity, positive predictive value (PPV), and negative predictive value (NPV) for the detection of new MS lesions (NMSL), MS lesion size changes (MSLSC), and reading time (RT). A senior neuroradiologist provided the reference standard.

Results:

CCR showed the highest combined sensitivity [97.5%] and specificity [85%] for NMSL detection with a PPV of 93.1% and NPV of 95.4%. Regarding MSLSC detection, both CCR [60.9%] and ICR [80%] showed a higher combined sensitivity compared to SBSR [49%]. The combined mean RT showed a significant decrease when CCR [71s] and ICR [88s] were compared to SBSR.

Conclusion:

CCR and ICR are valid alternatives to SBSR that reduce the RT while keeping or improving NMSL and MSLSC detection. Therefore such reading methods are recommended in everyday clinical routine.

Limitations:

The number of included patients is limited.

Ethics committee approval

The IRB approved the study and waived the need for informed consent.

Funding:

No funding was received for this work.

Purpose:

To investigate whether patients with MS present with significant cervical cord atrophy compared to healthy controls and to identify a cervical spinal cord area threshold that could discriminate patients with MS from healthy controls.

Methods and materials:

Literature databases were searched (2001-2018) for cohort and case-control studies reporting measures of cervical cord areas on MRI scans in MS patients and controls to be included in the meta-analysis. Articles evaluating the size or changes in the size of MS plaques, non-homogeneous patient categories, and patients with MS undergoing experimental therapies were excluded.

Results:

17 studies were included, resulting in a total cohort of 1,545 patients and 505 controls. A significant statistical difference between progressive forms of MS (PMS, PPMS, and SPMS) patients and healthy controls cervical cord areas (p<00001) was ascertained. Based on a ROC curve, a cervical cord threshold area of 71 mm² (sensitivity 94%, specificity 100%) to determine progressive forms of MS patients from non-progressive forms of MS and healthy controls was identified (Younden index: 94%).

Conclusion:

Cervical cord atrophy is statistically significant in PMS patients compared to healthy controls. The detection of a cervical cord trans-sectional area threshold of 71mm², measured by MRI in MS patients, could be used both in clinical practise to discriminate progressive forms of MS from healthy controls and non-progressive forms of MS and in clinical trials to validate the efficacy of established therapies and new therapies.

Limitations:

The relative small number of studies analysed and the heterogeneity of sequences and methods used for cervical cord trans-sectional area measurements by the included studies.

Ethics committee approval

The study was approved by the institutional ethics board and adhered to the tenets of the declaration of Helsinki.

Funding:

No funding was received for this work.

Purpose:

Evidence from different neuropathies demonstrated that peripheral nerve pathology can broadly influence brain connectivity, way beyond the involvement of the sensorimotor network. The aim of our study was to systematically investigate functional connectivity (FC) changes in the brain of Charcot-Marie-Tooth (CMT) 1A patients.

Methods and materials:

In this observational cross-sectional study, we enrolled 18 right-handed patients with genetically confirmed CMT1A, along with 20 age- and sex-comparable healthy controls (HC). Resting-state (RS) fMRI data was analysed with a seed-based approach sampling 32 ROIs included in the SPM-based CONN toolbox and characterising an extended set of classical RS networks. Between-group differences for the distinct seeds were tested using the standard general linear model implemented in SPM12, covarying for age, sex, and mean motion, and correcting for multiple comparisons at the cluster level. The results of each analysis were considered significant for p<0.05 and were Bonferroni-corrected for multiple comparisons (0.05/32).

Results:

In CMT1A patients compared to HC, several clusters emerged of increased FC relative to seeds within the default mode, dorsal attention, and language and salience networks, encompassing different supratentorial cortical areas (p<0.001).

Furthermore, a cluster of reduced FC relative to the occipital cortex was found in the left lentiform nucleus (p=3E-4).

Conclusion:

CMT1A patients show extensive rearrangements of brain connectivity.

FC increase could reflect a reduction of the physiological anticorrelation between the default mode (DMN) and other networks (e.g. increased FC between hubs of the salience and dorsal attention networks and DMN nodes, namely posterior cingulate cortex and inferior parietal lobule), as well as a compensatory mechanism (e.g. increased FC within the language network).

Reduced FC between the occipital cortex and lentiform nucleus could represent a maladaptive disruption of the visual corticostriatal functional loop.

Limitations:

n/a

Ethics committee approval

n/a

Funding:

No funding was received for this work.

Purpose:

Although it is primarily a peripheral nervous system disease, several anecdotal reports of central nervous system (CNS) involvement have been described in different forms of Charcot-Marie-Tooth disease (CMT), including its most common subtype (CMT1A). The aim of this study was to investigate the presence of structural damage in the brain of CMT1A patients.

Methods and materials:

20 patients with genetically confirmed CMT1A were enrolled along with 20 age- and sex-comparable healthy controls (HC). Brain MRI exams included a 3D-GE-T1w for the voxel-based morphometry analysis (VBM) and diffusion tensor imaging (DTI) data for the tract-based spatial statistics (TBSS) analysis. CMT1A patients also underwent clinical and electrophysiological examinations including the determination of the CMT neuropathy score (CMTNS) as a global measure of disease severity and the compound motor action potential (CMAP) as an index of distal arm axonal damage.

Results:

The VBM analysis revealed a cluster of significantly increased GM density in CMT1A patients compared to HC encompassing the right paravermian portions of the cerebellar lobules III, IV and V. No between-group differences emerged from the TBSS analysis when considering the DTI metrics. A significant negative correlation (r=-0.738, p=0.003) was found between the CMAP values and the age, sex, and TIV-adjusted z-scores of the first eigenvariate extracted from the cluster of significant between-group difference at the VBM analysis.

Conclusion:

CMT1A patients show cerebellar grey matter modifications, which occur independently from central white matter microstructural alterations, possibly representing a structural plasticity mechanism which compensates for the primary peripheral nerve damage.

Limitations:

n/a

Ethics committee approval

n/a

Funding:

No funding was received for this work.

Purpose:

Despite the use of advanced MRI techniques, imaging biomarkers often fail to accurately predict the accumulation of physical disability in multiple sclerosis (MS) patients. The aim of our study was to investigate the possible role of radiomics for the prediction of disability in MS by applying a machine learning approach to texture analysis (TA) parameters.

Methods and materials:

79 patients with MS were recruited. All patients underwent a neurological examination with the determination of the expanded disability status scale (EDSS) as an index of clinical disability along with an MRI protocol on a 3T scanner including 3D-T1-weighted and 3D-FLAIR sequences.

On FLAIR images, hyperintense lesions were segmented, while T1-weighted volumes were used to segment white matter (WM) and both cortical and deep grey matter (GM). Features were extracted using open-source software (Pyradiomics v2.1.2) with a resulting dataset that was processed with a data mining software (Weka v3.8), while a regression support vector machine was used for EDSS score prediction.

Model validation was performed with a train-test approach, with the test group constituting 25% of the total population.

Results:

MS patients (M/F=44/35; 41.5±11.0 years) showed a median EDSS score of 3.5.

The subset evaluator produced a dataset containing 47 features (22 from T1-weighted and 25 from FLAIR sequences, mainly derived from deep GM).

The support vector machine obtained a correlation coefficient of 0.87, with a mean absolute error of 0.5 and a root mean square error of 0.7.

Conclusion:

Our results demonstrate that a machine learning-derived model including TA parameters mainly related to deep GM is able to predict physical disability in MS patients with significant accuracy.

Limitations:

A single center, retrospective study.

Ethics committee approval

Study approved by the local ethics committee.

Funding:

No funding was received for this work.

Purpose:

Quantitative susceptibility mapping (QSM) is an emerging method used to characterise brain lesions in multiple sclerosis (MS). Our goals were to investigate the relation between QSM values and gadolinium (Gd)-enhancing MS lesions and to assess the relation between QSM values and lesion size.

Methods and materials:

We performed a retrospective study on 73 MS patients who underwent a single-time-point brain MRI at 3T (MAGNETOM Skyra, Siemens Healthcare, Erlangen, Germany), including 3D FLAIR, T1-MPRAGE pre/post-Gd, and double-echo GRE (TE=20/40ms) sequences. The fully automated prototype method LeManPV was used for the automated lesion segmentation on 3D FLAIR and T1-MPRAGE pre-Gd images. A set of lesions was selected manually according to specific criteria. Active lesions were identified using T1-MPRAGE pre/post-Gd data. QSM maps were estimated from GRE images using a standard post-processing pipeline including RESHARP and TVSB algorithms. QSM lesions values were extracted using LeManPV lesion masks and compared between non-enhancing and enhancing lesions with the Wilcoxon rank-sum test. The relation to size was evaluated with the Spearman coefficient.

Results:

3,894 lesions were found by the LeManPV method. We selected 985 lesions (25%), of which 35 (3.6%) were enhancing. We observed similar mean QSM values in enhancing and non-enhancing lesions (0.0026±0.0184 vs 0.0050±0.0176 ppm, p=0.32). For lesions with a volume greater than 70 µL, the difference was significant (-0.0044±0.0150 vs 0.0082±0.0177 ppm, p=0.028). In non-enhancing lesions, there was a very weak correlation between QSM values and lesions size for lesions ≥70 µL (rho=0.18, p=0.0015) but not for lesions <70 µL (rho=0.06, p=0.13).

Conclusion:

On a single time-point, QSM values may be influenced by lesion size and enhancement status. This should be taken into account when QSM values are used to identify active MS lesions.

Limitations:

n/a

Ethics committee approval

n/a

Funding:

No funding was received for this work.

Purpose:

Can tensor-valued diffusion encoding provide higher specificity in multiple sclerosis (MS)? In contrast to the traditional fractional anisotropy (DTI-FA), microscopic fractional anisotropy (μFA) obtained from tensor-valued diffusion encoding is not affected by fibre dispersion and might therefore be a better disease biomarker. We estimated DTI-FA and μFA in a group of relapsing-remitting (RRMS) and primary progressive (PPMS) MS patients and age-matched healthy controls (HC).

Methods and materials:

45 MS and 28 HC were scanned (14 PPMS, 26 RRMS, and 27 HC analysed) on a Philips Achieva 3T, 32-channel head-coil. DTI-FA from conventional DWI: 106 images 2x2x2 mm, 50 slices, scan time 21.5 min. µFA from tensor-valued DWI: 2.5x2.5x2.5 mm, 14 slices, scan time 16.2 min.

Results:

Tissue pathology in normal-appearing white matter (NAWM) was reflected by μFA, but not DTI-FA. μFA was lowest in PPMS and highest in HC. DTI-FA tended to be higher in HC than in RRMS, but counterintuitively, higher DTI-FA was observed in PPMS than in RRMS, probably due to degenerated crossing fibres. μFA provides stronger correlations with clinical scores. Total lesion volume correlated with μFA but not DTI-FA. μFA negatively correlated with age but not DTI-FA.

Conclusion:

Compared to traditional diffusion MRI, tensor-valued encoding can significantly improve the characterisation of pathology in MS.

Limitations:

Lower image resolution may lead to partial volume effects reducing specificity.

Ethics committee approval

Approval by the ethics committee of the Capital Region of Denmark (protocol H-15006964) and written informed consent was obtained.

Funding:

Danish Multiple Sclerosis Society (Grant Nr. A31910 and A27996).

VINNMER Marie Curie Industry Outgoing grant (Grant Nr. 013-04350).

Swedish Research Council (2014-3910)

Swedish Foundation for Strategic Research (Grant Nr. AM13-0090)

Lundbeck Foundation (Grant Nr. R186-2015-2138)

Purpose:

Brain volume change has been suggested as an MRI predictor of disability in multiple sclerosis (MS). However, volumetric measurements are not used routinely as no accepted methodology for clinical use exists to date. Easily applicable planimetric measurements are proposed for different neurologic disease entities. In this pilot study, we investigated whether planimetric measures predict the technically more complex volumetric measurements in MS patients.

Methods and materials:

Neural networks (NNs) were used to predict the volumetric measures (grey matter [GM], white matter [WM], and cerebrospinal fluid [CSF]) with planimetric measures. The planimetric input variables were preselected using Spearman's correlation (SC) coefficient.

Results:

Planimetric characteristics yielding a high correlation with volumetric characteristics were GM with frontooccipital diameter (FOD) (SC=0.98), midbrain diameter (M_d) (SC=0.92), pontine area (P_A) (SC=0.98), midbrain-to-pontine diameter ratio (SC=0.90), WM with transverse skull diameter (SC=0.89), FOD (SC=0.93), M_d (SC=0.85), P_A (SC=0.95), CSF with third ventricle width unaligned (SC=0.82), anterior segment of corpus callosum/genu (SC=-0.83), corpus callosum index (CCI) (SC=-0.83), and regional anterior CCI (SC=-0.92).

The NNs constructed based on these measures predicted the volumes of GM, WM, and CSF with errors below 11% for GM, 5% for WM, and 15% for CSF.

Conclusion:

Despite the small sample size, we observed high correlations between MR planimetric and volumetric measurements and were able to construct NNs with low prediction errors.

Limitations:

Larger patient cohorts will allow separation in training and test sets to investigate MR planimetry as an easy to apply and robust imaging biomarker to evaluate cerebral and ventricular volume in patients with MS with similar accuracy as, and strong correlations with, volumetric measurements.

Ethics committee approval

The study has been approved by the ethical committee of the Medical University of Innsbruck.

Funding:

No funding was received for this work.