Positron Emission Tomography with [18 F]-DPA-714 Unveils a Smoldering Component in Most Multiple Sclerosis Lesions which Drives Disease Progression.

OBJECTIVE: To determine the prognostic value of persisting neuroinflammation in multiple sclerosis (MS) lesions, we developed a 18 kDa-translocator-protein-positron emission tomography (PET) -based classification of each lesion according to innate immune cell content and localization. We assessed the respective predictive value of lesion phenotype and diffuse inflammation on atrophy and disability progression over 2 years. METHODS: Thirty-six people with MS (disease duration 9 ± 6 years; 12 with relapsing-remitting, 13 with secondary-progressive, and 11 with primary-progressive) and 19 healthy controls (HCs) underwent a dynamic [18 F]-DPA-714-PET. At baseline and after 2 years, the patients also underwent a magnetic resonance imaging (MRI) and neurological examination. Based on a threshold of significant inflammation defined by a comparison of [18 F]-DPA-714 binding between patients with MS and HCs, white matter lesions were classified as homogeneously active (active center), rim-active (inactive center and active periphery), or nonactive. Longitudinal cortical atrophy was measured using Jacobian integration. RESULTS: Patients with MS had higher innate inflammation in normal-appearing white matter (NAWM) and cortex than HCs (respective standardized effect size = 1.15, 0.89, p = 0.003 and < 0.001). Out of 1,335 non-gadolinium-enhancing lesions, 53% were classified as homogeneously-active (median = 17 per patient with MS), 6% rim-active (median = 1 per patient with MS), and 41% non-active (median = 14 per patient with MS). The number of homogenously-active lesions was the strongest predictor of longitudinal changes, associating with cortical atrophy (ß = 0.49, p = 0.023) and Expanded Disability Status Scale (EDSS) changes (ß = 0.35, p = 0.023) over 2 years. NAWM and cortical binding were not associated to volumetric and clinical changes. INTERPRETATION: The [18 F]-DPA-714-PET revealed that an unexpectedly high proportion of MS lesions have a smoldering component, which predicts atrophy and clinical progression. This suggests that following the acute phase, most lesions develop a chronic inflammatory component, promoting neurodegeneration and clinical progression. ANN NEUROL 2023;94:366-383.

Simultaneous assessment of blood flow and myelin content in the brain white matter with dynamic [11 C]PiB PET: a test-retest study in healthy controls.

BACKGROUND: Exploring the relationship between oxygen supply and myelin damage would benefit from a simultaneous quantification of myelin and cerebral blood flow (CBF) in the brain's white matter (WM). To validate an analytical method for quantifying both CBF and myelin content in the WM using dynamic [11C]PiB positron emission tomography (PET). METHODS: A test-retest study was performed on eight healthy subjects who underwent two consecutive dynamic [11 C]PiB-PET scans. Three quantitative approaches were compared: simplified reference tissue model 2 (SRTM2), LOGAN graphical model, and standardized uptake value ratio (SUVR). The sensitivity of methods to the size of the region of interest was explored by simulating lesion masks obtained from 36 subjects with multiple sclerosis. Reproducibility was assessed using the relative difference and interclass correlation coefficient. Repeated measures correlations were used to test for cross-correlations between metrics. RESULTS: Among the CBF measures, the relative delivery (R1) of the simplified reference tissue model 2 (SRTM2) displayed the best reproducibility in the white matter, with a strong influence of the size of regions analyzed, the test-retest variability being below 10% for regions above 68 mm3 in the supratentorial white matter. [11C]PiB PET-derived proxies of CBF demonstrated lower perfusion of white matter compared to grey matter with an overall ratio equal to 1.71 ± 0.09 when the SRTM2-R1 was employed. Tissue binding in the white matter was well estimated by the Logan graphical model through estimation of the distribution volume ratio (LOGAN-DVR) and SRTM2 distribution volume ratio (SRTM2-DVR), with test-retest variability being below 10% for regions exceeding 106 mm3 for LOGAN-DVR and 300 mm3 for SRTM2-DVR. SRTM2-DVR provided a better contrast between white matter and grey matter. The interhemispheric variability was also dependent on the size of the region analyzed, being below 10% for regions above 103 mm3 for SRTM2-R1 and above 110 mm3 for LOGAN-DVR. Whereas the 1 to 8-minute standardized uptake value ratio (SUVR1-8) showed an intermediary reproducibility for CBF assessment, SUVR0-2 for perfusion or SUVR50-70 for tissue binding showed poor reproducibility and correlated only mildly with SRTM2-R1 and LOGAN-DVR estimations respectively. CONCLUSIONS: [11C]PiB PET imaging can simultaneously quantify perfusion and myelin content in WM diseases associated with focal lesions. For longitudinal studies, SRTM2-R1 and DVR should be preferred over SUVR for the assessment of regional CBF and myelin content, respectively. TRIAL REGISTRATION: European Union Clinical Trials Register EUDRACT; EudraCT Number: 2008-004174-40; Date: 2009-03-06; https//www.clinicaltrialsregister.eu ; number 2008-004174-40.

Automatic segmentation of the choroid plexuses: Method and validation in controls and patients with multiple sclerosis.

Choroid Plexuses (ChP) are structures located in the ventricles that produce the cerebrospinal fluid (CSF) in the central nervous system. They are also a key component of the blood-CSF barrier. Recent studies have described clinically relevant ChP volumetric changes in several neurological diseases including Alzheimer's, Parkinson's disease, and multiple sclerosis (MS). Therefore, a reliable and automated tool for ChP segmentation on images derived from magnetic resonance imaging (MRI) is a crucial need for large studies attempting to elucidate their role in neurological disorders. Here, we propose a novel automatic method for ChP segmentation in large imaging datasets. The approach is based on a 2-step 3D U-Net to keep preprocessing steps to a minimum for ease of use and to lower memory requirements. The models are trained and validated on a first research cohort including people with MS and healthy subjects. A second validation is also performed on a cohort of pre-symptomatic MS patients having acquired MRIs in routine clinical practice. Our method reaches an average Dice coefficient of 0.72 ± 0.01 with the ground truth and a volume correlation of 0.86 on the first cohort while outperforming FreeSurfer and FastSurfer-based ChP segmentations. On the dataset originating from clinical practice, the method reaches a Dice coefficient of 0.67 ± 0.01 (being close to the inter-rater agreement of 0.64 ± 0.02) and a volume correlation of 0.84. These results demonstrate that this is a suitable and robust method for the segmentation of the ChP both on research and clinical datasets.

Online hard example mining vs. fixed oversampling strategy for segmentation of new multiple sclerosis lesions from longitudinal FLAIR MRI.

Detecting new lesions is a key aspect of the radiological follow-up of patients with Multiple Sclerosis (MS), leading to eventual changes in their therapeutics. This paper presents our contribution to the MSSEG-2 MICCAI 2021 challenge. The challenge is focused on the segmentation of new MS lesions using two consecutive Fluid Attenuated Inversion Recovery (FLAIR) Magnetic Resonance Imaging (MRI). In other words, considering longitudinal data composed of two time points as input, the aim is to segment the lesional areas, which are present only in the follow-up scan and not in the baseline. The backbone of our segmentation method is a 3D UNet applied patch-wise to the images, and in which, to take into account both time points, we simply concatenate the baseline and follow-up images along the channel axis before passing them to the 3D UNet. Our key methodological contribution is the use of online hard example mining to address the challenge of class imbalance. Indeed, there are very few voxels belonging to new lesions which makes training deep-learning models difficult. Instead of using handcrafted priors like brain masks or multi-stage methods, we experiment with a novel modification to online hard example mining (OHEM), where we use an exponential moving average (i.e., its weights are updated with momentum) of the 3D UNet to mine hard examples. Using a moving average instead of the raw model should allow smoothing of its predictions and allow it to give more consistent feedback for OHEM.