Gadolinium-based contrast agents and perivascular spaces in multiple sclerosis

<p dir="ltr"><b>Background and Purpose</b>: MRI plays a key role in the diagnosis and monitoring of MS. While GBCAs reveal active inflammation, concerns about long-term retention drive the search for safer alternatives. This thesis investigates the effects of Gd on brain MRI by: (a) comparing conventional T1-weighted (T1W) with advanced MT imaging to assess Gd-related signal changes; (b) assessing GBCAs influence on brain volumetrics; and (c) developing a new automated method for quantifying enhancing MS lesion. Additionally, it examines the relevance of VRS as imaging biomarker and their histopathological correlates (d). </p><p dir="ltr"><b>Study Population</b>: All these studies used cohorts of participants with confirmed MS diagnosis per current criteria. Additionally, in Studies I and IV, age- and sex-matched healthy control cohorts were also recruited. Study IV further included a validation cohort for the in vivo assessment, as well as an ex vivo cohort comprising both MS patients and healthy controls. </p><p dir="ltr"><b>Materials & Methods</b>: </p><p dir="ltr"><b>Study I</b>: A cohort of 23 MS patients with long-term GBCA exposure, followed prospectively for over 18 years, was compared with 23 age- and sex-matched controls. All underwent 1.5 T brain MRI including 2-dimensional T1W spin echo (2D T1W SE), 3-dimensional T1W gradient echo (3D T1W GRE) and 2D MT sequences. Signal intensity indices (SII) were calculated in the dentate nucleus (DN) and globus pallidus (GP) to evaluate gadolinium retention. </p><p dir="ltr"><b>Study II</b>: Brain MRI from 200 MS patients, each with two contrast-enhanced 3D T1-weighted scans over a mean interval of 2.0 ± 0.9 years, were retrospectively analyzed. Whole brain and regional brain volumes, including cortex, thalamus, gray matter (GM), and white matter (WM), were quantified before and after GBCA administration, using an automated multi-atlas segmentation software (cNeuro) and a correction method to reduce GBCA-related bias was evaluated. </p><p dir="ltr"><b>Study III</b>: To develop and validate an automated method for contrast-enhancing lesions (CELs) detection in MS, contrast-enhanced 3D T1W MRI from 1,768 exams (1,034 patients) were retrospectively analyzed. The algorithm, trained on 97 CELs from 10 scans and validated on 107 CELs from 16 scans, was than applied to the full dataset. All scans underwent standardized manually segmentation and neuroradiologist review as reference. </p><p dir="ltr"><b>Study IV</b>: This cohort study assessed associations between non-dilated and dilated VRS and clinical as well as MRI outcomes in 142 MS patients and 30 healthy controls, with validation in an independent cohort of 63 MS patients. Furthermore, post-mortem samples from 6 MS patients and 3 controls were examined using ex vivo MRI and histopathology to assess the tissue substrates underlying VRS. </p><p dir="ltr"><b>Results</b>: </p><p dir="ltr"><b>Study I</b> confirmed a dose-dependent association between Gd-administrations and SII-changes in the DN and globus pallidus GP on conventional T1W imaging. SII alterations related to GBCAs were most effectively detected using 3D T1W GRE sequences, with the most pronounced changes observed in the GP. In contrast, no evidence of Gd retention was detectable using the applied 2D MT technique. </p><p dir="ltr"><b>Study II</b> demonstrated that GBCAs significantly affects brain MRI volumetry, particularly in the thalamus and cortex, thereby complicating longitudinal assessments in MS. Applying the correction method to contrast-enhanced imaging improved atrophy detection, but sensitivity remained lower than with non-contrast imaging. </p><p dir="ltr"><b>Study III</b> showed that the fully automated CEL detection method matched manual performance, ensuring transparency, reproducibility, and easy integration. Lesion counts correlated strongly with manual segmentation, with even higher agreement for lesion volumes, demonstrating its promise for routine clinical detection of contrast-enhancing MS lesions. </p><p dir="ltr"><b>Study IV</b> found that higher counts of dilated VRS in the centrum semiovale strongly correlated with increased T1 and T2 lesion volumes. Histopathology showed that VRS mainly correspond to arteries affected by vascular pathology but no MS-specific tissue damage, challenging the idea that VRS dilation reflects pre-inflammatory immune cell influx. Conclusion: Developing and integrating advanced MRI techniques, automated lesion detection, and emerging biomarkers may enhance MS diagnosis, support monitoring of disease progression and Gd-related changes and ultimately improve clinical management and patient outcomes.</p><p dir="ltr"><b>Conclusion</b>: Developing and integrating advanced MRI techniques, automated lesion detection, and emerging biomarkers may enhance MS diagnosis, support monitoring of disease progression and Gd-related changes and ultimately improve clinical management and patient outcomes.</p><h3>List of scientific papers</h3><p dir="ltr">This thesis is primarily composed of the following four papers, which will be referred to in the text by their corresponding roman numerals:</p><p dir="ltr">I. MRI detection of brain gadolinium retention in multiple sclerosis: Magnetization transfer vs. T1-weighted imaging. <b>Cananau C,</b> Forslin Y, Bergendal Å, Sjöström H, Fink K, Ouellette R, Kristoffersen Wiberg M, Fredrikson S, and Granberg T Journal of Neuroimaging 2023; 33: 247-255. <a href="https://doi.org/10.1111/jon.13079">https://doi.org/10.1111/jon.13079</a></p><p dir="ltr">II. Impact and correction of gadolinium-based contrast agent effects on clinical T1-weighted brain MRI volumetrics in multiple sclerosis. <b>Cananau C,</b> Ouellette R, Fredrikson S, Fink K and Granberg T [Submitted]</p><p dir="ltr">III. Automated Detection and Segmentation of Contrast-Enhancing Lesions in Multiple Sclerosis on Brain MRI: Development of a new Deep Learning Method with Validation in a Clinical Cohort. <b>Cananau C,</b> Ouellette R, Fredrikson S, Fink K and Granberg T [Manuscript]</p><p dir="ltr">IV. Dilated Virchow-Robin spaces are a marker for arterial disease in multiple sclerosis. Ineichen BV, <b>Cananau C,</b> Platten M, Ouellette R, Moridi T, Katrin B. M. Frauenknecht KBM, Okar SV, Kulcsar Z, Kockum I, Piehl F, Reich DS and Granberg T eBioMedicine - The Lancet. 2023;92: 104631 <a href="https://doi.org/10.1016/j.ebiom.2023.104631" rel="noreferrer" target="_blank">https://doi.org/10.1016/j.ebiom.2023.104631</a></p>