Hidden Pitfall in Multiple Sclerosis Imaging: How Standard Susceptibility-Weighted Imaging (SWI) May Miss Paramagnetic Rim Lesions

Abstract

This case report shows that paramagnetic rim lesions (PRLs), markers of chronic active lesions in multiple sclerosis, vary in visibility depending on scan-parameters of susceptibility-weighted imaging (SWI). Routine SWI for microbleed detection with low flip angle (FA) failed to depict PRLs, while longer TE and higher FA improved visibility. Phase images consistently visualized PRLs. These findings underscore the need to optimize TE and FA, as suboptimal SWI settings may hinder PRL detection.

Introduction

Multiple sclerosis (MS) is a chronic inflammatory demyelinating disease of the central nervous system, with some lesions exhibiting paramagnetic rim lesions (PRLs). PRLs are demyelinating plaques showing chronic activity and have been established as in vivo imaging biomarkers of MS.¹ At the 40th European Committee for Treatment and Research in Multiple Sclerosis (ECTRIMS) held in September 2024, it was announced that PRLs and the central vein sign (CVS) were scheduled to be adopted as imaging biomarkers in the proposed revision of the McDonald criteria.²

Histologically, PRLs are characterized by iron deposition and high-density accumulation of macrophage/microglia at the periphery, and a hypocellular region with complete demyelination lacking remyelination at the center. Rim-like iron accumulation results from multiple pathophysiological processes including iron release from oligodendrocyte destruction and hemoglobin degradation from microhemorrhages.³

While PRLs were initially discovered on 7T MRI, they have since been shown to be detectable on 3T and 1.5T systems, expanding the possibilities for clinical application and enabling evaluation in routine clinical practice.⁴

According to a meta-analysis, the patient-level prevalence of PRLs ranges from 26% to 57%, accounting for 7%–14% of all MS lesions at the lesion level.⁵ With a diagnostic specificity of 99%, PRLs serve as important findings for differentiation from related demyelinating disorders such as neuromyelitis optica spectrum disorder and myelin oligodendrocyte glycoprotein antibody-associated disease.⁶

A strong positive correlation exists between PRL count and disability progression. A 10 year observational study showed that patients with PRLs had higher relapse rates, worse Expanded Disability Status Scale scores after 10 years, higher rates of conversion to secondary progressive MS, and a greater risk of progression independent of relapse activity episodes compared to patients without PRLs.⁷ Furthermore, PRL-positive cases have shown significant correlations with clinical and biological markers including decreased information processing speed, brain volume reduction, and cerebrospinal fluid abnormalities.⁸ Additionally, disappearance of PRL during treatment has been reported to correlate with decreased confirmed disability progression rates.⁹ As the presence of PRL can serve as a risk indicator for future clinical deterioration, early PRL detection may inform decisions for aggressive therapeutic intervention, underlining the importance of accurate diagnosis using MRI.

Imaging diagnosis of PRL requires meeting the following criteria: (1) Presence of discrete rims with paramagnetic properties on susceptibility-weighted MRI sequences at 1.5T or higher. This rim must be continuous over at least two-thirds of the white matter portion of the lesion’s outer edge (excluding boundaries with cortex or ependyma) on the slice where it is most clearly depicted. (2) The rim must exist at the same location as the margin of all or part of the lesion core showing high signal on T2-weighted images. (3) The rim must be identifiable on at least 2 consecutive slices (2D imaging) or 2 orthogonal sections (3D imaging), (4) There is no enhancement on post-contrast T1-weighted images.¹⁰

Susceptibility-weighted imaging (SWI) is an MRI technique that utilizes differences in tissue magnetic susceptibility and is excellent for visualizing intracerebral iron deposits, microhemorrhages, and venous structures. The iron-rich rim portion of PRL is clearly depicted as a result of phase difference on phase images from 3D gradient echo (GRE) sequences and visualized as a low-signal rim on SWI images.¹¹

Quantitative susceptibility mapping (QSM) is known as an alternative PRL detection method to SWI. QSM is a technique that quantitatively reconstructs susceptibility distribution from multi-TE GRE phase data, directly depicting iron-deposited portions of MS lesions as high susceptibility. Previous studies have reported that QSM detects PRL with higher agreement and sensitivity than SWI.¹² However, QSM reconstruction requires advanced image processing techniques and currently remains at the research stage; therefore, SWI is still widely used in clinical practice.

As described above, accurately diagnosing PRL is important from the perspectives of diagnosis, severity assessment, and treatment response evaluation. Although QSM shows promise as an effective technique, few clinical facilities currently have the capability to implement it in routine practice. In contrast, while SWI offers easier acquisition, researchers have yet to establish the optimal imaging parameters for visualizing PRL.

This report presents differences in PRL visualization on SWI acquired under 2 different imaging conditions and discusses optimal acquisition conditions.

Case Presentation

A 50 year-old woman, 10 years after MS diagnosis, with known presence of PRL lesions, underwent routine MRI follow-up. During this examination on a new 3T scanner (Canon Vantage Centurian), SWI imaging was performed under 2 different conditions. The first sequence (Condition A) utilized TR 29 ms, TE 20 ms, and flip angle (FA) 10° that had been routinely used for detecting microbleeds, while the second sequence (Condition B) used TR 32 ms, TE 23 ms, and FA 20°, which were set to show higher T2* weighting and T1-weighting, compatible with prior studies for investigating PRLs.^4,13 Both sequences shared common parameters including slice thickness of 1.5 mm, pixel spacing of 0.36 × 0.36 mm, and matrix size of 640 × 640.

MRI revealed 2 hyperintense lesions on fluid-attenuated inversion recovery (FLAIR) images located in the right periventricular white matter (Fig. 1). Under condition A, paramagnetic rims were not clearly visible in either lesion on SWI (Fig. 1). However, under condition B, low signal areas were observed surrounding the lesion margins, which were interpreted as PRL rims (Fig. 1). Phase images demonstrated clear identification of PRL lesions under both imaging conditions (Fig. 1). Prior MRI performed 6 months earlier revealed identical lesions with similar rim characteristics on both SWI and phase images (Fig. 1: acquired using SIEMENS Skyra 3T: TR 28 ms, TE 20 ms, FA 15°, slice thickness 1.5 mm, pixel spacing 0.36 × 0.36 mm, matrix 640 × 640).

Fig. 1.

Comparison of PRL visualization in SWI under different imaging conditions. (a) SWI under condition A (TE 20 ms, FA 10°): Rim structure is unclear. (b) Phase image under condition A: Rims are clearly visualized around the 2 lesions (arrows). (c) SWI under condition B (TE 23 ms, FA 20°): Clear low-signal rims around the lesions. (d) Phase image under condition B: Rims are clearly visualized around the lesions similar to condition A. (e) FLAIR image: Lesions depicted as high-signal areas. (f) SWI acquired 6 months prior: Rim structure is clearly depicted. (a–e) Canon Vantage Centurian 3T, slice thickness 1.5 mm, pixel spacing 0.36 × 0.36 mm, matrix size 640 × 640. (f) SIEMENS Skyra 3T, TR 28 ms, TE 20 ms, FA 15°, slice thickness 1.5 mm, pixel spacing 0.36 × 0.36 mm, matrix size 640 × 640.FLAIR, fluid-attenuated inversion recovery; PRL, paramagnetic rim lesion; SWI, susceptibility-weighted imaging.

Discussion

In the current case, we observed clear differences in PRL lesion visualization on SWI images acquired under different imaging conditions at the same time point in the same patient. Specifically, PRL was clearly depicted under conditions with larger FA (20°) and longer TE (23 ms), while visualization was unclear with smaller FA (10°) and shorter TE (20 ms). In contrast, lesions were clearly depicted on phase images under both conditions.

Slightly prolonging TE enhances T2* weighting, allowing greater signal loss from iron-induced field inhomogeneities, while increasing FA from 10° to 20° increased T1-weighting, reducing the signal of the entire lesion. Disadvantages of prolonged TE can include lower SNR and unwanted susceptibility artifacts, such as those arising from the skull base, but none appeared to be prominent within the TE range examined in this study. Previous studies have also reported that longer TE was useful for detecting CVS in MS.¹⁴ This also suggests that imaging with longer TE contributes to accurate MS diagnosis.

Phase images directly reflect differences in phase change due to susceptibility. In this case, the rim structure, which was unclear on the SWI acquired under condition A, was clearly depicted on the corresponding phase image. This finding suggests that phase images are indispensable for complementing information that may be missed or remain ambiguous on SWI alone, consistent with previous studies.¹⁵ However, because anatomical structures are poorly visualized on phase images, it is necessary to interpret them in conjunction with SWI and FLAIR images. Moreover, phase images may produce false positives. Absinta et al. reported a case where a lesion judged to have a paramagnetic rim on 3T was judged to have no paramagnetic rim on 7T, and they speculated that the vascular arrangement at the lesion margin appeared rim-like.¹⁵

Limitations of this case include that multiple parameters such as FA, TE, and TR were changed simultaneously, making it unclear which factor was the main cause of image differences. Furthermore, the SWI implementations provided by major MR vendors differ from the original method proposed by Haacke,¹¹ and, strictly speaking, the images obtained in this case using Canon’s MR system should be classified as an SWI-like sequence.¹⁶ Because of these inter-vendor differences in processing, further vendor-specific validation will be required to identify the optimal conditions for PRL depiction. Also, as this is a single case report, patient- and lesion-specific factors cannot be ruled out. Future systematic comparative studies of imaging conditions in multiple patients are needed.

The imaging findings presented in this case provide valuable insights into the previously insufficiently studied effects of SWI acquisition conditions on PRL visualization; suitable TE and FA potentially improve PRL visualization on SWI. Additionally, the combined interpretation of both SWI and phase images is essential for maximizing sensitivity. Optimization of these imaging methods is expected to improve detection of PRL.