Imaging cortical multiple sclerosis lesions with ultra-high field MRI

Abstract

Background: Cortical lesions are abundant in multiple sclerosis (MS), yet difficult to visualize in vivo. Ultra-high field (UHF) MRI at 7 T and above provides technological advances suited to optimize the detection of cortical lesions in MS.

Purpose: To provide a narrative and quantitative systematic review of the literature on UHF MRI of cortical lesions in MS.

Methods: A systematic search of all literature on UHF MRI of cortical lesions in MS published before September 2020. Quantitative outcome measures included cortical lesion numbers reported using 3 T and 7 T MRI and between 7 T MRI sequences, along with sensitivity of UHF MRI towards cortical lesions verified by histopathology.

Results: 7 T MRI detected on average 52 ± 26% (mean ± 95% confidence interval) more cortical lesions than the best performing image contrast at 3 T, with the largest increase in type II-IV intracortical lesion detection. Across all studies, the mean cortical lesion number was 17 ± 6 per patient. In progressive MS cohorts, approximately four times more cortical lesions were reported than in CIS/early RRMS, and RRMS. Yet, there was no difference in lesion type ratio between these MS subtypes. Furthermore, superiority of one MRI sequence over another could not be established from available data. Post-mortem lesion detection with UHF MRI agreed only modestly with pathological examinations. Mean pro- and retrospective sensitivity was 33 ± 6% and 71 ± 10%, respectively, with the highest sensitivity towards type I and type IV lesions.

Conclusion: UHF MRI improves cortical lesion detection in MS considerably compared to 3 T MRI, particularly for type II-IV lesions. Despite modest sensitivity, 7 T MRI is still capable of visualizing all aspects of cortical lesion pathology and could potentially aid clinicians in diagnosing and monitoring MS, and progressive MS in particular. However, standardization of acquisition and segmentation protocols is needed.

Keywords: 7T MRI; Cortical lesions; Multiple sclerosis; Systematic review; Ultra-high field MRI.

Fig. 1

Examples of different cortical lesion types depicted on different 7 T MRI sequences. Images are from a single relapsing remitting multiple sclerosis patient (male, 42 years, EDSS = 1.5, disease duration = 14 years). Images were acquired in-house and are unpublished. Abbreviations: fluid attenuated inversion recovery, MPRAGE: magnetization prepared rapid gradient echo, T2w: T2-weighted.

Fig. 2

Prisma reporting chart. Initial screening excluded all reviews and commentaries and all publications that did not include MS, structural MRI at 7 T or above, or included only non-human populations. Further articles were excluded if they did not assess cortical lesions. All remaining studies were included in the qualitative synthesis. Further quantitative analysis was performed by four synthesis points. Note that publications may be included in more than one of the quantitative synthesis points, shown in the bottom row.

Fig. 3

Overview of studies directly comparing cortical lesion detection counts between 3 T and 7 T MRI. Data shown reflect the percent difference in lesion counts relative to the 3 T sequence that detected the highest number of cortical lesions within each study. Abbreviations: T1w = T1-weighted, T2*w = T2*-weighted, MTR = Magnetization Transfer Ratio, FLAIR = Fluid attenuated inversion recovery, T2w = T2-weighted, DIR = Double inversion recovery, Multi-S = Multiple sequences (Fartaria et al., 2017).

Fig. 4

Mean cortical lesion counts per patient. A) 7 T cortical lesion counts of the best performing sequence reported by each study, sorted by sequence. B) 7 T cortical lesion counts per phenotype. Black vertical lines denote the mean cortical lesion count per sequence, numbers denote median and range. * = mean was estimated based on median calculations as proposed by Luo et al., (2018). Abbreviations: T1w = T1-weighted, T2*w = T2*-weighted, MTR = magnetization transfer ratio, FLAIR = fluid attenuated inversion recovery, Multi-S = multiple sequences, MP2RAGE = magnetization prepared 2-rapid gradient echo, WHAT = white matter signal attenuation, RRMS = relapsing remitting multiple sclerosis, eRRMS = early RRMS, PMS = progressive multiple sclerosis (Bian et al., 2016, Herranz et al., 2016, Kuchling et al., 2014, Sinnecker et al., 2012b, Zurawski et al., 2020).

Fig. 5

Lesion type distributions detected with 7 T MRI. A) Lesion type ratio between type I, leukocortical, and type II-IV, intracortical lesions, for each study sorted by sequence. High opacity denotes type I lesions. B) Lesion type ratio between type I, type II and type III/IV for each study reporting separate type II and type III/IV lesion counts. High opacity denotes type I lesions, medium opacity denotes type II and low opacity type III/IV. C) Lesion type ratios for the three MS phenotypes. Denotation is the same as in A. Abbreviations: T1w = T1-weighted, T2*w = T2*-weighted, MTR = magnetization transfer ratio, FLAIR = fluid attenuated inversion recovery, T2w = T2-weighted, Multi-S: Multiple sequences, MP2RAGE: Magnetization prepared 2-rapid gradient echo, PMS: progressive multiple sclerosis, RRMS: relapsing remitting multiple sclerosis, eRRMS: early RRMS.

Fig. 6

Lesion type distributions from 7 T MRI and selected histological studies. A) Mean prospective and retrospective sensitivity of cortical lesion detection with ultra-high field MRI compared to histological staining. B) Cortical lesion type distributions from selected post-mortem studies. C) Mean cortical lesion type distribution from included 7 T MRI studies and D) selected post-mortem studies. E) Mean lesion type distribution for different 7 T sequences. Abbreviations: T1w = T1-weighted, T2*w = T2*-weighted, MTR = magnetization transfer ratio, FLAIR = fluid attenuated inversion recovery, T2w = T2-weighted, Multi-S: Multiple sequences, MP2RAGE: Magnetization prepared 2-rapid gradient echo, DIR = double inversion recovery, WHAT = white matter signal attenuation.