Multiple sclerosis lesion detection with 3D double inversion recovery (DIR) as compared to 3D fluid low attenuation inversion recovery (T2-FLAIR): A systematic review and meta-analysis

Background: Several sequences have been developed to increase lesion detection in Multiple Sclerosis, including Double Inversion Recovery (DIR). The superiority of these pulse sequences is not well established compared to T2-FLAIR images. urpose: This meta-analysis aims to describe if lesion detection rates are higher in DIR compared to T2-FLAIR images in a 3D acquisition comparison. Data Sources: Search was performed of the PubMed/MEDLINE, EMBASE and Cochrane databases between January 1995 and December 2023. Study Selection: Studies identified were assessed independently by two physicians following PRISMA guidelines. Articles were screened to exclude duplicates, review articles, not performing direct 3D DIR and 3D T2-FLAIR comparisons and abstracts. Remaining articles were reviewed in a full-text review by two physicians independently. Data Analysis: Cortical lesion count ranged from to 12.4 to 40.0 for DIR; 5.25 to 27.9 for T2-FLAIR, with juxtacortical lesions ranging from 4.9 to 19.7 and 4.72 to 22.0 on DIR and T2-FLAIR respectively. Intracortical lesions varied from 1.2 to 8.0 for DIR and 1.1 to 3.1 for T2-FLAIR. Infratentorial lesions mean lesions count varied from 2.0 to 12.0 for DIR, as compared to 1.45 to 8.4 for T2-FLAIR. In the periventricular WM, results varied from 11.84 to 73 and 11.31 to 69 for DIR and T2-FLAIR respectively. Data Synthesis: R statistical software was used for data synthesis. Pooled estimates showed relative significant differences in lesion detection for intracortical (175.41 [95 % Confidence Interval (95 %-CI): 48.68; 410.16]) and infratentorial (30.56 [95 %-CI: 9.34; 55.91]) regions with the entire 95 % confidence intervals >0. Confidence intervals were <0 when counting differences for total cortical lesions (37.35 [95 %-CI:12.47; 115.54]), including juxtacortical (25.44 [95 %-CI:32.12; 131.81]) and for supratentorial WM lesions (1.93 [95 %-CI:14.41; 21.39]), including the periventricular WM (11.22 [95 %-CI:4.02; 28.90]). Limitations: The available number of studies was relatively low. Also, given significant heterogenicity for lesion load measurements in different patients for absolute differences, relative differences were only estimated from one study by the log-normal assumption. Conclusions: DIR acquisition allows higher detection in intracortical and infratentorial lesions compared to T2FLAIR.