Methods

Literature search, study selection, and data extraction

This meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [11]. A systematic literature search was performed in PubMed in December 2025 to identify studies investigating PRLs and/or CVS in POMS. The search strategy combined terms related to MS, pediatric or early-onset populations, and susceptibility-based MRI markers of PRLs and CVS (eTable 1). Google Scholar, Embase, Web of Science and Cochrane were additionally screened as supplementary sources.

Eligible studies were original research articles that (1) enrolled patients with POMS and (2) reported data on PRL and/or CVS prevalence. Case reports, conference abstracts, and non-English-language publications were excluded. For studies including mixed pediatric and adult cohorts, only studies reporting separate data for the pediatric population were considered eligible. The records were screened by title and abstract, followed by full-text assessment for eligibility. Study selection was performed independently by two reviewers (A.d.M. and A.C.), with discrepancies resolved by consensus. The study selection process is summarized in the PRISMA flow diagram (Fig. 1).

Fig. 1: PRISMA flow diagram of study selection. Flow diagram illustrating the identification, screening, eligibility assessment, and inclusion of studies, with reasons for exclusion reported at each stage

The same two reviewers independently extracted study-level data into a standardized database, with final entries agreed upon by consensus. When studies included additional populations in comparison with POMS, data from these populations were also extracted.

Study screening and data extraction were conducted using Rayyan [12]. Risk of bias and applicability concerns were assessed using the QUADAS-2 tool (eTable 2). A review protocol was not prospectively registered.

Outcomes

For PRLs, the primary outcome was the proportion of patients with ≥1 PRL. For CVS, outcomes included (1) the mean proportion of CVS-positive lesions per patient and (2) the proportion of patients meeting the 40%-CVS-positive lesion threshold, which was selected for quantitative synthesis because it was the most frequently reported threshold across the included studies. Additional extracted data included clinical and demographic characteristics, MRI acquisition protocols, MRI analysis criteria, fulfillment of the Select-6 and 50% CVS-positive thresholds, number of raters, and measures of intra- and inter-rater agreement, when reported.

Statistical analysis

All analyses were performed using R (v. 4.3.1). Meta-analyses were conducted using random-effects models to account for between-study heterogeneity.

The proportion of individuals exhibiting ≥1 PRL and fulfilling the 40%-CVS rule was synthesized using meta-analyses of proportions with logit transformation and inverse-variance weighting. Between-study variance was estimated using restricted maximum likelihood, and confidence intervals (CIs) were calculated using the Hartung–Knapp method, which provides more conservative inference when the number of studies is limited [13]. Sensitivity analyses included the Freeman–Tukey double arcsine transformation as an alternative to the logit transformation and leave-one-out analyses to assess the influence of individual studies.

The proportion of CVS-positive lesions per patient was pooled using a meta-analysis of means. When studies reported medians with corresponding measures of dispersion, values were approximated to means and standard deviations to allow harmonization across studies [14].