**4. Discussions**

Cognitive issues in pediatric MS have become a research priority in this field, and more studies surrounding cognitive evaluation for these patients were published over the past decade. Although routine cognitive screening is recommended for pediatric MS [26], and cognition has been incorporated into disease activity measure and treatment consideration [5,27,28], the best assessment tools for pediatric MS remain to be determined. Findings yielded by commonly used tools were sometimes discrepant across studies. For instance, SDMT has been recommended and widely used as a screening tool in adult and pediatric MS [26,29], whereas its sensitivity in pediatric MS was occasionally challenged [17,21,30,31], particularly in early stage of the disease. Brief international cognitive assessment for multiple sclerosis (BICAMS) and a Cogstate brief battery were shown to exhibit comparable sensitivity in detecting cognitive impairment in POMS [19]. These two batteries were subsequently employed in a study investigating the neuroanatomical correlates of cognitive impairment in POMS; however, the findings cast doubt on the discriminative power of BICAMS [31]. Trail making test (TMT)-B performance more readily di fferentiated between POMS and controls compared to TMT-A in earlier studies [13,30], whereas the reverse was found in later studies [21,32]. Bartlett et al. reported that verbal memory as assessed by Rey auditory verbal learning test (RAVLT) was impaired in POMS [31], while Storm Van's Gravesande et al. found no significant verbal memory deficit in POMS using Verbaler Lern- und Merkfähigkeitstest (the German version of RAVLT) [21]. Discrepant results were also reported with regard to visuospatial memory tapped by brief visuospatial memory test-revised (BVMTR) versus Rey–Osterrieth figure tests [15,21]. Collectively, these reports underscore the need for more systematic evaluation on the selection and performance of neuropsychological tests in pediatric patients with MS.

Some of the discrepancies mentioned above could be due to di fferences in patient characteristics across studies. For example, the average EDSS score of subjects involved in evaluating multiple sclerosis inventory of cognition for adolescents (MUSICADO) was only 0.65, which might explain why verbal and visuospatial memory were preserved in POMS compared to controls in that study [21]. Indeed, Amato et al. reported that di fferences in verbal and visuospatial memory performance between POMS and healthy controls were inconspicuous at baseline [30], ye<sup>t</sup> became significant at five-year follow-up [9]. Given that psychometric properties could be disease stage-dependent, and floor or ceiling e ffects could occur, longitudinal studies may be required to delineate the performance of various assessment tools along the disease course. This dimension has been less addressed so far, and existing studies were mostly cross-sectional in design.

Another issue concerns the validation of cognitive assessment tools in di fferent populations and language versions, as was being performed for application of BICAMS in adults [33,34]. This is particularly important for pediatric populations, as many of the cognitive assessment tools have been shown to be age-sensitive during preteen to adolescent periods [14,15,17,21], and the relationship between age and cognitive development may be nonlinear [16]. In addition, most of existing studies were performed in North America and Europe, where some ethnic groups may be underrepresented [15]. Therefore, validation of the tools, including establishment of the norms, remain to be carried out in di fferent populations.

Fatigue is relatively common in POMS, and it could be an important confounder for executive function or other aspects of cognitive performance in these patients [21,30,35–38]. Fatigue also poses practical limitation on the duration of assessment for these patients. It is intriguing to note that in the study evaluating computerized version of SDMT, POMS patients exhibited an "inverted U" pattern of performance over successive trials, in contrast to the progressive improvement observed in healthy controls [17]. This suggests a time-on-task e ffect, a psychological construct related to cognitive fatigue, which was often investigated using questionnaire [21,35]. A psychometric analysis of time-on-task effect in pediatric MS patients using cognitive tests involving repeated tasks (such as reaction time tasks) may deserve further research, as it could provide a complementary indicator of cognitive fatigue. The intra-individual variability, another neuropsychological metric of white matter pathology, could

also be explored using these tasks [39]. Overall, more studies are needed to clarify whether and how fatigue and cognitive performance interact in pediatric MS.

On the other hand, our review shows that dedicated studies concerning cognitive rehabilitation for POMS remain scarce. A search for relevant clinical trials (Table 4) also showed a paucity of research specifically focusing on cognitive issues for pediatric MS, though this could be an underestimation because some trials may be retrospectively registered. Most current trials of cognitive rehabilitation for MS aim exclusively at adult populations [40,41]. We identified only three published studies evaluating effects of cognitive rehabilitation for POMS, two of which aimed to improve working memory and one targeted attention [22–24]. It is remarkable that all of these studies used home-based computerized programs. Although group- or institution-based rehabilitation may have merits in some circumstances, there appears a trend toward a more flexible and easy-to-access way of cognitive remediation, and these preliminary results seemed encouraging with regard to feasibility and patient satisfaction.

Concerning the impact of cognitive rehabilitation for POMS, these studies also showed promising results. More or less improvements in targeted cognitive domains were reported in all three studies, although discrepancy was noted between subjective and objective measures [23]. Given the heterogeneity and limited number of studies and their small sample size, no recommendation can be made for specific type of cognitive rehabilitation for POMS. It is noteworthy that the study on attention training showed far transfer e ffect [24], which is plausible because di fferent facets of cognition could a ffect one another. These cross-modal e ffects deserve more exploration in future studies. Two studies of cognitive rehabilitation included neuroimaging evaluation [22,23], and one of them found correlation between working memory network activation and behavioral response [22], providing preliminary evidence that functional training is viable. Admittedly, more research is needed to resolve the controversial issue of functional training versus strategy training in cognitive rehabilitation for pediatric MS [42,43]. Sophisticated neuroimaging techniques may help to answer this question, as well as to clarify whether and how neuroplastic changes are facilitated by rehabilitation [44].

This scoping review has some limitations. First, we do not address social cognition, which appears more dissociable from other aspects of cognition and requires a separate approach [45,46]. Second, we do not examine studies of exercise training and its cognitive e ffects in pediatric patients with MS. There have been suggestions that physical activity may exert beneficial e ffects on cognition for both youth and MS patients, although more research in this direction is needed [47–49]. Third, given that our focus is on the cognitive assessment tools, we do not include studies aiming to characterize the cognitive profile of POMS. Nonetheless, we should acknowledge that in a broad sense many of those studies also contributed supportive evidence for the validity of various assessment tools.
