Current Status of Oral Disease-Modifying Treatment Effects on Cognitive Outcomes in Multiple Sclerosis: A Scoping Review

Introduction. Cognitive impairment represents one of the most hidden and disabling clinical aspects of multiple sclerosis (MS). In this regard, the major challenges are represented by the need for a comprehensive and standardised cognitive evaluation of each patient, both at disease onset and during follow-up, and by the lack of clear-cut data on the effects of treatments. In the present review, we summarize the current evidence on the effects of the available oral disease-modifying treatments (DMTs) on cognitive outcome measures. Materials and Methods. In this systematised review, we extract all the studies that reported longitudinally acquired cognitive outcome data on oral DMTs in MS patients. Results. We found 29 studies that evaluated at least one oral DMT, including observational studies, randomised controlled trials, and their extension studies. Most of the studies (n = 20) evaluated sphingosine-1-phosphate (S1P) modulators, while we found seven studies on dimethyl fumarate, six on teriflunomide, and one on cladribine. The most frequently used cognitive outcome measures were SDMT and PASAT. Most of the studies reported substantial stability or mild improvement in cognitive outcomes in a short-time follow-up (duration of most studies ≤2 years). A few studies also reported MRI measures of brain atrophy. Conclusion. Cognitive outcomes were evaluated only in a minority of prospective studies on oral DMTs in MS patients with variable findings. More solid and numerous data are present for the S1P modulators. A standardised cognitive evaluation remains a yet unmet need to better clarify the possible positive effect of oral DMTs on cognition.


Introduction
Multiple sclerosis (MS) is a chronic inflammatory demyelinating disease of the central nervous system, characterised by progressive neurodegeneration with neuroaxonal damage leading to pronounced brain matter atrophy [1,2]. Cognitive impairment (CI) is one of the clinical features of MS with a variable prevalence in the adult MS population ranging between 34% and 65%. CI is also observed in paediatric MS, in clinically isolated syndrome (CIS), and in radiologically isolated syndrome (RIS) [1][2][3]. These findings highlight the importance of CI as an early indicator of disease activity that sometimes may anticipate the classical clinical manifestations [4][5][6]. Longitudinal studies have shown that CI progresses over time in people with MS [7,8], and it has been observed that cognitive disorders have a higher prevalence and severity in progressive MS [9,10]. Not all patients show the same rate of progression of CI. Age at onset (younger or older), higher lesion load on a brain MRI, and higher rate of brain atrophy within the first two years after MS diagnosis represent some of the worst prognostic factors for CI [7,[11][12][13].

Results
The keywords used for the search were (multiple sclerosis) AND (fingolimod ozanimod OR siponimod OR teriflunomide OR cladribine OR dimethyl fumarate) A (cognit*). Out of the 818 detected articles, 715 were excluded for ineligibility concerning cognition; animal studies; not concerning oral DMTs; reviews; case repo book chapters; editorials). 103 articles were then analysed, of which 75 were exclu because they did not provide sufficient data on cognitive assessments. Therefore studies were included in this review ( Figure 1). We found 29 studies that longitudinally evaluated CI in MS patients treated with oral DMT. The vast majority (n = 20) provided data on S1P modulators [fingolimod (F ozanimod (OZN) and siponimod (SIP)], while we found six studies with data We found 29 studies that longitudinally evaluated CI in MS patients treated with an oral DMT. The vast majority (n = 20) provided data on S1P modulators [fingolimod (FTY), ozanimod (OZN) and siponimod (SIP)], while we found six studies with data on teriflunomide (TFL), seven on dimethyl fumarate (DMF), and one on cladribine (CLAD). Most of the studies (n = 25) evaluated a single oral DMT, while three papers included patients taking two different oral DMTs (1 FTY and DMF; 1 FTY and TRF; 1 FTY and SIP) and one evaluated patients on FTY, DMF, TFL, and CLAD.

Sphingosinte-1-Phosphate Modulators
The majority of studies providing data on S1P modulators concerned FTY (n = 16), while a minority analysed OZN (n = 2) and SIP (n = 2), with one of these studies collecting data on both FTY and SIP ( Table 1). The reviewed papers came from observational prospective studies (n = 13), randomised clinical trials, and their extension studies (n = 7) with a median duration of 21 months (range 6-120). The total number of patients was 4935 (2765 on FTY, 1272 on SIP, and 898 on OZN), of which 3360 had relapsing remitting MS (RRMS), 303 had primary progressive MS (PPMS), and 1272 had secondary progressive MS (SPMS). The median number of enrolled patients was 84 with a range of (8-1575). The most frequently used cognitive assessment tools were the SDMT (in five papers as the sole assessment tool, and in seven together with other tests) and the PASAT (alone in four papers, and in combination with other tests in 11). In all patients from baseline to 1.8 years WM volume, peripheral GM and whole brain volume decreased significantly.  In 16 studies (80%), there was a comparator, which was interferon in 7 studies (35%), placebo in 5 (25%), while in one study a pool of healthy controls was used as a comparator [55].
In 14 studies (70%), patients treated with S1P modulators showed a significant improvement in cognitive performance over time. Three of these studies [40,49,53] compared S1P modulators with placebo and reported a significant difference between groups, with S1P modulator-treated patients performing better than placebo. Four studies reported a better performance of S1P modulators compared to an injectable DMT (two studies with IFN; one with GA; and one with both IFN and GA) [39,43,46,50]. Other DMTs were used as comparisons such as Natalizumab (NTZ) (three studies) [41,47,56], Alemtuzumab (ATZ) (one study) [56], DMF (two studies) [56,57], and TFL (two studies) [45,56]. In these studies, no significant differences were found with these DMTs, with the exception of the study by Glasmacher et al. [56] where TFL and DMF were associated with significant worsening over time in cognitive assessment scores compared to FTY and CLAD.
MRI data on brain volume loss (BVL) were collected in 12 studies. Compared to placebo, S1P modulator treatment demonstrated a significant reduction in BVL in all of these studies. Regarding a direct comparison with other DMTs, three papers [39,42,43] found a lower BVL in the S1P modulator group compared to IFN and/or GA, while two other comparative studies vs. GA and NTZ, respectively [46,47], did not find any significant difference.
Moreover, two papers found that a lower BVL during S1P modulator treatment correlated with better cognitive test scores [53,55]. Two studies were mainly focused on thalamic atrophy, demonstrating that patients treated with S1P modulators have a smaller reduction in thalamic volume compared to placebo [33], which could also be predictive of CI [55].
Only two studies include data on neurofilaments (sNFL) [54,57]. In one of these studies, higher levels of sNFLs correlated with greater CI and BVL, and patients treated with S1P modulators had a significant reduction in sNFL levels compared to placebo [54].
Three studies included only patients exposed to DMF, while two studies also included other DMTs. The remaining two studies had a placebo control group and a healthy subject's control cohort.
Observation of the data shows that in four studies, patients presented a longitudinal improvement in cognitive performance over time [58,[60][61][62], while in two studies, cognitive assessment remained stable with no significant differences vs. FTY [57,59]. On the contrary, there was a significant worsening of PASAT in the study by Glasmacher et al., where DMF was analysed together with TFL as Category 3 vs. Category 1 (NTZ and ATZ) and Category 2 (FTY and CLAD) DMTs [56].
MRI data on BVL came from only one prospective study without comparator where there was a percentage reduction in brain volume from a baseline of 0.12% at 6 months and 0.24% at 1 year [62]. One study also collected data on sNFLs, showing higher levels in FTY-treated patients. sNFL levels were assessed with an average timeframe of 7.1 months from the first study enrolment visit, and there was no follow-up data to assess the trend of sNFLs over time [57].

Teriflunomide
Six studies evaluated clinical cognitive outcomes in MS patients during TFL treatment. Most of the papers included were observational prospective studies (n = 5), but there was also a post hoc analysis of the RCT TEMSO and its extension [63] with a median duration of 24 months (range 12-58) ( Table 3). There were a total of 1020 patients, with a median of 88 (range 4-594). All enrolled patients were diagnosed as RRMS. SDMT was the only cognitive assessment in two studies, while the other four used PASAT (n = 2), Brief International Cognitive Assessment for MS (BICAMS) (n = 1), and Rao's BRB (n = 1). Two studies evaluated patients on TFL treatment longitudinally without any comparator group. Coyle et al. reported substantial stability in the SDMT score during a two-year follow-up in a large cohort of patients, while Bencsik et al. found a mild improvement in BICAMS at 12-and 24-month evaluations [64,66].
The remaining four studies also evaluated patients who did not undertake TFL. One study included a healthy subject control group [65], while another included a placebo control group [63]. The last two studies included patients on other DMTs [45,56].
In the TEMSO RCT and its extension study, a mild improvement in PASAT Z score was found over time. Moreover, patients who first started TFL in the core phase of the trial better performed in PASAT 3 than patients who switched to TFL in the extension phase [63].
In their study, Corallo et al. found that the Rao's BRB scores remained stable over time in TFL-treated patients [65].
Regarding the potential comparison with other DMTs, Glasmacher et al. reported that TFL and DMF treatment was associated with a significant worsening of the PASAT score over time compared to the higher efficacy DMTs [56], whereas in the prospective study by Guevara et al.,8/45 patients of the full cohort (only four patients in TFL cohort) had a SDMT that decreased 4 points or more at year two [45]. However, TFL sample size was very low in both studies, and the reported data do not separate individual treatments.
An analysis of BVL was performed in three studies [45,63,65]. In the study by Guevara et al., a significant reduction from baseline in white matter (WM) volume, peripheral gray matter (GM), and whole brain volume was observed, without significant inter-group (different DMTs) differences [45]. Another study showed a mild increase in GM volume compared to the baseline in TFL-treated patients [65], whereas in the TEMSO study and its extension, lower BVL on TFL treatment correlated with higher PASAT-3 scores [63].

Cladribine
The data on Cladribine (CLAD) came from a single study. Three categories of DMTs were distinguished in the study design, where Category 1 included NTZ and ATZ, Category 2 included FTY and CLAD, and Category 3 included DMF and TFL. The total number of study subjects treated with CLAD was 13, and analyses of the results were performed per category of DMT. Category 2 (FTY and CLAD) was associated with significant improvement in the PASAT scores, while Category 3 was associated with significant worsening. Stability in the scores was seen in those treated with Category 1 DMTs [56].

Discussion
The focus on the non-motor symptoms of MS has grown considerably in recent years, as there has been an increased awareness of their impact on the daily-living activities of people with MS (PwMS). Certainly, cognitive disorders represent one of the most difficult challenges for patients and clinicians. The assessment of cognitive disorders in MS comprises several reliable and valid assessment tools. Among these, the Minimal Assessment of Cognitive Function in MS (MACFIMS) allows for a complete neuropsychological assessment of cognitive disorders associated with MS, although it has the disadvantage of a more time-consuming administration [67]. Other more rapidly used instruments, such as the Brief Repeatable Neuropsychological Battery (BRNB) [3], the Brief International Cognitive Assessment in Multiple Sclerosis (BICAMS) [68], and the Symbol Digit Modalities Test (SDMT) alone, on the other hand, do not allow for a complete neuropsychological assessment but can evaluate the cognitive functions most frequently affected in MS. BICAMS and SDMT have the shortest administration durations, offering an excellent balance between duration and clinical validity. Cognitive assessments with validated tests should be performed at baseline and once a year thereafter with the same assessment tool in all patients with clinical and radiological evidence consistent with MS [69].
MS patients with significant CI have less engagement in social activities, lower employment rates, and an increased risk of developing psychiatric diseases [20], highlighting the clinical and socio-economic importance of cognitive involvement as a marker of disease severity [89].
The amount of data on CI derived from observational studies and RCTs is constantly growing. The available DMTs for MS treatment have important anti-inflammatory properties with less pronounced neuroprotective effects [27][28][29][30]. Although there are still insufficient data on their impact on CI, in this review we aimed to give an overview of the available data regarding the effect of oral DMTs on CI.
Considering the longer experience with this drug class, S1P modulators are covered by a significantly higher number of studies prospectively evaluating cognition in PwMS. Moreover, only S1P modulators have data on progressive MS, FTY on primary progressive MS, and SIP on secondary progressive MS, respectively [33,49,54].
PwMS treated with S1P modulators seem to have slight improvement or stability in cognitive test scores over time. Moreover, three different studies report a superiority of FTY compared to injectable DMTs [43,46,52]. No clear difference has been seen compared to other oral DMTs nor compared to NTZ or ATZ [45,56,57]. Among the S1P modulators, FTY certainly has the most evidence, since it is the drug that has been on the market the longest. The CI data for SIP and OZN are substantially derived from the EXPAND and SUNBEAM RCTs, respectively [33,39,49,50,54].
More than half of the papers on S1P modulators provide data on brain atrophy. The evidence would seem to show that in patients treated with S1P modulators, there is a reduction in BVL over time compared to placebo and IFN, while showing discordant results for GA and no difference compared to NTZ [45,46,56]. BVL also seems to correlate with cognitive performance in two of the included studies [53,55]. On the other hand, data on NFL are less abundant, mostly deriving from the INFORMS (FTY) and EXPAND (SIP) RCTs and a real-world study in which exposure to FTY and SIP would appear to be associated with a reduction in pNFL levels over time [54,57].
With regard to DMF, we retrieved data from the RCTs DEFINE and CONFIRM and six prospective studies. These studies would seem to indicate that patients treated with DMF show longitudinal improvement or stability in cognitive performance over time. However, there is a study where DMF and TFL (Category 3) were associated with significant worsening of cognitive performances vs. Category 1 (NTZ and ATZ) and Category 2 (FTY and CLAD) DMTs [56]. No significant differences were shown in the only other study analysed in this review against a DMT (FTY) [57]. Only one of the included studies also reported data on brain atrophy, which found a BVL of less than 0.4% (NEDA-4 annualised BVL cut-off) after one year of treatment [62]. The same applies to NFL, of which we have data from only one study vs. FTY where they were analysed at a single timepoint after 7 months from the start of treatment, showing higher values in the FTY group [57].
Concerning TFL, on the other hand, data are available from five prospective studies and a post hoc analysis of the RCT TEMSO and its extension study. These studies suggest mild improvement or stability in cognitive scores over time in PwMS treated with TFL. Data including an active DMT comparator come from two studies with a very low sample size, so there are insufficient data [45,56]. On brain atrophy, the data mostly come from the RCT TEMSO and its extension, where an inverse correlation between BVL and PASAT score was shown [63].
Data on CLAD, on the other hand, have only been analysed in one paper, with a small sample size (n = 13) [56]. In this paper, Glasmacher et al. grouped different DMTs together without distinguishing the individual DMTs in their analysis, thus not permitting the evaluation of each treatment on cognitive outcomes. So, no conclusions can be drawn to date on the effects of this CLAD on CI.
Although the data seem encouraging, there are several limitations in these studies. For example, the cognitive assessments used are very heterogeneous, comprising both comprehensive batteries such as Rao's BRB-NT and more rapidly used tests such as the SDMT or PASAT. The PASAT and the SDMT were the most widely used tests, and although they are validated instruments in MS, they still have limitations. In fact, both PASAT and SDMT are focused on information processing speed, which despite being the most affected cognitive domain, is not the only one [3,[19][20][21][22][23]. Other evaluation tools such as Rao's BRB-NT were used in few papers, making it difficult to compare results across studies. Moreover, another factor to consider is that these assessment tools over time may be vulnerable to the practice effect that may overestimate performances [90,91].
Papers analysed in this review are a mixture of RCTs, observational prospective studies, and post hoc analyses. Study durations were not homogeneous, ranging from 6 to 120 months, generating potential variability in the results. In several studies, patients were evaluated longitudinally without a comparator group. That is likely to limit the ability to attribute changes in cognitive outcomes directly to the treatments under investigation. In addition, due to their design, observational studies and post hoc analyses are more susceptible to bias than RCTs. In fact, the most solid data, especially in the case of S1P modulators, come to us from RCTs and their extensions, where the patients had the longest follow-up. In these studies, the presence of a selection bias could be assumed, with better responding patients having the highest probability of continuing the study. On the other hand, the study durations (in most cases, lower than 24 months) can be inadequate to demonstrate a significant beneficial effect on the CI outcomes. As already mentioned, there are to date insufficient data about oral DMTs and their impact on BVL and sNFL. It is not possible at the present time to draw definitive conclusions on these outcomes.
In view of these considerations, we can state that cognitive disorders continue to be a large terrain that has not yet been fully explored and needs further analysis. The assessment of the cognitive sphere must unquestionably become part of the routine at MS centres, by carrying out annual assessments in all patients even in the early stages of the disease [69]. When facing an MS patient with cognitive disorders, a suitable strategy could be to flank DMT with the use of symptomatic therapies, cognitive remediation, and cognitive exercises.  Data Availability Statement: The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.