Statement of research significance
Cognitive impairment significantly impacts the quality of life in people with multiple sclerosis (pwMS), yet the potential combined effects of biological sex and cognitive reserve (CR) on cognitive outcomes remain underexplored. This study provides critical insights into how sex and CR independently influence cognitive performance in pwMS, highlighting that women generally exhibit higher CR levels, particularly in daily activities, hobbies, and social life. Despite observed sex differences in specific cognitive domains—verbal memory and fluency favoring women, and processing speed favoring men—the interaction between sex and CR on cognition was not statistically significant. These findings emphasize the need for tailored psychoeducational interventions to enhance CR, potentially mitigating/reducing the risk of cognitive deficits in MS. By addressing CR in both sexes, clinicians can develop more effective strategies for cognitive preservation and rehabilitation.
Introduction
Cognitive dysfunctions in multiple sclerosis (MS) are one of the most common features of the disease (Grzegorski & Losy, Reference Grzegorski and Losy2017; Sumowski et al., Reference Sumowski, Benedict, Enzinger, Filippi, Geurts, Hamalainen, Hulst, Inglese, Leavitt, Rocca, Rosti-Otajarvi and Rao2018). People with MS (pwMS) experience more difficulties, compared with their healthy peers, in several cognitive domains, particularly in information processing speed, memory, and executive functions, with repercussions on quality of life and activities of daily living. The scientific literature on MS found that several factors may influence cognitive abilities, and most studies focused on the role of central nervous system (CNS) burden, evaluated by means of Magnetic Resonance Imaging (MRI) measures such as brain atrophy, lesion burden, or functional connectivity. For example, in a recent meta-analysis of 21 studies and more than 3000 MS patients, T1 and T2 weighted lesions strongly correlated with cognitive functions, particularly with task assessing processing speed and attention (Nabizadeh et al., Reference Nabizadeh, Pirahesh, Azami, Moradkhani, Sardaripour and Ramezannezhad2024). Longitudinal changes in T2 lesions and atrophy were also associated with changes in cognition, particularly in processing speed, attention, and verbal fluency domains (Simani et al., Reference Simani, Molaeipour, Kian and Leavitt2024). Subcortical volumetry (i.e., hippocampal and thalamic volumes) and white matter integrity were also correlated to cognitive performance in MS (Bisecco et al., Reference Bisecco, Capuano, Caiazzo, d’Ambrosio, Docimo, Cirillo, Russo, Altieri, Bonavita, Rocca, Filippi, Tedeschi and Gallo2021; Hulst et al., Reference Hulst, Schoonheim, Van Geest, Uitdehaag, Barkhof and Geurts2015; Schoonheim et al., Reference Schoonheim, Popescu, Rueda Lopes, Wiebenga, Vrenken, Douw, Polman, Geurts and Barkhof2012, Reference Schoonheim, Vigeveno, Rueda Lopes, Pouwels, Polman, Barkhof and Geurts2014; Tóth et al., Reference Tóth, Faragó, Király, Szabó, Veréb, Kocsis, Kincses, Sandi, Bencsik, Vécsei and Kincses2019), highlighting these metrics’ critical role in predicting cognitive outcomes. Furthermore, several resting-state functional connectivity studies found a correlation between cognitive impairment and disruption of several networks, such as the default mode network and the frontoparietal network, as pointed out in a recent review (Mahmoudi et al., Reference Mahmoudi, McCarthy and Nelson2025). However, a recent study (Motyl et al., Reference Motyl, Friedova, Ganapathy Subramanian, Vaneckova, Fuchs, Krasensky, Blahova Dusankova, Kubala Havrdova, Horakova and Uher2024), when further examining the relationship between CNS burden and cognitive performance, found that MRI measures only partially explain interindividual differences in cognitive outcomes, suggesting the need to explore additional factors.
As for sociodemographic factors linked to cognitive performance in MS, it has been found that biological sex may cause differences in cognition between men and women with MS, as also noted in healthy people and other neurological diseases (Cerri et al., Reference Cerri, Mus and Blandini2019; Gibson & Galea, Reference Gibson and Galea2023; Li & Singh, Reference Li and Singh2014). Some studies on the prevalence and severity of cognitive impairment in MS reported sex differences: for example, one early study on this topic (Savettieri et al., Reference Savettieri, Messina, Andreoli, Bonavita, Caltagirone, Cittadella, Farina, Fazio, Girlanda, Le Pira, Liguori, Lugaresi, Nocentini, Reggio, Salemi, Tedeschi, Trojano, Valentino and Quattrone2004) found that cognitive impairment was more frequent in men with MS and it was associated with disease duration and physical disability; however, in women, no association between cognition and clinical variables was found. Similarly, another study pointed out that men tended to exhibit a more severe cognitive dysfunction (Schoonheim et al., Reference Schoonheim, Popescu, Rueda Lopes, Wiebenga, Vrenken, Douw, Polman, Geurts and Barkhof2012). However, other studies have found no significant sex differences in the prevalence of cognitive deficits in MS (Amato et al., Reference Amato, Portaccio, Goretti, Zipoli, Ricchiuti, De Caro, Patti, Vecchio, Sorbi and Trojano2006; Beatty & Aupperle, Reference Beatty and Aupperle2002) and suggested that biological sex may influence specific cognitive domains, but not overall prevalence.
As regards differences in specific cognitive domains, scores on verbal memory were consistently and significantly higher in women with MS (when compared with their male counterparts) in several studies (Altieri et al., Reference Altieri, Santangelo, Bisecco, D’Ambrosio, Rippa, Capuano, Giuliano, Matrone, Bonavita, Tedeschi and Gallo2021; Beatty & Aupperle, Reference Beatty and Aupperle2002; Donaldson et al., Reference Donaldson, Patel, Shammi and Feinstein2019; Motyl et al., Reference Motyl, Friedova, Ganapathy Subramanian, Vaneckova, Fuchs, Krasensky, Blahova Dusankova, Kubala Havrdova, Horakova and Uher2024). However, there have also been other studies that have found no differences (Amato et al., Reference Amato, Portaccio, Goretti, Zipoli, Ricchiuti, De Caro, Patti, Vecchio, Sorbi and Trojano2006). However, the analysis of sex differences in other cognitive domains provided more discordant results, which may be ascribed to different study designs and study populations, and to the use of neuropsychological batteries in some studies that were not validated specifically for MS (Altieri et al., Reference Altieri, Santangelo, Bisecco, D’Ambrosio, Rippa, Capuano, Giuliano, Matrone, Bonavita, Tedeschi and Gallo2021; Amato et al., Reference Amato, Portaccio, Goretti, Zipoli, Ricchiuti, De Caro, Patti, Vecchio, Sorbi and Trojano2006; Beatty & Aupperle, Reference Beatty and Aupperle2002; Donaldson et al., Reference Donaldson, Patel, Shammi and Feinstein2019; Motyl et al., Reference Motyl, Friedova, Ganapathy Subramanian, Vaneckova, Fuchs, Krasensky, Blahova Dusankova, Kubala Havrdova, Horakova and Uher2024).
There is evidence of other contributing factors to cognitive impairment in MS population, namely levels of depression, anxiety and/or fatigue. It has been found that higher depressive levels may exacerbate cognitive deficits through reduced motivation, impaired attention, and slower processing speed; moreover, longitudinal studies suggested that depression can also predict cognitive decline independent of disease burden, particularly in memory and executive functions (Feinstein et al., Reference Feinstein, Magalhaes, Richard, Audet and Moore2014; Rao et al., Reference Rao, Leo, Bernardin and Unverzagt1991). On the other hand, higher anxiety levels have been correlated with lower performance on tasks assessing attentional control, processing speed, and memory (Marrie et al., Reference Marrie, Patel, Figley, Kornelsen, Bolton, Graff, Mazerolle, Marriott, Bernstein and Fisk2019, Reference Marrie, Patel, Bernstein, Bolton, Graff, Marriott, Hitchon, Figley, Kornelsen and Fisk2021; Ribbons et al., Reference Ribbons, Lea, Schofield and Lechner-Scott2017), although there are also studies that do not confirm this correlation (Gill et al., Reference Gill, Santo, Blair and Morrow2019; Sandroff et al., Reference Sandroff, Pilutti and Motl2019; Wallis et al., Reference Wallis, Bol, Köhler and van Heugten2020). Regarding fatigue, one of the most debilitating symptoms of MS, trait, and cognitive fatigue were associated with poorer performance in sustained attention and information processing speed, working memory and executive functions (Guillemin et al., Reference Guillemin, Lommers, Delrue, Gester, Maquet and Collette2022). This association may be explained by overlapping pathophysiological mechanisms underlying fatigue and cognitive impairment, suggesting shared inflammatory and neurodegenerative processes (Chitnis et al., Reference Chitnis, Vandercappellen, King and Brichetto2022).
Another factor linked to cognition in MS is disease duration (DD); indeed, longer DD has been associated with a higher prevalence of cognitive impairment, reflecting the cumulative impact of neurodegeneration and neuroinflammation of the CNS. Early in the disease course cognitive deficits may limited to specific domains such as processing speed and attention (Rao et al., Reference Rao, Leo, Bernardin and Unverzagt1991); as the disease progresses, cognitive deficits tend to be more pronounced and widespread, encompassing memory, executive function, and visuospatial abilities (Chiaravalloti & DeLuca, Reference Chiaravalloti and DeLuca2008). However, the trajectory of cognitive decline does not follow a uniform pattern across all patients, and it was theorized that CNS burden, along with levels of cognitive reserve (CR) may be responsible for these different clinical outcomes (Eijlers et al., Reference Eijlers, van Geest, Dekker, Steenwijk, Meijer, Hulst, Barkhof, Uitdehaag, Schoonheim and Geurts2018; Sumowski et al., Reference Sumowski, Rocca, Leavitt, Riccitelli, Comi, DeLuca and Filippi2013).
CR is a set of cognitively enriching life experiences (formal education, work, speaking multiple languages, leisure, and social activities) that allow coping with brain damage (Stern, Reference Stern2009, Reference Stern2012). The role of CR on cognitive performance in pwMS has been explored in several studies and meta-analyses (Artemiadis et al., Reference Artemiadis, Bakirtzis, Ifantopoulou, Zis, Bargiotas, Grigoriadis and Hadjigeorgiou2020; Estrada-López et al., Reference Estrada-López, García-Martín and Cantón-Mayo2021; Santangelo, Altieri, Enzinger, et al., Reference Santangelo, Altieri, Enzinger, Gallo and Trojano2019; Santangelo, Altieri, Gallo, et al., Reference Santangelo, Altieri, Gallo and Trojano2019), revealing an association between higher levels of CR and better cognitive performance on tests evaluating several cognitive abilities, such as memory, processing speed, attention, and some executive functions. However, most studies employed educational attainment as a proxy of CR (Luerding et al., Reference Luerding, Gebel, Gebel, Schwab-Malek and Weissert2016; Machado et al., Reference Machado, Lima, d’Almeida, Afonso, Macário, Castelo-Branco, Sousa, Santana and Batista2021; Scarpazza et al., Reference Scarpazza, Braghittoni, Casale, Malagú, Mattioli, di Pellegrino and Ladavas2013), whereas only a small portion of studies used other proxies limited to specific sets of abilities or activities (i.e., premorbid IQ, leisure activities, vocabulary knowledge) (Ifantopoulou et al., Reference Ifantopoulou, Artemiadis, Bakirtzis, Zekiou, Papadopoulos, Diakogiannis, Hadjigeorgiou, Grigoriadis and Orologas2019; Maggi, Altieri, Risi, Rippa, Borgo, Lavorgna et al., Reference Maggi, Altieri, Risi, Rippa, Borgo, Lavorgna, Bonavita, Buonanno, D’Ambrosio, Bisecco, Santangelo, Tessitore and Gallo2024; Maggi, Altieri, Risi, Rippa, Borgo, Sacco, et al., Reference Maggi, Altieri, Risi, Rippa, Borgo, Sacco, Buonanno, D’Ambrosio, Bisecco, Santangelo and Gallo2024; Sumowski et al., Reference Sumowski, Wylie, Gonnella, Chiaravalloti and Deluca2010). These single indexes may not be able to fully explore engagement in a wide range of cognitively enhancing activities across several life stages; therefore, in recent years, researchers tried to produce questionnaires to evaluate a multitude of stimulating activities in adulthood, such as the Cognitive Reserve Scale (CRS) (Leon-Estrada et al., Reference Leon-Estrada, Garcia-Garcia and Roldan-Tapia2017), that explores the engagement of the individual on several sets of activities in various stages of adult life.
Several studies have highlighted the relationship between CR and the other abovementioned variables associated with cognitive performance in MS. For example, individuals with higher CR reported higher cognitive performance despite significant grey matter atrophy (Sumowski et al., Reference Sumowski, Rocca, Leavitt, Riccitelli, Comi, DeLuca and Filippi2013); moreover, there is evidence that CR may moderate the impact of subcortical grey matter atrophy on cognitive status (Machado et al., Reference Machado, Lima, d’Almeida, Afonso, Macário, Castelo-Branco, Sousa, Santana and Batista2021). Regarding the association between CR and functional connectivity, CR was associated with preserving brain network efficiency, providing a compensatory mechanism that buffers the effects of CNS burden on cognitive performance (Fuchs et al., Reference Fuchs, Benedict, Bartnik, Choudhery, Li, Mallory, Oship, Yasin, Ashton, Jakimovski, Bergsland, Ramasamy, Weinstock-Guttman, Zivadinov and Dwyer2019; Leavitt et al., Reference Leavitt, Dworkin, Kalina and Ratzan2024; Lopez-Soley et al., Reference Lopez-Soley, Solana, Martínez-Heras, Andorra, Radua, Prats-Uribe, Montejo, Sola-Valls, Sepulveda, Pulido-Valdeolivas, Blanco, Martinez-Lapiscina, Saiz and Llufriu2020). Resting-state functional MRI studies revealed that pwMS with high CR had reduced intrinsic functional connectivity within the dorsal anterior insula, a key hub of the salience network (Bizzo et al., Reference Bizzo, Arruda-Sanchez, Tobyne, Bireley, Lev, Gasparetto and Klawiter2021).
CR has also been examined in relation to psychological factors. For example, higher CR was linked to lower levels of depression (Artemiadis et al., Reference Artemiadis, Bakirtzis, Ifantopoulou, Zis, Bargiotas, Grigoriadis and Hadjigeorgiou2020); moreover, it may moderate the relationship between physical disability and depressive symptoms, suggesting that individuals with higher CR are less likely to experience depression even when they have a significant disability (Cadden et al., Reference Cadden, Guty and Arnett2019). Similarly, Bradson and colleagues (Bradson et al., Reference Bradson, Cadden, Riegler, Thomas, Randolph and Arnett2023) reported that CR seems to moderate the relationship between fatigue and depressive symptomatology, suggesting that CR may play a protective role against the psychological burden of MS.
Despite these advancements, the relationship between sex, CR and cognition (i.e., sex differences on the level of engaging in several sets of cognitively stimulating activities linked to CR), or how sex and the CR have an impact on cognition were not fully investigated. Considering the above-mentioned studies and the evidence in other conditions (e.g., subjective cognitive decline, Alzheimer’s disease) that show sex differences in CR (Giacomucci et al., Reference Giacomucci, Mazzeo, Padiglioni, Bagnoli, Ferrari, Bracco, Nacmias, Sorbi and Bessi2021; Letenneur et al., Reference Letenneur, Launer, Andersen, Dewey, Ott, Copeland, Dartigues, Kragh-Sorensen, Baldereschi, Brayne, Lobo, Martinez-Lage, Stijnen and Hofman2000), there is a need to fill the abovementioned gaps in literature.
Therefore, in a sample of pwMS, the present study aimed to explore sex differences in engaging in cognitively stimulating activities associated with CR, and to investigate the possible effect of sex and CR on performance on several cognitive domains.
Materials and methods
Participants
pwMS were enrolled at the MS Center of the I Division of Neurology, University of Campania “Luigi Vanvitelli”, Naples, Italy. We performed a consecutive sampling, in which every eligible participant who met the inclusion and exclusion criteria was enrolled in the study without omission, provided they were available during the recruitment period. To be included in the study, all participants had to meet the following inclusion criteria: a) diagnosis of MS according to the latest McDonald diagnostic criteria (Thompson et al., Reference Thompson, Banwell, Barkhof, Carroll, Coetzee, Comi, Correale, Fazekas, Filippi, Freedman, Fujihara, Galetta, Hartung, Kappos, Lublin, Marrie, Miller, Miller, Montalban and Cohen2018) and relapsing-remitting MS phenotype; b) age above 18 years old; c) no relapses or corticosteroids therapy within 1 month from the evaluation; d) no neurological comorbidities; e) no assumption of psychotropic drugs that could interfere with the neuropsychological examination. All participants were informed about the purposes of the study and signed an informed consent form before proceeding to the clinical and neuropsychological evaluation. The research was approved by the Local Ethics Committee and performed following the Declaration of Helsinki.
Clinical and neuropsychological evaluation
Each participant was evaluated by a trained neurologist with expertise in MS, who collected clinical data (i.e., disease duration, disability, disease modifying therapies); moreover, the Expanded Disability Status Scale (EDSS) (Kurtzke, Reference Kurtzke1983) was administered to evaluate the degree of physical disability. The Beck Depression Inventory, second edition (BDI-II) was completed by each participant to evaluate levels of depressive symptomatology. Moreover, a trained psychologist collected sociodemographic data (i.e., age, educational status, work activity based on the International Standard Classification of Occupations) and administered Rao’s Brief Repeatable Battery of Neuropsychological Tests (BRB-N) to assess cognitive performance (Amato et al., Reference Amato, Portaccio, Goretti, Zipoli, Ricchiuti, De Caro, Patti, Vecchio, Sorbi and Trojano2006; Rao et al., Reference Rao, Leo, Bernardin and Unverzagt1991; Tedone et al., Reference Tedone, Vizzino, Meani, Gallo, Altieri, D’Ambrosio, Pantano, Piervincenzi, Tommasin, De Stefano, Cortese, Stromillo, Rocca, Sala, Valsasina, Pagani, Preziosa, Giannì, Petsas and Filippi2024). BRB-N is a neuropsychological battery that includes several cognitive tests evaluating different cognitive domains:
The Selective Reminding Test (SRT) measures verbal learning and long-term verbal memory through a list-learning procedure over six trials. The participant is required to immediately recall as many words as possible from a list of 12 words that is pronounced in full only on trial 1. In trials 2–6 only the words that are immediately recalled by the participant are presented. Two scores are produced: Long term storage (SRT-LTS), which indicates the sum of words that are recalled without reminding in each trial, and the Consistent long term retrieval (SRT-CLTR) which indicates how much the retrieval of the words is organized and consistent. Moreover, after ∼ 10 minutes, the administrator asks the participant to recall das many words of the list as possible, without any cue. The score of the delayed recall of SRT (SRT-D) is the total number of recalled words. Higher scores of the SRT-LTS, SRT-CLTR and SRT-D are associated to better performance.
The Spatial Recall Test (SPART) is a measure of visuospatial learning and long-term visuospatial memory. The psychologist presents a visual stimulus (a grid with 10 painted black tokens) for ten seconds; after that, the participant is asked to place 10 tokens on an empty grid in the exact pattern that was shown to them. The stimulus is presented for 3 trials. The SPART score is calculated by summing up the number of the tokens placed in the correct position in each trial. As for the SRT, the SPART has also a delayed recall task (SPART-D), that it is administered after ∼ 10 minutes. The number of tokens put in the correct position is the raw score of SPART-D. Higher scores of the SPART and SPART-D are associated to better performance.
The Symbol Digit Modalities Test (SDMT) assesses attention and information processing speed, and it is considered one of the most sensitive tests to detect cognitive disorders in MS (Benedict et al., Reference Benedict, DeLuca, Phillips, LaRocca, Hudson and Rudick2017). The administrator presents a page headed with a key that associates numbers from 1 to 9 to nine different symbols. Below the key, there are several rows of symbols, and the patient is asked to orally report the correct number associated to each symbol. Only the first 10 items are completed with the guidance of the psychologist. The score of the SDMT is calculated by reporting the number of correct responses in 90 s; the higher the score, the better is the performance.
The Paced Auditory Serial Addition Test is employed to assess levels of auditory information processing speed, cognitive flexibility, and calculation ability. The participant listens to a tape recording of digits, and it is asked to add each number to the one that immediately precedes it. The task is presented twice: for the PASAT 3’’ task, each number is presented every 3 s, whereas in the PASAT 2’’ the presentation rate is every 2 s. The scores of PASAT 3’’ and 2’’ are the total number of correct responses for the two tasks; higher scores are indicative of a better performance.
The Word List Generation (WLG) is a task that evaluates semantic verbal fluency. The patient is asked to report all word that fall within the category “fruit and vegetables”. Score is calculated by summing up the correct words pronounced in 90 s. Higher scores of the WLG are associated to a better performance.
In addition to the BRB-N, the Stroop test (STROOP) was administered to measure sustained attention and some executive functions, such as inhibitory control. The STROOP presents 3 trials: in trial 1, the participant is asked to read some words enlisting different colors; in trial 2, the participant needs to report the name of the color of several rows of squares; in trial 3, called the interference task, there are several rows of words indicating different colors, but the meaning of the word and the ink employed to write the word does not match. The patient is asked to pronounce the colors that the words are printed in. The psychologist reports the time needed to reach half of the task (STROOP T ½; Amato et. al, Reference Amato, Portaccio, Goretti, Zipoli, Ricchiuti, De Caro, Patti, Vecchio, Sorbi and Trojano2006). For the STROOP 30’’, higher scores reflect a better performance; as for the STROOP T ½, higher scores indicate a worse performance. For this study, we employed the STROOP T ½ score.
CR was assessed through the Italian adaptation of the Cognitive Reserve Scale (I-CRS) (Altieri et al., Reference Altieri, Siciliano, Pappacena, Roldán-Tapia, Trojano and Santangelo2018). The I-CRS is a 24-item self-administered questionnaire that evaluates how much a person is involved in several cognitively stimulating activities. Based on the operational definition of CR provided by León-Estrada and colleagues (Leon-Estrada et al., Reference Leon-Estrada, Garcia-Garcia and Roldan-Tapia2017), the items of the I-CRS reflect 4 groups of activities that, taken together, form the CR construct. These factors are: a) daily activities (I-CRS_DA): activities linked to functional autonomy, such as the ability to manage engagements and appointments in everyday life, to do chores and to be able to use widespread technology; b) training-information (I-CRS_TI): activities related to the acquisition of formal and informal knowledge; c) hobbies (I-CRS_H): engagement in leisure activities; d) social life (I-CRS_SL): the ability to create and maintain social relationship.
I-CRS is based on a 5-point Likert scale; scores range from 0 to 4, with higher scores indicating a higher engagement in that specific activity. The level of cognitive stimulation is assessed in three life stages: young adulthood (18–35 years), adulthood (36–64 years), and late adulthood ( ≥65 years). Participants are required to complete the questionnaire once, twice, or three times, according to their age: adults and older adults need to complete the I-CRS for their current life stage and the previous ones. The total I-CRS score is obtained by producing the mean score on each item in every life stage. Moreover, scores of I-CRS subscales can be calculated by summing up the scores of the items of the single subscale.
Statistical analysis
Statistical analysis was performed with SPSS (SPSS Statistics, version 25.0). Means and standard deviations, or medians and minimum-maximum ranges were calculated as appropriate. The Chi square test was performed to examine the relationship between sex and overall cognitive impairment. pwMS defined as cognitively impaired had at least two tests failed with a Z score ≤ 1.5 according to the latest normative data (Tedone et al., Reference Tedone, Vizzino, Meani, Gallo, Altieri, D’Ambrosio, Pantano, Piervincenzi, Tommasin, De Stefano, Cortese, Stromillo, Rocca, Sala, Valsasina, Pagani, Preziosa, Giannì, Petsas and Filippi2024). The t-test for independent samples (or the Chi square test, as appropriate) were performed to compare sociodemographic variables, I-CRS and cognitive test scores between men and women with MS. To evaluate the effect of CR and sex, and a possible interaction between these factors on cognitive performance several models of multivariate analyses of covariance (MANCOVAs) were performed. Dependent variables were SRT-LTS, SRT-CLTR, SRT-D, SPART, SPART-D, SDMT, PASAT 3’’, PASAT 2’’, WLG, and STROOP scores; the independent variables were sex (men vs women) and CR (high vs low CR; each MANCOVA included one of the following: I-CRS total score, I-CRS_DA, I-CRS_TI, I-CRS_H, and I-CRS_SL). In each model, CR dichotomous level was calculated by computing the median score of I-CRS on the whole sample: pwMS with a CR score under or above the median value were allocated to the low CR or high CR group, respectively. Covariates included: age, EDSS, and BDI-II scores. A p-value of .05 was considered the cut-off for statistical significance. Bonferroni correction for multiple comparisons was applied to reduce the probability of incurring in type I error.
Results
The whole sample included 233 participants (95 men, 138 women) with MS, with age ranging from 18 to 64 years old, education ranging from 5 to 24 years and disease duration ranging from 0 to 32 years. Men and women with MS did not differ significantly for sociodemographic variables, except for frequencies in specific work activities: when compared to men with MS, women with MS were more likely to be unemployed or to be working as “Professionals”; on the other hand, men were more likely to be craft and related trades workers. As regards clinical variables, women reported higher scores at BDI-II compared to men. Demographic, clinical, and neuropsychological characteristics of the sample are shown in Table 1.
Table 1. Demographic, clinical and neuropsychological characteristics of the sample
Note: a = chi-square test was performed. Unless otherwise specified, all values are reported as mean (standard deviation). Statistically significant values are reported in bold. EDSS = Expanded Disability Status Scale, SRT-LTS = Selective reminding test long term storage, SRT-CLTR = Selective reminding test consistent long term retrieval, SRT-D = Selective reminding test delayed recall, SPART = Spatial recall test, SPART-D = Spatial recall test Delayed Recall, SDMT = Symbol Digit Modalities Test, PASAT 3’’ and 2’’ = Paced Auditory Serial Addition Test, 3 secondss and 2 secondss, WLG = Word List Generation, STROOP = Stroop Test, I-CRS = Cognitive Reserve Scale Italian version, BDI-II = Beck Depression Inventory, second edition.
Sex differences in cognition and CR levels
Cognitive impairment was present in 45.26% of men with MS and 39.85% of women with MS. The proportion of participants who showed cognitive impairment did not differ by biological sex (χ2 = .675, p = 4.11). The independent samples t-test revealed that women performed significantly better on WLG (t = −4.091, p <.001) and SRT-D (t = −2.145, p = .033), whereas men with MS showed higher scores on PASAT 3’’ (t = 2.326, p = .021). Women with MS reported higher scores on I-CRS than men with MS (t = −2.372, p = .019). As regards I-CRS subscales (see Figure 1), women reported higher scores on I-CRS_DA (t = −5.848, p <.001), I-CRS_H (t = −2.591, p = .010), and I-CRS_SL (t = −2.362, p = .011), whereas the comparison of scores of I-CRS_TI (t = 1.791, p = .075) revealed no difference between sexes.
Figure 1. Raincloud plots of independent samples t-tests regarding the comparison between men and women with MS on: a) scores of the daily activities subscale of I-CRS (I-CRS_DA); b) scores of the training-information subscale of I-CRS (I-CRS_TI); c) scores of the hobbies subscale of I-CRS (I-CRS_H); d) scores of the social life subscale of I-CRS (I-CRS_SL).
Multivariate analyses of sex and CR effects on cognitive performance
Interaction between CR (I-CRS) and sex
After the correction for multiple comparisons, the MANOVA (see Table 2) revealed a significant main effect of CR level (Λ= .876, F(10,154)= 2.185, p = .021, ηp 2 = .124) and sex (Λ= .836, F(10,260)= 3.017, p = .002, ηp 2 = .164) on cognitive performance, but the interaction between CR and sex did not reach statistical significance (Λ= .950, F(10,154)= .813, p = .617, ηp 2 = .050).
Table 2. Multivariate analysis of sex, CR, and their interaction effect on cognitive performance in people with MS
Note: Df = degrees of freedom, Λ = Wilks lambda, FF = Fischer test, pp = probability value, ηp 2 = partial eta squared. Significant values are reported in bold.
Univariate analyses revealed a significant main effect of CR on SRT-LTS (F(1,195) = 6.551, p = .011, ηp 2 = .039), SRT-CLTR (F(2,195) = 12.252, p = .011, ηp 2 = .039), SPART (F(2,195) = 5.482, p = .020, ηp 2 = .033), SDMT (F(2,195) = 9.582, p = .002, ηp 2 = .056), SRT-D (F(1,195) = 11.829, p = .001, ηp 2 = .068), SPART-D (F(1,195) = 4.503, p = .035, ηp 2 = .027), and WLG (F(1,195) = 5.724, p = .018, ηp 2 = .034) scores. Moreover, a significant main effect of sex was found on PASAT 3’’ (F(1,195) = 4.819, p = .030, ηp 2 = .029; men>women), and WLG (F(1,195) = 6.837, p = .010, ηp 2 = .040; women>men) scores.
Interaction between daily activities (I-CRS_DA) and sex
Daily activities (I-CRS_DA) and sex
A significant main effect of sex on cognitive performance (Λ=.858, F(10,154)= 2.549, p = .007, ηp 2 = .142) was found. I-CRS_DA subscale (Λ= .938, F(10,154)= 1.010, p = .438, ηp 2 = .062) and the interaction between I-CRS_DA and sex (Λ= .946, F(10,154)= .878, p = .555, ηp 2 = .054) were not significant. Univariate analyses revealed a significant main effect of sex on PASAT 3’’ (F(1,195) = 5.620, p = .019, ηp 2 = .033; men > women), and WLG (F(1,195) = 4.664, p = .032, ηp 2 = .028; women > men) scores.
Training-information (I-CRS_TI) and sex
The MANOVA revealed a significant main effect of sex on cognition (Λ= .798, F(10,154)= 3.902, p <.001, ηp 2 = .202); nor I-CRS_TI (Λ= .932, F(10,154)= .1.124, p = .348, ηp 2 = .068), nor the interaction between sex and I-CRS_TI (Λ= .915, F(10,154)= .1.422, p = .175, ηp 2 = .085) were significant. At the univariate level, sex had a significant impact on SRT-D (F(1,195) = 5.510, p = .020, ηp 2 = .033; women>men) and WLG (F(1,195) = 10.945, p = .001, ηp 2 = .063; women>men).
Hobbies (I-CRS_H) and sex
The results revealed a significant main effect of sex (Λ= .835, F(10,154)= 3.023, p = .002, ηp 2 = .165) and I-CRS_H (Λ= .791, F(10,154)= 4.042, p< .001, ηp 2 = .209) on cognitive performance; on the other hand, the interaction between sex and I-CRS_H was not significant (Λ= .969, F(10,154)= 487, p = .897, ηp 2 = .031). Univariate analyses were performed, and a main effect of sex was found on PASAT 3’’ (F(1,195) = 6.719, p = .010, ηp 2 = .040; men>women), PASAT 2’’ (F(1,195) = 3.987, p = .048, ηp 2 = .024; men>women) and WLG (F(1,195) = 5.380, p = .022, ηp 2 = .032; women>men) scores, whereas a main effect of I-CRS_H was found on SRT-LTS (F(1,195) = 24.840, p <.001, ηp 2 = .133; high I-CRS_H > low I-CRS_H), SRT-CLTR (F(1,195) = 29.116, p <.001, ηp 2 = .152; high I-CRS_H > low I-CRS_H), SPART (F(1,195) = 7.011, p = .009, ηp 2 = .041; high I-CRS_H > low I-CRS_H), SDMT (F(1,195) = 11.925, p = .001, ηp 2 = .069; high I-CRS_H > low I-CRS_H), SRT-D (F(1,195) = 28.280, p <.001, ηp 2 = .149; high I-CRS_H > low I-CRS_H), SPART-D (F(1,195) = 5.281, p = .023, ηp 2 = .032; high I-CRS_H > low I-CRS_H), WLG (F(1,195) = 19.826, p <.001, ηp 2 = .109; high I-CRS_H > low I-CRS_H), and STROOP (F(1,195) = 4.083, p = .045, ηp 2 = .025; high I-CRS_H > low I-CRS_H).
Social life (I-CRS_SL) and sex
A significant main effect of sex on cognitive performance (Λ= .861, F(10,154)= 2.472, p = .009, ηp 2 = .139) was found. I-CRS_SL subscale (Λ= .895, F(10,154)= 1.787, p = .067, ηp 2 = .105) and the interaction between I-CRS_DA and sex (Λ= .907, F(10,154)= .1.569, p = .121, ηp 2 = .093) were not significant. Univariate analyses revealed a significant main effect of sex on PASAT 3’’ (F(1,195) = 4.208, p = .042, ηp 2 = .025; men > women), and WLG (F(1,195) = 4.289, p = .040, ηp 2 = .026; women > men) scores.
Discussion
Our study aimed at evaluating, in a sample of pwMS, the relationship between biological sex, CR, and cognition; in particular, we analyzed possible sex differences in levels of engagement in cognitively enhancing activities (which reflect the construct of CR) and explored the possible effect of sex and CR on performance in several cognitive domains. In this study, CR was measured with a specific questionnaire, the CRS, that evaluates how much a person engages in activities related to functional autonomy, formal and informal knowledge, leisure and social activities.
As regards the possible sex differences in levels of engagement in cognitively enhancing activities, the comparison of levels of CR in women and men with MS revealed that women with MS showed significantly higher levels of CR when compared with their male peers. The comparison of groups of activities related to CR (by comparing single I-CRS subscales) revealed that women reported higher levels of engagement in daily, leisure, and social activities. In contrast, mean scores of the engagement in activities related to the acquisition of knowledge were comparable between the two sexes. As noted in previous studies on CR in other neurological diseases (i.e., Alzheimer’s disease) some proxies of CR were highly gendered (Subramaniapillai et al., Reference Subramaniapillai, Almey, Natasha Rajah and Einstein2021): for example, until a few decades ago men had easier access to formal education and acquisition of knowledge; on the other hand, women were more prone to engage in daily activities, mostly related to housekeeping. Modernization and evolution of gender norms may have had an impact on the engagement of specific cognitively stimulating activities (such as the acquisition of knowledge) thus reducing sex differences to a negligible level; on the other hand, the other sets of activities linked to CR would not have to be affected yet by the abovementioned factors, and therefore they remain highly gendered. Hence, CR activities, especially those related to functional autonomy, leisure, and social activities should be carefully assessed in pwMS, particularly in men, who may benefit from specific training to increase engagement in these activities.
Moreover, our results corroborated the hypothesis of sex differences in cognitive performance in pwMS. For example, women showed an advantage in the semantic verbal fluency task. This is in line with the literature on healthy participants (Capitani et al., Reference Capitani, Laiacona and Barbarotto1999, Reference Capitani, Barbarotto and Laiacona2005; Filippetti & Allegri, Reference Filippetti and Allegri2011; Munro et al., Reference Munro, Winicki, Schretlen, Gower, Turano, Muñoz, Keay, Bandeen-Roche and West2012) and in other studies on MS patients (Altieri et al., Reference Altieri, Santangelo, Bisecco, D’Ambrosio, Rippa, Capuano, Giuliano, Matrone, Bonavita, Tedeschi and Gallo2021; Amato et al., Reference Amato, Portaccio, Goretti, Zipoli, Ricchiuti, De Caro, Patti, Vecchio, Sorbi and Trojano2006). However, it should be noted that a recent meta-analysis revealed that sex differences in semantic fluency tasks may be category-dependent (Hirnstein et al., Reference Hirnstein, Stuebs, Moè and Hausmann2023): since our study employed the WLG from BRB-N, in which participants were required to list names of only one category (fruits and vegetables), future studies should employ tools that allow evaluating multiple categories to verify if differences between sexes are category-dependent in MS, too.
Furthermore, in our study women with MS showed higher scores on recall of verbal material when compared to men with MS, a finding that confirms previous MS studies (Altieri et al., Reference Altieri, Santangelo, Bisecco, D’Ambrosio, Rippa, Capuano, Giuliano, Matrone, Bonavita, Tedeschi and Gallo2021; Beatty & Aupperle, Reference Beatty and Aupperle2002; Donaldson et al., Reference Donaldson, Patel, Shammi and Feinstein2019). A recent meta-analysis of 617 studies with more than 1 million participants (without neurological diseases or pathologies that could affect neuropsychological performance) found a female advantage for several verbal memory tasks, such as the ability to remember words, sentences and prose, and nameable images; on the other hand, men were more capable in spatial memory tasks, such as remembering routes and abstract images (Asperholm et al., Reference Asperholm, Högman, Rafi and Herlitz2019). In neurological diseases, there is evidence that women outperform men in verbal memory in amnestic mild cognitive impairment, independently from levels of hippocampal volume loss, temporal lobe hypometabolism and cortical amyloid-β deposition; however, in patients with clinically defined Alzheimer’s disease, sex differences were attenuated (Sundermann et al., Reference Sundermann, Biegon, Rubin, Lipton, Landau and Maki2017). The authors explained this finding by postulating a form of sex-specific CR that protects from the effects of brain burden up to some point of the disease; however, when the threshold of protection is surpassed, cognitive decline in women with AD is more accelerated than in men, as previously demonstrated (Stern, Reference Stern2002). Since the cognitive profile of pwMS rarely evolves in frank dementia, our results may suggest a mechanism comparable to what is found in people with mild cognitive impairment. However, future studies are needed to further expand this aspect by employing neuroimaging techniques to evaluate lesion load in the MS population and a longitudinal study design to evaluate if sex difference in delayed verbal memory is reduced when the disease progresses.
Sex differences were also noted in auditory information processing speed and sustained attention, with men reporting higher scores than women. Studies on sex differences in PASAT scores reported a similar performance between men and women in healthy participants and pwMS (Amato et al., Reference Amato, Portaccio, Goretti, Zipoli, Ricchiuti, De Caro, Patti, Vecchio, Sorbi and Trojano2006; Wiens et al., Reference Wiens, Fuller and Crossen1997; Wingenfeld et al., Reference Wingenfeld, Holdwick, Davis and Hunter1999). However, other studies found an advantage in men in pwMS (Altieri et al., Reference Altieri, Santangelo, Bisecco, D’Ambrosio, Rippa, Capuano, Giuliano, Matrone, Bonavita, Tedeschi and Gallo2021). This discrepancy in our results may be due to different study designs and differences in the sociodemographic and clinical variables of the samples. Given that the PASAT is one of the most employed tests in MS, the possible sex difference in performance in this task deserves further investigation in future studies.
Our multivariate analyses revealed a significant effect of CR in cognition only in models with the I-CRS total scale and the I-CRS_H subscale: higher CR, and in particular higher engagement in leisure activities, was associated with a better performance in verbal and spatial memory, processing speed, verbal fluency and inhibitory control tests. Several leisure activities, such as reading and engaging in hobbies, have been related to increased performance in several cognitive domains, including verbal memory and processing speed (Sumowski et al., Reference Sumowski, Wylie, Gonnella, Chiaravalloti and Deluca2010, Reference Sumowski, Rocca, Leavitt, Riccitelli, Meani, Comi and Filippi2016). These results particularly emphasize the possibility for individuals to actively increase their levels of CR throughout life by engaging in a specific set of cognitively stimulating activities. The first (and most) studies on CR have focused on the role of formal education as a protective variable on cognitive performance. However, most adults acquire knowledge in formal settings (school, university) only in early adulthood. In this case, our results would confirm how even activities that can be easily performed at any stage of adult life can play a role in preserving cognitive performance. Therefore, clinicians should evaluate CR levels at the time of diagnosis of MS to more precisely indicate which activities should be most strengthened in the individual; moreover, during routine visits, they should encourage patients to engage in these activities and check their progress.
The possible interaction effect between sex and CR was revealed to be nonsignificant on neuropsychological tests administered in our study; that could mean that the impact of CR on cognitive performance did not depend on being a man or a woman with MS, but this finding must be taken with caution due to unavailability of other data (e.g, CNS burden metrics). However, due to the significant association between specific cognitively stimulating activities and cognitive performance in MS, our findings may prompt clinicians to address MS patients with cognitive impairment to specific tailored neuropsychological rehabilitation programs or to enable cognitive stimulation programs focusing on CR-building activities regardless of the sex of the MS patient.
Our study, however, is not exempt from limitations. Due to the cross-sectional nature of this study, it was not possible to verify if, and to what extent, CR protects from cognitive decline over time, and if an interaction between CR levels and sex can emerge in follow-up evaluations. Secondly, this study lacks a control group, so future studies may include a group of healthy peers to evaluate if the interaction effect between CR and sex on cognitive performance is present in healthy participants. Moreover, the impact of CR and sex differences on cognition in participants with progressive forms of MS was not evaluated in the present study. This could be an area of investigation for future studies. Finally, MRI data were unavailable, so we did not control for lesion load, cerebral volumes or CNS burden metrics. Studies focusing on the effect of CNS burden metrics and biological sex on cognition were discordant, with some studies revealing that sex differences in cognitive performance in pwMS were independent of CNS burden (Motyl et al., Reference Motyl, Friedova, Ganapathy Subramanian, Vaneckova, Fuchs, Krasensky, Blahova Dusankova, Kubala Havrdova, Horakova and Uher2024), and some other studies that demonstrated a better neurocognitive functioning and brain resilience (evaluated by using MRI metrics) in women (Leavitt et al., Reference Leavitt, Dworkin, Kalina and Ratzan2024); nonetheless, this is an area of paramount importance that deserves to be further investigated in future studies.
Taken together, our results highlight the association between CR and specific cognitive domains, and that this association is independent of the sex of the MS patient.
Acknowledgements
The authors thank Taysia del Pozzo, Genoveffa Maddaloni, and Arianna Mazzella for their support in acquiring neuropsychological data.
Funding statement
None.
Competing interests
None.
Ethical approval statement
The present study was approved by the Ethics Committee of the University of Campania “Luigi Vanvitelli” (Naples, Italy).
