Cognitive problems in patients with multiple sclerosis (MS) occur widely in up to 62% of cases, especially concerning executive functions, processing speed, attention, short-term memory and verbal fluency [1].

The use of disease modifying therapies (DMTs), such as interferon beta-1a (INFβ-1a) and interferon beta-1b (INFβ-1b) is considered to affect the cognitive performance of patients with relapsingremitting multiple sclerosis (RRMS) [2-4], although there are opposite opinions [5].

Cognitive assessment is performed using a battery of neuropsychological tests, such as the Paced Auditory Serial Addition Test 3 (PASAT-3) [6], and Symbol Digit Modalities Test (SDMT) [7-11]. In some studies with a small number of patients, p300 event-related brain potentials (ERP) are considered to be an indicator of objective neuropsychological cognitive functioning for measuring cognitive function in patients with MS [12].

The first objective of our study was to determine whether there are differences between groups of patients receiving therapy, patients without therapy and a control group, in terms of neuropsychological test results and ERP. The second objective was to determine factors that may serve to assess the risk of cognitive impairment in patients with RRMS.

Participants

In the period from 01.05.2016 to 01.07.2016 in the Department of Neurology of the Clinical Center Kragujevac and the Faculty of medical Sciences, University of Kragujevac, Kragujevac, Serbia, we examined a total of 81 patients with RRMS (71.7%) and 32 healthy controls (28.3%). We tested all patients from our center with RRMS who were on DMTs.

The aims of our study were explained to them and informed consent was signed by all participants and approved by the Ethics Committee of the Clinical Center, Kragujevac.

Diagnosis of definitive RRMS was made using the 2010 revisions to the McDonald criteria [13]. All patients fulfilled the inclusion criteria, i.e. they had RRMS without disability in a functional system that would prevent testing. The exclusion criteria for the group of patients were presence of other neurological diseases and psychiatric disorders, use of neuroleptics, anticholinergics, antidepressants and antiepileptic drugs, relapse and use of corticosteroid therapy in the last month before the tests. The inclusion criteria for the control group were absence of: neurological disease, psychiatric diagnoses, use of any drugs. The RRMS group and the control group were matched by age, gender and education.

Patients were divided into three subgroups: those who were not receiving DMTs, patients on INF β-1a and patients on INFβ-1b. Some patients were not receiving DMTs for economic (financial) reasons in Serbia. They had not been given other types of DMTs earlier.

Clinical and neuropsychological evaluation

Neurological disability was evaluated according to the Expanded Disability Status Scale (EDSS), by the same experienced EDSS certified neurologists [14].

Cognitive functions were evaluated using a standard PASAT 3 (form A and B) and SDMT. Means with standard deviation were calculated for SDMT and standard deviations ≤ 1.5 below the mean value of correct responses for the SDMT by the control group was used as the threshold value separating cognitively preserved (CP) and cognitively impaired (CI) MS patients [10].

Objective cognitive testing

ERP were determined for all patients and control subjects. ERP were registered from the scalp during the morning in the same soundproofed room. Participants listened to auditory stimuli through a headset and responded by pressing a button with their dominant hand. Bioelectrical activity of the brain was registered by monopolar electrodes placed on the center line of the frontal region (Fz), central region (Cz) and parietal region (Pz) using “Keypoint® Software ver. 3.00 with three dual-channels (Ag¬AgCl). Inactive electrodes were placed on the mastoid process, and the ground electrode was on the forearm of the subjects. Impedance was kept below 5 Ω after detailed preparation of the place where they were put. The resulting signals were amplified, filtered and sampled in the time frame of 1000 ms. We used the “oddball” paradigm with two tones: a “standard” tone of 90 dB and 1000 Hz and an “expecting“ tone of 90 dB and 2000 Hz. These were presented binaurally at irregular intervals and in incorrect order through a special headset. Participants were ordered to ignore low “standard” tones, and to react as quickly as possible by pressing the key with the dominant hand, every time they heard the “expecting” high tone. About 260 attempts were registered with the ratio “standard” 80%-“expecting” 20%. Processing ERP included individual identification of the highest positive wave in the window of 220 to 450 ms for each electrode and determination of the amplitude and latency of the p300 wave. Reaction time (RT) was also registered (time elapsed from the occurrence of the stimulus to the response of the patient by pressing the button).

Statistical analysis

Descriptive data are presented as medians with percentages. Kolmogorov-Smirnov and Shapiro-Wilk tests were used to test for normality of distribution of numerical data. For differences in parametric variables Student’s t-test and for nonparametric variables the Mann-Whitney U test was employed. Categorical variables were compared using the Chi-square test. Spearman’s test was used to determine correlations between variables for data that did not follow a normal distribution. ANOVA was employed to examine differences in latency and amplitude and RT of ERP between the groups of participants in the study. Univariate and multivariate logistic regression was performed to examine which factors influence the occurrence of cognitive impairment in patients with RRMS. Multivariate binary logistic regression allows creation of a model for assessing the risk of cognitive impairment. A model showing the influence of factors affecting the occurrence of cognitive impairment was made based on the significant factors obtained in the univariate and multivariate logistic regression analysis. Probability (P) values<0.05 were considered statistically significant. All statistical analyses were processed in the SPSS v.20 program.

Demographic, clinical and neuropsychological characteristics for RRMS patients compared to the control group are presented in Table 1. There were negative correlations between EDSS and the results of PASAT A (r=-0.289, p=0.010) and PASAT B (r=-0.319, p<0.001). The results of the SDMT were influenced by gender (women performed better p<0.05), age (p<0.05, r=-0.217), duration of education (p<0.01, r=0.460), educational level (people with a university degree performed better than those with secondary (p<0.05) and primary school education only (p<0.01). There was a significant negative correlation between SDMT and EDSS (r=-0.387, p=0.001). When patients were divided into groups of CP and CI participants, in the RRMS group the relative number with cognitive impairment was 53.2%. The difference between the mean values of EDSS between these groups was statistically significant (CP=2.2 ± 1.5, CI=2.6 ± 1.4, p<0.05). The differences in mean values for PASAT A (CP=45.8 ± 11.1, CI=37.4 ± 10.8, p<0.05) and PASAT B (CP=49.5 ± 9.5, CI=40.4 ± 11.5, p<0.01) between these two groups were statistically significant. CP was related to the level of education (p<0.05).

Table 1: Demographic, clinical and neuropsychological characteristics for RRMS patients compared to control group.

Our study showed that therapy did not affect performance in PASAT A and PASAT B (PASAT A: INFβ-1b vs. without DMTs p=0.739; INFβ- 1a vs. without DMTs p=0.776; PASAT B: INFβ-1b vs. without DMTs p=0.243, INFβ-1a vs. without DMTs p=0.844). Moreover, no differences between the groups on DMTs and patients without DMTs (INFβ-1b vs. no DMTs p=1.000; INFβ-1a vs. no DMTs p=0.676) or between the two groups receiving DMTs (INFβ-1b vs. INFβ-1a p=1.000) were found. There was also no significant difference in relation to DMTs between the CP and CI groups (χ2=6.876, p=0.076).

Comparison of ERP latency (ANOVA) between four groups of participants is presented in Figure 1. There were significant differences between the mean values of Pz (p<0.05) and Cz latency (p<0.05) between the four groups of study participants. Thus, Cz latency for the control group diverged from that for patients without DMTs (p<0.05), while Pz latency for the control group diverged from that for patients on INFβ-1b (p<0.01) (Figure 1).

Figure 1: The comparison of event related potentials latency, in milliseconds, between 4 groups of participants, from Fz, Cz and Pz electrode.

Correlations of ERP with demographic characteristics, neuropsychological tests and EDSS in the RRMS group are shown in Table 2.

Table 2: Correlation of P300 event-related potentials with demographic characteristics, neuropsychological tests and EDSS in the group of patients with RRMS.

Univariate binary logistic regression and multivariate binary logistic regression indicated factors that have an impact on the occurrence of cognitive impairment (Table 3).

Table 3: Univariate and Multivariate binary regression showed which factors influence the occurrence of cognitive impairment and the values of OR with 95% CI.

RRMS increased the risk of cognitive impairment by approximately 3.5 times. Each year of age raised the risk of cognitive impairment by 6.0%. Unit increase in level of education reduced the risk of cognitive impairment approximately 2.5 fold. Increase in the variable RT Cz by 1 ms elevated the risk of cognitive impairment by 0.5%. Multivariate binary logistic regression allowed creation of a model for assessing the risk of cognitive impairment. The model is a new variable calculated as follows:

Model =e^sum/(1 + e^sum)·100

sum=1.278*MS+0.058*age-0.913*level of education+0.005*RT Cz- 3.291.

The ROC curve showed that this model may be a good marker to separate participants into those with cognitive impairment and those without cognitive impairment (AUROC=0.803, p<0.01) (Figure 2).

Figure 2: ROC curve.

Cognitive impairment has been registered over a wide range of proportions of RRMS patients, i.e. from 9% to 62% [5,15-23]. In our sample of RRMS patients the percentage of CI (defined as ≤ 1.5 standard deviations below the mean value of correct responses for the SDMT by the control group) was 53.2%.

In our cross-sectional study therapy did not affect performance in PASAT A and PASAT B, while some longitudinal studies have shown that performance in PASAT improves slightly after one year of application of INFβ-1b, but not statistically significantly, while use of INFβ-1a led to a statistically significant improvement [3].

There were also marked improvements in the SDMT one year after the application of INFβ-1b and INFβ-1a [3], which was not confirmed in our cross-sectional study. Thus, no differences between the groups on DMTs and patients without DMTs or between the two groups receiving DMTs or in relation to DMTs in those with CP and with CI.

A small number of studies have aimed to investigate the effect of DMTs on latency, amplitude and RT of ERP in patients with RRMS. Thus, INFβ-1b was shown to reduce amplitude and latency [1]. In our cross-sectional study, there were no differences in amplitude and latency and RT between patients without DMTs and patients on INFβ- 1a and INFβ-1b (Figure 1).

Others have recorded significantly higher EDSS in CI patients [20,24], which was also the case in our group of participants. Cognitive assessment using PASAT-3 indicated better performance by our control group than the RRMS group (Table 1), in accordance with earlier findings [25,26].

Parmenter and Borghi showed high sensitivity and specificity of the SDMT demonstrating its efficiency as a screening test for cognitive impairment in patients with MS [15,27]. Thus our control group also achieved better results on SDMT than our patients, confirming the findings of others [25]. In our group of participants, women achieved better scores in the SDMT. According to our knowledge, no other study has demonstrated a difference in SDMT between the genders.

SDMT negatively correlated with EDSS during the entire follow-up of patients in Brochet’s study [7,25], similarly to our group of patients (p<0.01, rho=-0.363).

We have found that performance of neuropsychological tests is significantly affected by education level, in accordance with Luerding’s conclusion that cognitive reserve is the greatest among the highly educated. He found an especially strong relation of educational background to memory and executive functions [28].

The two main neuropsychological markers of cognitive functioning are ERP latency and ERP amplitude. Prolonged latency indicates a long information processing time [29]. Some research has confirmed a connection between ERP and cognitive impairment, defined by neuropsychological tests in MS patients [12,30-32] but this correlation is stronger in patients with Alzheimer’s disease where the cortical dementia concept is dominant [33-35]. The correlation between multifocal white-matter and gray-matter lesions in cognitive dysfunction pathology has laid the foundations of the theory of multiple disconnection. Disconnection occurs between cortical and subcortical regions. MS patients with CI manifested more cortical lesions than patients who were CP. The multiple disconnection syndrome means that more than one cognitive domain can be interrupted, producing a variety of neuropsychological deficits [36]. As far as we know there is no study connecting ERP and multiple disconnection syndrome.

Duration of education in our RRMS patients significantly correlated with Pz amplitude. Sundgren stated that half of the variance of cognitive functioning in patients with RRMS is explained by variations in ERP amplitude and RT, together with EDSS, but there are different opinions [30,31]. Positive correlations of EDSS with Cz and Pz latency, and a negative one with Pz amplitude, were also observed in our patients with RRMS. Others have shown that changes in ERP correlate significantly with changes in PASAT during a one year follow-up [37]. While Sundgren found that success in neuropsychological tests correlates with ERP from the parietal region, and not from the frontal region in patients with RRMS [31] we only obtained correlation of PASAT B and RT Fz (Table 2).

The risk of cognitive impairment was calculated in percentages, with a formula that includes factors of importance for the assessment of cognitive risk: the diagnosis of RRMS, age, education level and RT Cz. For example, participant number 38 without a diagnosis of RRMS, was 32 years old, has finished university, and his RT Cz was 414 ms. Our model calculated that the relative risk of cognitive impairment for this participant was 10.87%. This subject had no cognitive impairment. Participant number 41 had an RRMS diagnosis, was 43 years old, finished secondary school and his RT Cz was 674 ms. The risk of cognitive impairment for this individual was high, 88.34% and he did manifest cognitive impairment. In the model the significant factors for assessing cognitive impairment in RRMS patients did not include either latency or amplitude from ERP, so we conclude that ERP is not a reliable indicator of cognitive impairment in these patients.

The advantages of our research are the size of the sample of patients with RRMS, who were receiving DMTs, and who were tested for ERP. We are the first to form a model that calculates an adequate assessment of the risk of cognitive impairment for each RRMS patient. A limitation of this study is that some patients were not receiving DMTs for economic reasons in Serbia.

There were a significant differences between the mean values for Pz (p<0.05) and Cz latency (p<0.05) between four groups of study participants. Thus, Cz latency for the control group differed from that for patients without DMTs (p<0.05), while Pz latency for the control group differed from that for patients on INFβ-1b (p<0.01). Factors that can be used to assess the risk of developing cognitive impairment in patients with RRMS are age, education level, RT Cz. However, ERP (latency and amplitude) did not independently assess the risk of cognitive impairment in RRMS patients.

1. Achiron A, Chapman J, Magalashvili D, Dolev M, Lavie M, et al. (2013) Modeling of cognitive impairment by disease duration in multiple sclerosis: A cross-sectional study. PLoS One 8: e71058.
2. Flechter S, Vardi J, Finkelstein Y, Pollak L (2007) Cognitive dysfunction evaluation in multiple sclerosis patients treated with interferon beta-1b: An open-label prospective 1 year study. Isr Med Assoc J 9: 457-459.
3. Melanson M, Grossberndt A, Klowak M, Leong C, Frost EE, et al. (2010) Fatigue and cognition in patients with relapsing multiple sclerosis treated with interferon β. Int J Neurosci 120: 631-640.
4. Mokhber N, Azarpazhooh A, Orouji E, Rao SM, Khorram B, et al. (2014) Cognitive dysfunction in patients with multiple sclerosis treated with different types of interferon beta: A randomized clinical trial. J Neurol Sci 342: 16-20.
5. Gerschlager W, Beisteiner R, Deecke L, Dirnberger G, Endl W, et al. (2000) Electrophysiological, neuropsychological and clinical findings in multiple sclerosis patients receiving interferon beta-1b: A 1-year follow-up. Eur Neurol 44: 205-209.
6. Berard JA, Bowman M, Atkins HL, Freedman MS, Walker LA (2014) Cognitive fatigue in individuals with multiple sclerosis undergoing immunoablative therapy and hematopoietic stem cell transplantation. J Neurol Sci 336: 132-137.
7. Brochet B, Deloire MS, Bonnet M, Salort-Campana E, Ouallet JC, et al. (2008) Should SDMT substitute for PASAT in MSFC? A 5-year longitudinal study. Mult Scler 14: 1242-1249.
8. Benedict RH, Amato MP, Boringa J, Brochet B, Foley F, et al. (2012) Brief International Cognitive Assessment for MS (BICAMS): international standards for validation. BMC Neurol 12: 55.
9. Langdon DW, Amato MP, Boringa J, Brochet B, Foley F, et al. (2012) Recommendations for a Brief International Cognitive Assessment for Multiple Sclerosis (BICAMS). Mult Scler 18: 891-898.
10. Sonder JM, Burggraaff J, Knol DL, Polman CH, Uitdehaag BM (2014) Comparing long-term results of PASAT and SDMT scores in relation to neuropsychological testing in multiple sclerosis. Mult Scler 20: 481-488.
11. López-Góngora M, Querol, L, Escartín A (2015) A one-year follow-up study of the Symbol Digit Modalities Test (SDMT) and the Paced Auditory Serial Addition Test (PASAT) in relapsing-remitting multiple sclerosis: an appraisal of comparative longitudinal sensitivity. BMC Neurol 15:40.
12. Aminoff JC, Goodin DS (2001) Long-latency cerebral event-related potentials in multiple sclerosis. J Clin Neurophysiol 18: 372-377. 
13. Polman CH, Reingold SC, Banwell B, Clanet M., Cohen JA, et al. (2011) Diagnostic criteria for multiple sclerosis: 2010 revisions to the McDonald criteria. Ann Neurol 69: 292-302.
14. Kurtzke JF (1983) Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS). Neurol 33: 1444-1452.
15. Borghi M, Carletto, S. Ostacoli L, Scavelli F, Pia L, et al. (2016) Decline of Neuropsychological Abilities in a Large Sample of Patients with Multiple Sclerosis: A Two-Year Longitudinal Study. Front Hum Neurosci 10: 282.
16. Pérez-Martín MY, Eguia-Del Río P, González-Platas M, Jiménez-Sosa A (2016) Cognitive status in patients with multiple sclerosis in Lanzarote. Neuropsychiatr Dis Treat 12: 1553-1559.
17. Reuter F, Zaaraoui W, Crespy L, Faivre A, Rico A, et al. (2011) Frequency of cognitive impairment dramatically increases during the first 5 years of multiple sclerosis. J Neurol Neurosurg Psychiatry 82: 1157-1159.
18. Nogales-Gaete J, Aracena R, Díaz V, Zitko P, Eloiza C, et al. (2012) Neuropsychological assessment of patients with relapsing remitting multiple sclerosis prior to the use of immunomodulatory drugs. Rev Med Chil 140: 1437-1444.
19. Cáceres F, Vanotti S, Rao S. RECONEM Workgroup (2011) Epidemiological characteristics of cognitive impairment of multiple sclerosis patients in a Latin American country. J Clin Exp Neuropsychol 33: 1094-1098.
20. Patti F, Amato MP, Trojano M, Bastianello S, Tola MR, et al. (2009) Cognitive impairment and its relation with disease measures in mildly disabled patients with relapsing-remitting multiple sclerosis: baseline results from the Cognitive Impairment in Multiple Sclerosis (COGIMUS) study. Mult Scler 15: 779-788.
21. Polychroniadou E, Bakirtzis C, Langdon D, Lagoudaki R, Kesidou E, et al. (2016) Validation of the Brief International Cognitive Assessment for Multiple Sclerosis (BICAMS) in Greek population with multiple sclerosis. Mult Scler Relat Disord 9: 68-72.
22. Walker LA, Osman L, Berard JA, Rees LM, Freedman MS, et al. (2016) Brief International Cognitive Assessment for Multiple Sclerosis (BICAMS): Canadian contribution to the international validation project. J Neurol Sci 362: 147-152.
23. O'Connell K, Langdon D, Tubridy N, Hutchinson M, McGuigan C (2015) A preliminary validation of the brief international cognitive assessment for multiple sclerosis (BICAMS) tool in an Irish population with multiple sclerosis (MS). Mult Scler Relat Disord 4: 521-525.
24. Da Silva M.A, Cavaco S, Moreira I, Bettencourt A, Santos E, et al. (2015) Cognitive reserve in multiple sclerosis: Protective effects of education. Mult Scler 21: 1312-1321.
25. Zhang ML, Xu EH, Dong HQ, Zhang JW (2016) Information processing speed and influential factors in multiple sclerosis. Zhonghua Yi Xue Za Zhi 96: 1173-1177.
26. Uysal U, Idiman F, Idiman E, Ozakbas S, Karakas S, et al. (2014) Contingent negative variation is associated with cognitive dysfunction and secondary progressive disease course in multiple sclerosis. J Clin Neurol 10: 296-303.
27. Parmenter BA, Weinstock-Guttman B, Garg N, Munschauer F, Benedict RH (2013) Screening for cognitive impairment in multiple sclerosis using the Symbol digit Modalities Test. Mult Scler 13: 52-57.
28. Luerding R, Gebel S, Gebel EM, Schwab-Malek S, Weissert R (2016) Influence of Formal Education on Cognitive Reserve in Patients with Multiple Sclerosis. Front Neurol 7: 46.
29. Ivica N, Titlic M, Pavelin S (2013) P300 wave changes in patients with multiple sclerosis. Acta Inform Med 21: 205-207.
30. Sundgren M., Nikulin VV, Maurex L, Wahlin Å, Piehl F, et al. (2015) P300 amplitude and response speed relate to preserved cognitive function in relapsing-remitting multiple sclerosis. Clin Neurophysiol 126: 689-697.
31. Sundgren M, Wahlin Å, Maurex L, Brismar T (2015) Event related potential and response time give evidence for a physiological reserve in cognitive functioning in relapsing-remitting multiple sclerosis. J Neurol Sci 356: 107-112.
32. Honig LS, Ramsay RE, Sheremata WA (1992) Event-related potential P300 in multiple   sclerosis. Relation to magnetic resonance imaging and cognitive impairment. Arch Neurol 49: 44-50.
33. Lai CL, Lin RT, Liou LM, Liu CK (2010) The role of event-related potentials in cognitive decline in Alzheimer's disease. Clin Neurophysiol 121: 194-199.
34. Vecchio F, Määttä S (2011) The Use of Auditory Event-Related Potentials in Alzheimer's Disease Diagnosis. Int J Alzheimers Dis 2011: 653173.
35. Cecchi M, Moore DK, Sadowsky CH, Solomon PR, Doraiswamy PM, et al. (2015) A clinical trial to validate event related potential markers of Alzheimer's disease in outpatient settings. Alzheimers Dement (Amst) 1: 387-394.
36. Guimarães J, Sá MJ (2012) Cognitive Dysfunction in Multiple Sclerosis. Frontiers in Neurology 3: 74.
37.  Kiiski H, Reilly RB, Lonergan R, Kelly S, O'Brien M, et al. (2011) Change in PASAT performance correlates with change in P3 ERP amplitude over a 12-month period in multiple sclerosis patients. J Neurol Sci 305: 45-52.