Клинические и параклинические характеристики отличают усталость от депрессивных фенотипов при множественном склерозе.

Introduction

Fatigue and depression are common symptoms in multiple sclerosis (MS) 1 , 2 , considerably significant for the well-being of the patients and with a crucial economic burden 3 – 6 . While many patients remain untreated for those symptoms 7 , depression and fatigue might persist even when effective disease-modifying treatment (DMT) is applied 8 , 9 . Differentiating between them in the everyday clinical routine can be difficult. Often patients report similar, overlapping symptoms such as lack of energy, tiredness or low task performance, which might be explained by both fatigue and depression. Moreover, they seem to share at least some common pathophysiological mechanisms 10 , 11 . Combined with the usually very short time for clinical consultations, there is a significant chance that if detected, depressive symptoms might be mistaken for fatigue and vice versa. While current therapeutic concepts for depression in MS encompass both pharmacological (e.g., SSRI medication 12 ) and non-pharmacological (e.g. cognitive behavioural therapy 13 ) options, there are only limited therapeutical tools for fatigue. Exercise and behavioural therapeutic interventions might alleviate fatigue symptoms but there is no solid evidence on pharmacological treatments 14 . Identifying clinical and pathophysiological distinctive features of depression vs. fatigue symptoms might facilitate more specific treatment options.

Ideally, the distinction between a predominantly fatigue-associated vs. primarily depression-driven phenotype would allow also for a more sensitive and specific recognition of MS-related symptoms, better clinical evaluations and the development of patient- and problem-tailored treatments. To this end, we performed a study (Study 1) based on real-life routine data and investigated which clinical and paraclinical variables might be able to distinguish between (1) MS patients reporting mainly depressive symptoms, (2) MS patients predominantly complaining about fatigue, (3) MS patients reporting both symptoms and (4) MS patients with neither depression nor fatigue. To validate our results and to investigate the full cognitive profile of those groups, we repeated the analysis in a second study (Study 2), where a comprehensive neuropsychological test battery was applied. In a final step, we evaluated whether motor and cognitive fatigue correlate differently with the identified clinical features.

Materials and methods

Both Study 1 and Study 2 were performed in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki) for experiments involving humans and were approved by the local ethics committee of the University Hospital Frankfurt. In Study 1 all data were acquired during clinical routine and analyzed retrospectively. The retrospective analysis of anonymized clinical data was approved and the need for informed consent was waived by the local ethics committee of the University Hospital Frankfurt at the Goethe University Frankfurt (Nr. 2024-661). In Study 2, study-specific written informed consent was obtained before enrollment. All patients were diagnosed according to the revised 2017 McDonald criteria (Thompson et al., 2018).

Study 1

Study population

The population in Study 1 encompassed MS patients who underwent fatigue, depression, cognitive screening, clinical and lab testing as a part of the clinical routine at the Department of Neurology, Goethe University Frankfurt 01/2023-10/2023. Demographic and clinical characteristics were extracted from the database of the hospital. Physical disability was quantified by the Expanded Disability Status Scale (EDSS) and was always assessed by certified EDSS raters. Patients with the diagnosis of either CIS or RRMS according to the 2017 revision of the McDonald criteria were pooled into one category (“CIS/RRMS”). Patients with primary or inactive secondary progressive MS (SPMS) were categorized in “PMS” (progressive MS). DMTs were pooled into the categories “none”, “platform” (interferon-beta, glatiramer acetate), “oral” (teriflunomide, dimethyl fumarate, S1RP-modulators, cladribine), “natalizumab” and “anti-CD20 antibodies” (rituximab, ocrelizumab, ofatumumab). Disease activity in the last 6 months was defined by having either a clinical relapse or new, larger or contrast-enhancing lesions in the MRI imaging.

Neuropsychological testing

We employed Quick Inventory Depression Symptomatology (QIDS-SR16, a self-report questionnaire); Fatigue Scale for Motor and Cognitive Functions (FSMC, a questionnaire measuring motor and cognitive as well as total fatigue); and the Symbol Digit Modalities Test (SDMT, gold standard for screening for cognitive deficits in MS; it measures mainly information processing speed impairment; oral version). The cut-off for at least mild depression in QIDS-SR16 was 6 points. Scoring at least 43 points on FSMC classified as having at least mild fatigue. Subscale-specific cut-offs (motor and cognitive fatigue) were ≥ 22 points. To standardize the raw scores and allow for comparison of SDMT performance between patients of different age and educational levels, z-scores were calculated using normative data 15 .

Laboratory parameters

Laboratory and flow cytometry panel data were acquired from blood samples drawn for routine clinical monitoring. The following parameters were determined: C-reactive protein (CRP); white blood cells count (WBC) and lymphocyte count; flow cytometry: CD19 + and CD20 + B-cells, CD3 + and CD3 + HLA-DR + T-cells, CD3 + CD4 + T-helper cells, CD3 + CD8 + cytotoxic T-cells, CD3 + CD8 + CD38 + activated cytotoxic T-cells, CD3-CD16 + CD56 + natural killer cells (NK-cells) as well as CD4/8 ratio.

Statistical analysis

Based on the results from the QIDS-SR16 and FSMC testing, patients were grouped in reporting depressive symptoms only (QIDS-SR16 ≥ 6 and FSMC ≤ 42), fatigue symptoms only (QIDS-SR16 ≤ 5 and FSMC ≥ 43), both depressive and fatigue symptoms (QIDS-SR16 ≥ 6 and FSMC ≥ 43) or neither depressive nor fatigue symptoms (QIDS-SR16 ≤ 5 and FSMC ≤ 42). One-way analysis of variance (ANOVA) was employed to test for differences between the four groups with regard to each of the parameters in the study. Post-hoc t-tests were computed where relevant. For categorical variables, chi-square tests of independence were used to test for differences between the four groups. The full list of the computed measures is reported in Table  1 . In the final step, we entered variables exhibiting significant effects in the ANOVA or in the chi-square tests into a multinomial logistic regression model which tested for their utility as predictors of predominant self-reported symptomatology (depression only vs. fatigue only vs. both vs. none).

Table 1 Based on the results from the QIDS-SR16 and FSMC testing in study 1, patients were grouped in reporting depressive symptoms only (QIDS-SR16 ≥ 6 and FSMC ≤ 42), fatigue symptoms only (QIDS-SR16 ≤ 5 and FSMC ≥ 43), both depressive and fatigue symptoms (QIDS-SR16 ≥ 6 and FSMC ≥ 43) or neither depressive nor fatigue symptoms (QIDS-SR16 ≤ 5 and FSMC ≤ 42).).
N
Mean Std. Deviation p -value Age (years) Depression only 13 30.08 8.53
< 0.001
Fatigue only 25 46.12 12.18 None 40 37.53 12.12 Both 70 42.87 13.22 SDMT (z-score) Depression only 13 0.22 0.87
< 0.001
Fatigue only 24 − 0.99 1.34 None 40 0.15 1.08 Both 69 − 0.88 1.40 EDSS Depression only 13 1.54 1.35
< 0.001
Fatigue only 25 3.48 1.87 None 39 1.73 1.56 Both 70 3.27 1.55 QIDS-SR16 Depression only 13 8.08 2.69
< 0.001
Fatigue only 25 3.76 1.17 None 40 2.63 1.76 Both 70 11.69 4.43 FSMC total score Depression only 13 32.38 5.06
< 0.001
Fatigue only 25 59.80 16.56 None 40 28.35 6.38 Both 70 70.43 14.99 FSMC cognitive score Depression only 13 15.54 3.31
< 0.001
Fatigue only 25 27.28 9.33 None 40 12.60 2.93 Both 69 33.07 8.84 FSMC motor score Depression only 13 17.62 2.60
< 0.001
Fatigue only 25 32.88 9.12 None 40 16.25 5.09 Both 69 36.68 7.26 Diagnosis (n) CIS/RRMS PMS 0.053 Depression only 12 1 Fatigue only 13 12 None 29 10 Both 44 25 Sex (n) female male 0.689 Depression only 8 5 Fatigue only 15 10 None 21 19 Both 45 25 DMT (n) none platform oral NTZ antiCD20 0.612 Depression only 2 0 1 1 9 Fatigue only 5 1 2 0 17 None 12 2 2 2 20 Both 23 5 8 4 27 Disease activity in the last 6 months (n) no yes 0.765 Depression only 7 5 Fatigue only 16 8 None 20 18 Both 38 28 Disease duration (years) Depression only 13 6.14 4.50 0.084 Fatigue only 25 11.19 8.38 None 37 6.75 6.50 Both 68 8.98 8.08 CRP (mg/dl) Depression only 11 0.23 0.29 0.163 Fatigue only 20 0.16 0.29 None 33 0.13 0.12 Both 53 0.23 0.24 White blood cell count (x 10 9 /L) Depression only 12 6.92 2.05
0.042
Fatigue only 21 6.91 2.00 None 35 6.09 1.98 Both 56 7.52 2.59 Lymphocyte count (x 10 9 /L) Depression only 11 2.03 0.92
0.043
Fatigue only 20 1.58 0.56 None 33 1.54 0.56 Both 49 1.93 0.86 CD19 + cells (/µL) Depression only 10 62.36 192.44 0.290 Fatigue only 19 22.71 73.03 None 27 118.76 200.48 Both 40 106.08 201.07 CD20 + cells (/µL) Depression only 10 67.23 203.06 0.346 Fatigue only 19 33.28 91.92 None 27 127.35 207.39 Both 40 112.48 206.40 CD3 + cells (/µL) Depression only 9 1691.67 212.86 0.095 Fatigue only 15 1345.27 134.29 None 26 1215.65 80.79 Both 36 1488.94 102.99 CD3 + CD4 + cells (/µL) Depression only 9 1126.67 84.29
0.023
Fatigue only 15 866.40 94.25 None 26 759.65 52.26 Both 36 997.61 72.03 CD3 + CD8 + cells (/µL) Depression only 9 541.44 134.13 0.685 Fatigue only 15 446.53 55.72 None 26 419.62 38.22 Both 36 471.67 49.77 CD4+/CD8 + ratio Depression only 9 2.71 0.40 0.225 Fatigue only 16 2.06 0.21 None 26 2.11 0.22 Both 36 2.76 0.31 CD3-CD16 + CD56 + cells (/µL) Depression only 9 291.11 52.72 0.262 Fatigue only 14 230.57 22.84 None 26 245.31 19.28 Both 36 298.22 26.14 CD3 + CD8 + CD38 + cells (/µL) Depression only 9 480.44 125.56 0.586 Fatigue only 14 378.21 47.42 None 26 357.46 31.83 Both 36 375.47 41.27 CD3 + HLA-DR + cells (/µL) Depression only 9 179.33 22.98 0.844 Fatigue only 14 148.50 17.22 None 26 157.85 14.02 Both 36 162.92 16.20 One-way analysis of variance (ANOVA) was employed to test for differences between the four groups with regard to each of the parameters in the study. Post-hoc t-tests were computed where relevant. For categorical variables, chi-square tests of independence were used to test for differences between the four groups. SDMT Symbol Digit Modalities Test, EDSS Expanded Disability Status Scale, QIDS-SR16 Quick inventory of depressive symptomatology (16-item) (Self-Report), FSMC Fatigue Scale for Motor and Cognitive Functions, DMT disease-modifying treatment, CRP C-reactive protein.

Based on the results from the QIDS-SR16 and FSMC testing in study 1, patients were grouped in reporting depressive symptoms only (QIDS-SR16 ≥ 6 and FSMC ≤ 42), fatigue symptoms only (QIDS-SR16 ≤ 5 and FSMC ≥ 43), both depressive and fatigue symptoms (QIDS-SR16 ≥ 6 and FSMC ≥ 43) or neither depressive nor fatigue symptoms (QIDS-SR16 ≤ 5 and FSMC ≤ 42).).

Disease activity

in the last 6 months (n)

One-way analysis of variance (ANOVA) was employed to test for differences between the four groups with regard to each of the parameters in the study. Post-hoc t-tests were computed where relevant. For categorical variables, chi-square tests of independence were used to test for differences between the four groups.

SDMT Symbol Digit Modalities Test, EDSS Expanded Disability Status Scale, QIDS-SR16 Quick inventory of depressive symptomatology (16-item) (Self-Report), FSMC Fatigue Scale for Motor and Cognitive Functions, DMT disease-modifying treatment, CRP C-reactive protein.

Study 2

Study population

To validate our findings from Study 1 and examine the differences in the full neuropsychological performance of the investigated groups, we re-analysed the data from a previously published work, which was focused on cognitive deficits in MS 16 . MS patients were recruited in our outpatient clinic and performed a full neuropsychological testing including depression, fatigue and cognitive assessment as well as a behavioral task and an eye-tracking paradigm. The results from the analysis of the latter two tasks have been reported elsewhere 16 , 17 . EDSS measurement and group membership were defined similarly to Study 1. Patients were categorized into four categories: reporting depressive symptoms only (BDI-II ≥ 9 and FSMC ≤ 42), fatigue only (BDI-II ≤ 8 and FSMC ≥ 43), both depressive and fatigue symptoms (BDI-II ≥ 9 and FSMC ≥ 43) or neither depressive nor fatigue symptoms (BDI-II ≤ 8 and FSMC ≤ 42).

Neuropsychological testing

The full neuropsychological test battery consisted of the Rey Complex Figure Test (RCFT), written-version of SDMT, Verbaler Lern- und Merkfähigkeitstest (VLMT, a German adaptation of the Rey Auditory Verbal Learning Test), Paced Auditory Serial Addition Test (PASAT), Trail Making Test (TMT, versions A and B), Regensburger Wortflüssigkeits-Test (RWT, a German verbal fluency test), Wortschatztest (WST, a German vocabulary test), Beck Depression Inventory-II (BDI-II). Additionally, FSMC as well as the 9-hole peg test (9HPT) and a simple reaction time task to test for alertness were applied. The full list of the computed measures from the neuropsychological tests is reported in Table  2 .

Table 2 Based on the results from the BDI-II and FSMC testing in study 2, patients were categorized into four categories: reporting depressive symptoms only (BDI-II ≥ 9 and FSMC ≤ 42), fatigue only (BDI-II ≤ 8 and FSMC ≥ 43), both depressive and fatigue symptoms (BDI-II ≥ 9 and FSMC ≥ 43) or neither depressive nor fatigue symptoms (BDI-II ≤ 8 and FSMC ≤ 42).). Since only one patient reported depressive without any fatigue symptoms, this subcategory was removed from the further analysis. One-way analysis of variance (ANOVA) was employed to test for differences between the four groups with regard to each of the parameters in the study. Post-hoc t-tests were computed where relevant. For categorical variables, chi-square tests of independence were used to test for differences between the four groups.
N
Mean Std. Deviation p -value Age (years) Fatigue only 17 46.76 13.68 0.06 None 12 35.08 9.82 Both 25 44.24 14.41 SDMT (z-score) Fatigue only 17 −1.67 1.22
0.01
None 12 −0.23 1.01 Both 25 −1.01 1.08 EDSS Fatigue only 16 3.22 1.46
< 0.001
None 11 1.95 1.04 Both 22 4.05 1.38 BDI-II Fatigue only 17 5.06 2.11
< 0.001
None 12 4.33 2.39 Both 25 17.04 8.22 FSMC total score Fatigue only 17 66.00 10.06
< 0.001
None 12 30.58 6.42 Both 25 71.32 15.76 FSMC cognitive score Fatigue only 17 30.88 5.63
< 0.001
None 12 14.00 4.18 Both 25 34.48 9.97 FSMC motor score Fatigue only 17 35.12 6.01
< 0.001
None 12 16.58 4.12 Both 25 36.84 6.98 Diagnosis (n) RRMS PMS
0.04
Fatigue only 12 5 None 12 0 Both 15 10 Sex (n) female male 0.95 Fatigue only 13 4 None 9 3 Both 18 7 Disease duration (years) Fatigue only 17 9.55 9.57 0.18 None 12 4.75 5.86 Both 25 10.94 10.7 Alertness (reaction time, ms) Fatigue only 16 371.57 167.88 0.25 None 12 290.13 54.61 Both 24 358.97 135.96 9-hole peg test (dominant hand; s) Fatigue only 16 25.03 5.5 0.25 None 12 22.42 9.32 Both 25 29.26 15.8 9-hole peg test (non-dominant hand; s) Fatigue only 16 25.97 5.43 0.09 None 12 21.67 5.5 Both 25 30.13 14.88 RCFT (immediate recall; z-score) Fatigue only 16 −0.39 1.63 0.41 None 12 −1.00 1.29 Both 25 −0.98 1.46 VLMT verbal learning (z-score) Fatigue only 17 −0.14 1.5 0.38 None 12 0.45 0.86 Both 25 0.02 0.97 VLMT proportion remembered (z-score) Fatigue only 17 −0.39 1.21 0.27 None 12 0.21 0.7 Both 25 −0.10 0.92 VLMT free recall (z-score) Fatigue only 16 −0.25 1.56 0.05 None 12 0.76 0.67 Both 25 −0.16 1.07 PASAT (z-score) Fatigue only 16 −0.85 1.46 0.47 None 12 −0.20 1.45 Both 24 −0.60 1.26 TMT (version A; z-score) Fatigue only 17 −0.69 0.87 0.95 None 12 −0.63 0.61 Both 25 −0.71 0.65 TMT (B/A; z-score) Fatigue only 17 0.01 0.83 0.94 None 12 0.04 0.71 Both 25 −0.06 0.98 RWT (phonemic fluency; z-score) Fatigue only 16 −0.87 0.5 0.75 None 11 −0.71 0.75 Both 25 −0.82 0.52 RWT (semantic fluency; z-score) Fatigue only 16 −0.76 0.84 0.56 None 12 −0.51 0.82 Both 25 −0.78 0.63 Vocabulary test (z-score) Fatigue only 15 0.05 0.8 0.86 None 12 0.17 0.65 Both 25 0.16 0.63 SDMT Symbol Digit Modalities Test, EDSS Expanded Disability Status Scale, BDI-II Beck Depression Inventory-II, FSMC Fatigue Scale for Motor and Cognitive Functions, VLMT Verbaler Lern- und Merkfähigkeitstest, PASAT Paced Auditory Serial Addition Test, TMT Trail Making Test, RWT Regensburger Wortflüssigkeits-Test.

Based on the results from the BDI-II and FSMC testing in study 2, patients were categorized into four categories: reporting depressive symptoms only (BDI-II ≥ 9 and FSMC ≤ 42), fatigue only (BDI-II ≤ 8 and FSMC ≥ 43), both depressive and fatigue symptoms (BDI-II ≥ 9 and FSMC ≥ 43) or neither depressive nor fatigue symptoms (BDI-II ≤ 8 and FSMC ≤ 42).). Since only one patient reported depressive without any fatigue symptoms, this subcategory was removed from the further analysis. One-way analysis of variance (ANOVA) was employed to test for differences between the four groups with regard to each of the parameters in the study. Post-hoc t-tests were computed where relevant. For categorical variables, chi-square tests of independence were used to test for differences between the four groups.

VLMT proportion remembered

(z-score)

VLMT free recall

(z-score)

SDMT Symbol Digit Modalities Test, EDSS Expanded Disability Status Scale, BDI-II Beck Depression Inventory-II, FSMC Fatigue Scale for Motor and Cognitive Functions, VLMT Verbaler Lern- und Merkfähigkeitstest, PASAT Paced Auditory Serial Addition Test, TMT Trail Making Test, RWT Regensburger Wortflüssigkeits-Test.

Statistical analysis

Since only one patient reported depressive without any fatigue symptoms, this subcategory was removed from the further analysis. Thus, patients were grouped into reporting fatigue only vs. both depression and fatigue vs. neither depression nor fatigue. Similarly to Study 1, one-way ANOVA was employed to test for differences between the three groups with regard to each of the clinical and neuropsychological parameters in the study. Post-hoc t-tests were computed where relevant. For categorical variables, chi-square tests of independence were used to test for differences between the four groups. In the final step, we entered variables exhibiting significant effects in the ANOVA or in the chi-square tests into a multinomial logistic regression model which tested for their utility as predictors of predominant symptomatology (fatigue only vs. both vs. none).

For both studies statistical analysis was performed using IBM SPSS Statistics version 29.0.2.0 and GraphPad Prism 10.4.0. For descriptive purposes, mean and standard deviations were calculated for each parameter. All significance levels were set to p  < 0.05. Due to the exploratory nature of our investigation, no correction for multiple comparisons was applied.

Motor vs. cognitive fatigue

To test whether our main findings differ with regard to the two subscales of the FSMC instrument (motor vs. cognitive fatigue), we pooled together the data from the two studies and computed partial correlations between the FSMC motor/cognitive subscales with each of the variables which emerged significant from the analyses of Study 1 and 2, while controlling for the presence of self-reported depressive symptoms. Each pair of correlations (e.g. “FSMC motor subscale and age” vs. “FSMC cognitive subscale and age”) was compared using z-statistics. Finally, FSMC motor and cognitive scores were compared between the groups CIS/RRMS and PMS using one-way ANOVA.

Results

Study 1

In total n  = 156 patients were recruited for the study: n  = 105 CIS/RRMS ( n  = 103 RRMS, n  = 2 CIS) as well as n  = 51 PMS ( n  = 22 SPMS and n  = 29 PPMS). More detailed description of the clinical and paraclinical parameters of the cohort is given in Table  1 .

Patients reporting depressive symptoms only were younger in comparison to those reporting only fatigue symptoms (30.08 ± 12.18 vs. 46.12 ± 12.18 years, post-hoc t-test p  = 0.0005, Table  1 ; Fig.  1 A), had a higher information processing speed (SDMT z-score 0.22 ± 0.87 vs. −0.99 ± 1.34, p  = 0.007, Table  1 ; Fig.  1 B) and lower EDSS (1.54 ± 1.35 vs. 3.48 ± 1.87, p  = 0.001, Table  1 ; Fig.  1 C). Additionally, they had higher CD3 + CD4 + T-helper cell counts as compared to patients without depressive or fatigue symptoms (1126.67/µl ± 84.29 vs. 759.65/µl ± 52.26, p  = 0.02, Table  1 ; Fig.  1 D), while patients with fatigue only did not differ from the latter (866.40/µl ± 94.25 vs. 759.65/µl ± 52.26, p  = 0.36, Table  1 ; Fig.  1 D). A significant ANOVA effect was achieved also for white blood cell count (WBC). However, this difference was driven mainly by the significant effect of patients reporting both depressive and fatigue vs. those without any symptoms (WBC: 7.52 × 10 9 /L ± 2.59 vs. 6.09 × 10 9 /L ± 1.98, p  = 0.004, Table  1 ). Lymphocyte counts differed also significantly but mainly between patients with only depressive vs. those without any symptoms (2.03 × 10 9 /L ± 0.92 vs. 1.54 × 10 9 /L ± 0.56, p  = 0.02, Table  1 ). Differences in the frequencies of diagnoses (CIS/RRMS vs. PMS) across the four categories of symptomatology was very close to but failed to reach significance ( p  = 0.053, Table  1 ).

Association of clinical/paraclinical features with different MS phenotypes reporting mainly fatigue, depressive, both or none of those symptoms. Patients reporting depressive symptoms only were younger in comparison to those reporting only fatigue symptoms ( A ), had a higher information processing speed ( B ) and lower EDSS ( C ). Additionally, they had higher CD3 + CD4 + T-helper cell counts as compared to patients without depressive or fatigue symptoms, while patients with fatigue only did not differ from the latter ( D ).

A multinomial logistic regression was performed to create a model of the relationship between the predictor variables age, SDMT z-score, EDSS, CD3 + CD4 + T-helper cell counts, WBC and lymphocyte count and the four phenotypes. The fit between the model containing only the intercept and data improved with the addition of the predictor variables, χ 2
(18, 83)  = 39.16, Nagelkerke R 2  = 0.41, p  = 0.003. The likelihood ratio tests for each of the predictors demonstrated a significant effect for SDMT z-score only (χ 2  = 11.39, p  = 0.01). When comparing directly in the model patients who reported only depressive vs. no symptoms, CD3 + CD4 + T-helper cells count was the only significant predictor ( p  = 0.03). When distinguishing in the model between patients with fatigue only vs. no symptoms, SDMT z-score was the only significant predictor ( p  = 0.01). Adding diagnosis as a predictor revealed similar findings and no other predictors reached significance.

Study 2

In total n  = 54 patients were recruited for the study: n  = 39 RRMS as well as n  = 15 PMS. More detailed description of the clinical and paraclinical parameters of the cohort is given in Table  1 . Patients reporting fatigue only as compared to patients without depressive or fatigue symptoms had lower SDMT z-scores (−1.67 ± 1.22 vs. −0.23 ± 1.01, p  = 0.001, Table  2 ) and higher EDSS (3.22 ± 1.46 vs. 1.95, p  = 0.02, Table  2 ). Diagnosis distribution differed, too ( p  = 0.04): all PMS patients reported either fatigue or depressive symptoms, while 30.8% of the RRMS patients did not report any of those (Table  2 ).

We repeated the multinomial logistic regression with EDSS, SDMT z-score and diagnosis as predictors. The fit between the model containing only the intercept and data improved with the addition of the predictor variables, χ 2
(6, 48)  = 33.54, Nagelkerke R 2  = 0.563, p  < 0.001. The likelihood ratio tests for each of the predictors demonstrated a significant effect for both EDSS (χ 2  = 12.70, p  = 0.002) and SDMT z-score (χ 2  = 13.64, p  = 0.001) but not for diagnosis (χ 2  = 2.61, p  = 0.27). When comparing patients with fatigue only vs. no symptoms directly in the model, SDMT z-score ( p  = 0.02) and diagnosis ( p  < 0.001) were significant predictors. When distinguishing patients with both symptoms from those with none, the model revealed significant contributions from EDSS only ( p  = 0.013).

Interestingly, no neuropsychological test other than SDMT distinguished among the three groups (fatigue vs. fatigue and depression vs. no symptoms). The VLMT free recall trial was close to but did not reach significance ( p  = 0.05). Adding the VLMT free recall trial performance to the multinomial logistic regression and repeating the analysis did not reveal new significant predictors.

Pooled analysis

In total the data of n  = 207 MS patients from both studies was submitted to partial correlational analysis where reported depressive symptoms were controlled for. While age correlated with both FSMC motor ( r  = 0.375, p  < 0.001) and cognitive ( r  = 0.211, p  = 0.003) subscales, correlation between age and motor fatigue was stronger (z = 3.602, p  < 0.001). SDMT z-score correlated with both FSMC motor ( r =−0.396, p  < 0.001) and cognitive ( r =−0.400, p  < 0.001) scores and those two correlations did not differ from each other (z = 0.092, p  > 0.05). While EDSS correlated with both FSMC motor ( r  = 0.561, p  < 0.001) and cognitive ( r  = 0.365, p  < 0.001) scores, correlation between EDSS and motor fatigue was stronger (z = 4.633, p  < 0.001) (Fig.  2 ). PMS patients scored higher on both FSMC motor (35.62 ± 10.35 vs. 26.90 ± 10.55, p  < 0.001) and cognitive (28.59 ± 12.45 vs. 24.85 ± 11.03, p  = 0.034) scales compared to CIS/RRMS.

Association of clinical/paraclinical features with motor vs. cognitive fatigue scores. While age correlated with both FSMC motor and cognitive subscales, correlation between age and motor fatigue was stronger ( A ). SDMT z-score correlated with both FSMC motor and cognitive scores and those two correlations did not differ from each other ( B ). While EDSS correlated with both FSMC motor and cognitive scores, correlation between EDSS and motor fatigue was stronger ( C ).

Discussion

Our findings suggest that MS patients reporting predominantly depressive vs. fatigue symptoms have distinguishing clinical and paraclinical features, which might facilitate the implementation of a more specific treatment and extend the understanding of their pathophysiology.

Patients reporting predominantly fatigue were characterized by older age, lower information processing speed and higher degree of physical disability. However, they did not differ with regard to disease activity in the last 6 months as defined by relapse or MRI dynamics and the used DMTs did not have an effect on these findings. Progression independent of relapse activity (PIRA), driven mainly by central nervous system intrinsic inflammatory and neurodegenerative processes which are different from the autoimmune inflammation arising from the periphery might be discussed as a possible factor contributing to those findings. Diagnosis (PMS vs. CIS/RRMS) was very close to reaching significance as a distinguishing feature in the first study but reached significance in the second. Here, fatigue was more strongly associated with the progressive course of the disease. Furthermore, PMS patients scored higher than CIS/RRMS on both motor and cognitive fatigue. Remarkably, in the first study duration of the disease was also close to significance. This could be an indirect hint towards the role of accumulating PIRA and smouldering inflammation, since they reveal their effects more strongly on the long term. However, this tentative explanation needs further specifically designed studies and more evidence.

“Depression only” MS phenotype was characterized by younger age, higher cognitive processing speed, lower degree of physical disability, higher T h (CD3 + CD4+) cell count compared to fatigue-only MS patients and higher lymphocyte count in comparison to MS patients who reported neither depressive nor fatigue symptoms. Since peripheral inflammation is the hallmark of relapse-associated worsening (RAW) especially in the early stages of disease, the higher inflammatory cell count might be interpreted as an early subclinical disease activity which might underly depressive symptoms in MS. Any further considerations regarding this finding, however, should take into account the fact that the MS-associated inflammation is particularly complex and far from this simplified view. Furthermore, it would be of paramount interest to see whether more specific surface markers of peripheral immune cells are associated with depressive symptoms.

In a previous study we demonstrated that depressive symptoms in newly diagnosed MS patients are associated with peripheral inflammatory markers and gadolinium-enhancing lesions 18 . Furthermore, peripheral inflammation is a distinguishing characteristic of a subgroup of major depression patients who exhibit an “inflammation-driven” and particularly treatment-refractory disease course 19 . Importantly, higher lymphocyte and CD3 + CD4 + counts have been associated with current major depressive episodes 20 corroborating our findings and suggesting that the “depression only” MS phenotype might be associated with peripheral inflammation. This might be also the explanation why this MS phenotype is apparently less affected by the disease, as suggested by the higher information processing speed, less physical disability and younger age: peripheral inflammation is more predominant in early stages of the disease and thus younger patients, who have a lesser disease burden are more probable to report predominantly depressive symptoms, while fatigue might be hypothetically an early sign of PIRA. During the further course of the disease both peripheral and CNS-intrinsic inflammation unfold their effects and both depression and fatigue accumulate.

Since SDMT performance differentiated between fatigue-only and depression-only patients, we assessed the performance in a second cohort of MS patients undergoing a full neuropsychological evaluation. Interestingly, none of the further tests differed between the groups and only the VLMT free recall trial was close to significance ( p  = 0.05). This could indicate an overlap between fatigue and cognitive deficits which is more specific for information processing speed. Alternatively, since SDMT is a rather sensitive screening tool for cognitive decline in MS 21 , 22 , this might suggest that fatigue-associated cognitive impairment occurs first in the information processing domain before further, more strongly pronounced deficits unfold.

It remains challenging to disentangle fatigue from depression, since “reduced energy or fatigue” is one of the ICD-11 criteria for single episode depressive disorder. However, there are MS patients who do not meet the criteria for depression but report significant amounts of fatigue. Moreover, the fatigue findings of our study seem to be not motor- or cognitive-specific since both FSMC motor and cognitive subscales correlated with age, EDSS and SDMT performance, even when controlling for reported depressive symptoms.

Our study is not without limitations. While the second cohort had the advantage of having a full neuropsychological evaluation, it was very limited in size and did not allow us to investigate depression-specific symptoms. Larger sample size as well as follow-ups would allow for a more detailed insight into the different MS phenotypes. Due to the explorative nature of this study, we did not employ a standard diagnostic interview for confirming the presence or absence of a major depressive disorder, which might could have helped stratify the results and the two cohorts. Furthermore, since the data originated from the clinical routine, only a limited flow cytometry analysis was available. A more detailed investigation of the cell subpopulations contributing to the different symptoms would allow a better understanding of the involved inflammatory processes. Last but not least, we studied self-reported symptoms of depression but did not include diagnostic interviews and clinician ratings.

Our study suggests that MS patients reporting predominantly fatigue are characterized by older age, lower information processing speed, higher degree of physical disability and possibly by a progressive disease course. MS patients reporting mainly depressive but no fatigue symptoms are younger, perform better in SDMT and are less physically impaired, while exhibiting higher peripheral inflammatory cell counts. This allows a more precise identification of fatigue- vs. depression-predominant clinical phenotypes and possibly a better understanding of the underlying pathophysiological possibilities with fatigue possibly more closely associated with CNS-intrinsic, PIRA-related processes while depression being rather linked to peripheral inflammation.

Author contributions

Conceptualization: YY, IKP. Methodology: YY, MH, IKP. Investigation: YY, KK, JG, MH, MM. Writing—Original Draft: YY, IKP. Writing—Review and Editing: YY, KK, MH, IKP, MM. Funding Acquisition: YY. Supervision: YY.

Funding

Open Access funding enabled and organized by Projekt DEAL.

This research project was supported by the Deutsche Forschungsgemeinschaft (project number 511312822).

Data availability

Data that support the findings of this study are available upon reasonable request to the corresponding author.

Declarations

Competing interests

Yavor Yalachkov has been supported by travel grants from Alexion, Novartis and SanofiGenzyme and has received honoraria for active participation in advisory boards andspeaker/writing honoraria by Alexion, Amgen, Bristol Meyers Squibb, KassenärztlicheVereinigung Hessen, Landesärztekammer Hessen, Merck, Neuraxpharm, Novartis, RG Gesellschaft fürInformation und Organization mbH, Roche, Sanofi Genzyme, TEVA; his research is funded byDeutsche Forschungsgemeinschaft (DFG), Fachbereich Medizin der Goethe-UniversitätFrankfurt, Heinrich und Erna Schaufler Stiftung and Novartis. Kimberly Koerbel is a participantat the Junior Clinician Scientist Program funded by Goethe University Frankfurt and receivedspeaker honoraria by Novartis. Johannes Gehrig has been supported by a travel grant fromSanofi. Marion Hug reports no conflicts of interest. Iris‐Katharina Penner has receivedhonoraria for speaking at scientific meetings, serving at scientific advisory boards andconsulting activities from Adamas Pharma, Almirall, Bayer Pharma, Biogen, BMS, Celgene, Sanofi‐Genzyme, Janssen, Merck Serono GmbH (an affiliate of Merck KGaA), Novartis, Roche and Teva; she has received research support from the German MS Society, Celgene, Roche, Teva and Novartis. Michelle Maiworm received speaking honoraria from Roche, Synlab and Novartis; her research was funded by the Faculty of Medicine at the GoetheUniversity Frankfurt and the Goethe Research Academy for Early Career Researchers (GRADE) of the Goethe University Frankfurt, the Heinrich und Erna Schaufler-Foundation aswell as the MS Foundation of the State of Hesse.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

Associated Data

Data Availability Statement

Data that support the findings of this study are available upon reasonable request to the corresponding author.