Article Text

Endocrine and immune substrates of depressive symptoms and fatigue in multiple sclerosis patients with comorbid major depression
  1. Stefan M Gold1,2,
  2. Schulamith Krüger1,
  3. Kristin J Ziegler3,
  4. Thorsten Krieger4,
  5. Karl-Heinz Schulz3,
  6. Christian Otte5,
  7. Christoph Heesen1,2
  1. 1Institute for Neuroimmunology and Clinical Multiple Sclerosis Research (inims), University Hospital Hamburg, Germany
  2. 2Department of Neurology, University Hospital Hamburg, Hamburg, Germany
  3. 3Department of Medical Psychology, University Hospital Hamburg, Hamburg, Germany
  4. 4Department of Immunology, University Hospital Hamburg, Hamburg, Germany
  5. 5Department of Psychiatry, University Hospital Hamburg, Hamburg, Germany
  1. Correspondence to Dr S M Gold, Institute for Neuroimmunology und Clinical Multiple Sclerosis Research (inims), Falkenried 94, D-20251 Hamburg, Germany;{at}


Objective Depression and fatigue are among the most common symptoms of multiple sclerosis (MS). These symptoms frequently co-occur and partially overlap in MS but their underlying biological substrates are unclear. In this study, the relative role of cytokines and hypothalamic–pituitary–adrenal (HPA) axis activity in depression and fatigue were examined in patients with relapsing–remitting MS (RRMS).

Methods HPA axis function and frequency of stimulated cytokine (interferon γ (IFNγ) and tumour necrosis factor α (TNFα)) producing T cells was measured cross sectionally in 44 female patients with RRMS. All subjects completed a neurological examination, the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I) and self-report questionnaires.

Results 10 patients met diagnostic criteria for major depressive disorder (MDD). MS patients with comorbid MDD showed normal morning but elevated evening salivary cortisol levels, resulting in a flattened slope. While a higher frequency of cytokine producing CD8+ T cells was also seen in MS patients with MDD, these markers were more closely associated with fatigue than depression.

Conclusions This study supports a role for HPA axis hyperactivity in major depression in MS. In addition, inflammatory and neuroendocrine factors may differentially mediate fatigue and depressive symptoms.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Depression and fatigue are among the most common symptoms of multiple sclerosis (MS).1 Depression has a point prevalence of 13–30% and a lifetime risk of 25–50%,2 with a significant impact on cognitive function,3 quality of life,4 work performance5 and treatment compliance.6 It is also one of the strongest predictors of suicide.7 Fatigue is experienced by up to two-thirds of MS patients and often perceived as the most debilitating symptom.8 9 Importantly, fatigue represents the leading cause for absence from work.10 While this clearly illustrates the importance of depression and fatigue for the socioeconomic and psychosocial burden of MS, the underlying mechanisms are not well understood.

Earlier studies have found no evidence for an association of depression and fatigue in MS,11 12 however, later reports have usually confirmed a moderate correlation.13–17 This suggests that while sometimes linked in MS, fatigue and depression may be mediated by at least partially independent pathological mechanisms. Given the high clinical relevance of depression and fatigue in MS, a better understanding of potentially underlying biological pathways is crucial for developing effective therapeutic strategies.

Hyperactivity of the hypothalamic–pituitary–adrenal (HPA) axis is among the most consistently reproduced biological findings in psychiatric patients with major depressive disorder (MDD).18 A role for inflammation in the pathogenesis of depression has also been demonstrated.19 Fatigue and depressive symptoms often occur together as part of ‘sickness behaviour’ and may share some biological correlates20 but the differential role of inflammatory and neuroendocrine pathways remains to be elucidated.

Inflammation is the hallmark of MS and cytokines such as tumour necrosis factor α (TNFα), which have been implicated in major depression,21 22 play a key role in MS pathology. Associations between cytokines, such as TNFα and interferon γ (IFNγ), and fatigue have been described in MS.23 24 In addition, a growing number of studies provide evidence for HPA axis hyperactivity in up to 50% of MS patients,25 26 and gene variants involved in HPA axis regulation have recently been associated with MS.27 Clinical studies, including both relapsing–remitting (RRMS) as well as progressive MS patients, suggest that HPA axis hyperactivity in MS is associated with more progressive disease and global neurodegeneration.28–31 However, there are indications that during the earlier phases of MS, subtle HPA axis alterations may be a marker for depressive symptomatology.32

In the current study, we examined the role of HPA axis activity and cytokines (IFNγ, TNFα) in T cell subpopulations for major depression in patients with RRMS. We hypothesised that patients with comorbid MDD would exhibit HPA axis hyperactivity and higher frequencies of cytokine producing T cells. Finally, we expected neuroendocrine and immune markers to be differentially associated with depression and fatigue, respectively.


Participants and recruitment

Patients were recruited through the MS outpatient clinic at the University Hospital Hamburg Eppendorf. We contacted female patients from our database within the age range 20–50 years who had been diagnosed with clinically definite RRMS.33 Patients were excluded if during an initial phone interview they met exclusion criteria such as a medical history of endocrine abnormalities (diabetes mellitus, uncontrolled thyroid disease), neoplasms, addiction or substance abuse, schizophrenia or anorexia. Patients were excluded if they had a relapse or received steroid/immunosuppressive treatment in the previous 4 weeks or were currently taking benzodiazepines, neuroleptics or antidepressants. Patients were allowed to be on immunomodulatory therapies if the treatment had not been started or changed in the previous 6 months. In addition, patients were excluded if they were pregnant.

Recruitment of patients was completed over a time period of 14 months. Eligible patients were scheduled for an appointment at the MS outpatient clinic and underwent a standard neurological examination to obtain clinical information and disability rating on the Expanded Disability Status Scale (EDSS).34 All procedures were approved by the ethics committee of the Medical Board Hamburg and all subjects provided written informed consent prior to enrolment in the study.

Diagnosis of MDD and clinical questionnaires

Diagnosis of current MDD was established by an experienced PhD level clinical psychologist (KJZ) using the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I). Depressive symptoms were quantified using the Hospital Anxiety and Depression Scale,35 a 14 item screening tool specifically developed to assess depression in medical and surgical settings which excludes symptoms that might arise from somatic aspects of the illness, such as fatigue. Fatigue due to MS was assessed using a recently developed and validated German version of the Modified Fatigue Impact Scale and the Fatigue Severity Scale.36

HPA axis assessment

Salivary cortisol samples were collected using Salivette devices (Sarstedt AG, Nümbrecht, Germany). We obtained circadian profiles (awakening, 11am, 3pm, 8pm and 10pm) on two consecutive days. On each day, we also collected samples for the cortisol awakening response (CAR) (awakening, +15 min, +30 min, +45 min and +60 min).

In addition, a low dose (0.25 mg) oral dexamethasone suppression test was obtained, as described previously.37 At 11pm of day 2, subjects ingested a dexamethasone pill and salivary cortisol levels were measured the next morning. Cortisol levels were determined in duplicate by radioimmunoassay (DRG, Marburg, Germany).

Flow cytometry

Intracellular cytokines (IFNγ, TNFα) in CD4+ and CD8+ T cell populations were measured using commercially available flow cytometry kits (Becton Dickinson Fastimmune Cytokine Systems Beckton Dickinson, Franklin Lakes, NJ, USA). Whole blood (500 μl) was stimulated with 0.2 μl/ml phytohaemagglutanin (PHA) and 5 μl costimulatory antibodies (CD28/CD49d) in the presence of the 10 μl secretion inhibitor Brefeldin A at 37°C. After a stimulation period of 6 h, 50 μl of EDTA were added to the sample in order to arrest activation and to remove adherent cells from the activation vessel. This step was followed by the simultaneous lysis of erythrocytes and fixation of leucocytes using 1 ml of BD FACS Lysing Solution. Cells were then washed and permeabilised with BD FACS Permeabilising Solution. After an additional wash, surface and intracellular staining antibodies were added in a single staining step. Finally, the cells were washed and fixed for flow cytometric analysis (FACS Calibur, Becton Dickinson). The method used a three colour staining to identify CD4 T cell responses (anticytokine FITC, CD69 PE, CD4 PerCP-Cy5.5) and a four colour staining system to identify CD8 T cell responses (anticytokine FITC, CD69 PE, CD8 PerCp-Cy5.5, CD3APC). CD69 staining was used to allow better clustering of cytokine positive cells and to ensure that cells defined as antigen responsive have been sufficiently stimulated to express this activation marker. The Becton Dickinson Fastimmune CD8 Anti-HU-IFNγ detection kit includes CD3 APC to avoid misidentification of natural killer cell responses (CD8 dim) on antigenic stimulus. Results are presented as per cent positive cells.

Statistical analysis

Corresponding cortisol values were averaged across the 2 days and log transformed. The circadian slope was computed using linear regression, as described previously.32 As a measure of overall cortisol levels, we computed the area under the circadian profile using trapezoidal estimation. Similarly, the area under the curve was computed for CAR.

Group differences between depressed and non-depressed RRMS patients for clinical characteristics, immune and endocrine markers were evaluated using independent sample t tests. If Levene's test indicated a violation of the equality of variance assumption, degrees of freedom were adjusted accordingly. When appropriate, group differences were also tested using univariate ANOVAs with disease modifying therapy (DMT) or EDSS entered as a covariate. Three group comparisons (untreated vs IFNβ vs glatiramer acetate) were conducted using univariate ANOVAs. Associations between endocrine and immune measures were tested using Pearson's correlation coefficients. Stepwise linear regression models were used to test the relative contribution of cortisol and immune markers to depression and fatigue severity. Values of p<0.05 were considered statistically significant. All statistical analyses were conducted using Predictive Analytics Software Statistics 18.0 for Macintosh.


Forty-four female RRMS patients were enrolled in the study, 10 (23%) of who met diagnostic criteria for current MDD. Patients with comorbid MDD and non-depressed MS patients were well matched (see table 1). There were no differences in DMTs between the groups. As expected in this population, several patients were on symptomatic therapies, including stimulants (modafinil n=3), pain/spasticity medication (ibuprofen n=2, gabapentin n=1), anticholinergic drugs for bladder symptoms (oxybutynin n=1, throspiumchlorid n=1), β blocker eyedrops (timolol n=1), thyroid hormones (L-thyroxin n=3) or homeopathic substances (n=3). While there was a slight difference in disability (EDSS), disease severity (as measured by the EDSS disability increase per year since diagnosis) was not significantly different between the groups (see table 1). Disease duration was not significantly different and was also not correlated with HADS depression scores (r=−0.01, p=0.93) or fatigue (r=−0.03, p=0.88).

Table 1

Clinical characteristics of relapsing–remitting multiple sclerosis patients with or without comorbid major depressive disorder

As expected, patients with MDD had significantly higher levels of depressive symptoms, as measured by the HADS depression subscale, as well as higher levels of anxiety and fatigue (see table 1).

HPA axis activity

MS patients with comorbid MDD showed hyperactivity of the HPA axis, characterised by normal morning cortisol but elevated evening levels (figure 1A), resulting in higher cortisol values across the day (area under the circadian curve, MS 17.5+1.1; MS+MDD 22.0+1.5; p=0.04) and a significantly flatter slope (p=0.017, figure 1B). No significant differences were seen in the CAR (p=0.88, figure 1C). In addition, there were no significant differences in cortisol patterns after overnight dexamethasone suppression (p=0.27, figure 1D). Cortisol slope was not significantly associated with CAR (r=−0.12, p=0.43) or post dexamethasone cortisol (r=−0.16, p=0.30) but there was a significant association between CAR and post dexamethasone cortisol (r=0.45, p<0.01). The difference in cortisol slope between depressed and non-depressed patients remained significant after statistically controlling for EDSS (p=0.03).

Figure 1

Hypothalamic–pituitary–adrenal axis dysregulation in patients with comorbid relapsing–remitting multiple sclerosis (RRMS) and major depressive disorder (MDD). MS patients with MDD showed normal morning cortisol levels but elevated evening cortisol (A), resulting in a significantly flatter circadian slope (B). No significant differences were detected in the cortisol awakening response (C) or after overnight low dose dexamethasone suppression (D). Data are presented as mean±SEM. All statistical analyses were run using log transformed cortisol values.

Some evidence suggests that prolonged treatment with IFNβ may affect HPA axis activity in MS,38 39 and interferon α administration has been associated with a flattening of cortisol profiles in other patient groups.40 Thus we more closely examined the role of DMTs on HPA axis activity in our sample. Indeed, patients taking IFNβ had flatter cortisol slopes that showed a statistical trend (see supplementary table 1, available online) (p=0.07). However, the difference in cortisol slope between depressed and non-depressed patients remained significant after statistically controlling for DMT (p=0.02).

Immunological measures

MS patients with or without MDD did not differ in absolute numbers of lymphocyte subsets (table 2). However, there were differences in T cell function, as indicated by a significantly higher proportion of TNFα producing CD8+ T cells after PHA stimulation (p=0.049) and a trend towards higher frequency of stimulated IFNγ producing CD8+ T cells (p=0.08) in MS patients with comorbid MDD (figure 2). The differences in cytokine producing CD8+ T cell populations between depressed and non-depressed patients were similar when statistically controlling for EDSS (TNFα p=0.05; IFNγ p=0.048).

Table 2

Absolute counts of lymphocyte subpopulations (cells/μl) in relapsing–remitting multiple sclerosis patients with or without comorbid major depressive disorder

Figure 2

Immunological alterations in patients with comorbid multiple sclerosis and major depressive disorder. A higher frequency of tumour necrosis factor α (TNFα) producing CD8+ T cells (phytohaemagglutanin (PHA) stimulated) was observed in relapsing–remitting multiple sclerosis (RRMS) patients with comorbid major depressive disorder (MDD). In addition, there was a trend towards higher percentage of interferon γ (IFNγ) producing CD8+ T cells (PHA stimulated). However, no differences were found for IFNγ or TNFα producing CD4+ T cells. Data are presented as mean±SEM.

Frequencies of TNFα and IFNγ producing CD8+ T cells were significantly correlated but no significant associations were found between immune and HPA axis measures (see supplementary table 2 available online). IFNβ therapy was associated with significantly lower levels of TNFα and IFNγ producing CD8+ T cells (see supplementary table 1 available online), however, after controlling for DMTs, a statistical trend remained for both cytokines (IFNγ p=0.053; TNFα p=0.087).

Prediction of depression and fatigue severity

Using a linear regression model, cortisol slope was the only significant predictor of HADS depression severity (R=0.34, adjusted R2=0.09, p=0.03) while frequency of IFNγ producing CD8+ T cells was the only significant predictor of fatigue scores (R=0.44, adjusted R2=0.17, p=0.01).


In the present study, we found that RRMS patients with comorbid MDD exhibited hyperactivity of the HPA axis, as characterised by normal morning cortisol but elevated evening levels, resulting in a flattening of the circadian slope. MS patients with comorbid MDD also showed increases in cytokine producing CD8+ T cells but these immunological markers were more closely associated with measures of fatigue than with depression.

Using both a clinical diagnosis of MDD as well as a quantitative measure of depression severity that is not confounded by somatic aspects of the disease such as fatigue, our study indicates that increased evening cortisol may be clinically relevant and could represent a depression specific biological correlate in MS.

In humans, the circadian trough of cortisol secretion is largely mediated via the high affinity mineralocorticoid receptor (MR) while circadian peak levels are more strongly dependent on low affinity glucocorticoid receptors (GR).41 In line with this notion, we found no significant associations between cortisol slope and the dexamethasone suppression test, which reflects GR function. While post dexamethasone cortisol levels were associated with the awakening response, neither of these tests showed significant differences between depressed and non-depressed MS patients. Taken together, this suggests that major depression in MS may be mediated by MR dependent HPA axis hyperactivity. In line with this hypothesis, elevated evening cortisol levels in an independent sample of depressed MS patients were recently linked to atrophy in the cornu ammonis 2–3 and dentate gyrus subfield of the hippocampus,32 where MR is highly expressed.42 Thus novel strategies targeting MR signalling43 may be promising treatment options for depression in MS.

Cytokines such as TNFα and IFNγ have been implicated in the pathogenesis of ‘sickness behaviour’.19 Here we provide evidence that the frequency of TNFα and IFNγ producing CD8+ T cells is upregulated in MS patients with comorbid MDD but that these markers are more closely correlated with fatigue than depression. This is in line with our previous results on associations between IFNγ and TNFα with fatigue scores and daytime sleepiness in non-depressed MS patients.24 In psychiatric patients, cytokines are thought to be produced peripherally and affect CNS processes through several indirect pathways.22 MS presents a unique situation where cytokines are directly released in the CNS by invading peripheral immune cells. Although CD4+ T cells appear to be most important for the initial steps in MS pathogenesis, the majority of T cells found in acute and chronic lesions of MS are actually CD8+ T cells, outnumbering CD4+ T cells 3–10-fold in chronically inflamed MS plaques.44 Thus our finding of higher percentages of IFNγ and TNFα producing peripheral CD8+ T cells on stimulation ex vivo may reflect similar responses of these CD8+ T cells once they are restimulated in the CNS.

MS patients with MDD showed higher frequency of IFNγ producing CD8+ T cells despite elevated circulating cortisol levels. A similar lack of associations between increased evening cortisol, flattened cortisol curves and increases in cytokine levels have been recently shown in hepatitis C patients treated with interferon α.40 This argues for a decoupling of inflammatory mechanisms from HPA axis mediated regulation. Interestingly, decreased sensitivity of immune cells to GC inhibition has been described both in MDD45 and MS.46 Thus an overshoot of inflammation despite HPA axis hyperactivity may represent a shared biological mechanism that could contribute to the high incidence of depression in MS.

Some limitations of this study need to be considered. Firstly, our study did not include a healthy control group. Our main interest was to investigate biological substrates of depression within the MS population (ie, what biological characteristics are seen in MS patients who have high levels of depression compared with patients who do not). Pronounced HPA hyperactivity has been described in later stages of progressive MS but in the early stages of RRMS, HPA abnormalities are less obvious.25 In a recent study, with a sample of RRMS patients comparable with the current study, only patients with high levels of depressive symptoms showed flattened cortisol profiles while non-depressed patients had circadian profiles indistinguishable from matched healthy control subjects.32 Thus flattened cortisol slopes indeed seem to specifically identify depressed MS patients and are not characteristic for MS itself. This is corroborated and extended by our current data using a clinical diagnosis of MDD. Secondly, our data provide indirect evidence that depression in MS may be linked to MR, rather than GR, dysfunction. However, whether the observed HPA abnormalities are indeed linked to an imbalance in MR/GR signalling needs to be further examined using selective agonists/antagonists in vivo or in vitro. Thirdly, our study enrolled only female patients. This was done to eliminate the confounding impact of gender differences in endocrine regulation. While this approach increases the power of the study, especially in small samples, it also decreases the generalisability of the findings. Thus it remains to be determined if the same mechanisms can be observed in male patients. However, it should be noted that both RRMS as well as MDD have a female preponderance of 2:1 to 3:1, suggesting that our results may be relevant to the majority of patients. In addition, we have observed very similar HPA axis alterations in depressed RRMS patients in our previous study that enrolled both male and female patients.32 Fourthly, our sample of MS patients with comorbid MDD is comparatively small and these results need to be confirmed in larger studies. Lastly, due to the cross sectional nature of this study, we cannot infer a causal relationship between neuroendocrine and immunological abnormalities and depression or fatigue in MS. Further studies are needed to establish if changes in these systems are associated with changes in depressive symptoms or fatigue.

Taken together, our study suggests that although HPA axis hyperactivity and inflammation often coexist in depression, they may differentially mediate aspects of ‘sickness behaviour’ such as mood alterations or fatigue. Our approach also indicates that using the unique pathogenetic situation of MS as a paradigm can help to further enhance our understanding of the clinical relevance of neuroendocrine and inflammatory processes with regard to medical depression and ‘sickness behaviour’ in humans.


Supplementary materials


  • See Editorial commentary, p 709

  • Linked article 240507.

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the ethics committee of the Medical Board Hamburg.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles