Glial cell dysfunction in schizophrenia indicated by increased S100B in the CSF

SIR—Findings from neuropathological and imaging studies reveal a reduction of neuropil in schizophrenia that might even have a progressive component. Neuronal cell sizes as well as dendrite and synapse numbers are decreased while neuronal counts remain unchanged.1, 2 Until recently, astrocytes were merely regarded as metabolic supporters of neurons. Advanced research, however, has revealed evidence that a dysfunction of astrocytes can be directly responsible for neuronal malfunction and might therefore play a key role in the development of schizophrenia. Astrocytes influence dendrites and synapses mainly via glutamate-induced Ca2+ modulation (for a review, see Araque et al3).

Regarding schizophrenia, there is no increased astrogliosis within the cortex. However, recent studies point towards an astroglial loss in some areas of the brain. Signs of reactive and dystrophic changes with decreased intact mitochondria in the astrocytes have been reported, and astroglial function after injury has been shown to be impaired in schizophrenic brains. These findings suggest that there might be glia pathology in schizophrenia.

S100B as an astrocytic protein exerting paracrine and autocrine effects on neurons and glia is an indicator of astroglial function. From in vitro and in vivo animal experiments, it is known that S100B plays important roles in cell proliferation and differentiation, cellular energy metabolism and cytoskeletal modification. Those effects appear to be mediated primarily via binding to key synaptic proteins and inhibition of their phosphorylation (for a review, see Heizmann et al4, Rothermundt et al5).

In schizophrenia, several studies have demonstrated increased S100B serum concentrations in the acute stage of disease.6, 7, 8, 9 High concentrations were associated with deficit symptoms,7, 8, 9 and patients with high S100B serum levels showed slowed psychopathological improvement upon treatment.7, 8 These earlier studies provide evidence that S100B might be relevant for the pathophysiology of schizophrenia (for a review, see Rothermundt et al5). However, all previous studies were confined to the measurement of serum S100B concentrations. A concomitant investigation of CSF and serum would finally allow us to conclude whether serum S100B concentrations reliably reflect CSF and brain S100B concentrations.

After written informed consent according to the standards of the Declaration of Helsinki was acquired, CSF and plasma samples were taken by lumbar and venipuncture from 21 unmedicated inpatients (17 males, four females, mean age 32.5±13.0 years, 16 drug naive, five unmedicated >6 months), suffering from an acute schizophrenic episode (n=4 disorganized, n=17 paranoid subtype) and age- and sex-matched healthy controls. The control group for CSF measures (mean age 33.6±10.9 years) consisted of individuals who underwent an elective knee arthroscopy. The controls for serum measures were recruited from the blood donation service (mean age 32.2±11.8 years). All patients were diagnosed independently by two psychiatrists according to DSM-IV criteria. Systemic diseases, brain injury, comorbid psychiatric diagnoses and substance abuse were excluded. The controls had no lifetime history of any psychiatric disorder. The Positive and Negative Symptom Scale (PANSS) was used to evaluate psychopathology.

CSF and blood samples were centrifuged within 4 h, aliquoted, and frozen at −80°C until analysis. S100B concentrations were determined by applying the LIAISON Sangtec 100 assay (AB Sangtec Medical, Bromma, Sweden), a quantitative automated luminometric immunoassay, according to the manufacturer's instructions. Albumin and immunoglobulin were measured by immunonephelometry (Behring nephelometer). Owing to the equal distribution of the data, the T-test for independent variables and the Pearson Correlation Coefficient were employed for statistical evaluation as provided by the SPSS 10.0 program.

The schizophrenic patients (1.19±0.43 μg/l) showed a significantly higher mean CSF S100B concentration than the matched controls (0.93±0.15 μg/l; T=2.78, df=20, P=0.012). In the same manner, the serum S100B levels were increased in patients (0.065±0.031 μg/l) compared to healthy controls (0.038±0.008 μg/l; T=3.16, df=20, P=0.008). The mean CSF/serum ratio of the patients was 21.2±14.5 : 1. No significant differences could be detected between first episode and recurrent episode patients. No patient showed an impairment of the blood–brainbarrier (BBB). The PANSS score (mean=97.7±25.6) was positively correlated with the S100B serum level (rho=0.58, P=0.03).

Even though S100B is highly brain specific and it has been suggested that increased S100B levels detected in the serum of schizophrenic patients are derived from the brain, it remained unclear whether increased S100B serum concentrations actually reflect increased CSF S100B levels. In our study, CSF as well as serum S100B concentrations of patients suffering from schizophrenia were elevated compared to matched healthy controls. The CSF/serum ratio of 21 : 1 was comparable to the ratio reported for healthy individuals (18 : 1–20 : 1) with intact BBB.10

The serum findings reproduce the results of earlier studies. In an acute stage of disease, serum S100B was shown to be elevated in unmedicated as well as medicated schizophrenic patients and S100B concentrations in serum were positively correlated with a more severe psychopathology.6, 7, 8

Since S100B is a rather small protein (21 kDa in its dimeric form), it can easily pass the BBB. What has been demonstrated in healthy individuals as well as patients with various neurological diseases, that is, that serum measures reliably reflect the S100B concentration in the CSF, holds true for schizophrenia as well. Therefore, future studies on serum S100B levels in schizophrenic patients can still most likely be used to infer S100B concentrations in the brain. These data provide further evidence that a dysfunction of glia cells might present a pathogenic factor in schizophrenia.


  1. 1

    Selemon LD, Goldman-Rakic PS . Biol Psychiatry 1999; 45: 17–25.

  2. 2

    McGlashan TH, Hoffman RE . Arch Gen Psychiatry 2000; 57: 637–648.

  3. 3

    Araque A, Parpura V, Sanzgiri RP, Haydon PG . Trends Neurosci 1999; 22: 208–215.

  4. 4

    Heizmann CW, Fritz G, Schäfer BW . Front Biosci 2002; 7: 1356–1368.

  5. 5

    Rothermundt M, Peters M, Prehn JHM, Arolt V . Microsc Res Tech 2003; 60: 614–632.

  6. 6

    Lara DR, Gama CS, Belmonte-de-Abreu P, Portela LVC, Goncalves CA, Fonseca M et al. J Psychiatr Res 2001; 35: 11–14.

  7. 7

    Rothermundt M, Missler U, Arolt V, Peters M, Leadbeater J, Wiesmann M et al. Mol Psychiatry 2001; 6: 445–449.

  8. 8

    Rothermundt M, Ponath G, Glaser T, Hetzel G, Arolt V . Neuropsychopharmacology 2004; 29: 1004–1011.

  9. 9

    Schroeter ML, Abdul-Khaliq H, Frühauf S, Höhne R, Schick G, Diefenbacher A et al. Schizophr Res 2003; 62: 231–236.

  10. 10

    Reiber H, Thompson EJ, Grimsley G, Bernardi G, Adam P, Monteiro de Almeida S et al. Clin Chem Lab Med 2003; 41: 331–337.

Download references


This study was supported by the Innovative Medical Research Program of the University of Muenster (RO 2 1 99 19).

Author information

Correspondence to M Rothermundt.

Rights and permissions

Reprints and Permissions

About this article

Further reading