THERE IS SUFFICIENT evidence that astrocytes play an important role in the central nervous system (CNS) and are implicated in the pathogenesis of schizophrenia.[1, 2] For this reason, research about the S100B protein has gained increasing attraction in recent years with different research groups interested in the potential role of this protein in schizophrenia. Given the difficulty of studying cerebral S100B levels in vivo, psychiatrists need to focus on peripheral measures, such as blood or urine. The study of the cerebrospinal fluid (CSF) requires carrying out a lumbar puncture and many ethics committees would not allow this technique because of its invasive nature.
The research for peripheral biological markers of schizophrenia, although abundant, has been unfruitful. In the last 2 decades, the S100B protein has made its own room in this area of research. S100B is a calcium-binding protein that has been proposed as a marker of astrocyte activation and brain dysfunction. Research results on S100B concentrations and schizophrenia clinical diagnosis are very consistent; patients with schizophrenia have higher S100B concentrations than healthy controls. The results regarding schizophrenia subtypes and clinical characteristics are not as conclusive. Age of patients, body mass index, illness duration and age at onset have been found to show no correlation, a positive correlation or a negative correlation with S100B levels. With respect to psychopathology, S100B data are inconclusive. Positive, negative and absence of correlation between S100B concentrations and positive and negative psychopathology have been reported. Methodological biases, such as day/night and seasonal variations, the use of anticoagulants to treat biological samples, the type of analytical technique to measure S100B and the different psychopathological scales to measure schizophrenia symptoms, are some of the factors that should be taken into account when researching into this area in order to reduce the variability of the reported results. The clinical implications of S100B changes in schizophrenia remain to be elucidated.
S100B is a monomer that belongs to the S100 protein family. Two of the main monomers of S100 are the S100A1 and the S100B. Both are found mainly as dimers, either homo- (BB) or heterodimers (A1B). S100B is a calcium-binding protein that was first described in 1965 and influences many cellular responses along the calcium-signal-transduction pathway. S100B is a dimer with a molecular weight of about 21 kDa that is excreted by the kidneys and has a half-life of approximately 30 min. S100B is found in the CNS, mainly in the cytoplasm and nucleus of astrocytes. At physiological levels, inside the cells, S100B exerts a proliferative effect. Once S100B is released from cells, it has trophic or toxic effects according to its concentration. At pico- and nanomolar concentrations (physiologic concentrations) it exerts a neuroprotective effect (increasing the number of dendrites and synaptic plasticity). At micromolar concentrations it is neurotoxic and inhibits the proliferation and differentiation of neurons and induces neurodegeneration and apoptosis.[5-11] S100B is mainly produced and secreted by astrocytes, but is also produced in other neural cells, such as oligodendrocytes, the ependyma, choroid plexus epithelial cells, and a few neurons. Outside the CNS, S100B has been detected in other cells, such as adipocytes, chondrocytes, lymphocytes, bone marrow cells, dendritic cells, Langerhans cells, satellite cells of dorsal root ganglia, and Schwann cells of the peripheral nervous system. S100B-increased blood levels have been found after soft-tissue injury in multiple trauma, cardiac surgery or injured myocardium. S100B protein released from soft tissue is probably not comprised from S100BB homodimers but rather from other species of S100B containing the B subunit in a heterodimer (S100A1B). The available assays systems for S100B use luminometric immunoassay, enzyme-linked chemiluminescent immunosorbent assay or enzyme linked immunosorbent assay. These methods are based on antibodies highly specific for the S100B monomer, but also bind to S100A1B. Thus, there may be a bias if the S100 heterodimer (A1B) produced by peripheral tissues (soft tissues, damage, multiple trauma, cardiac surgery, etc) is also detected.[2, 7] This bias could be avoided by using a technique only sensitive to the S100B homodimer. Furthermore, according to recent research, extracranial sources of S100B (especially from adipose tissue) do not significantly affect serum levels, therefore the diagnostic value of S100B is not compromised in the clinical setting (subjects did not have traumatic brain or bodily injury from accident or surgery). From our point of view, although the study of Pham et al. elicited interesting results, it needs to be replicated.
S100B regulates cell cycle, cytoskeleton, energy metabolism, cell communication, intracellular signal transduction and cell growth.
Astrocytes play a supporting role in the spatial distribution of neurons, formation of synapse networks, neurotransmitters release and in the immune response. Furthermore, astrocytes are related to the pathogenesis of various diseases, such as epilepsy, migraine, leukodystrophy, inflammatory demyelinating diseases, infections, brain injury, neurodegenerative disorders and schizophrenia.
S100B readily crosses the blood–brain barrier (BBB), and it has been used as a marker of BBB disruption (due to increased permeability).[17, 18] There is a positive correlation between CSF and serum S100B levels, although the levels of S100B in serum are lower than in the CSF, as has been shown in rats (ratio CSF/serum 3 to 5). This correlation has also been found in patients with schizophrenia. Multiple studies have demonstrated increased S100B in CSF and plasma or serum after cerebral ischemic injury or trauma, concluding that the increase of S100B would result from the destruction of astrocytes.[19, 20] S100B is linked to neurodegenerative diseases, such as Alzheimer's disease, and to psychiatric diseases, such as schizophrenia and mood disorders (major depression and bipolar disorder), so it is not specific of them. Therefore, S100B can be considered as a marker of astroglial damage that can be predictive of an unfavorable evolution after brain damage, and can hence be a future biomarker of CNS diseases, as C-reactive protein in plasma is a biomarker of low-grade systemic inflammation. Increases in S100B concentrations have been considered as a component of the neuroinflammatory response. In addition to these findings, variations in serum S100B levels related to age, from infants to adults, have also been found. Neonates' levels are much higher than in the elderly. S100B levels decline in the first 2 decades of life and remain constant throughout adult life. Such high levels in infants may be linked to neurodevelopmental processes as well as glial cell maturation and synapse formation. The possible neurotrophic activity of S100B has been pointed out in research, showing a positive correlation between S100B levels in umbilical cord and brain maturation.
As it is considered that normal levels of S100B exclude major pathology of the CNS, S100B levels have been used as a prognostic factor in cases of minor brain damage.
Other studies support a potential therapeutic role of the S100B protein. After head injury in rats, and posterior infusion of intraventricular S100B, there is an enhancement of hippocampal neurogenesis from day 5, with persisting neurogenesis up to 5 weeks after.
S100B and schizophrenia
With regard to schizophrenia, there is sufficient evidence that its pathogenesis is related to both neurodevelopmental and neurodegenerative processes. Also, the immune role in relation to maternal infections during pregnancy, such as influenza, Borna disease and human endogenous retroviruses, is noteworthy. This can contribute to developing schizophrenia in later life, as indicated by the high levels of antibodies found in patients with schizophrenic psychoses. Microarray studies have shown a decreased brain density of glial cells, and alterations in genes associated with astrocytes and oligodendrocytes in schizophrenia patients.
Therefore, it is important to measure the functionality of astrocytes in patients with schizophrenia through astrocytic markers that can be detected in both CSF and serum. The S100B protein concentrations can be considered as a potential marker of astrocytic response in schizophrenia.
In recent years there have been many studies supporting a link between alterations in S100B levels and schizophrenia, most showing an increase in S100B in acute stages of the disease.[16, 31] It seems that S100B should decline with the progression of schizophrenia, being more likely to be a marker of disease state (e.g. as in incipient psychosis) than a trait marker. It is not known whether S100B is increased in the prodromal phase preceding the onset of a full-blown psychosis, and if so, S100B would serve as a marker to help recognizing prodromic symptoms. It is unclear whether changes in S100B are due to treatment or disease progression. There are few studies showing links between S100B and untreated psychosis, duration of disease or type of antipsychotic prescribed.
Regarding studies of schizophrenia and S100B in CSF, increased S100B concentrations in CSF of schizophrenic patients in acute episodes of relapse have been reported when compared to healthy controls;[7, 33] however, Steiner et al. suggested that this increase in S100B appears to be caused by active secretion of this protein by astrocytes, as there were no differences in oligodendrocytes or glial fibrillary acidic protein (GFAP), which is a marker of astrocytes destruction.
When comparing schizophrenia and S100B in post-mortem brain studies, it has been found that the density of S100B immunopositive glial cells in the cortical regions of patients with paranoid schizophrenia is higher than in those with residual schizophrenia. Furthermore, the levels in healthy controls were lower. Those differences are most pronounced in the dorsolateral prefrontal cortex. The explanation for the S100B increase in paranoid schizophrenia may be due to an activation of astrocytes and oligodendrocytes. By contrast, the decrease in residual schizophrenia may be due to damage in white matter or glial dysfunction associated with S100B release into bodily fluids.
We have conducted a systematic review using the PubMed database. The strategy of the search was as follows: the intersection of the key word ‘schizophrenia’ with the key words ‘S100B or S100-B’. The only restriction criteria were selecting studies in humans and a temporal criterion, from 1970 to 2012 (base data accessed on 2 June 2012). The search elicited 40 papers of which eight were reviews.[2, 9, 22, 32, 34-37] From the 32 non-review papers, eight were focused on genetic studies,[38-45] one paper was focused on the role of S100B protein as a marker of suicidality, another paper focused on S100B and cerebral palsy, one paper was written in Japanese and was related to the levels of S100B in healthy subjects, and another was on histological distribution on S100B immunopositive glia. The latter 12 studies were excluded from the present review. Additional research papers have been included and extracted from the references of the 20 original research papers from the PubMed search.
Serum S100B and schizophrenia
With respect to research in schizophrenia and S100B, most of the studies were carried out in serum. In general, authors have found increased S100B levels in patients (chronic schizophrenia, acute relapse and in first-episode psychosis) compared to healthy controls.[6, 7, 16, 18, 31, 49-51] In plasma, the results differ from that of serum levels. Gattaz et al. compared plasma S100B concentrations of 23 schizophrenic patients treated with antipsychotics (16 of them with clozapine) with plasma S100B levels of healthy controls, finding lower levels of plasma S100B concentrations in patients than in the healthy control group. The decrease of S100B is also supported by another study, where astrocytic C6 cells and oligodendrocytic OLN-93 cells cultured with haloperidol and clozapine decreased the release of S100B; however, the decreased levels of S100B reported by Gattaz et al. should be interpreted with caution because the concentration of S100B was measured in plasma (with citrate as anticoagulant) and it has been reported that plasma S100B concentrations are about 10 times higher than those measured in serum.
In 2001, in a study with 20 schizophrenic patients without antipsychotic treatment, the serum S100B concentrations found in the patient group were higher than in the healthy control group. The authors did not find correlations between S100B levels and the Positive and Negative Syndrome Scale (PANSS) scale or duration of the disease.
In another study with schizophrenics treated for 3 weeks (16 on antipsychotic treatment and 14 untreated), S100B was higher in the treated group compared to untreated and controls. Also, in those schizophrenics with predominantly residual or negative symptoms, S100B levels were higher. Rothermundt et al. carried out a longitudinal study of 26 patients with untreated schizophrenia, where S100B was measured at baseline and after 6 weeks of antipsychotic treatment. At baseline, S100B levels were higher in patients than in healthy controls, but at week 6 this difference was not present; however, in a subgroup of patients with persistent negative symptoms, S100B concentrations remained high. One possible explanation may be that antipsychotics increase the levels of S100B in the first weeks of treatment and then S100B remains elevated only in those patients with predominantly negative symptoms. Studies with longer follow-up periods are required.
Another longitudinal study with 98 schizophrenic patients with predominantly negative symptoms over 24 weeks of treatment with risperidone or flupenthixol was published in 2004. It was concluded that negative symptoms predict the concentration of S100B in serum, that is, those with higher levels of S100B had more persistent negative symptoms.
Twenty-three healthy controls and 41 elderly chronic schizophrenics (14 were on haloperidol or typical antipsychotics, 15 on clozapine and eight on haloperidol or clozapine combined with typical antipsychotics) were compared in another study. It was found that there were higher levels of serum S100B in schizophrenic patients and there was a negative correlation between the Scale for the Assessment of Negative Symptoms (SANS) score and S100B. Also, S100B correlated positively with age in schizophrenic patients (but not in controls). These results disagree with previous studies in younger patients where there was a positive correlation between the PANSS negative subscale and serum S100B levels.[7, 16, 55] This is the only study that has found S100B sex differences between men and women in schizophrenic patients, with women having lower levels of S100B than men, while most of the studies have not reported this difference.[7, 33, 49, 55, 56]
In 2007, Sarandol et al. reported that serum S100B levels were higher in patients with negative symptoms than those with positive symptoms and the controls. These levels were significantly reduced after 6 weeks of antipsychotic treatment.
In 2009, a relapsed schizophrenic group was compared to healthy controls. S100B analytical monitoring was performed weekly from admission to discharge. They also measured the neuron-specific enolase (NSE), which is found in the cytoplasm of neuronal cells and is not secreted actively, thus, it may be considered as a marker of neuronal destruction or brain damage. Higher S100B levels were found in the schizophrenic group compared to controls, with significant differences for NSE. The same group carried out a meta-analysis of published studies from 1970 to 2007 (12 studies in total), concluding that S100B in schizophrenics was higher than in controls, with no effect of antipsychotic treatment on S100B concentrations. The authors suggest that S100B levels are increased by active secretion of astrocytes and BBB dysfunction in schizophrenia.
Chronic schizophrenics treated with clozapine or typical antipsychotics have also been compared to healthy subjects. Higher levels of serum S100B were found in the schizophrenic group. There were no differences in S100B levels between the types of antipsychotic. There was no correlation between PANSS total score and S100B levels.
Recently, the role of the soluble receptor for advanced glycation end-products (sRAGE) in schizophrenia has become the object of research. RAGE is the main S100B receptor, which is located in neurons, glia, T cells and endothelial cells. RAGE is also the main receptor for amphotericin and β-amyloid, and contributes to the development of atherosclerosis, rheumatoid arthritis, diabetes and Alzheimer's disease. sRAGE counteracts the action of RAGE. Steiner et al. investigated whether sRAGE increased in the recovery period of the paranoid schizophrenic patients and contributed to the normalization of S100B levels. They concluded that sRAGE increased over the 6-week follow up and that it negatively correlates to S100B levels. Therefore, the authors consider sRAGE as a reducing factor of S100B levels and suggest that in the near future, sRAGE may be considered for new treatment strategies for schizophrenia.
A comparison of the serum S100B levels in untreated recent-onset psychotics, chronic schizophrenics under treatment and healthy controls showed that the serum S100B levels in the untreated recent-onset psychotics were higher than in the chronic treated schizophrenics. The authors suggested that low levels of S100B in treated patients are due to the decrease of neurodegeneration caused by antipsychotics. No differences in S100B levels between the different types of antipsychotic drugs, smoking, age or sex were found.
This research group also found elevated serum S100B concentrations in a sample of schizophrenic patients with tardive dyskinesia. They link this phenomenon to neurodegeneration because patients with tardive dyskinesia had higher levels of S100B than both the schizophrenics without dyskinesia and the healthy controls.
Data about psychopathology and S100B concentrations are controversial. A positive correlation between the third Brief Psychiatric Rating Scale (BPRS) item score (thought disturbance) and the concentration of S100B and a positive correlation between total PANSS scores and S100B levels have been reported. The same group reported a negative correlation between S100B levels and PANSS negative scores. The patients of this study were selected with prominent negative symptoms, so this bias may explain the results. Also, no correlations between S100B levels and PANSS total, positive, negative and general scores have been reported.[49, 58] In our opinion, biases in the selection of the patients, for example those with prominent negative symptoms, and the use of different scales to measure psychopathology (BPRS, PANSS, SANS, etc.) are partly responsible for the contradictory results.
In vitro studies have reported that both haloperidol and clozapine decrease the release of S100B by astrocytic and oligodendrocytic cells; however, studies carried out on schizophrenic patients are not so clear. Increased[49, 59] and decreased S100B levels have been reported. Regarding the distinction between typical and atypical antipsychotics, there do not seem to be differences between atypical and typical antipsychotics; indeed some authors have reported that patients present higher levels of S100B than controls, independently of whether they were treated with typical or atypical antipsychotics. To have a definitive opinion on this subject, studies with bigger samples of patients treated with only one antipsychotic should be carried out. In a clinical setting, however, monotherapy is not the usual rule and this is the reason why different antipsychotics are converted into chlorpromazine-equivalent doses so that there are enough patients to compare. As a consequence of this, specific details, such as the differences between typical and atypical antipsychotics, are lost.
There is agreement with respect to the effect of illness duration and age of illness onset on S100B levels. It has been reported that neither are related to S100B concentrations.[16, 21, 33, 49, 52, 55, 56, 61] There is only one exception, which reported a negative correlation between illness duration and S100B levels. We do not have an explanation for this result.
No sex differences in S100B levels have been reported in patients with schizophrenia.[7, 33, 49, 55, 56] Only one paper has reported that schizophrenic men had higher levels of S100B than schizophrenic women.
In healthy subjects, S100B plasma levels have been reported as not correlated to age[23, 49, 50] as well as having a positive correlation with age. In patients with schizophrenia, no link between age and serum levels of S100B has been reported,[7, 16, 21, 49, 50, 55, 56, 61] although others have found a positive correlation between age and S100B serum levels.
Other variables that may affect S100B concentrations are body mass index and nutrition. It is noticeable that those variables are not routinely included in the S100B schizophrenia protocol studies. A positive and an absence of correlation have been reported in healthy subjects and patients with several pathologies. Due to the scarcity of publications on this topic, we believe that more research is needed before reaching a definitive conclusion. It has been reported that chronic starvation (anorexia nervosa) is related with low S100B levels and a posterior weight gain is characterized by a recovery of the S100B levels; however, no differences in S100B concentrations have been reported between subjects following a normal diet and subjects following a diet supplemented with mixed-grain. More research on the relation between nutrition and S100B levels in both healthy and diseased subjects is needed.
Adipocytes are an extracerebral source of S100B, therefore the increase in serum S100B could be due to an increase or a dysfunction of adipose tissue. S100B secretion by adipocytes is reduced by insulin and activated by physiological factors, such as stress, cathecolamines and fasting. In healthy subjects, a positive correlation between body mass index and S100B levels has been reported. Patients with schizophrenia have an increased risk for type 2 diabetes mellitus and metabolic syndrome, with increased insulin resistance (mediated in the brain) and therefore they present an increase of S100B levels. It is not clear if the S100B protein may be a marker of metabolic syndrome in schizophrenia.[62, 69] Therefore, obesity and insulin resistance may affect S100B serum levels.
It should be pointed out that the lack of agreement between different studies may be due to methodological errors. The type of anticoagulant with which blood is treated, such as sodium heparin, sodium citrate and ethylenediaminetetraacetic acid (EDTA), can alter S100B levels. Tort et al. compared S100B levels of the same subjects with and without anticoagulant. They found a positive correlation between serum S100B and S100B in plasma collected with heparin. They obtained an increase of S100B in the samples of heparin and citrate, while no differences with those of EDTA and serum. Since citrate and EDTA are calcium chelators, they have more interference with the analytical results; on the other hand, heparin always showed a strong positive correlation with serum. Thus, if blood needs to be anticoagulated, heparin is preferred. Ling et al. heparinized their plasma samples before measuring S100B levels. They reported a positive correlation of S100B plasma levels and PANSS total score at admission, and this correlation mainly existed between S100B levels and the PANSS negative subscore.
Circadian rhythmicity should also be taken into consideration. Our group carried out a study with 43 schizophrenic patients admitted to a psychiatric ward because of an acute relapse. At admission, S100B serum daytime levels were higher than night-time levels, and this pattern disappeared when the patient was clinically stabilized before discharge.
Regarding seasonal variation, after examining blood samples from 33 healthy subjects, serum S100B levels were significantly higher in the samples drawn in summer when compared to winter (results from our own research pending publication).
Recent research links the neuroinflammation hypothesis of schizophrenia with microglia. S100B acts as a cytokine after secretion from glial cells, CD8 + lymphocytes and natural killer (NK) cells, activating monocytes and microglial cells.[72-74] It is known that activated microglia release pro-inflammatory cytokines and free radicals, and these cause neuronal degeneration and decrease neurogenesis. Recent results of Bianchi et al.[72, 73] suggest that S100B may participate in the pathophysiology of brain inflammatory disorders via RAGE regulation with activation and migration of microglia. Furthermore, S100B upregulates cyclo-oxygenase-2 expression in microglia in a RAGE-dependent manner. The minimum levels of RAGE that are necessary for the S100B to exert a toxic or neuroprotective effect are still unknown; however, according to Steiner et al., in schizophrenic patients, S100B + NK cell counts are correlated with the free cortisol index (marker for stress axis activity) but not with the S100B serum concentrations. These results suggest that NK are probably not a major source of S100B in the blood of schizophrenia patients.
In summary, glial dysfunction appears to play a role in the pathogenesis of schizophrenia. Therefore, S100B seems to be an important marker of both state and prognosis of schizophrenia and other disorders with brain damage. Some studies link S100B to negative symptoms and a poor response to antipsychotic treatment. S100B is also related to obesity and insulin resistance. The generalized use of the measurement of serum S100B levels might help in clarifying the role of the S100B protein in the clinical setting. It is still important to do further research on the possible therapeutic role of S100B. It is also necessary to do more studies on the possible links to the 5HT1A receptor and its potential therapeutic value in schizophrenia, as well as its association with specific neurocognitive functions and antipsychotic treatment. Recent studies on S100B and sRAGE indicate that the systemic administration of sRAGE might be a future treatment in schizophrenia that may work by normalizing elevated levels of S100B. Therefore, it is important to do further studies with larger samples of schizophrenic patients and longer follow-up periods to try to replicate previous results. The control of confounding variables, such as obesity, diabetes mellitus, circadian changes and seasonal changes, is strongly advisable.
This study was partly supported by a grant (PI:08/115) from the Fundación Canaria de Investigación y Salud (FUNCIS). None of the authors has anything to disclose.