Introduction

Schizophrenia (SCZ), bipolar disorder (BD), and major depressive disorder (MDD) are three types of common neuropsychiatric disorders that are characterized by severity and recurrence and are among the leading causes of serious self-harm or even suicidal behavior in young people [1,2,3]. In the early stages of the disease, they are usually difficult to distinguish by clinical diagnosis with symptoms overlapping across diagnoses, and shared phenotypes [4]. Although the etiology of these psychiatric disorders is still unknown and there are no effective drugs for treatment, several neuropathology [5], oligodendrocyte abnormalities [6], and metabolic disturbances [7] have been proposed. Among these underlying biological factors, several studies support the role of neuroinflammation in the pathogenesis of these mental disorders [8,9,10].

Neuroinflammation is an inflammatory response within the central nervous system (CNS) characterized by the proliferation and activation of glial cells (e.g., microglia and astrocytes) [11, 12]. Microglia are macrophages in the central nervous system that mediate innate and adaptive immune responses in the brain [13]. Under abnormal conditions such as brain infection, injury, or disease, microglia change from ramified (“resting”) state to an “activated” state, releasing proinflammatory cytokines such as interleukin (IL)-1β, IL-6, tumor necrosis factor (TNF)-α, interferon (IFN)-γ, or several chemokines [14]. Proinflammatory cytokines released from microglia can activate astrocytes, which are generally manifested by increased glial fibrillary acidic protein (GFAP) expression [8].

Given the pathology of inflammation in the brain, most studies have attempted to elucidate the link between inflammation and psychiatric disorders during the past 20 years. For example, advances in molecular biology and genetics have shown that genes involved in regulating the immune system are highly associated with the risk of SCZ, BD, and MDD [15]. Inflammatory biomarkers derived from peripheral blood of major psychiatric diseases have been investigated by several studies, which is due to the easy accessibility of the “blood and periphery as a window to the brain” hypothesis [16]. Elevated serum and plasma levels of proinflammatory cytokines, such as IL-1β and TNF-α, have been found in SCZ and BD [17, 18]. In addition, a recent meta-analysis focused on cerebrospinal fluid (CSF) cytokines in patients with SCZ, BD, and MDD and found that CSF levels of IL-6 and IL-8 were similarly elevated in these patients [19].

Although the above studies have suggested overall similarities in the pattern of blood cytokine alterations in patients with SCZ, BD, and MDD and raise the possibility that inflammation is involved in a potential brain pathologic pathway for these mental disorders, peripheral inflammatory markers are not representative of cerebral inflammatory markers since the CNS is the ultimate site of disease. However, in the literature related to major psychiatric disorders such as SCZ, BD, and MDD, there have been a large number of reports evaluating markers related to inflammation in the brain. This makes it possible to further explore the connection between inflammation in the brain and disease. Therefore, we systematically reviewed the literature on brain inflammatory markers measured in SCZ, BD, and MDD postmortem brains to identify more elevated inflammatory markers in the postmortem brain of these patients, and to provide a preliminary conclusion on the inflammatory pathways by which postmortem brain samples of these diseases are affected.

Methods

We performed this systematic review as stated in a prospective protocol following guidelines that are recommended by the PRISMA Statement (Preferred Reporting Items for Systematic reviews) [20].

Literature Search Strategy

We performed a literature search for records indexed within PubMed (1974 to 8th May 2022) and Web of Science (1985 to 8th May 2022) using the following search terms: “(schizophrenia or bipolar disorder or major depressive disorder or depression) and (inflammation or cytokine or chemokine or interleukin or interferon or tumor necrosis factor or colony-stimulating factor) and (postmortem or brain sample).”

Eligibility Criteria

Studies were screened for relevance based on their title and abstract by two researchers independently. The full text of potentially relevant articles was retrieved and screened against the following inclusion criteria: (1) studies that focused on postmortem brain samples, including SCZ, BD, and MDD; (2) measurement of inflammation-related markers, including any kind of inflammatory cytokine/chemokines, and other related markers were considered if the authors mentioned their role in neuroinflammation; and (3) matched healthy controls were included. Duplicates and articles that did not meet the above criteria were excluded. In addition, review articles, in vitro studies, and animal studies were excluded. Finally, conference abstracts and non-English papers were also excluded.

Data Extraction

Eligible studies were assessed, the data were extracted into an Excel spreadsheet by the researcher, and any disagreements were resolved by discussion. For the eligible studies, the first author’s name, publication year, brain bank, sample size, sex, age, and death from suicide were extracted as background information. In addition to inflammatory markers measured, measuring techniques and in which brain regions the measurements were made were all extracted, along with comparative results between the patient subjects and the healthy controls. Graphical data were extracted using a Web plotter (https://apps.automeris.io/wpd/).

Given the differences in the brain regions measured, we decided that if studies measured the same parameters in the cortical (or subcortical) layer and at least two sets of data were available, we performed a systematic review of these studies. The effective size (ES) was used except where stated. ES was produced by sample size, mean concentration, and standard deviation (SD), or by sample size and P value if the data of mean concentration were not available [21]. This excluded several studies (e.g., which used medians and interquartile ranges).

Results

Our search strategy resulted in the identification of 2166 unique studies from the initial search. After screening the titles and abstracts for relevance, 77 articles were full-text screened against the inclusion criteria. Out of the 77 articles, 31 articles were excluded because they did not measure inflammation-related markers (12 studies); data was not separable from other diagnostic groups (8 studies); postmortem brain from the dataset (2 studies); and incorrect measurement methods, such as cDNA microarray experiments, and gene network analysis (9 studies). Thus, a total of 46 studies were ultimately included in this review (Fig. 1).

Fig. 1
figure 1

Flow chart of the systematic search

Full size image

Characteristics of the Included Studies

Table 1 summarizes the basic demographics of these included studies. The incorporated studies contained relatively small numbers of subjects with BD, which may be due to the BD postmortem brain samples being scarce. Most studies involving donor subjects were defined according to the Diagnostic and Statistical Manual of Mental Disorders (DSM)-III or DSM-IV, but the remaining studies used other criteria or were not specified and were therefore not included in the statistics. Although several studies included one or more diagnostic groups, our results were still discussed by disease classification and summarized in Tables 2, 3 and 4, highlighting the main results the authors reported as being statistically significant in their study.

Table 1 Demographic characteristics of the studies included in the systematic review
Full size table
Table 2 Inflammatory markers of the included studies in postmortem schizophrenia brain
Full size table
Table 3 Inflammatory markers of the included studies in postmortem bipolar disorder brain
Full size table
Table 4 Inflammatory markers of the included studies in postmortem major depressive disorder brain
Full size table

Studied Brain Areas

The regions of the brain studied in the included studies mainly included the anterior cingulate cortex (ACC; Brodmann area (BA)24), dorsolateral prefrontal cortex (DLPFC; BA46), frontal cortex (FC), hippocampus, orbitofrontal cortex (OFC; BA11), prefrontal cortex (PFC; BA9), medial frontal gyrus (MFG), superior frontal gyrus (SFG), and superior temporal gyrus (STG).

Immune/Inflammation Response, Cell Regulatory Proteins, Glia/Macrophage Proliferation, Metabolic Pathway, and Chemokines in Postmortem Studies of SCZ

A total of 39 studies were included to measure the difference between SCZ and control subjects. In detail, immune/inflammation response (Arion et al. [22], Dean et al. [23], Durrenberger et al. [24], Fillman et al. [25], Fillman et al. [26], Foster et al. [27], Harris et al. [28], Hoseth et al. [29], Hwang et al. [30], Iwamoto et al. [31], Izumi et al. [32], Kim et al. [33], Kindler et al. [34], Lanz et al. [35], López-González et al. [36], Maida et al. [37], Murphy et al. [38], Murphy et al. [39], Pandey et al. [40], Saetre et al. [41], Schmitt et al. [42], Toyooka et al. [43], Volk et al. [44], Volk et al. [45], Yokota et al. [46], Zhang et al. [47]), cell regulatory proteins (Abdolmaleky et al. [48], Catts et al. [49], Gibbons et al. [50]), glia/macrophage proliferation (Busse et al. [51], Purves-Tyson et al. [52], Sneeboer et al. [53], Zhang et al. [54], Zhang et al. [55]), metabolic pathway (Afia et al. [56], Tang et al. [57]), and chemokines (Hill et al. [58], Nakatani et al. [59], Volk et al. [60]) were measured in postmortem brains of SCZ. The characteristics of included studies are summarized in Table 2.

As shown in Table 2, the main regions of the postmortem brains of patients with SCZ consisted of the DLPFC, PFC, CB, HPC, temporal lobe, cortical gray, STG, etc. The different regions of brain were analyzed using polymerase chain reaction (PCR) , Western blot (WB), enzyme linked immunosorbent assay (ELISA), and immunohistochemical (IHC) methods. Therefore, the analysis of different regions of the brain showed differential expression of inflammatory factors. For example, 8 studies reported the expression of TNF-α protein levels; 3 studies on DLPFC and one for cortical gay and STG respectively found no difference in expression; one study on PFC, STG, and FC respectively found an increase. Furthermore, other inflammatory factors including IL-1β, IL-4, and IL-10 had similar results.

Immune/Inflammation Response, Cell Regulatory Proteins, Glia/Macrophage Proliferation, and Chemokines in Postmortem Studies of BD

Regarding BD, 18 studies were included. The immune/inflammation response (Bezchlibnyk et al. [61], Dean et al. [23], Fillman et al. [26], Foster et al. [27], Hoseth et al. [29], Iwamoto et al. [31], Kim et al. [62], Kim et al. [63], Kim et al. [33], Lanz et al. [35], Maida et al. [37], Nascimento et al. [64], Rao et al. [65]), cell regulatory proteins (Abdolmaleky et al. [48], Catts et al. [49]), glia/macrophage proliferation (Zhang et al. [55]), and chemokines (Hill et al. [58], Nakatani et al. [59]) were measured in postmortem brains of BD (Table 3).

Table 3 shows that the main regions of the postmortem brains of patients with BD consisted of DLPFC, FC, OFC, ACC, MFG, CB, etc. The different regions of brain were analyzed using PCR, WB, ELISA, and IHC methods. Few studies were included in the analyses of postmortem brains of BD. Therefore, the analysis of inflammatory factors was less than that of SCZ. Unlike SCZ, the postmortem brains of BD showed different expressions in the same regions of the brain in different studies. For example, two studies analyzed the expression of TNF-α protein levels in DLPFC, one study an increase, and one study no difference. This may be due to the number of deaths from suicide, technique, and/or other factors affecting the heterogeneity of the study.

Immune/Inflammation Response, Cell Regulatory Proteins, Glia/Macrophage Proliferation, and Metabolic Pathways in Postmortem Studies of MDD

As shown in Table 4, there were 21 studies included for MDD, immune/inflammation response (Böttcher et al. [66], Dean et al. [67], Dean et al. [23], Foster et al. [27], Khundakar et al. [68], Lanz et al. [35], Maida et al. [37], Martín-Hernández et al. [69], Morrison et al. [70], Pandey et al. [71], Pandey et al. [72], Pandey et al. [73], Pantazatos et al. [74], Thomas et al. [75], Torres-Platas et al. [76], Wang et al. [77], Zhao et al. [78]), cell regulatory proteins (Tanti et al. [79]), glia/macrophage proliferation (Zhang et al. [55]), and metabolic pathway (Clark et al. [80]) were measured in postmortem brains of MDD.

The main regions of the postmortem brains of patients with MDD consisted of frontal the lobe, temporal lobe, thalamus, subventricular zone, DLPFC, BA, and OC. The methodological approaches mainly used CyTOF measurements, GC/MS, PCR, WB, and IHC. Similar to SCZ, the analysis of different regions of the brain showed different expressions of inflammatory factors. For example, 10 studies reported the expression of TNF-α protein levels; four studies on the DLPFC and one for the frontal lobe, ACC, and PC respectively found no difference in expression; one study on the ACC, VLPFC, and PFC respectively found an increase, and one study for the VLPFC found a decrease. Moreover, other inflammatory factors included HLA-DR, TNFR1, and NF-κB, but only one study reported it, and therefore, it cannot be summarized effectively.

Discussion

SCZ, BD, and MDD have been linked to neuroinflammation and metabolic disorders [81], which have been shown to have aberrant blood cytokines in blood [15, 82]. This study systematically reviewed the literature reporting brain inflammatory markers in the postmortem brains of SCZ, BD, and MDD patients.

Multiple studies have evaluated neuroinflammation markers, chemokines, and microglial activation in postmortem brain samples of SCZ [8, 83, 84]. However, it is impossible to determine whether there are the abovementioned facts in postmortem brain samples of SCZ due to a large number of null studies. For example, 39 studies were included to measure the differential expression between SCZ and control subjects. Out of 26 studies that evaluated inflammatory markers, 12 examined IL-1β. Therefore, whereas eight studies found no differences, three found a decrease, and one had elevated IL-1β expression. Similarly, six studies evaluated the anti-inflammatory cytokine IL-10, three found no effect of SCZ, two studies found a decrease, and one study found an increase. Previous studies have implicated proinflammatory profiles in psychiatric disorders, where the most consistent findings were alterations in TNF-α and related pathways [85,86,87], which have been reported in peripheral blood. Thus, the researchers set out to examine TNF pathway-related molecules at the protein and mRNA levels in the postmortem brain of SCZ patients in search of a larger association. TNF-α protein levels or mRNA expression were determined in eight studies, six had no effect, and two studies found an increase. Cytokine modulators (Toll-like receptors, colony-stimulating factors, and members of the complement system) have been evaluated in several studies. Three studies reported TLR4 in the postmortem brains of SCZ patients; one study found an increase and two studies found a decrease. The analysis of different regions of the brain is one of the heterogeneous variables in studies. For example, studies evaluating IL-1β expression have analyzed nine brain regions, including the PFC, DLPFC, cortical gray, STG, FC, HPC, STC, and CFC. Therefore, eight of them with no differences included the PFC, DLPFC, cortical gray, STG, and HPC. For the PFC analysis, four studies found no differences and one found a decrease. Nevertheless, although more studies have indicated no difference in IL-1β expression in the PFC, not all studies.

SCZ is a common mental illness associated with suicide [88]. A previous study found that there was a trend in microglial density and elevated proinflammatory cytokines in SCZ [89, 90]. In this study, fewer included studies analyzed the differential expression of inflammatory factors between suicide and nonsuicide. However, there is evidence that there is a difference in the expression of inflammatory factors between suicide and nonsuicide SZ patients [40]. Previous studies have shown an effect of SCZ on neurokinin receptors compared to suicide victims, which may confound the results [91]. Although some studies have considered the impact of suicide on the measurements of inflammatory factors, many studies have not reported these data or included it in statistical analysis, which makes it a limitation. Treatment for SCZ might reduce proinflammatory markers [92]. These findings may be associated with potential effects on neuroinflammatory markers in SCZ in postmortem brains. This is noteworthy because not all studies measured antipsychotic levels at death or corrected for this potential confounding factor. Furthermore, the separation of antipsychotics was not considered in the statistical analysis. In addition, subjects in the control group were not exposed to antipsychotics, which may cause confusion between the HC group and the SCZ group.

The same effect was also observed for BD and MDD, and it is impossible to determine whether inflammatory factors were significantly expressed in postmortem brain samples of BD and MDD. For example, seven of the included studies that measured the differential expression of inflammatory markers between the BD and control groups examined TNF-α. Therefore, six studies found no differences, and one had elevated TNF-α expression. Similar to BD, the MDD results found that six studies found no effect, one decreased, and three studies found an increase. Previous studies have indicated that inflammation is documented extensively in BD and MDD [93, 94]. The heterogeneity of our systematic review may be explained by these results. One of the heterogeneous variables may be the brain region. Significant functional and structural alterations in the neural circuits of emotion or reward processing may explain the heterogeneity. During emotional, reward, and/or cognitive related tasks, different activation patterns occur in the neural network including the amygdala, ACC, PFC, and striatum [95]. In addition, different stages of BD and MDD have distinct neurobiological changes in the related brain regions [96, 97]. Similar to SCZ, the treatments for inflammation also influence the expression of neuroinflammation markers. Aspirin has beneficial effects in clinical trials of mood disorders; it inhibits the inflammatory response and reduces the levels of inflammatory biomarkers, including C-reactive protein, TNF-α, and IL-6 [98]. These missing treatment statistics may influence the effects on neuroinflammatory markers in BD and MDD in postmortem brains. Another variable may be the differences in the methodological approaches. Although most techniques of the included studies were PCR and WB, other detection methods such as CyTOF measurements may contribute to the heterogeneous results. A previous study showed that the kynurenine pathway in the CNS of suicide attempts is chronically dysregulated, and an increase in inflammatory cytokines is associated with more severe symptoms [99]. In addition, a similar study also reported on BD, which is related to baseline biomarkers of suicide attempts with clinical outcomes [100]. Therefore, it is important to consider elevated proinflammatory cytokines in postmortem brains of suicide victims, which may confound the results. Several other confounding factors, such as age, lifestyle choices, and brain banks (diagnostic methods, storage, inclusion and exclusion criteria, and many other variables), may also need consideration.

These findings indicate that inflammatory markers appear to have different expression patterns in each psychiatric disease, which is of great significance for us to realize the pathophysiology of inflammatory markers in major psychiatric diseases and provide new directions for therapy. However, because the samples all came from postmortem brains, there was no record of the use of antipsychotic drugs before death. It is a limitation of this paper that the influence of antipsychotic drugs on the expression of the protein and molecules in the samples cannot be taken into account in statistics.

In conclusion, although numerous included studies have noted a lack of changes in neuroinflammatory markers in postmortem brain samples of SCZ, BD, and MDD, there are still multiple studies that have indicated an increase or decrease in neuroinflammatory markers. The heterogeneity is not clear at present, and may be caused by several factors, including the measured brain region, disease stage, brain source, medication, and other factors. The expression of neuroinflammatory markers was different, which means that inflammation was accompanied by the occurrence of neuropsychiatric disorders. Whether the inflammation in the brain is the pathogeny of schizophrenia, bipolar disorder, and major depressive disorder or the pathological manifestations of these diseases. According to some preclinical studies [101, 102], after some anti-inflammatory treatment, the neuropsychiatric disorder symptoms have improved significantly. This finding revealed that inflammation in the brain may be the pathogenesis of schizophrenia, bipolar disorder, and major depressive disorder.

In the future, these potential sources of heterogeneity should be considered to measure neuroinflammatory markers in postmortem brain samples for patients with SCZ, BD, and MDD, which will contribute to the successful construction of a similar study.