Introduction
Glioma is a frequent tumor in the central nervous system (CNS), representing approximately 25% of primary brain tumors and 80% of intracranial malignancies in adults [
1]. The nature of aggressive growth and inherent resistance to conventional therapies make glioma one of the most lethal diseases. The median overall survival (OS) is only less than 2 years for the most common subtype glioblastoma multiforme (GBM) [
2]. Over the last four decades, considerable endeavors have been undertaken to ascertain the etiological factors contributing to gliomagenesis. Noteworthy risk factors, such as radiation exposure and genetic alterations have been linked to susceptibility to glioma [
3]. However, the etiology of gliomas has not been fully elucidated.
Previous epidemiological studies have suggested potential associations between psychiatric disorders and the susceptibility to brain cancers. Grinsphpoon et al. showed that men with schizophrenia were less likely to develop brain cancers [
4]. Dalton et al. reported that dysthymia increased the risk of brain cancers, with a standardized incidence ratio (SIR) of 1.18 (95% CI: 1.13–1.23) [
5]. A regional population-based study yielded no evidence of an association between bipolar disorder and the incidence of brain cancers [
6]. However, conventional observational studies have been constrained by the poorly-defined histological classification of brain cancers and potential methodological biases, leading to inconclusive or conflicting conclusions regarding the causal connections between psychiatric disorders and glioma.
Mendelian randomization (MR) is a powerful epidemiological approach that uses genetic variants robustly associated with exposure to estimate the causal effect on outcome. MR studies are less susceptible to reverse causality and unmeasured confounding due to the fixed nature of genetic variants at conception. Accordingly, MR analyses have been widely employed to investigate the causal relationship between risk factors and diseases.
To shed light on the causal effect of mental illness on glioma, we applied a two-sample bidirectional MR analysis and explored the associations of single nucleotide polymorphisms (SNPs) related to 5 frequent psychiatric disorders and glioma risk by using summary data from Psychiatric Genomics Consortium (PGC) and 8 glioma GWAS datasets.
Discussion
Until now, etiological factors for glioma have not been fully understood. By using the two-sample bi-directional MR analysis, we investigated the causal relationship between genetically predicted five psychiatric disorders and glioma risk. In the current study, we identified the genetic predisposition to SCZ increased the risk of non-GBM glioma, whereas genetically predicted MDD, BD, ASD and PD were not associated with glioma risk. Our findings provide evidence for the etiological basis of glioma and implicate the necessity for medical professional to be cautious for the occurrence of non-GBM glioma in patients with a history of SCZ.
The susceptibility of patients with SCZ to malignancies has been a subject of longstanding debate. However, accumulating evidence indicates that the causal relationship between SCZ and cancer risk is contingent upon the specific disease context. In individuals with SCZ, there is an elevated incidence of breast cancer, ovarian cancer, and thyroid cancer, while a reverse association has been reported for melanoma and bladder cancer [
18‐
20]. The association between SCZ and glioma has been inconclusive. One assumption posits that SCZ exerts a tumor-suppressive effect on glioma. This hypothesis relies on a limited number of observational studies and the underlying assumption that molecular alterations in shared genes and their regulatory networks have contrasting roles in the development of SCZ and glioma [
21,
22]. However, our results did not support the protective effect of SCZ against glioma. We found that the causal effect of SCZ on glioma was differently related to glioma subtypes. Specifically, SCZ was found to increase the risk of non-GBM glioma (OR = 1.13, 95% CI: 1.03–1.23,
P = 0.0096), while no significant influence was observed for GBM. The distinct causal relationship between SCZ and glioma subtypes is likely attributed to connectomic and genetic features specific to certain brain regions. SCZ has been well-recognized as a neurodevelopmental abnormality, which frequently disrupts highly connected hub nodes that functionally integrate anatomically disparate neural systems in brain network. Comprehensive studies on topology of the normative connectome and the anatomy of brain disorders demonstrated that SCZ lesion-concentrated hubs exhibit a non-random spatial distribution, with a notable preference in areas such as anterior cingulate, medial frontal and parahippocampal regions [
23,
24]. Meanwhile, recent works also revealed that brain network features are associated with gliomagenesis. Cerebral regions with highly functional hubness are more vulnerable for the occurrence of glioma, partially due to their elevated metabolic turnover. The vibrant energetic demands and increased reactive oxygen species generated by metabolic stress enhance the likelihood of oncogenesis in these regions [
25]. In addition, the combined neuroimaging and transcriptomic analysis unveiled a grade-specific pattern of regional susceptibility to gliomas. The constructed grade-related expression map demonstrated that genes overexpressed in GBM are preferentially enriched in parietal and occipital cortices. By contrast, genes overexpressed in non-GBM are found to be expressed in hub nodes such as anterior cingulate, motor, parahippocampal and entorhinal regions, which are commonly involved by SCZ [
26]. The accumulation of genetic aberrations specific to non-GBMs might guide the development of gliomas in brain regions frequented by SCZ.
MDD is a major mental illness that leads to disabilities. Chronic depression has been found to undermine immune response through the disturbed function of hypothalamic-pituitary-adrenal axis and elevated levels of proinflammatory cytokines, which also accounts for the development and progression of various cancers [
27]. Although observational studies have indicated an association between MDD and glioma, the causal relationship remains inconclusive due to inherent methodological biases and challenges in establishing the chronological order of events [
28,
29]. However, our MR showed that genetic predisposition to MDD does not confer an increased risk for glioma.
Our study has several strengths. To the best of our knowledge, this is the first MR study to investigate a causal link between a range of psychiatric disorders and glioma. Secondly, our analyses were based on information from the largest glioma GWAS datasets to date that compose more than 30,000 participants. In addition, the leverage of multiple germline genetic variants as proxies for psychiatric disorders reduces potential bias in conventional epidemiological studies. Moreover, we performed various sensitivity analyses, such as exploring horizontal pleiotropy by using MR-PRESSO approach, to support the robust causal inference [
14]. However, several limitations should be acknowledged. Firstly, analyses in this study investigated GWAS data derived from participants of European ancestry. Validation of our findings in other ethnic groups is warranted. Secondly, the presence of potential confounders, including hidden population stratification, cannot be entirely ruled out, which may introduce bias in the risk associations [
30].
In conclusion, by using a comprehensive Mendelian randomization approach, our study contributes evidence supporting the potential causal impact of mental illness on glioma. Specifically, we identified an increased risk of non-GBM glioma associated with schizophrenia. Genetic susceptibility in the overlapping of connectomic distribution of SCZ and non-GBM glioma may account for the association. In term of clinical pratice, it is important for healthcare providers to be vigilant to the potential occurrence of non-GBM glioma in patient with a history of SCZ. Meanwhile, further prospective studies and mechanistic investigations were warranted for a better understanding of the causative effect of SCZ on glioma.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.