Post-neurosurgical bacterial meningitis (PNBM) is a severe complication in patients receiving neurosurgical treatments, with the reported incidence rate ranging from 0.3% to 10% in different neurosurgical diseases (Blomstedt, 1985; Zhang et al., 2017). Previous studies have demonstrated that stroke-induced immunosuppression increased the risks of systemic and neurological infections (Sarrafzadeh et al., 2011). In an epidemiological study, the incidence rate of PNBM in cerebrovascular disorder was reported to be 4.9% (Zhang et al., 2017). In comparison with community-acquired bacterial meningitis (CABM), the clinical symptoms of PNBM patients are usually concealed by primary diseases or surgery-induced injuries, thus providing limited information to support the diagnosis. Furthermore, the prophylactic application of antibiotics during the perioperative period also has a significant impact on pathogenic identification in cerebrospinal fluid (CSF). So far, an accurate and prompt diagnosis of PNBM is still difficult in clinical practice. Therefore, this study aimed to develop novel biomarkers in CSF with robust diagnostic potential to satisfy the urgent needs in the clinical diagnosis of PNBM.

Current studies have mostly focused on the diagnostic role of neuroinflammation-related cytokines and neurocyte-related biomarkers in central nervous system (CNS) infection, such as TNF-α, IL-1β, and GFAP (Baran et al., 2023; Caragheorgheopol et al., 2023). However, these molecules were also reported to participate in the pathogenesis of stroke. A neurotransmitter is a bioactive chemical that transmits information between neurons or between neurons and effector cells. Previous studies have found that neuroinflammation and infection in the CNS changed the neurophysiological environment and consequently influenced the levels of neurotransmitters (Wilson et al., 2002; Kong et al., 2022). Linthorst et al. (1995) reported that LPS also induced increased levels of extracellular 5-hydroxytryptamine (5-HT) and its metabolite 5-hydroxyindole acetic acid (5-HIAA) in the hippocampus region of rats. Furthermore, because neurotransmitter changes in the CNS not only reflect the pathological condition but also indicate potential neurological deficit, it might be theoretically considered a candidate biomarker in neurological disorders. However, we only found one clinical study concerning the association between community-acquired meningitis and CSF neurotransmitters.

However, in PNBM, the pathological process of infection might be diverse because of its secondary insult characteristics to the brain compared to CABM. Therefore, the neurotransmitter profile has diagnostic potential for PNBM. Moreover, it may also help the understanding of pathophysiological alterations of neural circuits in the complicated coincidence of primary neurological disease with secondary infection.

In this study, we performed a targeted metabolomics analysis of neurotransmitter precursors and metabolites in CSF samples from patients with stroke and aimed to analyze the diagnostic potential of neurotransmitters in PNBM. Furthermore, this study also offers information on the potential mechanisms of neural circuit changes in the pathological condition of PNBM.

In this study, we performed a targeted metabolomics analysis of neurotransmitter precursors and metabolites in CSF. The results indicated that 14 CSF neurotransmitters were downregulated but DOPAC was upregulated in patients with PNBM. These neurotransmitters were mainly involved in the pathways of tyrosine metabolism; butanoate metabolism; histidine metabolism; alanine, aspartate, and glutamate metabolism; glycerophospholipid metabolism; arginine and proline metabolism; and tryptophan metabolism.

Previous studies have demonstrated that many neurotransmitters are produced by the commensal bacteria in the gastrointestinal tract, such as dopamine, 5-HT, and GABA. We thought pathogenic bacteria in the CNS might not directly produce neurotransmitters. Although some evidence has shown that CNS infection probably influences bacterial composition and abundance in the gut, we mainly aimed to investigate the diagnostic potential of neurotransmitters for PNBM. Some biomarker candidates have been presented in clinical and laboratory investigations with diagnostic potential for PNBM, such as lactate acid and cytokines (Zhang et al., 2017; Kul et al., 2020), but there is still a lack of a robust tool for the accurate and timely diagnosis of infection. In our previous studies, we have explored the potential diagnostic roles of glycomics and immunity-related protein biomarkers for PNBM (Ye et al., 2023; Zhao et al., 2023). However, these molecules provided little information about the infection-induced neurological alterations. Neurotransmitters transmit information between neurons and effector cells, inducing nerve impulses and regulating neuronal excitability and inhibition. A pathological condition influences the neurotransmitter levels and neural circuits, consequently leading to alterations in neurological functions (Fiore et al., 2023; Zhang et al., 2023). Meloni et al. (2015) found that poly-arginine and arginine-rich peptides are highly neuroprotective in an animal model of stroke. Furthermore, Rachedi et al. (2024) demonstrated that the catabolism of glutamine and serine sustained proline and glycine anabolism and promoted collagen biosynthesis, consequently controlling pathologic vascular stiffness. Therefore, the alterations in neurotransmitters can not only provide characteristic changes in nerve function of the patients with PNBM but also serve as a treatment basis.

Irazuzta et al. (1999) reported that bacterial meningitis that was induced by Group B Streptococcus type III in rabbits upregulated the levels of glutamate and aspartate in CSF samples. These excitatory neurotransmitters increased neuronal stress and led to excitotoxic effects on the brain. Furthermore, Taj and Jamil (2017) found infection-induced significant changes in the catecholamine level (including DA, DOPAC, and HVA) in the brains of rats. Moreover, they validated the results in clinical samples (Taj and Jamil, 2018) and found the absolute levels of DA, DOPA, HVA, and 5-HIAA in CSF were elevated in the patients with CNS infections with Neisseria meningitidis, Listeria monocytogenes, and herpes simplex virus, respectively. Furthermore, the upregulations of glutamine and arginine in the CSF were also reported in patients with tuberculous meningitis (Nayak and Bhat, 2005; Parihar et al., 2022). In our study, we found that the levels of 14 neurotransmitter precursors and metabolites in CSF were downregulated but DOPAC was upregulated. The results were discrepant with those of studies in CABM. We hypothesized that the primary neurological disorder also had an impact on the changes in neurotransmitters. In clinical studies, decreased levels of neurotransmitters were reported to be associated with stroke, including dopamine, glutamine, histidine, and L-arginine (Armengou et al., 2003; Lin et al., 2018; Rout et al., 2023). Furthermore, mechanism studies also indicated that these neurotransmitters are involved in the pathogenesis of stroke, even providing a therapeutic effect. Park et al. (2003) used a dopamine infusion method in a neonatal bacterial meningitis animal model and found that dopamine restored the activity of cerebral cortical cell membrane Na⁺ and K⁺-ATPase and decreased lipid peroxidation products. Furthermore, dopamine also helps maintain adequate cerebral perfusion pressure to prevent cerebral ischemia, reduce cerebral energy depletion, and consequently attenuate brain injury. This infers that stroke may also have a significant effect on the metabolism of neurotransmitters and a different pathophysiological characteristic might exist between PNBM and CABM. Therefore, it is necessary to reconsider the applicability of the biomarkers that have been used in clinical routine tests for the diagnosis of CNS infection.

Among the differentially expressed neurotransmitters, the upregulated changes in DOPAC levels were of interest to us. As the relevant molecules in dopamine (DA) metabolism, the levels of HAV and 5-MT were downregulated. Meanwhile, DA levels had no statistical difference due to infection status. These neurotransmitters are involved in the tyrosine metabolism pathway. Among them, DA metabolism in glial cells is illustrated in Figure 4 . DOPAC and 5-MT, both the metabolic products downstream of DA metabolism, are the intermediate products and consequently metabolize to HAV. Our results indicate that the DA-5-MT-HAV pathway was inhibited in PNBM, while the DA-DOPAC pathway was activated. In this process, DOPAC was accumulated but did not then metabolize to HVA. We hypothesized that it would have contributed to the inhibition of catechol-O-methyltransferase (COMT) which can transfer DA to 5-HT and DOPAC to HVA, respectively. Taj and Jamil (2018) also found that in CNS infection by Listeria monocytogenes, a high rate of synthesis of DA was observed in comparison with its further degradation into products and the absence of HVA indicated the complete inhibition of COMT. Additionally, although scarce evidence has shown an association between CNS infection and COMT expression, some studies have shown that COMT level might be influenced by inflammation in vitro or infection at other organs in vivo. Ogburn et al. (2006) found that prostaglandin J2 decreased COMT expression level, inducing the sequestration of COMT into large aggregates and a decline in COMT activity in human neuroblastoma SK-N-SH cells. Additionally, Zhou et al. (2023) demonstrated that colonic COMT levels were significantly decreased and correlated with post-infectious states in diarrhea-predominant irritable bowel syndrome patients as compared to recovered patients and control individuals. Furthermore, some investigations also found associations between comt gene variants and infection-promoted risks in certain diseases, such as schizophrenia (Rovira et al., 2022) and HIV infection-related complications (Horn et al., 2017; Saloner et al., 2019). However, we did not test the COMT level or its gene variants in this study. Whether COMT plays an important role in DA metabolism in glial cells needs to be further validated in animal and cytological experiments.

In the bioinformatic analysis, we found that several pathways might be involved in the pathogenesis of PNBM. The results may provide potential information to further investigate how these neurotransmitters exert biological impacts on PNBM. However, after reviewing the published articles on mechanism studies, we found a lack of evidence of a direct association between neurotransmitters and PNBM. We hypothesized that neurotransmitters played a role in biological functions by regulating enzymatic activities. Sokolova et al. performed cytological experiments and found that protein tyrosine kinases, which linked signals from integrins to intracellular signaling pathways, were essential for both bacterial internalization and cytokine secretion by human brain microvascular endothelial cells (Sokolova et al., 2004). Additionally, in an animal model of meningitis, Tumani et al. demonstrated that the inability of hippocampal glutamine synthetase to metabolize excess amounts of glutamate contributed to neuronal apoptosis in the hippocampal formation (Tumani et al., 2000), as these precursors of neurotransmitters could metabolize into different bioactive compounds. After considering the possible differences in mechanism between PNBM and CABM, a further study should be performed on the neurotransmitters with differential expression characteristics in these two diseases from the perspective of mechanism research.

Some limitations should be noticed. First, we performed a preliminary single-center study which contained a limited sample size, and the statistical power value was just 0.053. The results may reflect local information about neurotransmitter changes in PNBM. In future research, the sample size should be augmented using a multi-center, cross-geographical approach to enhance the universality and persuasiveness of the findings. Additionally, because of the low positivity of Gram staining and bacterial culture, we lacked information on pathogenic types. Therefore, the results may be restricted to the general bacterial infection types post-neurosurgery. However, specific types of infections also need to be investigated to find out the particular metabolic style in CNS, which would facilitate a better understanding of the bacterial behavior of certain pathogens and help develop a proper treatment method to prevent infection-related complications. Furthermore, the control samples were extracted from stroke patients without PNBM, and we lacked absolute disease-free samples. Because CSF extraction from healthy subjects may have medical risks, i.e., reducing cranial pressure, causing infection, cerebral hernia, or even death, it is difficult to collect the samples. However, a comparison between the CSF samples from patients with PNBM and healthy subjects would be helpful. Moreover, the result of our study indicated that the potential biological functions of COMT in the pathogenesis of PNBM might be a key point of difference from CABM. To confirm this hypothesis, the pathway of dopamine metabolism in glial cells should be investigated in a mechanism study.

In summary, we found that the downregulated levels of D-glutamine, Boc-D-Tyr-OH, and L(+)-arginine and upregulated level of DOPAC in CSF had strong diagnostic efficiencies for PNBM in patients with hemorrhagic stroke. The result also offers potential targets in the treatment of PNBM and its complications.