The genomic alterations in glioblastoma influence the levels of CSF metabolites

Carnitine and 2-methylbutyrylcarnitine were found to be present at significantly higher levels in the CSF of GBM patients compared to controls (Fig. 3A, 3B). GBM cells have been shown to exhibit elevated levels of carnitine [15], and although this was found in our study, it is unclear what role the GBM size, GBM location, and other factors may play. Carnitine and acyl-carnitines are crucial components of the carnitine shuttle system (CSS), which facilitates the transport of fatty acids into the mitochondria for fatty acid oxidation (FAO) (Sup. Material 3A) [8, 20, 35]. Recent studies have implicated FAO as an important energy production pathway in GBM as well as a key mediator of GBM plasticity and adaptability [26, 31, 49]. An increase in the CSF levels of carnitine found in this study may indicate increased reliance on lipid metabolism in GBM. Organic cation transporter OCTN2 (SLC22A5) is the primary transporter of dietary carnitine throughout the body, and OCTN2 is highly expressed in the brain of GBM patients compared to healthy brains [15, 24, 29]. Consequently, the higher carnitine levels observed in the CSF of the GBM patients evaluated here may be facilitated by upregulation of OCTN2, which in turn could potentially be a response to greater demand for carnitine by enhanced fatty acid oxidation in GBM.

In our previous study of CSF metabolites, we found an association between higher CSF levels of shikimate and IDH-wildtype gliomas compared to controls [2], a finding recapitulated in Fig. 3C. CSF shikimate levels have also been found to be elevated in patients with autism spectrum disorder [5]. Shikimate is an intermediate in the shikimate pathway. This pathway is not present in animals but it is used by plants, bacteria, algae, and other microorganisms in the biosynthesis of aromatic compounds (Sup. Material 3B) [19]. While the shikimate pathway is active in some bacteria present in the human gut microbiome, transcriptome analysis shows that most gut bacteria do not possess a complete shikimate pathway [36]. However, computational modeling reveals that the gut bacteria Akkermansia muciniphila expresses all the genes encoding the necessary enzymes in the shikimate pathway and accounts for a significant portion of the gene expression associated with the shikimate pathway in the human gut microbiome. These data suggest that A. muciniphila could be responsible for a significant amount of total shikimate produced by human gut microbiome [36]. We have previously identified higher levels of the Akkermansia genus in the gut microbiome of glioma patients and a glioma mouse model, compared to controls [44]. Therefore, the elevated CSF levels of shikimate in GBM patients identified in the current study may be influenced by the gut microbiome, in particular Akkermansia sp., suggesting a possible relationship between the gut-brain axis and CSF metabolites. Additionally, recent studies demonstrate unique microbial signatures within cancer types, including GBM [39]. To determine the source of the CSF shikimate identified in this study, future studies could determine whether shikimate-producing bacteria are present in human GBM tissue [39]. In order to detect remnants of shikimate-producing bacteria in the CSF, metagenomic sequencing could be employed, in a manner similar to but more specific than that used clinically in patients with meningitis or encephalitis [58].

We found uridine CSF levels to be greater in pre-treatment GBM samples compared to control samples. Uridine is a pyrimidine molecule that is an intermediate in nucleoside synthesis, particularly in the catabolism of L-glutamine to B-alanine (Sup. Material 3C) [62]. Consistent with our results, uridine concentration was found to be higher in tissue samples of GBM, IDH-wildtype tumors compared to controls [25].

Previous studies have demonstrated differences in the levels of metabolites based on genetic alterations in tumors. For example, diffuse gliomas have been shown to exhibit distinct tissue metabolite profiles depending on IDH1 mutation status [55]. Moreover, our previous study revealed higher levels of D-2-hydroxyglutarate in the CSF of patients with IDH-mutant gliomas compared to patients with IDH-wildtype gliomas [17]. In the current study, we have identified novel differences in CSF metabolites between patients with TP53-mutant and TP53-wildtype GBM. CSF from patients with TP53-wildtype GBM contained higher levels of several metabolites including lactate, GABA, choline, carnitine, and carnitine derivates. In contrast, the levels of lactate, GABA, and choline were increased in the CSF of patients with GBM-PTEN-mutant, compared to GBM-PTEN-wildtype (Fig. 5).

We identified carnitine and carnitine derivatives as key metabolic biomarkers in the CSF of patients with GBM. Carnitine and its acyl derivatives emerged as significant factors in our dataset, with increased abundance in the CSF of GBM patients, particularly those with TP53-wildtype status. Carnitine, acylcarnitines 2-methylbutyrylcarnitine, propionylcarnitine, isobutyryl-L-carnitine, and deoxycarnitine were significantly more abundant in TP53-wildtype, compared to TP53-mutant, GBM patients (Fig. 6). In GBM patients, TP53 potentially plays an indirect yet critical role in regulating FAO and may subsequently affect carnitine and acylcarnitine levels in CSF. More specifically, p53-responsive elements have been identified in the first intron of the carnitine palmitoyltransferase IC (CPT1C) gene; studies in cell lines have shown that p53 can regulate CPT1C expression [48]. The effects of increased CPT1C expression on tumor progression and FAO activity is well-documented in the literature [11, 57, 60, 63]. While the exact mechanism for reduced levels of carnitine compounds in the CSF of patients with TP53-mutant GBM remains to be elucidated, impaired TP53 regulation of CPT1C expression in TP53-mutant GBM may be a contributing factor. When the TP53 mutations occur, FAO is reduced [48], and the tumor cells’ ability to use FAO as an alternative energy production pathway might be affected. This may result in a reduced demand for carnitine in TP53-mutant GBM and lower concentrations of carnitine and carnitine derivatives in the CSF. The present findings support this hypothesis, as the relative levels of carnitine, propionylcarnitine, and isobutyryl-L-carnitine were significantly increased in TP53-wildtype GBM but were not significantly different from control patients or patients with TP53-mutant GBM (Fig. 6).

The metabolites lactate, GABA, and choline are clinically relevant because they are measured in imaging studies through magnetic resonance spectroscopy (MRS) [56]. Some studies show that tumor mutation status can potentially influence MRS results [30]. Our results reveal that the levels of these metabolites can be influenced by the mutations present in the tumor (Fig. 7). Lactate, GABA, and choline were present at significantly lower levels in the CSF of TP53-mutant GBM patients when compared to that of TP53-wildtype GBM patients (Fig. 7A and 7C). On the contrary, the relative values of each of the three MRS-measurable metabolites were significantly higher in the CSF of PTEN-mutant GBM patients compared to that of PTEN-wildtype GBM patients (with the exception of lactate, which follows a similar trend) (Fig. 7D and 7F). Absence of p53 has been associated with downregulation of choline levels in human embryonic stem cells [65], suggesting a potential link between choline metabolism and p53.

In addition, TP53 is known to regulate glycolysis and influence the Warburg effect by negatively regulating lactate dehydrogenase [64]. Similarly, it has been proposed that mutant TP53 can stimulate the Warburg effect in cultured cells [61]. Interestingly, we observed lower lactate levels in the CSF of patients with mutant TP53. The relationship between lower lactate levels in CSF and the predicted effects of mutant TP53 in regulating the Warburg effect is unclear and requires further investigation. Given that the levels of lactate, GABA, and choline levels appear to be influenced by GBM mutations, it may be important to consider tumor mutations when interpreting MRS results. To evaluate this further, it would be necessary to correlate the CSF metabolite findings with MRS imaging and perform receiver-operator curve analysis to detect the sensitivity and specificity of a given cut-off on MRS that reliably predicts TP53 and PTEN mutation status when GBM tissue is not available.

Differences between male and female patients with glioblastoma have been described. For example, the incidence of glioblastoma is 1.6 times higher in men than women [51]. Also, men have worse OS than women. Although differences in the proportion of MGMT promoter methylation between male and female and female with glioblastoma have been proposed [9], it appears that the difference is not statistically significant [16]. In addition, studies have shown differences in the expression of glycolytic genes between males and females, with potential association to overall survival [23]. In our analysis, we identified some CSF metabolites that were significantly different between male and patients, consistent with the idea that physiological and hormonal differences associated with sex can influence the levels of metabolites in CSF in patients with GBM. This is important when trying to identify metabolites that can serve as GBM biomarkers, because it might be required to use different cutoffs for males and females, when using a CSF metabolite as biomarker to identify patients with GBM.

We identified four metabolites that showed a statistically significant association with OS (Fig. 8). Increased levels of 2-methylbutyrylcarnitine and aminobutanal were found to be associated with the presence of GBM, and higher levels were associated with worse OS (Figs. 3B and 3D and 8A and 8B). Aminobutanal is involved in the synthesis of the neurotransmitter GABA [27] and GABA metabolism has been linked to GBM survival [4]. Higher levels of acetylcholine in CSF were found to be associated with worse OS (Fig. 8C). Acetylcholine is a well-defined neurotransmitter with a wide variety of roles in the CNS [37]. GBM invasion was found to be significantly enhanced in brain regions with activation of acetylcholine receptors, and it has been suggested that acetylcholine autocrine signaling facilitates GBM invasion through brain tissue [54]. The same study also found that increased acetylcholine receptor expression in GBM correlated with worse OS.

One limitation of this study is that only CSF samples acquired post-treatment have tumor mutation data available for analysis. Another limitation is the small number of female patients in the post-treatment group. Although, the lower proportion of female patients in the post-treatment GBM group is a limitation, comparison of only the male GBM patients before and after treatment showed significant differences in CSF metabolites. This supports the result that there are differences between the pre- and post-treatment GBM samples that are not due to differences in the number of female patients between the groups. Additional studies with a larger number of patients, including CSF samples acquired before therapy, will provide further confirmation for some of our results. Also, how the time between treatment and CSF collection influences the levels of CSF metabolites is unknown. Despite these limitations, the results of this study increase our understanding of CSF metabolites in GBM patients.