Glioma is the most common (~75%) group of primary brain tumours in adults with an overall survival of less than 20% over 5 years, and therefore there is a need to better identify biomarkers for prediction and to support early diagnosis (1). Diagnosis can be difficult as many glioma patients present with non-specific symptoms such as headaches (2). Currently the gold standard test for identifying glioma is magnetic resonance imaging (MRI) and computerized tomography (CT) scans, however these tests cause significant burden to the NHS in terms of cost and long wait times (2). Liquid biopsies such as a blood test could be used as a tool to triage patients for further scans (2). Moreover, a liquid biopsy could have utility for disease monitoring and prediction of disease recurrence (3).

Previous literature has identified putative predictive factors which are associated with glioma, some of these factors include atopic disease and diabetes (4–6). A number of nested case-control studies have attempted to define pre-diagnostic biomarkers for glioma; some of which are informed by putative factors and others carry out a hypothesis-free investigation. However, in some instances it can be difficult to be establish whether the pre-diagnostic biomarker truly predates the onset of glioma or if it represents a change seen earlier in the natural history of disease. In this review we consider any pre-diagnostic biomarker, including proteins and metabolites, which is identified before glioma diagnosis.

Proteomics characterise the global protein landscape within liquid biopsies. Protein alterations linked to cancer could provide novel biomarkers for monitoring disease and early detection of cancer (7). Studies measuring protein marker levels in pre-diagnostic glioma patients compared to control individuals could inform pre-diagnostic biomarker discovery ( Figure 1 ). A number of pre-diagnostic protein biomarkers have been suggested to associate with glioma, these are outlined and considered below ( Table 1 ).

Epidermal growth factor receptor (EGFR) is an oncogene implicated in glioma initiation, tumour growth and progression (18, 19). Späth et al. measured pre-diagnostic serum levels of EGFR and human epidermal growth factor receptor 2 (HER2) in glioma patients. EGFR (Odds Ratio (OR)=1.58, 95% confidence interval (95% CI)=1.13-2.22) and HER2 (OR=1.63, 95% CI=1.09-2.44) were found to be elevated in serum samples with an increased risk of glioblastoma development. However only elevated HER2 levels (OR=1.39, 95% CI=1.00-1.93) were found to be associated with increased glioma risk (10). Insulin-like growth factor (IGF-1) has similar downstream signaling pathways to EGF (20). Some studies found no evidence of insulin-like growth factor 1 (IGF-I), insulin-like growth factor binding protein 1 (IGFBP1), IGFBP2, IGFBP-3, or IGF-I/IGFBP3 ratio being associated with glioma risk (9, 12, 21, 22). However, Lönn et al. found conflicting evidence with IGF-I/IGFBP-3 ratio found to be associated with glioma risk, although the sample size was small (9).

Schwartzbaum and colleagues investigated other growth factors from different pathways, with transforming growth factor-beta 2 (TGFB2) found to be inversely related to glioblastoma (OR=0.87, 95% CI=0.76-0.98) in the Janus Serum Bank (JSB) (11). Although, this relationship was not identified in all glioma cases as evidenced in other studies (11, 12). Increased circulating vascular endothelial growth factor (VEGF) was found to be associated with pre-diagnostic glioma (OR=1.46, 95% CI=1.18-1.82) in the JSB, whilst two other studies did not find the same relationship within the Department of Defense Serum Repository (DoDSR) and Northern Sweden and Disease Study cohort (NSHDS) cohorts (8, 12, 13). Increased levels of soluble vascular endothelial growth factor receptor 2 (sVEGFR2) was found to be associated with increased glioma risk (OR=2.44, 95% CI=1.29-4.61) (13). No associations were found between placental growth factor (PLGF), hepatocyte growth factor (HGF), transforming growth factor beta 1 (TGFβ1), fibroblast growth factor 2 (FGF2) and transforming growth factor alpha (TGFα) levels and pre-diagnostic glioma patients (8, 12, 13).

Many epidemiological studies have suggested that allergic and atopic conditions have an inverse relationship with glioma risk (5, 23). Most of these papers examine the association between immunoglobulin E (IgE) antibodies which have a role in type I immediate allergic response (24). Schlehofer et al. investigated atopic status through IgE levels in pre-diagnostic glioma cases’ serum versus controls in the European Prospective Investigation into Cancer and Nutrition. High-grade gliomas were found to have a borderline association with IgE levels, with the greater intensity of the IgE, the lower the OR. Interestingly, only women were found to have an inverse association between allergic sensitization (via IgE levels) and all glioma risk, this association was stronger in high-grade gliomas (25). Schwartzbaum et al. identified the same pronounced association between elevated levels of IgE and glioma risk also observed within women in the JSB. This study also observed an association between testing positive for total IgE and a decreased risk of glioma at least 20 years before diagnosis (26). Conversely, Calboli and colleagues did not find an association between elevated IgE levels and risk of glioma, but there was suggestive evidence of an association in the Nurses’ Health Study, Women’s Health Study (WHS), HPFS, and Physicians’ Health Study (PHS) cohort (27). A receptor for IgE, soluble cluster of differentiation 23 (sCD23), was also investigated but no relationship was found with pre-diagnostic glioma in the JSB and NSHDS cohort (12, 13).

Previous literature has postulated an inverse relationship between immune response derived from a virus. These include a history of chickenpox, shingles, and herpes virus and glioma risk (28–30). Sjöström et al. examined immunoglobulin G (IgG) antibodies for cytomegalovirus, varicella-zoster virus (VZV), adenovirus, and Epstein-Barr virus in the NSHDS, Malmö Diet and Cancer Study (MDCS), and the Diet, Cancer, and Health cohort from Copenhagen. While no overall associations were discovered between IgG levels for the viruses and pre-diagnostic glioma, an inverse relationship was found between VZV IgG levels and pre-diagnostic glioma more than two years before diagnosis, and also an inverse association between positive VZV IgG levels and glioblastoma risk (31). However, Brenner et al. identified no relationship between glioma and prior immune-related conditions (this included any allergy, autoimmune disease or a combination) (8).

Allergy related transcription factors were investigated to identify potential links in pre-diagnostic disease of glioma in the JSB. Elevated beta-Catenin levels in serum were found to be associated with pre-diagnostic glioma (OR=1.86, 95% CI=1.28-2.71) (12). However, no other associations with pre-diagnostic glioma were found in any of the other allergy related transcription factors that were investigated (forkhead box p3 (FOXP3), signal transducer and activator of transcription 3 (STAT3) and STAT6) (11, 12).

Interleukins (ILs) have key roles in the development, progression, and control of cancer (32). Brenner at al. found increased levels of IL15 and IL16 to be inversely associated with glioma risk in pre-diagnostic individuals in the DoDSR (8). However another study did not find this same association in the JSB (12). Schwartzbaum and colleagues found pre-diagnostic glioma to be associated with increased IL4 levels (OR=1.13, 95% CI=0.90-1.43) in the JSB (12). In their previous paper there was only a suggestive relationship between IL4 and risk of glioblastoma, however this was with decreased levels with IL4 less than or equal to 5 years before diagnosis (11). Studies from Schwartzbaum and Wu et al. also investigated the relationship between other ILs and pre-diagnostic glioma cases; however, these were found not to be associated in the JSB and NSHDS (11–13).

Soluble interleukin receptors have also been investigated to identify their relationship with pre-diagnostic glioma. Schwartzbaum and colleagues found an association with reduced soluble interleukin 4 receptor alpha (sIL4RA) (OR=0.92, 95% CI=0.76-1.12) and elevated IL4-sIL4RA (OR=1.37, 95% CI=1.16-1.61) with increased risk of glioma in the JSB (12). This evidence is also suggested in their previous paper using the same cohort with sIL4RA (OR=0.80, 95% CI=0.65-1.00) and IL4-sIL4RA (OR=1.02, 95% CI=1.01-1.04) (11). Elevated levels of sIL2RA (OR=1.48, 95% CI=1.01-2.16) and sIL6R (OR=1.90, 95% CI=1.14-3.17) was suggested to be associated with glioma risk in the NSHDS, but this relationship was not significant after multiple testing and was not consistent in the JSB (12, 13). Decreased levels of sIL10RB (OR=0.69, 95% CI=0.55-0.87) and Leukemia inhibitory factor (LIF) (OR=0.47, 95% CI=0.23-0.94) was associated with glioma 5 or less years before diagnosis (12). SIL13RA2 and IL12p40 were found to have no relationship with glioma risk in two separate cohorts (8, 11, 12).

Chemokines are small, secreted proteins that make up one of the largest families of cytokines, and have critical roles in immune function (33). Schwartzbaum et al. identified the only pre-diagnostic chemokine which was C-C motif chemokine 22 (OR=1.45, 95% CI=1.07-1.96 and associated with glioma risk greater than 15 years before diagnosis) in the JSB (12). The other pre-diagnostic chemokines analysed showed no association with glioma, these included monocyte chemoattractant protein 1, thymus and activation regulated chemokine, these were assessed in the JSB or DoDSR cohorts (8, 12). Similarly, monocyte chemoattractant protein 3, macrophage inflammatory protein 1 alpha and beta, fractalkine and chemokine (C-X-C motif) ligand 13 were assessed in JSB and NSHDS with no association (12, 13).

Interferons (IFNs) are a type of cytokine which have been implicated in cancer progression (34). However, to date no association has been found with pre-diagnostic glioma. Brenner and Schwartzbaum et al. investigated interferons in the DoDSR and JSB. Overall, IFN gamma, IFN beta and IFN-alpha/beta receptors were found to have no relationship with glioma risk (8, 11, 12).

Tumour necrosis factors (TNFs) are key regulators in immune and inflammatory response to cancer (35). Only tumour necrosis factor alpha (TNF-α) and TNF receptors have been explored pre-diagnostic bloods of glioma patients. However, no relationship was identified between TNF-α and glioma risk in the JSB and DoDSR (8, 12, 13). Interestingly, while no relationship was found between TNF-α, Schwartzbaum et al. investigated two soluble receptors, soluble tumour necrosis factor receptor 2 (sTNFR2) was found to have an association with glioma risk (OR=1.72, 95% CI=1.01-2.93) but there was no association with sTNFR1 (13). Two soluble molecules in the tumour necrosis receptor family were also studied, sCD27 and sCD30, but no relationship was found with pre-diagnostic glioma in the JSB or NSHDS cohort (12, 13).

Chronic inflammation is well-established driver of carcinogenesis, it can lead to tumour progression and aid metastasis (36, 37). There are multiple biomarkers which signal systemic inflammation, this is usually due to injury or stress in an individual. These include C-reactive protein (CRP), white blood cell count (WBC) and neutrophil to lymphocyte count (NLR) (38–40). Cote and colleagues published the only study, to our knowledge, looking at biomarkers of inflammation in pre-diagnostic glioma. Data obtained from the UK Biobank cohort suggested a borderline relationship between NLR (hazard ratio=1.54, 95% CI=1.00-2.39, p-trend=0.05) and pre-diagnostic glioma. No other relationship was identified between WBC or CRP and glioma risk (22).

Metabolomics is a tool which measures the broad landscape of metabolites within biological samples such as blood, urine, and saliva, and may be detected using techniques such as mass spectrometry and nuclear magnetic resonance (41). An understanding of the pre-diagnostic metabolic signature in liquid biopsies prior to development of cancer compared to control cases could provide insights to novel candidate pre-diagnostic biomarkers. Several pre-diagnostic metabolite markers have been postulated to associate with glioma, which are summarized and discussed below ( Figure 1 , Table 1 ).

Previous studies have suggested an inverse relationship between diabetes and glioma risk (6). Schwartzbaum et al. identified a similar inverse association between pre-diagnostic diabetes and glioma risk in the Apolipoprotein-related Mortality Risk (AMORIS) cohort. The same study identified glucose levels in serum and plasma of pre-diagnostic individuals to have an inverse association with glioma risk within the AMORIS and Metabolic Syndrome and Cancer project (Me-Can) cohorts (17). In contrast, Björkblom and colleagues did not find the same association between glucose levels and pre-diagnostic glioblastoma in the JSB (14).

Deficiencies in vitamin A isoforms (including retinol, retinoic acid and retinal) and vitamin D (most abundant form in the blood is 25-hydroxyvitamin D (25(OH)D)) have been evidenced to increase the risk of cancer in individuals. Whereas vitamin E, which comprises of two groups, tocopherols and tocotrienols each with four isomers (α, β, γ and δ) is considered to have an inverse role in cancer prevention (42). A number of vitamin isoforms (25(OH)D, retinol, α-tocopherol, and γ-tocopherol) were assessed in pre-diagnostic glioma within UK Biobank, Nurses’ Health Study, and Health Professionals Follow-Up Study (HPFS) by Yue et al. However, there was no evidence of an association in 25(OH)D in serum, and α-tocopherol, γ-tocopherol and retinol in plasma and glioma risk (43). However, Björkblom et al. identified an association in increased levels of α-tocopherol (OR=1.7, 95% CI=1.0-3.0) and γ-tocopherol (OR=2.1, 95% CI=1.2-3.8) related to glioblastoma risk in the JSB (14). Alternatively, Huang et al. reported decreased levels of α-tocopherol (OR=0.65, 95% CI=0.44-0.96) associated with glioma risk in the Alpha-Tocopherol, Beta-Carotene Cancer Prevention (ATBC) study.

The amino acid tryptophan is metabolized through the kynurenine pathway in more than 90% of cases and the downstream products of kynurenine pathway have been implicated in carcinogenesis (44). Samanic et al. investigated the levels of tryptophan, kynurenine, and the ratio of kynurenine to tryptophan derived from plasma in pre-diagnostic glioma cases but found no association with circulating plasma in glioma risk within the Nurses’ Health Study and HPFS (45). Similarly, no association was found between tryptophan and glioblastoma risk in the JSB (14).

Björkblom et al. investigated 180 small molecular compounds in pre-diagnostic glioblastoma cases compared to matched control individuals in the JSB. This study found elevated levels of γ-tocopherol (OR=2.1, 95% CI=1.2-3.8), α-tocopherol (OR=1.7, 95% CI=1.0-3.0), 2-keto-L-gluconic acid, erythritol, N-acetyl-L-alanine, xylose and erythronic acid to be associated with glioblastoma risk (14). Jonsson et al. carried out a similar study investigating 142 metabolites in a different cohort of pre-diagnostic glioma patients compared to controls in the NSHDS. This study identified elevated plasma levels of myo-inositol, scyllo-inositol, cysteine, glycine, glyceric acid, aceturic acid, phosphate, proline, 4-hydroxyphenylacetic acid, erythronic acid, erythritol, N-acetylglucosamine, creatinine, uric acid and urea to be associated with glioma risk (16). Interestingly, both studies found elevated levels of erythronic acid (P-value: 0.0043 and 0.0391) and erythritol (P-value: 0.013 and 0.0221) related to an increased glioma risk in pre-diagnostic liquid biopsies within the JSB and NSHDS. Huang et al. performed an analysis on 730 known metabolites in pre-diagnostic glioma patients and controls in the ATBC study. In this study a total of 43 metabolites were found to be associated with overall glioma risk before diagnosis. The strongest associations were found between lower levels of 2-oxoarginine (OR=0.56, 95% CI=0.37-0.85) and cysteine (OR=0.39, 95% CI=0.19-0.77) with overall glioma risk and decreased glycocholenate sulfate (OR=0.39, 95% CI=0.19-0.79) and methyl 4-hydroxybenzoate sulfate (OR=0.47, 95% CI=0.26-0.83) with high-grade glioma risk. The other evidenced associations between metabolites and glioma risk have been noted in Table 1 (15). Notably, in two separate cohorts (ATBC study and NSHDS) cysteine was found to be associated with glioma risk (P-value: 0.0069 and 0.014) (15, 16).

The progression of biomarker analysis through multi-omics technologies has allowed novel protein and metabolite biomarkers to be discovered in liquid biopsies (46). Despite common practice of these techniques in biomarker research, there are still some limitations to the use of metabolomics and proteomics platforms.

Firstly, within metabolomics and proteomics there are multiple platforms and assays to detect biomarkers in liquid biopsies. Depending on the method there are different limitations based on sensitivity, reproducibility, or cost of equipment which need to be considered in biomarker analyses (46, 47).

There are some protein isoforms and metabolites which are present in low abundance, so it can be much harder to detect using current technology. The presence of abnormal/malfunctioning proteins in the body tend to be degraded, which means these proteins are also not measured in proteomic analyses (46).

Metabolites and proteins are dynamic molecules which tend to be difficult to detect due to their complex and changing structures. Because of this, it is unfeasible to accurately measure the entire metabolome and metabolite analyses that are carried out require huge amounts of data (46, 48, 49). Protein and metabolite levels can alter in the body day to day, and throughout the life course depending on many factors including age and lifestyle (50). As this review only found publications using a nested case-control study design, most of the liquid biopsies were taken at a single time point only providing a snapshot of the dynamic molecules at that specific time.

In the current literature, metabolomic and proteomic studies investigating pre-diagnostic biomarkers of glioma have only been carried out in the blood (51). Further analysis of other bodily fluids remains an area for future exploration. Some of the likely candidates for pre-diagnostic biomarker discovery in glioma patients include cerebrospinal fluid (CSF) and urine as these liquid biopsies have previously identified potential biomarkers in glioma (52–54). CSF is more likely to show higher levels of central nervous system (CNS) specific biomarkers due to its proximity to the brain, but to obtain the biopsy this requires an invasive procedure which is not routinely performed (54, 55). In contrast, urine is easily accessible and non-invasive, however it may yield much lower levels of CNS specific biomarkers due to proximity, the blood-brain barrier and glomerular filtration (52, 54, 56, 57). Other bodily fluids which could be considered for biomarker discovery include saliva, stool, and breath (58–60). All of these liquid biopsies could be utilized to discover novel pre-diagnostic biomarkers for glioma or to validate markers that have been suggested in this review.

Additionally, there are many other analytes which can be detected in liquid biopsies and utilized in biomarker discovery. These include microRNA, extracellular vesicles, cell-free RNA, circulating tumour cells, and circulating tumour DNA (61–65). However, the potential of these analytes in the early detection of cancer is unknown (66). Some of these suggested analytes may only be detected in a later stage of glioma progression and therefore may not be suitable as pre-diagnostic markers.

Glioma is a devasting cancer with survival rates less than 20% over 5 years. It is often diagnosed at a late stage of disease progression and usually leads to significant mortality and morbidity. Currently little is known about pre-diagnostic biomarkers which predate glioma detection, but this could improve the earlier detection of glioma. To our knowledge, this review outlines all of the current literature examining studies where biomarkers were assessed pre-diagnosis. Limitations of metabolomics and proteomics in biomarker detection were considered and the future directions for the discovery and validation of pre-diagnostic biomarker for glioma were suggested.

LA was responsible for article conception, article structure, creating figures and tables, and drafting the manuscript. KK was responsible for drafting the article abstract and providing critical revision of the manuscript. PD and CH provided revisions to the manuscript. All authors contributed to the article and approved of the submitted version.