Using the Epitranscriptome to Diagnose and Treat Gliomas

Diffuse gliomas are among the most common tumors of the central nervous system, with high morbidity and mortality and very limited therapeutic possibilities. Diffuse gliomas are characterized by great variability in terms of age at diagnosis, histological and molecular features, classification, ability to progress to a higher grade and/or to disseminate in the brain, response to treatment and patient outcome. One of the major challenges in the management of diffuse gliomas is related to the heterogeneity of tumor behavior within the same tumor subgroup. Although efforts have been made in recent decades to improve tumor characterization and classification, with the integration of molecular markers (e.g. Isocitrate DeHydrogenase (IDH) mutation), it remains difficult to predict treatment response and patient outcome at the individual level. Yet accurate tumor classification is of paramount importance in choosing personalized therapy or avoiding unnecessary treatments. At present, the main diagnostic methods for detecting gliomas are based on histopathological features, mutation detection or chromosome copy number variation.

However, difficulties remain, particularly with tumor classification, due to tumor heterogeneity and sampling bias for tumors obtained from small biopsies. In particular, grade 2 ("low-grade") and grade 3 ("high-grade") gliomas cannot be easily distinguished, as intratumoral tumor grade heterogeneity is not uncommon in patients treated with extensive surgical resection. Another challenge posed by gliomas is longitudinal monitoring of disease progression, which currently relies mainly on repeated brain MRI scans, with no return to the tumor itself due to the difficulty of obtaining new tumor samples in this setting. New tools to detect tumor changes in plasma, before imaging changes occur, would be useful. However, circulating markers present a real challenge, as the detection of markers readily used in other cancer types (e.g. circulating free DNA and circulating tumor cells) is hampered by a lack of sensitivity in gliomas.

Several genetic, epigenetic, metabolic and immunological profiles have been established in gliomas, considerably expanding the knowledge of the biological characteristics of these tumors and helping to identify potential treatments. Recently, the world of RNA has emerged as a promising area to explore for cancer therapy, particularly since the (re)discovery of chemical modifications of RNA (epitranscriptomics). To date, over 150 types of post-transcriptional modification have been reported on various RNA molecules. This landscape complex of chemical marks embodies a new, invisible code that governs the post-transcriptional fate of RNA: stability, splicing, storage, translation. Importantly, RNA epigenetics has emerged as a new layer of gene expression regulation in healthy tissues as well as in other pathologies such as cancer.

Chemical markers are associated with cancer evolution and adaptation, as well as with response to conventional therapies. Based on these observations, it is envisaged that: (1) the RNA epigenetic landscape evolves with cancer progression, establishing a "chemical signature" that could be exploited for diagnostic, prognostic and treatment response prediction purposes; (2) several chemical marks are not mere "transient" alterations but rather "driving" alterations of the tumorigenic process; (3) unlike unmodified nucleosides, modified nucleosides are preferentially excreted as metabolic end products in urine after circulating in the blood. Consequently, altered RNA markers in cancerous tissues can be detected in urine and blood and exploited for diagnostic purposes. An original approach recently published combines multiplex analysis of RNA marks by mass spectrometry with bioinformatics and machine learning. Using total RNA samples extracted from an existing cohort of patients (59 grade 2, 3 and 4 gliomas; 19 non-cancerous control samples), a first "chemical signature" capable of predicting glioma grade with remarkable efficiency and accuracy has been established.

N6, 2'-O-dimethyladenosine (m6Am), the most up-regulated marker in glioblastoma (GBM), is a driver of colorectal cancer aggressiveness. Located at the 5' end of messenger RiboNucleic Acid (mRNA), m6Am can influence mRNA stability and translation efficiency. This chemical tag is deposited by the Phosphorylated Carboxyl terminal domain Interacting Factor 1 (PCIF1), also known as CAPAM (PCIF1/CAPAM) methyltransferase (writer) and removed by the Fat mass and Obesity-associated protein (FTO) demethylase (eraser). FTO is down-regulated in colorectal cancer stem cells (CSCs), consistent with m6Am accumulation. High levels of m6Am significantly enhance CSC properties such as in vivo tumor initiation and chemoresistance, without significant changes to the transcriptome. This aggressive phenotype can be reversed by inhibition of PCIF1, demonstrating the potential of targeting epigenetic RNA effectors. The preliminary data on patient-derived glioma cell lines suggest a similar mechanism in glioma, where down-regulation of FTO promotes sphere-forming capacity in suspension culture of GBM stem cells.

(3) A method has been established to detect RNA markers in plasma samples that yielded favorable results after analysis of plasma samples from a colorectal cancer cohort. The same process was used to obtain preliminary data by analyzing plasma samples from grade 2 glioma patients vs. healthy donors. This experiment confirmed the possibility of detecting and quantifying 20 circulating nucleosides in blood. Significant changes were demonstrated between healthy donors and glioma patient samples for some of the circulating nucleosides. Some were up-regulated (e.g. n6,2'-O-dimethyladenosine (m6Am), 1-methylguanosine (m1G)) while others were down-regulated (e.g. adenosine (A), 5-methoxycarbonylmethyl-2-thiouridine (mcm5s2U)). Importantly, not all the tagged RNAs detected were altered (e.g. N1-methyladenosine (m1A); 5-methylcytosine (m5C)). If confirmed by a larger cohort, these changes could constitute an epitranscriptomics-based circulating signature for early disease detection. This preliminary experience reinforces the interest in m6Am.

Finally, changes were also observed in the serum of the same patients compared to healthy donor subjects, but from other nucleosides. This underlines the importance of studying circulating markers in blood for the diagnosis of gliomas.

• Sensitivity of RNA-modified nucleoside expression marks for glioma diagnosis vs controls in urine for patients in cohort 1. [Time frame: At baseline, 3 months, 9 months and 18 months]
• Specificity of RNA-modified nucleoside expression marks for glioma diagnosis vs controls in urine for patients in cohort 1. [Time frame: At baseline, 3 months, 9 months and 18 months]
• Positive Predictive Value (PPV) of modified nucleoside expression marks for the diagnosis of glioma vs. controls in urine for patients in cohort 1. [Time frame: At baseline, 3 months, 9 months and 18 months]
• Negative Predictive Value (NPV) of modified nucleoside expression marks for the diagnosis of glioma vs. controls in urine for patients in cohort 1. [Time frame: At baseline, 3 months, 9 months and 18 months]
• Progression-free survival [Time frame: Time from histological diagnosis to date of progression according to the Response Assessment in Neuro-Oncology Criteria (RANO 2.0) or death from any cause, assessed up to 18 months.]
• Global survival [Time frame: Time from histological diagnosis to the date of death, whatever the cause, assessed up to 18 months.]
• The best response according to RANO 2.0 [Time frame: Time from surgery to magnetic response imagery (MRI) showing best response, assessed up to 18 months.]
• Quantitative value obtained by Liquid Chromatography-Mass Spectrometry (LC-MS) for each post-transcriptional modification (mark) of RiboNucleic Acid (RNA) in cohort 3 [Time frame: At baseline, 3 months, 9 months and 18 months]
• Immunohistochemical detection of the Alpha-thalassemia-X-linked intellectual disability (ATRX) protein [Time frame: At surgery, Day 0]
• Tumor grading and classification [Time frame: At surgery, Day 0]
• Immunohistochemical detection of kiel 67 (KI67) protein [Time frame: At surgery, Day 0 of our timeline]
• Sensitivity of RNA-modified nucleoside expression marks for glioma diagnosis vs controls in tumoral tissue for patients in cohort 1 and 2. [Time frame: At Surgery, day 0 of our timeline]