Measuring Cancer Hallmark Mediation of the TET1 Glioma Survival Effect with Linked Neural-Network Based Mediation Experiments

Abstract

This paper examines the effect of TET1 expression on survival in glioma patients using open-access data from the Genomic Data Commons. A neural network-based survival model was built on expression data from a selection of genes most affected by TET1 knockdown with a median cross-validated survival concordance of 82.5%. A synthetic experiment was then conducted that linked two separately trained neural networks: a multitask model estimating cancer hallmark gene expression from TET1 expression, and a survival neural network. This experiment quantified the mediation of the TET1 survival effect through eight cancer hallmarks: apoptosis, cell cycle, cell death, cell motility, DNA repair, immune response, two phosphorylation pathways, and a randomized gene sets. Immune response, DNA repair, and apoptosis displayed greater mediation than the randomized gene set. Cell motility was inversely associated with only 12.5% mediated concordance. We propose the neural network linkage mediation experiment as an approach to collecting evidence of hazard mediation relationships with prognostic capacity useful for designing interventions. Gliomas represent approximately 75% of primary brain tumors in adults. The median survival of adult patients with gliomas is less than five years, with overall survival of 8-14 months for patients with glioblastoma multiforme (GBM) 1. The current standard of care includes surgery followed by radiation therapy and treatment with temozolomide. The therapy causes cells to die as a result of extensive DNA damage. Factors that contribute to recurrence are glioma invasiveness, the growth of cell populations resistant to radiation and temozolomide, and the blood brain barrier 2,3. Classifications of gliomas have been adding molecular features to histology for better diagnosis, especially under circumstances when histological phenotype is unclear. For example, the revised World Health Organization classifications now include 1p/19q co-deletion as a defining feature of oligodendroglial tumors 4. The use of molecular features will result in more precise prognosis and in the design of treatments tailored to the tumor. Resistance to the alkylating agent temozolomide is associated with promoter methylation of the gene encoding the DNA repair protein O6-methylguanine-DNA methyltransferase 5. We recently reported that glioma cell lines deficient with Tet methylcytosine dioxygenase 1 (TET1) exhibited greater genomic instability and were more resistant to ionizing radiation therapy 6. TET1 belongs to a family of enzymes comprised of three members that catalyzes the conversion of 5-methylcytosine (5-mC) to 5-hydroxymethylcytosine (5-hmC), which is the initial step of DNA demethylation 7. We found that TET1-deficient cells displayed an attenuated DNA damage response (DDR). TET1-deficient cells fail to undergo apoptosis in response to ionizing radiation and have attenuated DNA repair resulting in high numbers of DNA strand breaks in the survival cell population 6,8. There has also been evidence indicating the involvem...

Characterization of the m6A/m1A/m5C/m7G‐related regulators on the prognosis and immune microenvironment of glioma by integrated analysis of scRNA‐seq and bulk RNA‐seq data

“…42 In addition, low TET1 expression could lead to compromised genome integrity and induce more mutations in gliomas, thereby increasing the heterogeneity of tumor cell populations and tumor resistance to treatment. 43 A growing number of studies are finding possible links between different RNA modification regulators. 17,18,[44][45][46][47] NSUN2 catalyzes m5C modification and METTL3/METTL14 m6A modification in the three "non-coding" regions of p21 mRNA, in addition, methylation of METTL3/METTL14 promotes methylation of NSUN2 and vice versa.…”

“…TET1 was also downregulated in glioblastoma, knocking down TET1 ‐promoted proliferation, migration and invasion of GBM cells 42 . In addition, low TET1 expression could lead to compromised genome integrity and induce more mutations in gliomas, thereby increasing the heterogeneity of tumor cell populations and tumor resistance to treatment 43 …”

Applications for open access normalized synthesis in metastatic prostate cancer trials

“…Machine learning models like those demonstrated here can identify biomarker-outcome patterns that are too complex for humans or remain unobserved due to human bias (Maertens et al, 2020 ). Machine learning models can identify causal relationships, an important aspect when considering disease progression and treatment effects (Luechtefeld et al, 2020 ). Guidelines for using algorithms in trial designs indicate that models could be used to refine eligibility criteria or stratify patients based on model outputs from biomarker data.…”