File(s) under embargo





until file(s) become available

Mathematical Modelling of MGMT Dynamics in Glioblastoma

posted on 23.03.2022, 12:00 by Ayoub Lasri DoukkaliAyoub Lasri Doukkali
At the intersection of systems biology and bioinformatics, this work aims to use a combination of mathematical modelling and reverse engineering techniques applied to GRNs (gene regulatory networks) to gain a better understanding of the mechanisms of drug resistance driven by TMZ (temozolomide) in GBM (glioblastoma) and thus propose strategies to overcome this resistance. Several studies associated MGMT (O6-methylguanine-DNA methyltransferase) expression or promoter methylation status with a poor prognosis. However, the mechanisms through which MGMT contribute to drug resistance in GBM are not well investigated. Since very few models of MGMT dynamics have been developed, the first step in this study was to build a mathematical model that captures the existing biological findings about the dynamics of GBM cells, the expression of MGMT and the mode of action of TMZ while also calibrating and validating such a model with novel experimental data. The model was then used to investigate the relationship between MGMT expression and TMZ driven drug resistance through careful examination of the model parameter relationships. We found that, in the absence of any other assumption regarding epigenetic adaptations, noise at the transcription level may explain the observed resistance. Similarly, we also studied the contribution of cell death to the selection of a certain phenotype, in this case, a selective upregulation of MGMT in response to TMZ, and found that cell death plays a major role in phenotypic selection. Our mathematical model was capable of producing a selective up-regulation of MGMT in response to TMZ, suggesting that cells with high levels of MGMT divided and survived, i.e., phenotypic selection through cell death. Hence, the first chapter of this work offers an insight into MGMT expression in relation to GBM resistance and brings forward a possible explanation of the observations reported in several experimental studies where the authors detected an upregulation of MGMT in response to TMZ. Recently, several studies showed a discordance between MGMT expression and methylation status and called for further investigations to clarify the contribution of MGMT methylation and expression to GBM drug resistance. Therefore, the second step of this study was to build on the previously developed model to account for MGMT methylation dynamics and also incorporate recently unearthed global growth-transcription relationships in mammalian cells. This more complex model was also calibrated to relevant experimental data and used to investigate TMZ-driven drug resistance. More specifically, we probed the model’s ability to exhibit three different resistance mechanisms: phenotypic selection, methylation downshift, or mixed case (simultaneous phenotypic selection and methylation downshift). We were able to find each time a realistic parameter set that can lead to the aforementioned drug resistance mechanisms. Thereafter, we used this model to test different achievable strategies to overcome resistance by maximising cell death, consequently finding that inhibiting the MGMT translation rate is a robust strategy. Expanding the model to account for MGMT regulation was the final step of our study. In line with the modelling approach adopted in this thesis, scRNA-seq represents an appropriate data type to be used to infer MGMT regulation in GBM. Inability to identify a data set that contains sequenced tumours before and after TMZ treatment, we shifted our attention to understanding MGMT regulation in different subtypes of GBM. We faced many unanswered questions related to scRNA-seq data preprocessing and decided to answer these research questions by conducting a systematic synthetic study of the impact of imputation on network inference. For this study, we adapted our modelling approach developed in the first two chapters into a novel software tool for generating synthetic scRNA-seq data, which we made available online as Biomodelling.jl. Finally, we applied the lessons learned from this synthetic study to a recently published GBM scRNA-seq data set to estimate the gene regulation of MGMT in GBM cells.


First Supervisor

Dr Marc Sturrock

Second Supervisor

Dr Alexander Kel


Submitted for the Award of Doctor of Philosophy to the Royal College of Surgeons in Ireland, 2021

Published Citation

Lasri, A,. Mathematical Modelling of MGMT Dynamics in Glioblastoma [PhD Thesis] Dublin: Royal College of Surgeons in Ireland; 2021

Degree Name

Doctor of Philosophy (PhD)

Date of award



  • Doctor of Philosophy (PhD)

Research Area

  • Cancer