BENEVOLENTAI DRUG PIPELINE

Glioblastoma Multiforme

ONCOLOGY

End-to-end mechanistic drug discovery for Glioblastoma

What is GBM?

Glioblastoma Multiforme (GBM) is a complex, aggressive and lethal brain tumour with a devastatingly short average survival time due to its resistance to conventional radiation and cytotoxic chemotherapies.

Our work

GBM has defied conventional research efforts due to the complexity, diversity and rapid growth of GBM tumours. Our work focuses on mechanisms that may be contributing to the progression and relapse of the disease.

Progress so far

We used our target identification tools to identify several therapeutic targets and, after assessing safety considerations in GBM patients, we advanced one of the most promising candidates through to lead optimisation.

GBM is the most common of all primary brain tumours.(1)

It tends to occur in adults between the ages of 45 and 70 years old(2) but can also rarely occur in children.(3)

It has a poor prognosis with median survival time of 14-15 months after diagnosis.(4)

There is an urgent need to develop new more effective treatments, especially for temozolomide-resistant patient subgroups.(5)

Our work

Identifying therapeutic targets that modulate Glioblastoma stem cells

Despite intense research and over 1000 clinical trials, GBM is still one of the oncology indications with the highest unmet need. The highly aggressive and invasive tumour infiltrates into the brain tissues, making it difficult to remove through surgery. Survival rates are further lowered by the tumour-generating GBM stem cells, which can self-renew and are resistant to conventional treatments such as chemotherapy or radiotherapy. Finally, it is incredibly difficult for drugs to penetrate the tumour cells due to the blood-brain barrier, a protective membrane that separates the brain’s blood supply from the main circulation. Drugs that are usually effective in treating cancer are unsuitable as GBM treatments because they typically cannot pass through this membrane.

Consequently, for a treatment to be efficacious, it must work across three levels; to identify the responder patients for whom a new treatment would be effective, to specifically target the mechanism that leads to tumour growth and recurrence, and to design drugs optimised for passage into the brain.

Knowledge: expanding our disease knowledge

The process starts by enriching our proprietary Biomedical Knowledge Graph with data extracted from GBM literature. We use patient-level multi-modal datasets such as genomics and transcriptomics data to ensure that we capture the disease biology and heterogeneity of GBM, in addition to data from clinical trials, patents and other biomedical sources. This way we ensure that our machine learning models are trained on all available biomedical and clinical data, as well as disease-relevant context, with emphasis on the factors considered important by our GBM scientists.

Target identification: finding the right target

We use our AI platform to explore specific genes or pathways that affect the growth and proliferation of chemo- and radiotherapy resistant stem cells. Using AI enhances our capacity to identify potential molecular targets that will modulate the disease as single agents, or in synergy with the current standard of care treatments. Our predictive target identification tools identified targets relating to mechanisms that, when modulated, will have an impact on the disease. Target triage tools then display the biological rationale behind predicted targets, allowing the drug discovery experts to make data-driven decisions over which candidates to take forward to progress into experimental validation.

Precision medicine: finding the right patient

A further class of AI models are trained on specific data derived from GBM patient tumours. This approach finds hidden patterns within the data that could represent underlying disease mechanisms and could be predictive of GBM subtypes or patient groups most likely to respond to a particular treatment.

What's next

"BenevolentAI uses its AI platform to identify therapeutic targets that have the most potential disease-modifying effect and improve patient outcomes. We are working to ensure that our most promising target for GBM, which is currently in lead optimization, is rapidly and efficiently progressed through to preclinical testing."

Arpita Ray  —  Senior Principal Scientist, BenevolentAI
References

(1) AANS. Glioblastoma Multiforme. Available from: https://www.aans.org/en/Patients/Neurosurgical-Conditions-and-Treatments/Glioblastoma-Multiforme . [Accessed April 2021].

(2) AANS. Glioblastoma Multiforme. Available from: https://www.aans.org/en/Patients/Neurosurgical-Conditions-and-Treatments/Glioblastoma-Multiforme . [Accessed April 2021].


(3) The Brain Tumor Charity. Glioblastoma prognosis. Available from: https://www.thebraintumourcharity.org/brain-tumour-diagnosis-treatment/types-of-brain-tumour-adult/glioblastoma/glioblastoma-prognosis/. [Accessed April 2021].


(4) Thakkar et al., 2014. Available from https://cebp.aacrjournals.org/content/23/10/1985. [Accessed April 2021].

(5) Global Data Report on Glioblastoma, “Glioblastoma Multiforme (GBM): Opportunity Analysis and Forecasts to 2027”, 2018. [Accessed April 2021].