APBN New Site

APBN Developing Site

Brain Tumours: Identifying Patients for Targeted Therapy

Using gene signatures with a unique characteristic pattern of gene expression seen in aggressive brain tumours, scientists were able to develop and identify key elements for precision medicine in the treatment of these tumours.

Neuro-Oncology Research at the National Neuroscience Institute (NNI), Singapore

The Neuro-Oncology Research Laboratory at NNI started in mid-2005, co-led by Associate Professor Christopher Beng Ti Ang, and Dr Carol Tang. A/Prof Ang is a Senior Consultant and Head of Neurosurgery at the Singapore General Hospital campus, whilst his long-time friend and former college classmate, Dr Tang, is a scientist by training. Their complementary expertise facilitates the translation of bench-to-bedside discoveries, with a goal towards identifying patients for targeted therapy. The team also engages industry partners to develop novel therapeutic strategies for the treatment of brain tumour patients. This is particularly important because not all patients can undergo repeat surgeries due to the invasiveness of cancer cells, and location of tumour growth at eloquent areas of the brain. In 2016, the NNI neuro-oncology team comprising clinicians and scientists, was awarded the Translational and Clinical Research (TCR) Flagship Program grant, the nation’s highest level of research funding accorded to a clinical institute for its leadership role in research and management of the disease. The laboratory research team has since grown to six scientists with varied expertise in computational biology, animal modelling and biological experimentation using clinical material procured from patients at the time of their surgery or follow-up visits to the clinic.


The Study: Patient Stratification and Implementation of Precision Medicine Approaches

The team recently published a 5-year effort, aimed at stratifying patients so as to identify individuals most likely to receive treatment benefit from a particular class of drugs against the STAT3 protein.1 Brain tumour patients, especially those with grade IV glioblastoma (GBM), often survive no more than fifteen months after the first diagnosis. This is due to the tumour acquiring a more resistant profile, making recurrence a dreaded outcome. For the past decade, brain tumour patients have been diagnosed based on the pathologist’s microscopic examination of resected tissue. Although GBM tumours appear histologically identical, variation in treatment responses to the current standard of care drug, temozolomide, is frequently observed. We now know from several public research efforts led by the US that the molecular fingerprint of such tumours can differ between GBM patients, and even within the same patient as the disease progresses. There is thus a paradigm shift in how diagnosis should now incorporate molecular parameters that guide the physician in treatment decision.3

The study was premised on STAT3 being the final molecular switch that must be activated prior to the tumour transiting to its therapy-resistant state. Using artificial intelligence (AI)-driven algorithms, the investigators first established a unique STAT3-related gene classifier to predict patient cohorts that would most likely respond to drugs targeting STAT3. The rationale is that if STAT3 activation could be mitigated, the patient would be expected to experience a longer lasting and curative outcome. Further validation was carried out in cellular and animal models established using the patient’s tumour material. Specifically, mouse models bearing orthotopic tumours created from clinical material were used to test the computational predictions. Although AI-driven technologies have seen a surge in many aspects of healthcare, its implementation in drug discovery has yet to bear fruit. BenevolentAI in London, a company that has attained the unicorn status with valuation at USD 1 billion, collaborates with NNI to identify drug targets and specific chemical entities for advancement into clinical trials. Much of AI technologies in drug discovery lack biological validation.

This is where the NNI Brain Tumour Resource complements BenevolentAI’s operating model, where researchers here attempt to assign a biological / clinical phenotype to the otherwise voluminous data emerging from AI predictions. Such an approach creates a threshold at which AI-derived data correlates with an observable disease profile. The investigators then went on to analyse recurrent tumours and their shift in molecular patterns. Indeed, recurrent tumours adopted the STAT3 activation profile. Collectively, the underlying message of the study reinforces the importance of serially monitoring the patient’s tumour, with molecular profiling necessary to guide the implementation of STAT3 drugs in clinical trials. Currently, there are several STAT3 drug candidates in clinical trials for various diseases including haematological malignancies. However, none of these studies to-date have applied stratification methods to identify potential responders and non-responders. This would ideally minimize the side effects and financial cost to the non-responder cohort, whilst providing significant survival benefit to the responding patients. The team plans to advance their study with the N-of-1 precision oncology trial, where the patient is closely monitored at each state of disease progression. Treatment decision will utilize a targeted approach to achieve maximum survival benefit.


The NNI Brain Tumour Resource

Over the past fifteen years, the NNI team of researchers has established a tumour resource comprising clinical material procured from patients at the time of surgery.1,2,4-7 Fluid samples such as blood is also collected during surgery, and at subsequent follow-up visits to the clinic. Cell lines and orthotopic mouse tumour models are also created from such clinical material for the purpose of facilitating preclinical studies. In addition, the team has shown that molecular profiles acquired from these patient-derived animal tumours recapitulate the patient’s original tumour morphology and gene expression patterns. This resource thus represents a core capability of precision medicine where prospective tumours are re-established for drug testing endeavours yet have retrospective clinical information for correlative analyses.


Future Goals

The STAT3 study highlights the importance of targeted therapy and patient stratification to improve survival outcome, and consequently quality of life. Brain tumours are especially devastating because with each surgery, the patient almost always experiences cognitive decline. In the actual clinical scenario, serial monitoring based on tumour tissue may thus not always be possible, and as such, monitoring the STAT3 molecular pattern using the patient’s blood or cerebrospinal fluid (CSF) becomes critical. These alternate methods though technically challenging would provide a less invasive approach with real-time documentation of the patient’s disease progress, with treatment decision still possible in the absence of repeat surgeries.

The team includes academic and clinical collaborators from the National University of Singapore, Duke-NUS Medical School and major clinical institutes in Thailand and US. This multi-disciplinary group is currently evaluating a microfluidic chip platform capable of multiplexing, with measurement of the STAT3 molecular pattern using a small amount of blood plasma. The technology hinges on effective detection of a nucleic acid signature derived from circulating exosomes in the plasma. These exosomes are shed from cancer cells in the brain. Moving forward, the study team will advance this chip detection platform into preclinical validation using patients with recurrent tumours. At each follow-up visit, the patient’s blood and CSF will be collected and the STAT3 signature validated. The research team seeks to provide proof-of-concept for advancement into clinical trial. Clinical collaborators in Thailand’s Siriraj Hospital and US will provide multi-site patient recruitment. This effort will facilitate the first STAT3-related therapeutic approach based on a stratified patient population, with rapid and effective serial monitoring of a patient’s fluid biopsy to implement appropriate therapy. [APBN]


  1. Tan, M., Sandanaraj, E., Chong, Y.K., Lim, S.W., Koh, L.W.H, Ng, W.H., Tan, N.S., Tan, P., Ang, B.T., Tang, C. A STAT3 based gene signature stratifies glioma patients for targeted therapy. Nature Communications (2019).
  2. Xu, L., Chen, Y., Mayakonda, A., Koh, L., Chong, Y. K., Buckley, D. L., Sandanaraj, E., Lim, S. W., Lin, R. Y., Ke, X. Y., Huang, M. L., Chen, J., Sun, W., Wang, L. Z., Goh, B. C., Dinh, H. Q., Kappei, D., Winter, G. E., Ding, L. W., Ang, B. T., Berman, B. P., Bradner, J. E., Tang, C. & Koeffler, H. P. Targetable BET proteins- and E2F1-dependent transcriptional program maintains the malignancy of glioblastoma. Proc Natl Acad Sci U S A 115, E5086-E5095 (2018).
  3. Louis, D. N., Perry, A., Reifenberger, G., von Deimling, A., Figarella-Branger, D., Cavenee, W. K., Ohgaki, H., Wiestler, O. D., Kleihues, P. & Ellison, D. W. The 2016 World Health Organization Classification of Tumors of the Central Nervous System: a summary. Acta Neuropathol 131, 803-820 (2016).
  4. Chong, Y. K., Sandanaraj, E., Koh, L. W., Thangaveloo, M., Tan, M. S., Koh, G. R., Toh, T. B., Lim, G. G., Holbrook, J. D., Kon, O. L., Nadarajah, M., Ng, I., Ng, W. H., Tan, N. S., Lim, K. L., Tang, C. & Ang, B. T. ST3GAL1-Associated Transcriptomic Program in Glioblastoma Tumor Growth, Invasion, and Prognosis. J Natl Cancer Inst 108 (2016).
  5. Koh, L. W., Koh, G. R., Ng, F. S., Toh, T. B., Sandanaraj, E., Chong, Y. K., Phong, M., Tucker-Kellogg, G., Kon, O. L., Ng, W. H., Ng, I. H., Clement, M. V., Pervaiz, S., Ang, B. T. & Tang, C. S. A distinct reactive oxygen species profile confers chemoresistance in glioma-propagating cells and associates with patient survival outcome. Antioxid Redox Signal 19, 2261-2279 (2013).
  6. Ng, F. S., Toh, T. B., Ting, E. H., Koh, G. R., Sandanaraj, E., Phong, M., Wong, S. S., Leong, S. H., Kon, O. L., Tucker-Kellogg, G., Ng, W. H., Ng, I., Tang, C. & Ang, B. T. Progenitor-like Traits Contribute to Patient Survival and Prognosis in Oligodendroglial Tumors. Clin Cancer Res 18, 4122-4135 (2012).
  7. Chong, Y. K., Toh, T. B., Zaiden, N., Poonepalli, A., Leong, S. H., Ong, C. E., Yu, Y., Tan, P. B., See, S. J., Ng, W. H., Ng, I., Hande, M. P., Kon, O. L., Ang, B. T. & Tang, C. Cryopreservation of neurospheres derived from human glioblastoma multiforme. Stem Cells 27, 29-39 (2009).