Key members of the interdisciplinary multi-centre research team: (from left) Professor Gilberto Leung Ka-kit, Professor Zhang Gao and Professor Liu Lunxu.
An interdisciplinary multi-centre research team led by the LKS Faculty of Medicine (HKUMed) and Faculty of Dentistry at the University of Hong Kong has constructed the world's largest multi-omics atlas of brain metastases. This comprehensive analysis included 1,032 brain metastasis samples from diverse primary tumours, together with 82 matched primary tumours and 20 glioblastomas (a highly malignant type of brain tumour) as controls. The findings provide a novel framework for classifying brain metastases and establish a foundation for the development of personalised treatment strategies, advancing the field of precision oncology. This breakthrough was published in the international journal Nature Communications [link to the publication].
Cancer cells interact with brain environment: a paradigm shift in treatment approaches
Brain metastases occur when cancer cells spread from primary sites, such as the lung, breast or skin, to the brain. Despite advances in surgery, radiotherapy and systemic therapies, brain metastases exhibit significant heterogeneity, making it challenging to identify treatments that are effective for all patients. Consequently, brain metastases remain a leading cause of cancer-related deaths, affecting up to 30% of patients with advanced‑stage solid tumours.
Previous research focused primarily on the characteristics of the primary tumour, but this approach overlooks the unique biological properties that cancer cells develop after interacting with the brain microenvironment. 'Our hypothesis is that once cancer cells spread to the brain, they evolve into several molecular subtypes, regardless of their site of origin,' stated Professor Zhang Gao, Associate Professor in the HKU Faculty of Dentistry and co-leader of the study. 'Importantly, these subtypes are not shaped by the primary tissue but by the unique microenvironment of the brain. This represents a major paradigm shift in how brain metastases should be treated—moving away from treating brain metastases solely based on the primary tumour type to developing treatment plans based on the distinct molecular characteristics that tumours acquire in the brain.'
Four major subtypes reveal new treatment directions
The team integrated genomic, transcriptomic (RNA-level), proteomic, targeted metabolomic, single-nucleus RNA sequencing, and spatial data from 1,032 brain metastasis samples, and identified four distinct brain metastasis subtypes: Neural-like (BrMS1), Immune-infiltrated (BrMS2), Metabolic (BrMS3) and Proliferative (BrMS4). Each subtype exhibits unique immune landscapes, metabolic programmes, biological features and potential therapeutic targets:
- The 'Neural-like subtype' expresses neural-related genes, displaying neural-like characteristics associated with the brain microenvironment and is sensitive to radiotherapy.
- The 'Immune-infiltrated subtype' features abundant immune cell infiltration, exhibits the longest overall survival, and may respond better to immunotherapy.
- The 'Metabolic subtype' exhibits abnormally active energy metabolism pathways, indicating that metabolism-targeted therapies may effectively control this tumour type.
- The 'Proliferative subtype' is characterised by high cell proliferation and poorer prognosis. Its activated proliferation-related pathways suggest potential for targeted therapeutic interventions.
Decoding the brain's immunosuppressive environment to enhance precision in therapies
Building on the unique biological characteristics of these four molecular subtypes, the research team explored the possibility of developing personalised treatment strategies tailored to each subtype.
The study found that the 'Neural-like' and 'Immune-infiltrated' subtypes had significantly higher levels of cytotoxic T lymphocytes (CTLs), cells responsible for attacking tumours, and more active immune checkpoint molecules, such as PD-L1 and CTLA4. These features suggest that both subtypes may respond better to immune checkpoint blockade therapy. The 'Immune-infiltrated subtype' has the most abundant immune cells, which can proliferate extensively and restore their tumour-fighting capacity when stimulated by immunotherapy drugs. Therefore, patients with the 'Immune-infiltrated subtype' may have the strongest sensitivity to immunotherapy among the four subtypes. Additionally, the study found that CTL infiltration level is an independent prognostic indicator. The more abundant CTLs in the tumour, the longer the patient's overall survival rate.
To validate the differences in sensitivity to targeted therapies across subtypes, the research team employed patient-derived organoid models for targeted drug screening. Functional experiments showed that the 'Metabolic subtype' responded more favourably to mTOR inhibitors, while the 'Proliferative subtype' showed greater sensitivity to CDK4/6 inhibitors.
Professor Gilberto Leung Ka-kit, Tsang Wing-Hing Professor in Clinical Neuroscience and Clinical Professor in the Department of Surgery, School of Clinical Medicine, HKUMed, stated, 'The brain's immunosuppressive environment and the blood-brain barrier make treating metastases particularly challenging. Our research demonstrates how different tumour subtypes interact with brain neurons and immune cells, paving the way for developing innovative combinations of targeted drugs, immunotherapy and radiotherapy to find effective treatments for these tumours.'
Professor Liu Lunxu from West China Hospital of Sichuan University and co-leader of the study, added, 'This groundbreaking research exemplifies the power of multi-institutional collaboration. By bringing together expertise from across the Chinese Mainland and Hong Kong, we were able to assemble an unprecedented dataset. This successful partnership demonstrates how collaborative science can accelerate progress in precision oncology and benefit patients with brain metastases.'
About the research team
The research was co-led by Professor Gilberto Leung Ka-kit, Tsang Wing-Hing Professor in Clinical Neuroscience and Clinical Professor, Department of Surgery, School of Clinical Medicine, HKUMed; Professor Zhang Gao, Faculty of Dentistry, HKU; Professor Liu Lunxu, West China Hospital of Sichuan University; Professor Mou Yonggao, Sun Yat-sen University Cancer Center; and Professor Jia Wang, Beijing Tiantan Hospital. Co-first authors include Dr Yang Zhenyu, Dr Wei Shiyou, Dr Duan Hao, Wang Xiuqi, Dr Zhang Dainan, Dr Karrie Kiang Mei-yee, Dr Deng Yulan, and Dr Yang Yuanzhong.
Acknowledgements
This research was supported by the Research Grants Council, the Government of the Hong Kong Special Administrative Region of the People's Republic of China, the National Natural Science Foundation of China, and other funding bodies. Tumour tissue samples were provided by Queen Mary Hospital in Hong Kong, Sun Yat-sen University Cancer Center, Beijing Tiantan Hospital, and West China Hospital of Sichuan University.
