Paediatric brain tumour types revealed mid-surgery with nanopore sequencing and AI
A new study published in Nature showcases the release of “Sturgeon”, a pioneering neural network that harnesses the power of Oxford Nanopore sequencing for real-time molecular classification of central nervous system tumours during surgery
Researchers from UMC Utrecht in the Netherlands have unveiled a workflow that combines rapid sequencing and deep learning with the potential to transform neurosurgeries in the future. This development leverages Oxford Nanopore’s real-time DNA sequencing technology to address the challenges posed by central nervous system (CNS) tumours, one of the most lethal type of tumour, especially among children.
In a new Nature paper published on Wednesday, the researchers showed that they were able to successfully use nanopore sequencing during surgery paired with AI-powered analysis to determine tumour types in 20-40 minutes.
CNS tumours pose a significant challenge for neurosurgeons, as they navigate the fine line between removing as much of the tumour as possible and preserving neurological function to prevent complications. Surgeons often operate with limited knowledge of the exact tumour type before surgery, relying on previous preoperative imaging and intraoperative tissue analysis, which are not always conclusive, to guide decision making.
According to Prof. Dr Eelco Hoving, paediatric neurosurgeon and clinical director of neuro-oncology at the Princess Máxima Center, this can mean removing too much tissue, impacting healing, or too little, leaving behind some tumour tissues to prevent neurological damage. In some cases, a second surgery may be necessary “again creating risks and stress for the child and family,” he said in a press release.
‘Sturgeon’ is a deep-learning neural network trained to classify CNS tumours based on sparse methylation profiles that is now being used for research. Oxford Nanopore’s rapid sequencing technology can identify these profiles accurately and in real-time, offering critical insights into the molecular characteristics of the tumour. By combining these two innovative technologies, the researchers described how the time required for the entire process, from taking the biopsy to determining the tumour, has been dramatically reduced from one week to 60-90 minutes. The significance of this advancement lies in its ability to potentially allow surgeons to tailor their approach and surgical techniques to the specific characteristics of the tumour during surgery.
Over the next six months, further refinements to the process will be made, with the goal of validation and making this technology standard practice, leading to significant potential cost savings and improved patient outcomes.
Gordon Sanghera, CEO, Oxford Nanopore Technologies, commented: “Congratulations to the team at UMC Utrecht for this important research showing how advances in AI combined with Oxford Nanopore’s real-time sequencing have the potential to transform neurosurgical decision-making. By being able to access real-time insights during surgery, this study showed that neurosurgeons might in the future be better able to make critical decisions, eliminating the need for additional surgeries and ultimately improving patient outcomes. Although this was carried out in a research context, we are excited by the potential and look forward to this workflow becoming more available to clinicians and their patients globally.”
Dr. Jeroen de Ridder, research group leader within UMC Utrecht and Oncode researcher, commented: “Recently, technology to read DNA in real time has become available. By learning from millions of simulated realistic 'DNA snapshots’, an algorithm was created that determines the tumour type in 20 to 40 minutes. And that is fast enough to adjust the surgical strategy if necessary.”