Meningiomas account for about 40% of all central nervous system tumors. A relatively new classification system can help focus clinicians’ attention on the best treatment approach.
By John Halamka, M.D., M.S., Dwight and Dian Diercks President, Mayo Clinic Platform and Paul Cerrato, MA, senior research analyst and communications specialist, Mayo Clinic Platform
While most patients who develop this type of tumor respond well to surgery, meningiomas are “notoriously heterogenous,” according to specialists who manage the disease. In other words, several patients experience aggressive growth and too often suffer serious complications or die. Although the World Health Organization has a classification system for the disease, it doesn’t enable clinicians to determine an individual’s prognosis with any certainty or establish the optimal treatment protocol. With this challenge in mind, several investigators have identified unique molecular markers and are using AI models to refine the classification of these tumors in the hope of identifying the most aggressive forms.
DNA methylation has surfaced as one of the most promising of these molecular markers. Methyl groups are short chemical fragments composed of -CH3. When a methyl group attaches itself to a strand of DNA, it can turn off specific genes, including tumor suppression genes, or activate oncogenes. Both biochemical reactions increase the risk of cancer. Work by Mayo Clinic researchers Gelareh Zadeh, MD, and Kenneth Aldape, MD, demonstrated that DNA methylation profiling could stratify meningiomas into biologically distinct groups associated with recurrence risk and chromosomal instability, providing some of the first evidence that epigenetic signatures capture clinically meaningful heterogeneity beyond conventional histopathologic grading. Concurrently, German researchers who conducted a multicenter retrospective analysis discovered six unique methylation classes associated with genetic mutations and gene expression patterns. These new markers were better able to pinpoint patients at risk of meningioma progression. They concluded that: “DNA methylation-based meningioma classification captures clinically more homogenous groups and has a higher power for predicting tumour recurrence and prognosis than the WHO classification.”
Building on these early findings, Mayo Clinic researchers and others have likewise explored the possible role of DNA methylation as a prognostic signpost. Initially, they analyzed genetic markers, including DNA somatic copy-number aberrations, DNA somatic point mutations, DNA methylation, and messenger RNA abundance. As a result, they were able to develop a meningioma classification system consisting of four molecular groups: immunogenic, benign NF2 wild-type, hypermetabolic and proliferative. Mayo Clinic investigators then collected nearly 1,700 records of patients with the brain tumors and developed a machine learning based algorithm that only required DNA methylation. Like the original classification that required four genetic signals, the DNA methylation-only approach generated useful results that can help clinicians predict the likely outcomes for patients with meningioma. For patients who fell into the hypermetabolic group, progression-free survival was 7.4 years, while those with proliferative tumors only survived 2.5 years.
More recently, researchers demonstrated that deep learning applied to routine hematoxylin and eosin (H&E) whole-slide images can accurately infer meningioma molecular groups, chromosomal alterations, and clinical outcomes, achieving prognostic performance independent of established clinicopathologic factors and providing a scalable alternative to resource-intensive genomic profiling (Landry et al., Lancet Digital Health, 2026, In press).
Their research has several take-aways: Most hospitals that manage patients with meningioma don’t have access to the kind of laboratory resources needed to perform DNA methylation, but on a more positive note, the cost of such genetic sequencing is gradually coming down, which should eventually make it with reach. For those healthcare systems that can perform the analysis, the results will enable physicians to select adjuvant radiotherapy for patients with resected WHO grade 2 tumors. Equally important, Mayo Clinic clinicians have found that a “molecular group is strongly predictive of response to radiotherapy, the ability to classify prospective cases using DNA methylation alone will have immediate practice clinical impact to patient care.”
Finally, the ability to more accurately classify meningioma patients into risk groups will enable researchers who are recruiting patients into clinical trials to put them into homogenous subgroups. That in turn will address the misleading results from previous clinical trials that did not take into account these low and high-risk patient groups.
Of all the human cancers, brain cancer is one of the most frightening for patients and one of the most challenging for their physicians. The latest AI-powered research offers measurable hope for patients and clinicians alike.
Share this:
