Improving the Diagnosis of Pediatric CNS Tumors
Post by Lincoln Tracy
The takeaway
Integrating multi-omic approaches in the neuropathological assessment of pediatric CNS tumors improves diagnostic accuracy.
What's the science?
The World Health Organization has recently released an updated classification system for the diverse range of central nervous system (CNS) tumors that occur in children and adolescents. However, the wide variety of different tumors out there have an equally diverse range of outcomes, which makes it challenging for clinicians to make an accurate diagnosis. This week in Nature Medicine, Sturm and colleagues sought to improve diagnostic accuracy in pediatric neuro-oncology by developing a next-generation sequencing gene panel and introducing a DNA methylation-based classification for pediatric CNS tumors.
How did they do it?
The authors collected CNS tumor tissue samples from 1204 patients seeking treatment in 65 hospitals across Germany, Switzerland, Australia, and New Zealand. They first classified the CNS tumors according to the WHO classification system and the newly developed DNA methylation-based class prediction algorithm, before comparing the similarity of the two classification systems. The authors then integrated next generation sequencing (i.e., a large-scale DNA sequencing technology) to detect relevant genetic alterations in the tumors.
What did they find?
The authors made a confident diagnosis for 87% of CNS tumors using the WHO classification system, while three percent of tumors could not be assigned to any existing category. Low-grade glial/glioneuronal tumors were the most common (38%), followed by medulloblastomas and high-grade gliomas (both 16%). In contrast, 79% of tumors could be confidently assigned to a DNA methylation class. Low-grade glial/glioneuronal tumors were again the most common classification (29%), followed by medulloblastomas (16%) and high-grade gliomas (10%). While there was a strong correlation between the two classification systems, a large portion of WHO-based tumor types could not be classified by the DNA methylation-based algorithm, including 34% of WHO-defined high-grade gliomas and 20% of low-grade glial/glioneuronal tumors. Additional visualization of DNA methylation patterns suggested there were novel molecular classes not represented by the original reference cohort. Genetic alterations were detected in 60% of tumors, most commonly in the BRAF gene, although 42% of tumors had a diagnostically-relevant mutation and 15% had a therapeutically-relevant mutation.
What's the impact?
The data support incorporating DNA methylation-based classification approaches in the WHO classification for CNS tumors, showing it is a useful tool for diagnosing many types of tumors – especially those that are otherwise difficult to diagnose. This study provides an initial example of how neuropathological multi-omic approaches can be integrated into clinical practice.