Integrated morphometric and molecular classification of central nervous system cancers in 10 seconds using a unified platform with picosecond infrared laser mass spectrometry

The World Health Organization (WHO) now advocates for an integrated approach in the diagnosis of Central Nervous System (CNS) cancers. However, the proposed WHO workflow is complex, relying heavily on multiple instruments-like genetic sequencers and immunohistochemistry scanners-with variable, often lengthy, turnaround times. Leveraging the link between lipid metabolism and cancer, we demonstrate that Picosecond InfraRed Laser Mass Spectrometry (PIRL-MS) can profile the tissue lipidome in just 10 seconds, achieving over 80% sensitivity and specificity for CNS tumor type prediction. This was validated across a large cohort of 150 pediatric (3 morphometrically distinct types, 7 molecular classes) and 1,142 adult (24 morphometric, 8 molecular classes) brain tumour types. While initial results provided strong classification, falling short of robust diagnostic criteria, the ongoing incorporation of machine learning has increased sensitivity and specificity to over 94% in select cases. The capabilities of PIRL-MS to deliver a 10-second diagnosis-based on downstream phenomic (lipidomic) changes-for both morphologically and molecularly distinct tumor (sub)types (e.g., IDH mutation status, medulloblastoma subgroups) offers a clear path toward a unified diagnostic platform. This approach can integrate morphometric and molecular diagnoses on a single instrument in seconds, drastically reducing the need for tissue manipulation and lengthy preparation steps, a feat unmatched by current methods.