Assessing the Limit of Tumor Infiltration Detection Using Averaged Mass Spectra from a Handheld Picosecond Infrared Laser Mass Spectrometry Probe: A Proof-of-Principle Study

Abstract Utilizing picosecond infrared laser mass spectrometry (PIRL-MS), we set out to determine the limit of tumor infiltration detection offered by this ambient ion source. Using murine models of brain cancer subjected to handheld PIRL-MS, as well as spatially resolved PIRL-MS imaging and various tissue models, we estimated the limit of tumor infiltration detection to be approximately 15% using physically combined glioblastoma tissue sections and approximately 4% using computational pixel mixing of orthotopic medulloblastoma imaging data. These estimates are broadly consistent in magnitude with independent iKnife-REIMS studies that employed different sampling methods and experimental models. At these detection limits, PIRL-MS predicted positive margins with a sensitivity of 84.21% and a specificity of 86.43%. This observation is based on the analysis of n = 38 orthotopic glioblastoma murine tissue sections bearing various tumor ranges and utilized n = 178 classifiable mass spectral data points. As a proof-of-principle study, these estimates derive from a limited number of biological replicates, including n = 3 subcutaneous xenografts used to construct the tissue model, n = 5 orthotopic animals for validation, and n = 2 tumors underlying the imaging-based estimate and, therefore, should be regarded as preliminary rather than definitive detection limits. These findings highlight the potential of PIRL-MS as a rapid, label-free intraoperative tool for margin assessment and provide insight into the practical limits of tumor infiltration detection in heterogeneous brain tumors.