OS03.2 CSF metabolomic profiles can discriminate patients with leptomeningeal carcinomatosis from patients having high risk for leptomeningeal metastasis from brain metastasis or brain tumors

摘要

>Introduction: Leptomeningeal carcinomatosis (LMC) is terminal stage cancer disease that resist to conventional chemo-/radiation therapy and quickly devastating patient’s performance. Early diagnosis is necessary for challenging this formidable disease. However, cerebrospinal (CSF) cytology frequently results in false negative. Metabolic profile reveals tumor environment and has been tried to diagnose cancers at early stage. Here, we analyzed the CSF metabolome profiles and evaluate if they can be used differentiate patient with LMC from patients without LMC but having a risk factor for developing LMC. >Materials and Methods: We prospectively collected CSF from patients at time of spinal anesthesia, craniotomy, and CSF drainage after permission from both patient and institutional review board. CSF archive was consisted of 6 different groups; 1) group 1a; CNS tumor free with systemic cancer, 2) group 1b; no tumor, 3) group 2; LMC, 3) group 3; brain metastasis, 4) group 4; brain tumor other than brain metastasis, 5) group 5; CNS inflammatory disease without cancer. From the CSF archive, 199 samples were selected by following criteria; 1) in LMC group, CSF cytology proven at the time of sampling, 2) in the high risk group of brain metastasis and brain tumor, CSF cytology was proven to be negative and followed up at least 6 months. All metabolites in CSF samples collected from 6 different groups were detected as low-mass ions (LMIs) using triple-TOF mass spectrometry. Principal component analysis-based discriminant analysis (PCA-DA) and two search algorithms were used to select LMIs for differentiating healthy controls from patient group or for separating patients with LMC from other patient groups. >Results: A total of 10,905 low mass ion was evaluated thorough PCA-DA and can define group 2 of LMC at a sensitivity of 85% and specificity of 91%. Group 1b of no tumor can be differentiated with other groups at sensitivity of 83% and specificity of 84%. After selecting 33 low mass ion including indoleacrylic acid through algorithm 2, CSF metabolomics profile showed sensitivity of 100% and specificity of 92% for discriminating group 1b from others. After selecting 21 low mass ion including diacetyl spermine, CSF metabolomics profile can differentiate LMC patients from the high risk group of brain metastases and brain tumors at both sensitivity and specificity of 100%. >Conclusion: CSF metabolite profiles can differentiate these 6 groups of different cancer status. We evaluate these metabolic analysis can diagnose LMC and exclude the high risk group at an accuracy of 100%. We expect CSF metabolic profiles can replace CSF cytology for the diagnosis of LMC and in the future trial, can be used to differentiate false positive such as postoperative CSF cytology from truly proliferating LMC.