Cerebrospinal fluid metabolomic profiles can discriminate patients with leptomeningeal carcinomatosis from patients at high risk for leptomeningeal metastasis

// Byong Chul Yoo 1 , Jun Hwa Lee 1 , Kyung-Hee Kim 1 , Weiwei Lin 2 , Jong Heon Kim 2 , Jong Bae Park 2 , Hyun Jin Park 3 , Sang Hoon Shin 4 , Heon Yoo 4 , Ji Woong Kwon 4 and Ho-Shin Gwak 2, 4 1 Biomarker Branch, Research Institute, National Cancer Center, Goyang, Republic of Korea 2 Department of Cancer Biomedical Science, National Cancer Center, Graduate School of Cancer Science and Policy, Goyang, Republic of Korea 3 Center for Pediatric Cancer, National Cancer Center, Goyang, Republic of Korea 4 Neuro-oncology Clinic, National Cancer Center, Goyang, Republic of Korea Correspondence to: Ho-Shin Gwak, email: nsghs@ncc.re.kr Keywords: cerebrospinal fluid, metabolome, leptomeningeal carcinomatosis, profile, diagnosis Received: July 28, 2017      Accepted: August 28, 2017      Published: September 18, 2017 ABSTRACT Purpose: Early diagnosis of leptomeningeal carcinomatosis (LMC) is necessary to improve outcomes of this formidable disease. However, cerebrospinal fluid (CSF) cytology is frequently false negative. We examined whether CSF metabolome profiles can be used to differentiate patients with LMC from patients having a risk for development of LMC. Results: A total of 10,905 LMIs were evaluated using PCA-DA. The LMIs defined Group 2 with a sensitivity of 85% and a specificity of 91%. After selecting 33 LMIs, including diacetylspermine and fibrinogen fragments, the CSF metabolomics profile had a sensitivity of 100% and a specificity of 93% for discriminating Group 1b from the other groups. After selecting 21 LMIs, including phosphatidylcholine, the CSF metabolomics profile differentiated LMC (Group 2) patients from the high-risk groups of Group 3 and Group 4 with 100% sensitivity and 100% specificity. Materials and Methods: We prospectively collected CSF from five groups of patients: Group 1a, systemic cancer; Group 1b, no tumor; Group 2, LMC; Group 3, brain metastasis; Group 4, brain tumor other than brain metastasis. All metabolites in the CSF samples were detected as low-mass ions (LMIs) using mass spectrometry. Principal component analysis-based discriminant analysis (PCA-DA) and two search algorithms were used to select the LMIs that differentiated the patient groups of interest from controls. Conclusions: Analysis of CSF metabolite profiles could be used to diagnose LMC and exclude patients at high-risk of LMC with a 100% accuracy. We expect a future validation trial to evaluate CSF metabolic profiles supporting CSF cytology.