Abstract 2728: Multivariate analysis of cerebrospinal fluid protein biomarkers in central nervous system lymphoma patients and controls

Abstract Establishing the diagnosis of primary central nervous system lymphoma (PCNSL) is often difficult as the clinical presentation and radiographic features are nonspecific. The gold standards for diagnosis are cytologic analysis of the CSF, which has poor sensitivity, or brain biopsy, which may be associated with significant morbidity and a high rate of diagnostic failure. We are testing the hypothesis that the cerebrospinal fluid (CSF) proteome contains information which may facilitate early and noninvasive diagnosis of CNS lymphoma as well as provide prognostic information. We evaluated the CSF concentrations of three candidate CNS lymphoma protein biomarkers: CXCL-13, a B-cell homing chemokine, Interleukin-10 (IL-10), an autocrine B-cell growth factor, and antithrombin-III (ATIII), a candidate tumor biomarker which we previously identified by mass spectrometric analysis of the CSF. Determinations were performed by ELISA of CSF diagnostic specimens from 228 patients: 75 specimens were from patients with a confirmed diagnosis of non-Hodgkin's lymphoma of the brain and 153 controls including 15 specimens from patients with astrocytic neoplasms, 16 specimens from secondary brain tumors including metastases from breast and lung cancer, and 122 other controls including patients with established systemic lymphoma who staged negative for CNS involvement. Mean concentrations of CXCL-13, IL-10 and ATIII were each significantly elevated in the CSF of CNS lymphoma patients compared to the controls (p<0.0003). By contrast, mean CSF albumin concentration, measured to control for disruption of the blood-brain barrier in the two groups, was similar in CNS lymphoma versus controls (p=0.17). The potential utility of each of these candidate biomarkers for CNS lymphoma was evaluated by means of receiver operating characteristic (ROC) curve analysis, both individually as well as in combination. Bivariate determination of CXCL-13 and IL-10 concentrations yield the best potential discrimination of CNS lymphoma versus all other controls. Detection of IL-10 in combination with CXCL-13 in CSF identified CNS lymphoma with the highest accuracy with area under the ROC curve equal to 0.858 (79% sensitivity and 92.2% specificity), (p<0.0001). In addition, we noted a potential relationship between high CSF concentration of CXCL-13 and short survival at first relapse. These preliminary data support a role for CXCL-13 and IL-10 in the pathogenesis of B-cell non-Hodgkin's lymphoma involving the brain and support our hypothesis that molecular signals within the CSF may be rapidly evaluated in combination to facilitate early and relatively non-invasive diagnosis. This approach may also be applied with other diagnostic modalities such as neuroimaging and flow-cytometry. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 2728.