Identification of CSF Biomarkers for the Diagnosis of CNS Lymphoma.

CNS lymphoma is an aggressive form of non-Hodgkin’s lymphoma. The current means for establishing its diagnosis are:

1. brain biopsy, associated with risk of brain hemorrhage;

2. cytological analysis of the cerebrospinal fluid (CSF), which is insensitive in over 50% of cases.

The identification of novel, sensitive and specific biomarkers are required to facilitate early, non-invasive diagnosis of lymphomatous involvement of the CNS. Moreover, elevation of CSF protein is an established adverse prognostic variable in patients with primary CNS lymphoma; however specific peptide markers which relate to prognosis have not previously been identified within the CSF. We are testing the hypothesis that peptide biomarkers for CNS lymphoma are present in the CSF and that these may facilitate noninvasive, early diagnosis, provide prognostic information and enhance clinical monitoring in the setting of minimal residual disease. We describe a two-dimensional liquid chromatographymass spectrometry-based sensitive method for differential quantification and identification of several hundred CSF proteins. Proprietary spectral interpretation and intensity-normalization software was used to quantify differential expression of proteins between controls and lymphoma patients in CSF. An application of this approach to biomarker discovery in CSF is presented. We performed two sets of analyses using independent CSF specimens. In the experimental set we compared the pattern of expression in 9 control vs 9 cases of CNS lymphoma. 130 peptides were differentially expressed between the two groups (p<0.001). In a second, validation set of cases consisting of 7 controls and 7 cases of CNS lymphoma, we observed 131 peptides which exhibited significant differential expression between the two groups (p<0.001). At least 28 candidate biomarker proteins were found to exhibit differential expression in both experimental and validation sets of cases. Unsupervised clustering of the relative expression of these candidate biomarkers identifies peptides which correlate with adverse prognosis (p<0.003). In addition, we have validated the differential expression of four candidate biomarkers by immunoblot analysis of CSF. Quantitative measurement of two such novel CNS lymphoma peptide biomarkers has also been determined by ELISA methods. One of these biomarkers, a serine protease inhibitor, termed L1, was shown to be expressed within CNS lymphoma tumors by microarray as well as by quantitative RT-PCR. Immunohistochemistry localizes L1 expression to tumor cells and tumor vasculature. In addition, both gene expression analyses and ELISA reveal that high L1 concentrations in CSF correlate with significantly worse prognosis. Receiver operator characteristic curve analysis of L1 suggests that this marker may distinguish CNS lymphoma from a variety of non-malignant neurologic conditions with promising sensitivity and specificity. A second novel biomarker which we have identified and validated by immunoblot and by ELISA is a lymphoid chemokine, also found to be expressed in CNS lymphomas by gene expression analysis. Elevated concentrations of this chemokine in the CSF were also shown to be associated with adverse outcome in CNS lymphoma patients.