Cerebrospinal Fluid Immunophenotyping in B-Cell Lymphomas: A Single Centre Experience on 99 Samples.

Central nervous system involvement in B-cell lymphomas is almost always lethal. Recognized risk factors include Burkitt histology and high LDH level, multiple extranodal sites or an age-adjusted International Prognostic Index of > 1 in the case of diffuse large B-cell lymphoma (DLBCL) (van Besien et al. Blood 1998, Haioun, Ann Oncol 2000). Standard diagnostic approaches such as imaging (MRI, CT) and cytological study of the cerebrospinal fluid (CSF) are flawed with a large number of false negative results. Accordingly, currently used prophylactic approaches based on these uncertain criteria carry a risk of over- or undertreating patients. Lately, flow cytometry (FCM) has been shown to detect a high incidence of “occult” CNS disease in patients at risk (Hedge et al. Blood 2005), thus paving the way towards a more specific management of these cases. We performed an exploratory single-center study on 99 CSF samples from patients with B-cell lymphomas. FCM results did not influence the therapeutic strategy. CSF samples were processed as soon as possible following lumbar puncture (always within 3hrs). They were washed in saline buffer, counted in Trypan Blue (to evaluate cellularity & viability), and 1/3 was labelled with the following antibodies: CD45, CD3, CD14, CD19, CD5, and CD10. Two thirds of the remaining cells were either labelled with the same antibodies (in case paucicellularity did impede interpretation of the first labelling on one third of the sample), or with one of two complementary antibody combinations: CD20, CD22, CD19 or Kappa/Lambda/CD19 according to cellularity and clinical data. From July 2005 to April 2007, 99 samples from 87 consecutive patients were analyzed. Median age was 61 years and 64% patients had an IPI score of 3 or more. Forty six patients were devoid of neurological symptoms at the time of analysis. Forty had DLBCL, 6 Burkitt. Thirteen out of the 99 samples could not be analyzed because they were massively contaminated with blood, 7 because they showed non specific fluorescence, thus leaving 79 (80%) analyzable samples from 75 patients. Although some samples contained an excess of B-cells (22% with ≥ 5% and < 10% CD19+ cells, 22% with ≥10% CD19+ cells), none showed a clearly phenotypically abnormal B-cell population. On a clinical basis, all were treated with curative intent and 82% had meningeal prophylaxis. No patient developed CNS localization with a median follow-up of 15 months.Twenty-five samples were taken from 19 patients with neurological symptoms (cranial nerve/radicular palsies n=9, CNS tumors n=6, epiduritis n=4). Three cases showed a clearly phenotypically abnormal B-cell population (1 case of Burkitt lymphoma, 2 cases of lymphoplasmacytoid lymphoma), all being also positive by cytology.In conclusion, although preliminary, these results suggest that in our hand FCM appears not more sensitive that cytology in detecting CSF infiltration by neoplastic B-cells in aggressive B-cell lymphoma when performing a single exploratory lumbar puncture. A larger cohort of patients needs to be explored with standardization of FCM immunophenotyping technique in order to improve further diagnostic accuracy of CSF localization in lymphoma.