Central Nervous System Involvement in Chronic Lymphocytic Leukemia: Diagnosis and Treatment in a Retrospective Cohort of Thirty Patients

Introduction: Chronic lymphocytic leukaemia (CLL) with central nervous system involvement (CNSi) i.e. cerebrospinal fluid (CSF) infiltration and/or invasion of brain parenchyma is rare and poorly documented. There is currently no consensus either for diagnosis criteria or therapeutic management of CNSi in CLL. In order to characterize the clinical and biological features and determine the outcome after treatment, we present a retrospective cohort of 30 CLL patients with CNS involvement.

Patients and methods: We collected CLL patients fulfilling criteria for CNSi and treated within thirteen centres of the French Innovative Leukaemia Organization (FILO group) between 1995 and 2015.

CNSi was defined as lymphocytic documentation either leptomeningeal by positive CSF cytology and flow cytometry, or histopathological by CNS biopsy. Patients were excluded if CNS involvement was made of large cells in the context of Richter transformation.

Results: Thirty patients were included. Median age was 69 (range 42-87). Median time from CLL diagnosis to CNSi was 76 months (0-301). At the time of CNSi, Binet stage was A, B and C in 13, 9 and 6 patients respectively. There were 2 Richter syndromes. CLL was not progressive in 16 patients. FISH analysis showed 17p deletion in 5 patients. Clinical manifestations were heterogeneous, with both central (mostly visual disturbance, headache) and peripheral (mainly cranial nerve involvement) symptoms, leading to a diagnosis delay for 14 patients (median delay 6 months; 5 patients diagnosed > 1 year after first neurological symptoms). Diagnosis was based on CSF flow cytometry for 29 patients and 4 had histological diagnosis by biopsy. Lumbar puncture was positive in all cases with large range in cell count as well as in percentage of CLL cells. Median CSF cell count was 36/mm³ (range 1-1980) with predominance of lymphocytes (76 to 99%). Median percentage of CLL cells by flow cytometry was 16 (range 0.5-93). Magnetic resonance imaging was informative in 10 out of 27 patients. MRI abnormalities were FLAIR hyperintensity, pachymeningitis or mass syndrome. From CNSi, 5 years overall survival (OS) and progression free survival (PFS) was 71% and 56% respectively. At the time of CNSi, 20 patients had never been treated for CLL. Eleven patients received rituximab plus fludarabine/cyclophosphamide or bendamustine, with (n=7) or without (n=4) intrathecal chemotherapy (IT) as first line treatment for CNSi. Seven out of 9 patients evaluable for response are in persistent response after a median follow up of 55 months. In contrast, 5 out of 7 evaluable pre-treated patients receiving heterogeneous treatment didn't achieve CR nor PR. Five patients received ibrutinib with or without IT or rituximab. Four were previously treated (median 6 lines, range 2-8). Three out of 4 evaluable patients are in neurological CR. For treatment naïve and previously treated patients, 5 years OS was 80% and 56% (p=0.15), and 5 years PFS was 80% and 0% (p=0.006) respectively.

Conclusion: To our knowledge, this is the largest cohort of CLL patients with specific CNSi. Several important conclusions arise from this study: Symptoms were very heterogeneous and might lead to diagnosis delay. Diagnosis relayed on CSF flow cytometry including very low percentage of CLL cells. CNSi was the only sign of CLL progression in more than half of the cases. CNSi should be recognized as an initiation criterion of systemic treatment. Immunochemotherapy with RFC with or without IT, or R Bendamustine + IT was efficient in treatment naïve patients. Ibrutinib led to neurological response with or without IT in highly pre-treated patients. Prognosis seemed to be directly related to CLL prognosis and CNSi by itself did not appear as a poor prognostic factor.