Evaluation of intracranial cerebrospinal fluid cytology in staging pediatric medulloblastomas, supratentorial primitive neuroectodermal tumors, and ependymomas

Object The objective of this study was to determine the role of intracranial CSF examination in detecting true cases of early tumor dissemination. Cerebrospinal fluid dissemination is an ominous feature of pediatric brain tumors, occurring in as many as 30% of medulloblastomas, 25% of supratentorial primitive neuroectodermal tumors (PNETs), and 5% of ependymomas at diagnosis. Detecting early dissemination is important for determining both treatment and prognosis. Dissemination can be detected by evaluating imaging of the full neuraxis and by examining CSF cytology. Neuraxis MR imaging and lumbar CSF cytology evaluation are widely accepted methods for determining dissemination. However, the value of examining intracranial CSF cytology in detecting early dissemination is uncertain. Methods Under an institutional review board–approved protocol, medical records, pathology reports, and radiology reports for 150 patients who had undergone resection of brain tumors (88 with medulloblastomas, 21 with supratentorial PNETs, and 41 with ependymomas) and who had been evaluated using neuraxis MR imaging studies in the last 15 years were retrospectively reviewed. Radiology results were compared with the CSF cytology results and long-term disease outcomes. Results Between lumbar and intracranial CSF cytology results, 7 of 40 were discordant: in 2 intracranial CSF was negative and lumbar CSF was positive, and in 5 the reverse was true. The discordance percentage was 18%, with a kappa statistic of 0.36. Between MR imaging and lumbar CSF cytology results, 11 of 65 were discordant: in 9 the lumbar CSF was negative and MR imaging was positive, and in 2 the reverse was true. The discordance percentage is 17%, with a kappa statistic of 0.27. Between MR imaging and intracranial CSF cytology results, 8 of 52 were discordant: in 3 intracranial CSF was negative and MR imaging was positive, and in 5 the reverse was true. The discordance rate was 15%, with a kappa statistic of 0.41. Patients with positive and negative results on perioperative neuraxis MR imaging studies had a median survival of 26.8 and 33.1 months, respectively (p = 0.02). Patients with positive and negative results on perioperative lumbar CSF cytology had a median survival of 20.1 and 31.4 months, respectively (p = 0.11). Patients with positive and negative results on intracranial CSF cytology had a median survival of 31 and 31.4 months, respectively (p = 0.84). Conclusions Discordance exists between the results of neuraxis MR imaging and lumbar and intracranial CSF cytology in perioperative detection of tumor dissemination for pediatric medulloblastoma, supratentorial PNETs, and ependymoma. In 1 case in this series, perioperative dissemination was detected by intracranial CSF cytology, but not by lumbar CSF cytology or neuraxis MR imaging. Isolated intracranial CSF cytology positivity may represent an earlier stage of disseminated disease. Complementary use of perioperative neuraxis MR imaging and lumbar and intracranial CSF cytology can reduce the incidence of missed diagnoses of dissemination. Survival analysis revealed that perioperative neuraxis MR imaging findings are correlated with survival, whereas perioperative lumbar and intracranial CSF cytology findings are not.