Characterization and diagnosis spectrum of patients with cerebrospinal fluid pleocytosis

Abstract

Purpose There is an overlap in the cerebrospinal fluid (CSF) characteristics of patients presenting with different etiologies of CSF pleocytosis. Here, we characterized patients with CSF pleocytosis treated in a large hospital. Methods A retrospective cohort study of 1150 patients with an elevated CSF leukocyte count > 5 cells/µl treated at a university hospital in Germany from January 2015 to December 2017 was performed. Information on clinical presentation, laboratory parameters, diagnosis and outcome was collected. Clinical and laboratory features were tested for their potential to differentiate between bacterial meningitis (BM) and other causes of CSF pleocytosis. Results The most common etiologies of CSF pleocytosis were CNS infections (34%: 20% with detected pathogen, 14% without), autoimmune (21%) and neoplastic diseases (16%). CSF cell count was higher in CNS infections with detected pathogen (median 82 cells/µl) compared to autoimmune (11 cells/µl, p = 0.001), neoplastic diseases (19 cells/µl, p = 0.01) and other causes (11 cells/µl, p < 0.001). The CHANCE score was developed to differentiate BM from other causes of CSF pleocytosis: Multivariate regression revealed that CSF cell count > 100 cells/µl, CSF protein > 100 mg/dl, CRP > 5 mg/dl, elevated white blood cell count, abnormal mental status and nuchal rigidity are important indicators. The CHANCE score identified patients with BM with high sensitivity (92.1%) and specificity (90.9%) (derivation cohort: AUC: 0.955, validation cohort: AUC: 0.956). Conclusion Overall, the most common causes for CSF pleocytosis include infectious, neoplastic or autoimmune CNS diseases in ~ 70% of patients. The CHANCE score could be of help to identify patients with high likelihood of BM and support clinical decision making.

Author reply to “reporting standards for cerebrospinal fluid studies: more transparency of laboratory data is needed” by JL Frater

“…Therefore, time intervals between acquiring and analyzing CSF samples are usually low. Despite this optimal setup, a significant number of cells was found to be destroyed in n = 4 CSF samples of our study [ 5 ]. We are unable to determine the actual reason for this, but we assume that CSF was sent to the laboratory with delay in these cases.…”

“…We are unable to determine the actual reason for this, but we assume that CSF was sent to the laboratory with delay in these cases. However, as the portion of such incidences was low (0,3% of all cases), we consider it unlikely that preanalytical time periods significantly affected the results of our study [ 5 ].…”