Carcinomatous meningitis is a condition in which tumor cells spread to the subarachnoid space. Leukocyte counting and typing of cerebrospinal fluid (CSF) cell components are performed manually or using flow cytometry. However, a detailed analysis of these variables using cytological specimens has not yet been reported. The present study analyzed cytological specimens using Giemsa staining and whole slide imaging with computer-assisted image analysis (CAIA) to clarify the characteristics of the leukocyte population in CSF, especially in carcinomatous meningitis. Manual evaluation was performed using 280 Giemsa-stained cytological CSF specimens. For 49 samples, CAIA was used for the whole area of Papanicolaou (Pap) staining, and Giemsa-stained specimens of the same samples were imaged using a virtual slide scanner. The nuclear morphology of the leukocytes was assessed, and the total leukocyte and leukocyte subset (lymphocytes, neutrophils and macrophages) counts were evaluated. Then, the number and percentage of each leukocyte subset population were evaluated. The total leukocyte count was significantly higher in Giemsa-stained specimens compared with in Pap-stained specimens. The percentage of macrophages was significantly higher in samples from patients with non-hematological tumors compared with in samples from patients without tumors, which was confirmed by manual evaluation of the specimens. In addition, the cut-off value of the percentage of macrophages that could discriminate between the tumor history negative cases and cytologically tumor positive cases was determined, revealing that a higher proportion of macrophages reflected the existence of atypical/malignant epithelial tumor cells in CSF samples. Thus, atypical cell screening and analysis of the background characteristics of the leukocyte population should be the focus of cytological specimen screening, especially not to miss carcinomatous meningitis.
Keywords: Giemsa staining; cerebrospinal fluid cytology; computer-assisted image analysis; leukocyte; whole slide imaging.
Copyright: © Kobayashi et al.