Clinical application of an algorithm to screen for malignant cells in body fluids using an automated hematology analyzer.

Abstract

The detection of malignant cells in body fluids (BF) with an automated hematology analyzer has been proposed as an alternative to morphological examination owing to its various advantages; however, its limitations have also been highlighted. In this study, we devised a practical algorithm to screen for malignant cells in BFs using an automated hematology analyzer. A total of 558 BF samples, including 232 cerebrospinal fluid (CSF) samples and 326 non-CSF samples, were consecutively collected. Thereafter, the results obtained using the BF mode of Sysmex XN-350 (Sysmex, Kobe, Japan) were compared with the cytological diagnosis. A cutoff was also established to screen for malignant cells using receiver operating characteristic (ROC) curve analysis based on the final clinical judgment. The automated hematology analyzer showed a moderate correlation or good agreement with the existing cytological diagnosis. Further, of the ROC curves for detecting malignant cells, the absolute value of highly fluorescent cells on BF (HF-BF) in total body fluids showed the highest area under the curve (0.85 [95% confidence interval 0.82-88], p<.0001, Youden index >7×106 /L, sensitivity 93%, and specificity 65%). An automated hematology analyzer could function as a complement to cytological examination. We propose a practical and comprehensive algorithm for cytological examination that requires low- and high-resolution microscopy based on the absolute value of HF-BF in BF samples suspected of malignancy. This algorithm can more usefully detect malignant cells while taking advantage of the automated analyzer and cytological examination.