RESUMO
In patients with hematologic malignancies, multiparameter flow cytometry (FCM) offers greater sensitivity than cytology in detecting malignant cells in the initial cerebrospinal fluid (CSF) specimen. However, the role of FCM in assessment of subsequent specimens is unclear. We developed an algorithm to reduce the number of low-yield FCM tests without significant impact on clinically meaningful results. Patients with hematologic malignancies were studied in a derivation cohort, and the following algorithm was developed: (1) cytology and FCM on all initial samples, (2) cytology on all subsequent samples, and (3) FCM on subsequent samples only if previous FCM was positive. A separate population served as the validation cohort. The derivation cohort included 197 patients representing 1157 cytology and 543 FCM samples. Common malignancies were B-Cell ALL (25.3%), diffuse large B cell lymphoma (29.4%), and Burkitt lymphoma (7.7%). In the derivation cohort, the algorithm yielded a sensitivity of 90.0% (95% CI, 81.2-95.6%) and a specificity of 100% (95% CI, 98.9-100.0%). The validation cohort included 132 patients with 563 cytology and 602 FCM samples. In the validation cohort, the testing algorithm yielded a sensitivity of 87.5% (95% CI, 75.9-94.8%) and a specificity of 100% (95% CI, 99.1-100.0%). Of the 15 samples that were missed by the algorithm, FCM findings did not impact patients' management because of known CNS disease (seven patients) or they were responding to treatment (eight patients). CSF testing in hematologic malignancies using the proposed algorithm presents an evidence-based approach to reduce the number of unnecessary FCM tests of CSF without compromising patient care.