ANALYTICAL AND CLINICAL EVALUATION OF SYSMEX UF1000I FOR AUTOMATED SCREENING OF CEREBROSPINAL FLUIDS

Background: We evaluated the performance of Sysmex UF1000i for cell counting and differential cell count, as well as for assessment of bacteria load in cerebrospinal fluid (CSF), as a potential approach for the rapid screening of meningitis or bacterial encephalitis. Methods: We analyzed 77 consecutive CSF samples, 34 of which (44%) displayed leukocyte count > 5 white blood cell (WBC)/mu L with optical microscopy. Results on the UF-1000i were compared with those obtained by microscopic analysis. Imprecision was evaluated by testing three CSF samples with leukocyte values between 3.5 and 28.8 WBC/mu L in 10 replicates. Carry-over was evaluated with the Broughton formula on three CSF pools with leukocyte counts between 93.5 and 132.5 WBC/mu L. Linearity was assessed according to CLSI document EP6-A. In the presence of bacteria, identification and antibiogram were performed with Vitex (Biomerieux), except for Neisserie meningitidis (ApiNH, Biomerieux). Sensitivity tests were performed with Vitex and disc diffusion. Results: Optimal correlation was found between UF-1000i and optical microscopy, displaying Pearson's correlation of 0.99 and mean bias of -3.5 WBC/mu L (95% Cl, from -7.0 to 0.0 WBC/mu L). Imprecision varied between 12 and 16%. Linearity was excellent, 4-278 WBC/mu L. Carry-over was negligible. ROC analysis yielded AUC of 0.99 for both WBC and bacterial counts. The agreement at threshold > 4 WBC/mu L was 0.91, with sensitivity and specificity of 1.00 and 0.84. At <= bacteria/mu L cut-off, accuracy was 0.98, sensitivity 1.00 and specificity 0.97. Conclusions: According to these results, CSF screening with UF-1000i seems a reliable approach in terms of instrument performance, turnaround time and overall laboratory efficiency.