Abstract
More aggressive therapies used for treatment of oncohematological malignancies or control of immune responses are resulting in an increased frequency of platelet counts below the 50 x 109/L limit. The recommended reference method for platelet counts was tedious and showed low reproducibility until now. In the last 2 years, flow cytometry based techniques combined with specific monoclonal antibodies (MoAbs) have been accepted as reference method. We have evaluated the accuracy for low platelet counts of several hematologic analyzers currently used in our laboratories.
The new reference method approved by ISLH, ICSH y NCCLS is based on double labeling of platelets using MoAbs directed to CD41 and CD61 followed by flow cytometry analysis. Absolute platelet counts are calculated using a ratio with red blod cell (RBC) counts provided by the hematological analyzers. In our studies, 50 blood samples with platelet counts ranging from 1.5 to 39.4 x109/L were processed in duplicate through 1 Advia 2120 (Bayer Diagnostics), 2 Advia 120 (Bayer Diagnostics) and 2 Pentra 120 DX (Horiba-ABX Diagnostics). Advia analyzers use laser-based technology while Pentra analyzers use impedance one for cell counting. All samples were also processed through the reference flow cytometric method, being platelets identified by their double labeling for CD41 and CD61. The minimal number of platelets acquired in the platelet region was established at 1000. Absolute platelet counts were calculated using RBC counts provided by the respective analyzers. All blood samples were processed within 6 hours from phlebotomy. Statistical methods applied included: coefficient of variation (%CV), coefficient of correlation (r ), linear regression, Passing-Bablock (P-B) regression and Bland-Altman test. Precision of each analyzer was obtained by processing in 10 times different blood samples with counts from 4 to 39 x 109/L. Global results were evaluated, though special attention was paid to subgroups of results below or above 20 x 109/L.
Correlation between reference values and counts provided by the Advia 2120 was 0.945 with a linear regression of 0.987x+2.9. P-B correlation was good and the average difference was 2.7 x 109/L. In the subgroup of samples with counts below 20 x 109/L correlation was 0.874 with 1.00x+2.7. P-B was correct and the average difference was 2.8 x 109/L. Results with Advia 120 were always similar to those calculated with the Advia 2120, though the average difference was slightly lower with a value of 1.7 x 109/L. Precision (CV) was 16% for platelet count levels at 4 x 109/L, 12% for those at 13 x 109/L and 4% for those at 39 x 109/L. Correlation with Pentra 120 Dx was 0.937 with a linear regression of 0.894x+2.7, the P-B was acceptable with an average difference of 1.2 x 109/L. Correlation index was 0.824 with a linear regression of 0.88x+2.8 for platelet counts below 20 x 109/L, average difference was of 1.4 x 109/L and a correct P-B. Precision (CV) ranged from 26% at 4 x 109/L and 8% at 20 x 109/L platelet counts.
Our data demonstrate that hematological analyzers evaluated provided very reliable results at low platelet counts. Advia and Pentra analyzers investigated tend to over calculate the number of platelets (2.5 and 1.4 x109/L respectively). Correlation scattering was slightly superior with the Pentra analyzer. Overall reproducibility for low platelet counts was excellent for both laser and impedance technologies tested.
Author notes
Corresponding author
