HemoScreen hematology analyzer compared to Sysmex XN for complete blood count, white blood cell differential, and detection of leukocyte abnormalities

Abstract We compared a point‐of‐care HemoScreen hematology analyzer to an automated Sysmex XN analyzer for complete blood count (CBC) and white blood cell (WBC) differential, and evaluated its capacity to detect leukocyte abnormalities. A total of 100 K2‐EDTA whole blood samples, median age 56 years (2 months to 92 years), were compared. For CBC and WBC differential we compared 74 samples with no confirmed abnormal leukocytes. For 26 samples both analyzers gave flagging regarding leukocytes and the accuracy of the flagging was compared. Abnormal leukocytes were confirmed with manual microscopy (200 cells). HemoScreen CBC and WBC differential were highly comparable to Sysmex XN for most of the essential parameters (r = 0.909–0.975). More variation was seen for basophil and monocyte counts (r = 0.452 and 0.753, respectively). Sysmex XN gave more false WBC abnormal flagging (n = 15 altogether) compared to HemoScreen. In addition, Sysmex XN, as well as HemoScreen, gave false WBC flagging for eight samples confirmed normal. The samples verified by microscopy review to truly contain leukocyte abnormalities (n = 18) were flagged abnormal with both analyzers. The specificity for analyzer flagging was 72% and 88% for Sysmex XN and HemoScreen, respectively. HemoScreen hematology analyzer is essentially comparable to Sysmex XN for CBC and WBC differential analysis. Most importantly, HemoScreen detected all the samples confirmed to include abnormal leukocytes. HemoScreen was less prone for false WBC flagging compared to Sysmex XN, thereafter requiring less microscopy review. These abilities increase its utility in small health care units. Studies with a larger number of abnormal leukocyte samples are needed to confirm HemoScreen performance.


INTRODUCTION
An automated hematology analyzer should provide repeatable results for complete blood count (CBC). The measurement range should be wide enough to cover cytopenias and cytoses, and other clinically relevant hematological conditions. In addition, the analyzer's ability to provide reliable white blood cell (WBC) five-part-differential and to detect blood cell abnormalities is of high relevance. Samples with possible WBC abnormalities should be reviewed by manual microscopy to correctly identify patients with acute hematological diseases, such as acute leukemia. Other blood cell abnormalities have importance, for example, in diagnosis of the type of anemia.
Most hematology point-of-care (POC) devices provide CBC, or parts of CBC, such as hemoglobin concentration, and some give WBC five-part-differential [1][2][3]. However, their ability to detect blood cell abnormalities is often limited. HemoScreen POC device is a rather novel hematology analyzer, in which the enumeration and identification of blood cells is based on flow cytometry and digital imaging in a single plane using microfluidic viscoelastic focusing 4-6. The analytical technology combined to artificial intelligence (AI) and computational algorithms identify the cells based on their morphological properties, such as nuclear lobulation, cytoplasmic granulation and cell size. Cells not filling the built-in algorithms are classified abnormal and the analyzer will give alarms (flagging) concerning these samples [5,7] The technology based on imaging each cell passing the detector in a plane is shown to achieve high and stable focusing with less interference compared to traditional flow cytometry [5,6].
Studies on the performance of HemoScreen show that the analyzer is accurate and precise in analyzing red cell parameters compared to Sysmex XN [8]. With intensive care patients, HemoScreen gave good correlation for platelet, WBC and red blood cell (RBC) concentrations compared to Sysmex XN, although the bias was positive at higher platelet and WBC concentrations [9]. Another study with acute leukemia patients showed similar trend for platelets and WBC with a good correlation for complete blood count (CBC), absolute neutrophil count and hemoglobin compared to Sysmex XN [10]. None of these studies investigated the performance on full WBC differential or the analyzer's ability to detect WBC abnormal-  [11][12][13][14][15][16][17][18].
In this method comparison study, we present the performance of HemoScreen POC device compared to Sysmex XN analyzer. In addition to analytic performance with samples with normal CBC and WBC differential, we focused on the ability of HemoScreen to detect abnormal leukocytes compared to Sysmex XN interpretive program messages and manual microscopy.

MATERIALS AND METHODS
Within run and total method repeatability were assessed with two level

RESULTS
HemoScreen repeatability is presented in  (Figure 1 and Table 2).
For 26 samples both analyzers gave flagging regarding WBC or HemoScreen did not give any result for CBC or WBC differential.
These samples were excluded from automated method comparison but were used to evaluate the congruity of the two analyzers in detecting leukocyte abnormalities. Of these, 18 samples were verified to include abnormal leukocytes with manual microscopy (Table 3). In eight samples both analyzers gave flagging, but the microscopy review showed no abnormal cells and the Sysmex XN result was reported (Table 4).
Thereafter, specificity for analyzer flagging was 72% for Sysmex XN and 88% for HemoScreen in these study samples. Sensitivity for autovalidated samples could not be confirmed, as the samples with no flagging were not further reviewed according to routine autovalidation practices.

DISCUSSION
Other studies have described the HemoScreen analyzer to be userfriendly and suitable for health care units requiring fast results [6].
However, none of the studies compared analyzer's capacity to detect leukocyte abnormalities. This technical ability is of high clinical relevance in patient care and differential diagnosis, as, for example, CBC alone in acute leukemia differential diagnosis may be uninformative.
This study supports the previous findings of HemoScreen analyzer's good repeatability and analytical stability. Data showed slight difference for CBC parameters and WBC differential in samples without any abnormalities. Hemoglobin concentration and platelet count were slightly, but not statistically significantly lower with HemoScreen. For some parameters, such as MCV and HCT, the two analytical methods showed statistically significant difference. In spite of this, these differences in individual samples were minor and the statistical deviation can be considered acceptable in a clinical setting. In general, all parameters were in good agreement to Sysmex. However, there was one sample with deviating CBC results ( Figure 1). This sample was from a child 3 months of age, with no apparent clots and WBC 3.1 × 10 9 /L, RBC 5.4 × 10 12 /L, HGB 145 g/L, HCT 46%, and PLT 227 × 10 9 /L with Sysmex   Table 3) Hemo-Screen was not able to perform WBC differential, probably due to high WBC count (139 × 10 9 /L) exceeding linearity range (80 × 10 9 /L). For another sample with 4% blast cells (Sample 2, Table 3