The Sysmex XN‐L (XN‐350) hematology analyzer offers a compact solution for laboratories in niche diagnostics

Abstract Introduction In 2015, Sysmex launched a new series of hematology analyzers (XN‐L Series) designed to fulfill the needs of niche laboratories in areas such as pediatrics, dialysis, neurology, and oncology while providing a compact solution. In this study, we evaluate the whole blood and body fluid modes of one of these analyzers, the XN‐350. Methods A total of 300 residual EDTA samples were measured on the XN‐350 in whole blood mode and the XN‐1000 to evaluate method comparison, flagging sensitivity, repeatability, reproducibility, linearity, carryover, and stability. In addition, 191 samples were obtained and processed in body fluid mode which included, cerebrospinal fluid (CSF), continuous ambulatory peritoneal dialysis (CAPD), ascites, synovial, and pleural fluid to perform method comparison, repeatability, reproducibility, linearity, limit of quantitation, and carryover studies. Results Strong agreement was shown between the XN‐350 and XN‐1000 for both whole blood and body fluid modes in results and flagging. Linearity results in both modes on the XN‐350 showed a high R 2 value (>.99). For WBC, RBC, HGB, and PLT, the carryover results were well within the predetermined criteria of ≤0.5% for whole blood and ≤0.3% for CSF. Repeatability and reproducibility were acceptable for both modes, and there were no significant deviations present in stability for whole blood. In addition, there was high agreement in all body fluid types evaluated. Conclusion The performance of the XN‐350 is comparable to the XN‐1000 in both whole blood and body fluid modes, making it a reliable alternative to larger analyzers for smaller, niche laboratories.

in open mode only.
To date the XN-L series was evaluated in several studies, however, no combined studies can be found since only whole blood mode or body fluid mode was evaluated. A good correlation was generally observed for all parameters, except for basophils. Tailor et al. focused especially on the different platelet counting methods (impedance and optical) of the XN-550 and observed reliable results especially in the low counts (<40 × 10 9 /L) when triggers for preventive platelet transfusions are established. 11 The whole blood module of the XN-350 was evaluated for determination of routine hematology parameters in hematopoietic progenitor cell apheresis products. This study gave reliable results, but the RBC counts were overestimated, possibly due to interference of WBC in the impedance counting of RBCs. 12 The XN-350 body fluid mode has been evaluated in two studies. The first study established a cut-off for the detection of high fluorescence cells in pleural and ascites fluids as an indicator for malignancy 13 and the second, more recent study evaluated the analytical performance of the body fluid mode with two other systems, the UniCel DxH800 and the UF-5000. The  showed the most comparable results to those of manual differential counting. 14 However, there has yet to be a comprehensive study of the Sysmex XN-350 hematology analyzer and its performance for both whole blood and body fluid analysis combined. In this study, we evaluated the XN-L (XN-350) by comparing all common whole blood and body fluid parameters between the XN-350 and the XN-1000. Furthermore, for both whole blood and body fluid analysis we performed an analytical evaluation for repeatability, reproducibility, linearity, carryover, lower limit of detection as well as flagging performances and stability for whole blood analysis.

| Sample collection
Only residual samples were included in this study as approved by

| Method comparison
Method comparison studies were performed to assess the performance of the XN-L series (XN-350) analyzer compared with the XN-1000 analyzer for whole blood and body fluids. For whole blood, a total of 300 whole blood samples were included. All samples were run within two hours after venipuncture and within two hours of both runs on each analyzer. Samples covered clinical decision levels and the full reportable measuring ranges of the XN-Series analyzers including immature granulocytes, abnormal lymphocytes, atypical lymphocytes, and blasts. Samples were processed on the XN-350 in the whole blood mode in the CBC + DIFF + RET channel profile and measured on the XN-1000 in the full channel profile (CBC + DIFF + RET + PLT-F + WPC). All samples were run in duplicate and the comparisons were performed with the average results. The following parameters were measured on both sys-

| Flagging performance
The overall flagging performance of XN-350 vs XN-1000 was evaluated for immature granulocytes (IGs), blasts, abnormal lymphocytes, atypical lymphocytes, and left shift on the same 300 samples used for method comparison. Smears were made on each sample and a manual differential was performed on the CellaVision DM-96 digital cell morphology system (CellaVision AB). Following the DM-96 results, smears were then reviewed by trained medical technologists for the final determination of all cell types including abnormal cells present such as bands, blasts, atypical lymphocytes, and immature granulocytes. Various medical technologists completed a proficiency examination to minimize variability in smear review results. An abnormal manual differential was defined according to the criteria for action following automated CBC and differential WBC differential analysis as suggested by the international consensus group for hematology of the International Society for Laboratory Hematology (ISLH). 15 In short: immature granulocytes ≥1%, blasts ≥1% and/or abnormal lymphocytes ≥5% (which included plasma cells), atypical Lymphocytes ≥5% and left shift was band ≥5% and/or IG ≥1%. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and efficiency were calculated for each individual flag that relates to the morphology or presence of abnormal cells.

| Repeatability and reproducibility
Repeatability studies were performed using residual EDTA whole blood samples covering clinical decision levels and the upper and lower limit of the analytical measuring range. Twenty replicates of each sample were tested in the whole blood manual mode on the XN-350. The mean, standard deviation (SD), and coefficient of variation (CV) were calculated for each sample. After processing the samples in whole blood mode, the same samples were diluted with Cellpack DCL (1:7) and measured in the pre-diluted mode.
To determine repeatability of body fluids, one sample of each type of body fluid was processed ten times consecutively to determine if the variability in results exceeded the acceptance criteria of being within 20% for any particular body fluid type specifically. Each sample was divided into two tubes and processed five times each for a total of ten runs. In addition, three different concentrations of CSF were included with a final concentration of WBC ≦4/μL (CSF4), 10/ μL (CSF10) and 50/μL (CSF50).
The reproducibility was performed by using XN CHECK levels 1, 2, and 3 for the whole blood mode and XN CHECK BF levels 1 and 2 for the body fluid mode. All levels were processed twice a day in triplicate across a five-day period in order to provide additional reproducibility data accounting for time and day variability.
Studies were performed in accordance with the recommendations in CLSI_EP05-A3.

| Linearity
Seven serial dilutions of known, high concentration of whole blood samples in EDTA were prepared with the lowest dilution reaching In addition, linearity for WBC and RBC in body fluid mode was determined by selecting a CSF sample with a high concentration of cells to be serially diluted in PBS buffer. Samples with approximately 500 WBC/μL and >1000 RBC/μL were diluted 1:10, 1:20, 1:50, and 1:200 with PBS and each of these dilutions was measured 5 times consecutively on the XN-350.

| Carryover
Carryover was evaluated by measuring whole blood samples with high target values (HTV) for WBC, RBC, HGB, and PLT counts three consecutive times (H1, H2, and H3) followed immediately by testing samples with low target values (LTV) around clinical decision levels consecutively, three times (B1, B2, and B3). Carryover effect was calculated for each parameter using the Broughton method, [(B1-B3)/(H3-B3)] × 100%. In addition, carryover was assessed for body fluids on CSF samples containing high and low WBC and RBC counts. The Sysmex standard is a carry-over ratio coefficient <0.3% or a maximum difference of 1 cell/μL between B1 and B3. The carryover study design was performed in accordance with CLSI_H26-A2.

| Stability
Stability of whole blood was determined using residual samples from five normal individuals and five patients with abnormalities. The 5 abnormal blood samples contained one of the following parameters with a value well above or below the reference value, RBC, WBC, HGB, PLT, HCT, and RDW-CV. The other 5 normal blood samples had either normal values within the reference intervals or values that fell slightly below or above the reference value, but were not regarded as abnormal. All samples were run on the XN-350 (time point 0 hour) and subsequently aliquoted in two sets of six aliquots.
One set was stored at room temperature (RT) and the second at 4°C.
It was determined that 4°C would be used for stability based on ICSH guidelines. 16

| Limit of quantitation
Due to the potentially low WBC and RBC values in body fluid samples, a limit of quantitation needed to be determined on the XN-350. Six CSF samples were selected with very low concentrations (1 WBC/μL and 10 RBC/μL), and samples were diluted to achieve such results if they were not found naturally. Each sample was processed 5 times in order to obtain 30 analytical results in total. The limit of quantitation (LoQ) was determined to be at the concentration of the samples at which the coefficient of variation (CV) of <20% is achieved. The study was performed in accordance with CLSI EP17-A2, and the CV was calculated for each sample separately.

| Statistical analysis
For all statistical analysis, the absolute cell counts were used for all measurements. Data analysis was performed by using In addition, the coefficient of determination (R 2 ) was calculated for linearity to determine variation across the measured range.
Statistical significance was based on the 95% confidence intervals (CI). A significant proportional or constant bias was noted when the 95% CI of the slope did not encompass 1, and the 95% CI of the mean difference, limits of agreement (LoA) or intercept did not encompass 0, respectively. Spearman's correlation coefficient (r s value) was used in the method comparison analysis. Acceptance criteria are according to the manufacturer's criteria in the Sysmex Instructions for Use.

| Method comparison
For the evaluation of whole blood, a total number of 300 residual EDTA whole blood samples were analyzed on both the XN-350 and the XN-1000. Table 1 shows the correlation and the estimated bias for the 22 parameters measured and Figure 1 shows the Passing-Bablok plots for all directly measured parameters. The plots for the remaining calculated are in Figure S1.  Table 2 and Passing-Bablok plots for WBC and RBC are shown in Figure S2.

| Linearity results
For both whole blood and body fluid analysis, the XN-350 demonstrated to be linear from lower limit to upper limit and remained within the allowable maximum % diff for each interval.  showed that the carryover is less than 0.3% and is therefore negligible for WBC in CSF.

| Stability results
Seven parameters were measured in aliquots of whole blood, stored at 4°C and RT, of the same sample at the initial time point (0 hour) and time point 4, 8, 12, 24, 48, and 72 hours. Table 5

| Limit of quantitation results
As can be seen in Table S8A,B, all variation coefficients fall within the 20% limit and were ≤5.0 WBC/µL for body fluids. By using BF XN

| D ISCUSS I ON
The XN-350 is the smallest of the XN-L series and its compact presentation makes it an essential tool that fits the needs of (satellite) laboratories in specialized outpatient centers, for example, pediat-  In the case of repeatability and reproducibility, the measurements were carried out in the pre-diluted whole blood, and body fluid modes. For whole blood in the repeatability test, the measurements had a slightly smaller standard deviation and variance coefficient compared with the measurements in pre-diluted mode.
Although this difference is not very large, it may have to do with manual error from pipetting the dilutions. In the pre-diluted mode, the blood samples are naturally diluted 1:7 with DCL Cellpack. In addition, the XN-350 did indeed give a repeatability problem in whole blood mode with the basophils and the eosinophils due to the low cell numbers, which was expected. Variability was also seen in some body fluid types, especially #PMN, which was also due to the high were acceptable. A limitation of this study was the manual manipulation of the RBC and HGB for whole blood. Ideally, these samples would have occurred naturally, but the results were acceptable. In addition, the LoQ of the XN-350 in terms of CSF is comparable to that of the XN-1000 at ≤5.0 WBC/µL and for RBC at 2.0 × 10 3 /µL, which is better than most other body fluid analyzers on the market. 6 For most of the parameters, the stability remains 8h after blood sampling, which is the preferred maximum time elapsed after venipuncture before routine processing. All evaluated parameters, except MCV, are stable 12 hours after sampling when stored at room temperature. It is known that MCV changes more rapidly than other parameters over time, 22 but based on the study results MCV should be measured within 4h, which is acceptable especially for smaller laboratory settings as there is less time between venipuncture and processing. This emphasizes the importance of refrigerating samples within 4 hours of venipuncture for most accurate results, especially when further RBC parameter tests may be needed.
The aim of this study was to evaluate the performance of the XN-350 analyzer against XN-1000 and evaluate the interchangeability of these analyzers in terms of quality of the results as it meets the needs of niche laboratories. The compactness makes it suitable for remote, specialty laboratories in areas such as pediatrics, neurology, and oncology to eliminate the need for additional send outs and waiting time. The accurate leukocyte count, particularly neutrophils, makes it a reliable tool for chemotherapy monitoring in oncology clinics. Not only are the whole blood features and measurements comparable to that of the much larger, XN-1000 analyzer, it also has the ability to process body fluids, which is essential for use in dialysis clinics and neurology clinics in an accurate and convenient way. In conclusion, this study shows good performance results for the XN-350 analyzer and should be adopted as reliable and comparable to the XN-1000 analyzer.

ACK N OWLED G EM ENTS
We acknowledge the efforts of both Dr. Anne Ammerdorfer and Bea van den Berg for their technical assistance throughout the study.
There is no financial or material support to be mentioned.

CO N FLI C T O F I NTE R E S T
The authors have no competing interests.