Use of Sysmex XN‐10 red blood cell parameters for screening of hereditary red blood cell diseases and iron deficiency anaemia

Abstract Introduction In daily practice in haematology laboratories, red blood cell (RBC) abnormalities are frequent and their management is a real challenge. The aim of this study is to establish a “decision tree” using RBC and reticulocyte parameters from the SYSMEX XN‐10 analyser to distinguish between patients with a hereditary RBC disease from iron deficiency anaemia and other patients. Methods We analysed results of complete RBC counts in a cohort composed of 8217 adults divided into 5 different groups: iron deficiency anaemia (n = 120), heterozygous haemoglobinopathy (n = 92), sickle cell disease syndrome (n = 56), hereditary spherocytosis (n = 18) and other patients (n = 7931). A Classification And Regression Tree (CART) analysis was used to obtain a two‐step decision tree in order to predict these previous groups. Results Five parameters and the calculated RBC score were selected by the CART method: mean corpuscular haemoglobin concentration, percentage of microcytes, distribution width of the RBC histogram, percentage of nucleated red blood cells, immature reticulocytes fraction and finally RBC Score. When applying the tree and recommended flowchart, 158/166 of the RBC hereditary disease patients and 114/120 iron deficiency anaemia patients are detected. Overall, the correct classification rate reached 99.4%. Sensitivity and specificity for RBC disease detection were 95.2% and 99.9%, respectively. These results were confirmed in an independent validation cohort. Conclusion Based on the XN‐10 RBC and reticulocyte parameters, we propose a two‐step decision tree delivering a good prediction and classification of hereditary RBC diseases. These results can be used to optimize additional reticulocyte analysis and microscopy review.


| INTRODUC TI ON
According to the World Health Organization (WHO), anaemia is a global public health problem affecting 1.6 billion people around the world, which corresponds to approximately 25% of the world population. 1 Different aetiologies lead to anaemia or red blood cell (RBC) abnormalities. Iron deficiency anaemia (IDA) is considered to be the main cause of anaemia. Indeed, the proportion of anaemia associated with iron deficiency was 25% for preschool children and 37% for nonpregnant women of reproductive age. 2 Besides iron deficiency, hereditary red blood cell diseases including haemoglobinopathy, red cell membrane pathology and enzymatic deficiency are also the common origin of RBC parameter disorders. The current estimation of individuals carrying a significant variant of haemoglobin is up to 323 million (5.2%) in the world. 3 The estimated number of births with a serious inherited haemoglobin disorder is still up to 330 000 per year (83% sickle cell disease, 17% thalassaemias) and contributes to the equivalent of 3.4% of mortality in children aged under 5 years worldwide or 6.4% in Africa.
In daily laboratory practice, the use of automated haematology analysers brings the reliability of the results but no analyser can determine properly RBC morphological abnormalities and cytomorphological examination of the blood smear is necessary to interpret automated results for unknown patients. The decision of blood smear examination is triggered by one or more alarms generated by the analyser or quantitative and/or qualitative criteria decided by the laboratory performing the analysis. These criteria are not always relevant hence the interest to associate several RBC and reticulocyte parameters in order to improve the specificity and sensitivity of microscopy examination of a blood smear.
Different studies have been already described for discriminating between iron deficiency anaemia and thalassaemia trait, but none of them take into account more clinical situations including both constitutional and acquired anaemia. The aim of this study was to establish a decision tree using RBC and reticulocyte parameters from the SYSMEX XN-10 to distinguish between patients with hereditary RBC disease from iron deficiency anaemia and other patients.
Another goal was to focus on the patients flagged for RBC disease in order to specify RBC pathologies.

| Materials and Methods
The study was approved by the local Ethics Committee of Assistance Publique des Hôpitaux de Marseille (Marseille, France).

Blood samples were collected in EDTA K3 Becton Dickinson
Vacutainer TM tubes (Franklin Lakes, NJ, USA) and analysed within 6 hours after collection.
Complete blood counts were performed using Sysmex XN-10 (Sysmex Corporation TM , Kobe, Japan) analysers equipped with a reticulocyte analysis module.
The RBC count was measured using the impedance variation method after hydrodynamic focusing. The haematocrit (HCT) was  Consequently, all these mentioned tests (Ferritin, EMA test..) were not performed for all patients but only according to the call points and the context. If data were not available or if patient were receiving therapy affecting erythropoiesis (Iron therapy, blood transfusion, EPO), they have been excluded from the cohort.

| Validation cohort
The decision tree was subsequently tested in an independent validation cohort of 14 705 patients (>15 years old) from private Laboratory Ketterthill (Esch SUR/Azeltte, Luxembourg) with a recruitment exclusively issued from general practician collected over a two-month period between 1st of July and 3rd of September 2019.
Patients were divided into the same five groups, and diagnosis was obtained via similar reference methods. The research for validation cohort was exempted of IRB approval.

| Statistical Evaluation
RBC parameters were analysed as quantitative variables and summarized as median and range.
The association between RBC and reticulocyte parameters and groups was performed using the Kruskal-Wallis test with Dunn's multiple comparison test. In the group called "OTHERS", 4008 patients were fitting the normal reference values, 5 while 3923 presented with at least one abnormality from the panel of RBC parameters (Table S1). All patients were older than 15 years, and the sex ratio was between 0.60 and 1.04 depending on the groups.

| Red blood cell parameters
Results of the complete RBC count were analysed in this cohort and summarized in Table 1

| Reticulocyte parameters
In this second step, RET analysis was added for patients who were

| CART tree RET
Two parameters were selected for the second decision tree: RBC score and IRF%. RBC score and its cut-off value (0.15) are part of the CBC-O application and cannot be changed. The second parameter was selected by the CART method. All 138 patients were distributed into 3 leaves (Figure 2). In leaf 7, 48 patients belonged to the SCD group. One HGB HTZ and one HS with an extremely high reticulocyte count (783 G/L) were present. One patient from "OTHERS" was as well found. This patient was suffering from an autoimmune haemolytic anaemia.
Leaf 8 was mainly composed of HS patients (11/14). Additionally, one SCD and two "OTHERS" were present without any special observation.
Among 73 patients classified in this leaf 9, 71 were confirmed as "OTHERS." One SCD and one HS were also found. Both of them presented with a normal reticulocyte count and absence of FRC%, leading to a normal RBC score.
Overall, two leaves predicted RBC disease, SCD in leaf 7 and HS in leaf 8. Leaf 9 removed RBC disease false positives from the CART RBC.
In conclusion (

| Validation cohort
The

| D ISCUSS I ON
The aim of our study was to establish a decision tree using RBC and   Figure 3).
Of course, some limitations of our study can be underlined.
Our algorithm is only applicable to patients whose blood count has been performed on the Sysmex XN-10 series because some specific parameters are included in this tree therefore not transposable to another type of automaton. It should also be noted that this study is designed to isolate RBC disease versus IDA when present rather than giving a focus on discrimination between heterozygous haemoglobinopathy and IDA in the limited context of microcytic anaemia.
As such, this tree is not supposed to detect all patients presenting with iron deficiency.
Overall, these two-step decision tree allow us to reach a very good classification rate (99,4%) and these results were also confirmed by an external cohort. Our study takes into account many clinical situations and represents a daily university laboratory routine practice. This algorithm is able to well differentiate several causes of anaemia, both acquired and constitutional, and not only IDA or thalassaemia. Only few "other" patients or iron deficiency anaemia interfere with the RBC disease detection. Up until now, few investigators have also included subjects with other types of haemoglobinopathy, such as HbE 11 and HbS, both sickle cell thalassaemia and sickle cell diseases 12 or inflammatory anaemia.