Diagnostic performance of reticulocyte hemoglobin equivalent in assessing the iron status

Abstract Background Measurement of reticulocyte hemoglobin equivalent (RET‐He) is rapid, convenient, and cost‐effective. Yet, researches on its performance in diagnosing iron deficiency with concurrent inflammation are limited. Hence, this study investigated RET‐He value in various states, including inflammation, and evaluated its diagnostic performance in iron status assessment. Methods Retrospectively, 953 clinical data and laboratory results—complete blood count, reticulocyte count, RET‐He, and serum ferritin—were reviewed. Patients on iron therapy were excluded. Iron status was defined by serum ferritin as the reference method. RET‐He among populations was investigated. Its diagnostic performance and optimal cutoff were determined by ROC analysis. Results Three population groups were classified: healthy control, iron deficiency anemia (IDA), and non‐ID anemia. Significantly, RET‐He value in IDA was lower than that of healthy control, anemia of inflammation, and chronic kidney disease (P < .0001). Low RET‐He was also observed in IDA with concomitant inflammation despite normal‐to‐high serum ferritin levels. No significant difference was observed between RET‐He values in pure IDA and thalassemia (P = .57). ROC curve analysis revealed AUC of 0.876 (P < .0001) at cutoff 30 pg, by which IDA was discriminated with 74.2% sensitivity and 97.4% specificity. Applying cutoff ≤30 pg, IDA can be diagnosed with 96% sensitivity, 97.4% specificity, 80% PPV, and 99.6% NPV. Hence, RET‐He >30 pg signifies a non‐IDA state. Conclusion In addition to convenience and cost‐effectiveness, RET‐He cutoff >30 pg can be potentially used to exclude IDA due to its excellent diagnostic sensitivity and specificity.


| INTRODUC TI ON
Iron deficiency (ID) is well recognized as the most common nutritional deficiency 1,2 disorder in the world 3 and the principal cause of global anemia, 4 leading to adverse sequelae, namely, growth retardation, 5,6 neurocognitive deficit, 1 impaired immune system, 3 and increased risk of prematurity and maternal mortality. 7 Early detection is, thus, crucial for proper and timely management to prevent such consequences.
Presently, there is still no single best test for diagnosing this condition.
Bone marrow iron staining, the gold standard method for diagnosing iron deficiency, is invasive [8][9][10][11] and subjective. 9,10 Conventional iron biomarkers are also affected by acute-phase responses, 8,12 and influenced by diurnal variation and dietary intake. 12,13 Due to these diagnostic difficulties, an alternative marker is sought.
Reticulocyte hemoglobin equivalent (RET-He) is one of potential markers.
Its measurement is not only rapid but also convenient and cost-effective.
Introduced in 2005, studies evaluating RET-He for its performance in diagnosing iron deficiency and application in various clinical settings are numerous. [13][14][15][16][17][18][19] However, investigations on RET-He in the diagnosis of ID with coexistence of inflammatory conditions are limited. 4,20,21 This study, thus, aimed to investigate RET-He value in various conditions, including inflammation-related states, and evaluate its diagnostic performance in assessing the iron status. A diagnostic algorithm was, then, proposed.

| Study design
This was a retrospective analytical study, performed at Ramathibodi Hospital, Bangkok. An ethical approval was obtained from the Ethics Committee for Human Research of Ramathibodi Hospital, Mahidol University (MURA 2017/509). Nine hundred and fifty-three clinical data-collected from April 2017 to May 2018-were reviewed in conjunction with laboratory results, that is, complete blood count (CBC), reticulocyte count, RET-He, and serum ferritin. Patients on iron therapy (n = 15) were excluded.

| Definition of iron status and anemia
Iron status was defined by serum ferritin as the reference method.
Low serum ferritin (<33.71 pmol/L or 15 ng/mL according to the World Health Organization [WHO]; 1 ng/mL = 2.247 pmol/L) signified ID in both women and men. 22 Anemia was defined, following the WHO criteria, by hemoglobin (Hb) level <120 g/L in women (W) or <130 g/L in men (M). Iron deficiency anemia (IDA) was ID-associated anemia. IDA with concomitant inflammation (IDA-inf) and anemia of inflammation (AI) were similar in terms of anemia with normal or high serum ferritin level (≥33.71 pmol/L), and a history of associated inflammatory conditions, namely, chronic infections or non-infectious inflammation, and malignant diseases. Differences between the two were the characteristic of anemia: microcytic hypochromic in IDAinf versus normocytic normochromic anemia in AI. Thalassemia was defined by Hb typing and/or DNA analysis results, demonstrating alpha and/or beta thalassemia.

| Study populations
By the above criteria, three population groups were classified, for instance, healthy control (n = 155), IDA (n = 253) including pure IDA

| Laboratory methods
CBC and reticulocyte analysis were performed on Sysmex XN-10 hematology analyzer (Sysmex Corporation). Serum ferritin was quantified by using ARCHITECT ferritin assay, a chemiluminescent microparticle immunoassay (Abbott Laboratories). Determination of reticulocyte hemoglobin content is conducted in the reticulocyte channel, composing of two processes. Firstly, nucleic acid is stained 23 with polymethine dye, 19,24 which is specific for RNA/DNA. 24 Then, measurement of reticulocyte cellular hemoglobin was done based on fluorescent flow cytometry technique. 19,23,24 The 2D-scattergram depicts the measurement of mature erythrocytes and reticulocytes. On the y-axis, the forward light scatter demonstrates an individual cell size, plotted against fluorescence intensity-a measure of RNA content-on the x-axis. The average value of forward-scattered light intensity of reticulocytes is expressed as a log transformation of Ret-Y. 23 The results are presented as picograms (pg) of Hb per reticulocyte. 8 This obtained reticulocyte hemoglobin content is termed RET-He parameter.

| Precision studies
Precision analysis for hematology analyzer was carried out, following the International Council for Standardization in Haematology (ICSH) guidelines, 25 prior to sample testing. Repeatability (within-run precision) test was performed by a single run of 20 measurements 26 on each of the three levels (low, normal, and high) of quality control materials (XN CHECK). Between-run precision test was carried out by running a single measurement on the above control materials daily over a period of 20 days. 25,26

| Statistical analysis
Data analysis was carried out using MedCalc statistical software version 18 (MedCalc Software). Data normality was assessed by Kolmogorov-Smirnov test. Non-parametric data were described as median and interquartile range (IQR). Comparison of data between two groups employed Mann-Whitney U test and Kruskal-Wallis test for multiple groups. P value <.05 indicated statistical significance. ROC analysis was performed to evaluate the diagnostic performance of RET-He and determine its optimal cutoff value in distinguishing IDA.

| Precision studies
Precision analysis for RET-He revealed a low imprecision for the hematology analyzer employed in this study. A standard deviation of <0.5 pg with %CV of <2% in every level of the tested control materials was observed, falling within 5% limit specified by the manufacturer. Table 1 illustrates the demographic data and laboratory parameters of the study populations. Overall, most of the healthy control and patients in every group were women. Evidently, the prevalence of IDA was much higher in women than in men, ranging from two folds (80 women vs 40 men) in IDA-inf to four folds (108 women vs 25 men) in pure IDA. Also, IDA was observed most in women of

| RET-He values among the population groups
Comparison analysis revealed that RET-He in healthy control dif-

| Optimal RET-He cutoff for IDA detection
By ROC analysis, the optimal cutoff for RET-He in IDA detection was generated from the best combination of sensitivity and specificity. ROC curve revealed area under the curve of 0.876 (95% CI 0.854-0.897; P < .001) at cutoff ≤30 pg (Figure 3), at which IDA was distinguished with 74.2% sensitivity and 97.4% specificity.

| D ISCUSS I ON
Reticulocyte hemoglobin content has been reported as a marker that provides a snapshot of iron availability for erythropoiesis 13,23,27,28 in the bone marrow, 13,27 an indicator of iron therapy response, [28][29][30] and an early marker of iron-deficient erythropoiesis. 20,29,31 It enables an early detection 23  Clinical data and laboratory results are integrated for the analysis. Apparently, the finding that IDA possesses low RET-He is consistent with previous studies, supporting its clinical usefulness reported earlier. In this research, at RET-He cutoff >30 pg, IDA can be excellently ruled out. In contrast, at RET-He ≤30 pg, the finding is slightly complicated. Because, in the real world, IDA can occur not only singly but also commonly with several non-ID conditions, RET-He value ≤30 pg cannot evidently distinguish IDA from these situations. As such, a diagnostic algorithm is proposed.

RET-He is incorporated as the initial test in the algorithm because
of not only its outstanding diagnostic performance but also several analytical advantages. That is, its determination is fully automated 14 with only 2-step process and can be performed on the same specimen as for CBC. 23 This renders simplicity and convenience, 17 rapidity 14,17 with measurement in <2 min, 14 and cost-effectiveness. 23 The result is, then, readily reported as a part of CBC. Aside from up-to-3-days stability, RET-He measurement demonstrates low total, analytical, and biological variations (median 2.1%, 1.6%, and 1.7%, respectively). 37 The precision studies performed in this research also show good reproducibility and low imprecision in RET-He determination.
Consistent with the proposed diagnostic algorithm (Figure 4 Despite the mentioned potentials, some disadvantages of RET-He need to be stated. Initially, although RET-He is hypothetically able to indicate an early response to iron therapy according to previous studies, the endpoint of treatment monitoring or follow-up is the restoration of iron store, in which ferritin evaluation is indispensable.
Next, though this parameter can be employed as a marker for predicting iron depletion in otherwise healthy population, confirmation essentially requires determination of serum ferritin. What is more, in countries where thalassemia is prevalent like Thailand, this parameter may not be suitable for assessing the iron status in the presence of alpha and beta thalassemia. 27,40 This is supported by a previous study, summarizing that sensitivity and specificity of reticulocyte hemoglobin content cutoff were insufficient to distinguish thalassemia or hemoglobinopathy carriers from IDA.y. 41 As such, further studies are necessary.
To our knowledge, this is the first study that collects such a large sample size for the investigation. This is also one of a few researches that inflammation is considered for RET-He assessment. Despite such strengths, limitations exist. Firstly, this study is conducted on Sysmex XN-series hematology analyzer, which might not be available in many institutions, particularly those providing primary health care. This is probably the reason why the use of this parameter is still of limit despite numerous studies over years. Next, clinical diagnoses of the patient are extremely diverse, and more than one diagnoses are concurrently found in each patient, making the analysis difficult.
Lastly, the current collection of data limits the utilization of RET-He as a tool for iron therapy monitoring due to its retrospective nature.
Yet, potential still exists for it to be employed as a follow-up parameter. Further study is warranted to investigate and validate its feasible utility in this clinical setting.

| CON CLUS ION
In addition to its rapid, convenient, and cost-effective measurement, RET-He >30 pg is a potential marker in ruling out IDA with an excellent diagnostic sensitivity and specificity. However, when IDA co-exists with other non-ID conditions, which is common in actual clinical practice, evaluation of serum ferritin remains necessary for making the diagnosis.