Abstract

For many years, application of RBC indices has been recommended for discriminating between subjects with iron deficiency from those with thalassemia. However, application of the algorithms resulted in only 30% to 40% of subjects being appropriately classified. The aim of the study was to establish the efficacy of algorithms for anemia screening including new hematologic parameters such as percentage of hypochromic and microcytic RBCs and hemoglobin content of reticulocytes. Subjects with iron deficiency anemia (IDA) (n = 142) and subjects with β-thalassemia (n = 34) were enrolled in a European multicenter study. Apparently healthy subjects were used as a reference group (n = 309). Hemocytometric investigations were performed on a Sysmex XE5000 hematology analyzer. The algorithms for IDA discrimination yielded results for area under the curve, sensitivity, specificity, and positive and negative predictive values of 0.88, 79%, 97%, 74%, and 98%, respectively. The algorithms for β-thalassemia discrimination revealed similar results (0.86, 74%, 98%, 75%, and 99%, respectively). We conclude that the advanced algorithms, derived from extended RBC parameters provided by the Sysmex XE5000 analyzer, are useful as laboratory anemia screening devices.

The application of RBC indices has been recommended for discriminating between subjects with iron deficiency and subjects with thalassemia.1–4 However, application of the England and Fraser formula (mean corpuscular volume [MCV] – 5 × hemoglobin [Hb] – RBC) and Mentzer formula (MCV/RBC) resulted in only 30% to 40% of subjects being appropriately classified.

Additional application of protoporphyrin content in RBCs was recommended for classifying microcytic RBC disorders.4–11 Multivariant discriminant analysis of algorithms including MCV, mean corpuscular hemoglobin (MCH), RBC, and red cell distribution width (RDW) was useful for the differential diagnosis of α- or β-thalassemia and iron-deficiency, but resulted in an inconclusive diagnosis in several cases.4,12

Measurement of the percentage of hypochromic and microcytic RBCs (%HypoHe and %MicroR) has demonstrated usefulness for detecting small changes in the amount of RBCs with inadequate hemoglobinization.13–15

Iron deficient erythropoiesis was characterized by the production of RBCs with decreased hemoglobin content resulting in an increased result for %HypoHe.16,17 As severity of anemia progresses, results for %MicroR will increase. However, subjects with β-thalassemia showed erythrocytosis and a high score for microcytosis. RBCs in case of subjects with β-thalassemia have a decreased volume because of impaired hemoglobin synthesis.

We decided to study the efficacy of new hematologic parameters for RBCs, like %HypoHe and %MicroR, and parameters for hemoglobinization of reticulocytes (Ret-He and δ-He) to validate the application of discriminating algorithms for screening subjects for iron deficiency anemia (IDA) and β-thalassemia.

The study objectives included (1) establishing the sensitivity and specificity of new algorithms in a cohort of subjects with IDA, a group of subjects confirmed to have β-thalassemia, and a control group of healthy subjects, and (2) comparing the algorithms with current existing formulas for discrimination.

Materials and Methods

Study Design

Our European multicenter study included 3 subject groups, namely, subjects with IDA, subjects with β-thalassemia, and apparently healthy hospital employees.

Hemocytometry

Blood samples were collected in Vacutainer tubes with K2EDTA as anticoagulant (Becton Dickinson, Plymouth, England) and analyzed within 4 hours after collection. For hemocytometric investigations, we used the Sysmex XE5000 automated hematology analyzer (Sysmex, Kobe, Japan). Samples were selected from the daily workload and analyzed with full hemocytometric parameter profile (CBC + differential count + nucleated RBCs + reticulocytes) to ensure that results of all parameters were available. Parameters of particular interest were hemoglobin, RBC, MCV, RDW–standard deviation (SD), RDW–coefficient of variation (CV), reticulocytes Ret hemoglobin equivalent (He), and δ-He. In addition, %HypoHe and %MicroR were quantified, indicating the percentage of hypochromic RBCs with a hemoglobin content of less than 17 pg and the percentage of microcytic RBCs with a volume of less than 60 fL.

ZPP/Hb Ratio

Zinc protoporphyrin (ZPP/Hb ratio) was measured with front surface illumination fluorometry using a dedicated hematofluorometer (Aviv Biomedical, Lakewood, NJ).18

Hemoglobin Electrophoresis

Hemoglobin electrophoresis was used to diagnose subjects with β-thalassemia. Increased HbA2 content (>3.2%) was considered to confirm β-thalassemia. In case of an abnormal hemoglobin variant and an increased fetal hemoglobin fraction in the electropherogram (>1.5%), the sample was excluded from the study.

Statistical Evaluation

The statistical software package MedCalc, version V11.5.1 for Windows, was applied for statistical analysis of results (MedCalc, Mariakerke, Belgium).

Receiver operating characteristic (ROC) curve analysis, including calculation of the area under the curve (AUC), was used to evaluate the diagnostic performance of the algorithms. Cut-off values were established based on the optimal combination of sensitivity and specificity. Sensitivity (sens%), specificity (spec%), positive predictive value (PPV), and negative predictive value (NPV) were calculated as follows:  

Sensitivity=[true positive/(true positive+false negative)]×100
 
Specificity=[true negative/(true negative+false positive)]×100
 
Positive predictive value=[true positive/(true positive+false positive)]×100
 
Negative predictive value=[true negative/(true negative+false negative)]×100

Mathematical formulas and cut-offs were previously published elsewhere.1–3

Results

Subjects With IDA

One hundred forty-two subjects with IDA (6 male and 135 female) were enrolled in the study based on suspect results for MCV (n = 132 MCV ≤ 80 fL; n = 9 MCV 80–86 fL) and ZPP/Hb ratio greater than or equal to 100 μmol per mol of heme. All selected subjects demonstrated results for %MicroR of 3 or higher. This limit was used as the first precondition step for discrimination using microcytic erythropoiesis.

The subjects were subsequently divided into 3 groups with normal to slightly decreased MCV (<85 and ≥ 75 fL), moderately decreased MCV (<75 and ≥ 65 fL), and severely decreased MCV (<65 fL). These subgroups were used in the second precondition step.

In addition, for the subgroup with MCV (<85 and ≥ 75 fL), ROC curve analysis (AUC) for %MicroR demonstrated that the optimal cut-off for %MicroR was 5 Table 1, numbers 1–3). As shown in Table 1, new algorithms, including conventional and advanced hematologic parameters, were created for these subgroups. Results of ROC curve analysis (AUC) were applied to establish optimal cut-offs for concerning parameters.

Subjects With β-Thalassemia

Thirty-four subjects with β-thalassemia (18 male and 16 female) were enrolled in the study. We used an increased HbA2 content (>3.2%) to confirm β-thalassemia. Subjects with results for iron metabolism markers beyond the reference range were excluded. All subjects demonstrated %MicroR results of 3 or higher. This limit was used as the first precondition step for discrimination of microcytic erythropoiesis.

The subjects were subsequently divided into 3 subgroups as follows: (1) normal to slightly decreased MCV (<85 and ≥ 75 fL), RDW-SD values less than 44.7 fL, and RBC values greater than or equal to 3.50 × 1012/L; (2) moderately decreased MCV (<75 and ≥ 65 fL) and RBC values greater than or equal to 3.50 × 1012/L; and (3) severely decreased MCV (<65 fL). These subgroups were used in the second precondition step (Table 1, numbers 4–6).

Table 1

Preconditions and Algorithms for Screening for Microcytic Anemia in Patients With IDA and β-Thalassemia*

As shown in Table 1 (numbers 4–6) new algorithms, including conventional and advanced hematologic parameters, were created for these subgroups. Results of ROC curve analysis (AUC) were applied to establish optimal cut-offs for concerning parameters.

To clarify the decision-making process for screening for IDA and β-thalassemia, we used the preselection criteria and algorithms depicted in Figure 1. Our algorithms for IDA and β-thalassemia discrimination, including the cut-off levels, were compared with previously published discriminating algorithms listed in Table 2.

Apparently Healthy Subjects

Three hundred nine apparently healthy hospital employees (133 male and 176 female, 16–63 years of age) without any clinical symptoms of disease were used as a reference group. Of these subjects, 3 scored a positive result for one of the IDA algorithms. These subjects demonstrated ferritin results of less than 10 μg/L. All other subjects showed results within the reference ranges for hemocytometric parameters, including %MicroR less than 3, and iron metabolism markers (data not shown).19 None of the subjects scored a positive result with the algorithms for β-thalassemia.

Figure 1

Flow diagram for screening on IDA and β-thalassemia. CV, coefficient of variation; Hb, hemoglobin; IDA, iron-deficiency anemia; MCV, mean corpuscular volume; %MicroR, percentage of microcytic RBCs; %HypoHe, percentage of hypochromic RBCs; RDW, red cell distribution width; Ret, reticulocytes; SD, standard deviation.

Figure 1

Flow diagram for screening on IDA and β-thalassemia. CV, coefficient of variation; Hb, hemoglobin; IDA, iron-deficiency anemia; MCV, mean corpuscular volume; %MicroR, percentage of microcytic RBCs; %HypoHe, percentage of hypochromic RBCs; RDW, red cell distribution width; Ret, reticulocytes; SD, standard deviation.

Table 2

Previously Described Discrimination Formulas*

The new algorithms for IDA discrimination demonstrated results for AUC, sensitivity, specificity, and positive and negative predictive values of 0.88, 79%, 97%, 74%, and 98%, respectively. The algorithms for β-thalassemia discrimination showed comparable results of 0.86, 74%, 98%, 75%, and 99%, respectively Table 3.

Discussion

Application of RBC indices has been recommended for discriminating between subjects with IDA and subjects with thalassemia.1–3 In this study, application of these formulas resulted in only 30% to 40% of subjects with β-thalassemia in a proper classification.

More recently, Urrechaga et al17,20,21 reported discriminating formulas with novel hematologic parameters such as %HypoHe and %MicroR combined with RDW-CV. Measurement of %HypoHe and %MicroR was demonstrated to be useful for detecting small changes in the amount of RBCs with inadequate hemoglobinization.

Because of the lifespan of circulating mature RBCs, %HypoHe yields information on iron status in the preceding 2 to 3 months; this parameter is shown to be a sensitive indicator for detecting functional iron deficiency.

In subjects with β-thalassemia, %HypoHe results are below the reference range because of ineffective erythropoiesis resulting from reduced production of intact hemoglobin. %HypoHe results in subjects with β-thalassemia are not different from those observed in subjects with IDA.

Results for %MicroR are increased in β-thalassemia subjects compared with those with IDA.

A European multicenter evaluation was performed to establish advanced algorithms for anemia discrimination. The study revealed algorithms with advanced hematologic parameters, such as %HypoHe, %MicroR, Ret-He, and δ-He, for discrimination of IDA and β-thalassemia. The algorithms demonstrated excellent diagnostic efficacy, compared with conventional discriminating formulas. In our study, application of the Urrechaga formulas resulted in about 50% of subjects with IDA and only 10% to 20% of subjects with β-thalassemia being appropriately classified.17,20,21

Table 3

Efficacy of the Newly Developed Algorithms Compared With the Previously Published Conventional Formulas*

The purpose of using formulas in anemia discrimination is to detect subjects who have a high probability of requiring appropriate follow-up to reduce unnecessary investigations and costs. In future, screening and reporting results of various parameters will be insufficient. Reduction of healthcare budgets and increasing numbers of parameters available in laboratory hematologic analyses make it necessary to provide support and interpretation for a correct clinical diagnosis. Algorithms for discrimination purposes must have high sensitivity scores to detect the maximum number of subjects of interest. On the other hand, adequate screening algorithms should be able to eliminate as many “other” subjects (high specificity) as possible to avoid further analysis (false positives). Using our recently developed algorithms for anemia screening, about 75% of subjects with IDA and subjects with β-thalassemia were classified properly. High sensitivity and specificity scores were demonstrated compared with conventional formulas.1–3 The high sensitivity and specificity scores of the algorithms demonstrate that the discriminating algorithms are appropriate devices for microcytic anemia screening.

None of the algorithms had 100% sensitivity and 100% specificity in discriminating between subjects with iron deficiency from subjects with β-thalassemia. Therefore, after screening with the 6 algorithms, confirmatory testing should be performed for proper diagnosis. The efficacy of the algorithms should be confirmed in a subsequent study with prospectively selected subjects with microcytic anemia because of other causes such as α-thalassemia or combinations of IDA with thalassemia.

We conclude that the advanced algorithms, derived from extended RBC parameters provided by the Sysmex XE5000 analyzer, are useful as laboratory devices for anemia screening.

References

1.
England
JM
Fraser
PM
.
Differentiation of iron deficiency from thalassaemia trait by routine blood count
.
Lancet
 .
1973
;
i
:
449
452
.
2.
Green
R
King
R
.
Red cell discriminant function, β-thalassaemia minor versus iron deficiency
.
Blood Cells
 .
1989
;
15
:
481
495
.
3.
Mentzer
WC
.
Differentiation of iron deficiency from thalassaemia trait
.
Lancet
 .
1973
;
i
:
882
.
4.
Bartels
PCM
Schoorl
M
Schoorl
M
.
Application of haemocytometry in combination with reticulocyte count and red cell zinc protoporphyrin for appropriate classification of microcytic anemia
.
Sysmex J Int
 .
2003
;
13
:
78
82
.
5.
Stockman
JA
Weiner
LS
Simon
GE
et al
The measurement of free erythrocyte protoporphyrin (FEP) as a simple means of distinguishing iron deficiency from beta thalassaemia trait in subjects with microcytosis
.
J Lab Clin Med
 .
1975
;
85
:
113
119
.
6.
Koenig
HM
Lightsey
AL
Schanberger
JE
.
The micro-measurement of free erythrocyte protoporphyrin as a means of differentiating alpha thalassaemia trait from iron deficiency anemia
.
J Pediatr
 .
1975
;
86
:
539
541
.
7.
Han
P
Fung
KP
Teo
CP
et al
Alpha-thalassaemia and beta-thalassaemia traits: biological difference based on red cell indices and zinc protoporphyrin
.
Clin Lab Haematol
 .
1990
;
12
:
169
176
.
8.
Haska
J
Lasserre
JJ
Schwarzbeck
A
et al
Laboratory test of iron status: correlation or common sense?
Clin Chem
 .
1996
;
42
:
718
724
.
9.
Labbé
RF
Vreman
HJ
Stevenson
DK
.
Zinc protoporphyrin: a metabolite with a mission
.
Clin Chem
 .
1999
;
45
:
2060
2072
.
10.
Harthoorn-Lasthuizen
EJ
Lindemans
J
Langenhuijsen
MM
.
Combined use of erythrocyte zinc protoporphyrin and mean corpuscular volume in differentiation of thalassaemia from iron deficiency anemia
.
Eur J Hematol
 .
1998
;
60
:
245
251
.
11.
Akai
Y
Kubota
F
Nakamoto
H
et al
A study of β-thalassaemia screening using an automated haematology analyzer
.
Sysmex J Int
 .
1998
;
8
:
110
114
.
12.
Eldibany
MM
Totonchi
KF
Joseph
NJ
et al
Usefulness of certain red blood cell indices in diagnosing and differentiating thalassaemia trait from iron deficiency
.
Am J Clin Pathol
 .
1999
;
111
:
676
682
.
13.
Navarro
JF
Macia
ML
.
Hypochromic red cells as an indicator of iron deficiency
.
J Rheumatol
 .
1997
;
24
:
804
805
.
14.
Macdougall
IC
.
Merits of hypochromic red cells as a marker of functional iron deficiency
.
Nephrol Dial Transplant
 .
1998
;
13
:
847
849
.
15.
Thomas
C
Thomas
L
.
Biochemical markers and hematologic indices in the diagnosis of functional iron deficiency
.
Clin Chem
 .
2002
;
48
:
1066
1076
.
16.
Brugnara
C
.
Iron deficiency and erythropoiesis: new diagnostic approaches
.
Clin Chem
 .
2003
;
49
:
1573
1578
.
17.
Urrechaga
E
Borque
L
Escanero
JF
.
Potential utility of the new Sysmex XE5000 red blood cell extended parameters in the study of disorders of iron metabolism
.
Clin Chem Lab Med
 .
2009
;
47
:
1411
1416
.
18.
Blumberg
WE
Eisinger
J
Lamola
AA
et al
Principles and applications of haematofluorometry
.
J Clin Lab Autom
 .
1984
;
4
:
29
42
.
19.
Pekelharing
JM
Haus
O
de Jonge
R
et al
Haematology reference intervals for established and novel parameters in healthy individuals
.
Sysmex Diagn Perspect
 .
2011
;
1
:
1
11
.
20.
Urrechaga
E
.
Discriminant value of % microcytic/% hypochromic ratio in the differential diagnosis of microcytic anemia
.
Clin Chem Lab
 .
2008
;
46
:
1752
1758
.
21.
Urrechaga
E
Borque
L
Escanero
F
.
The role of automated measurement of RBC subpopulations in differential diagnosis of microcytic anemia and β-thalassemia
.
Am J Clin Pathol
 .
2011
;
135
:
374
379
.