Prof. A. P MacPhail, Department of Medicine, University of the Witwatersrand Medical School, 7 York Road, Parktown, 2193, South Africa. E-mail: firstname.lastname@example.org
Laboratory tests used in the diagnosis of iron status lack specificity in defining iron deficiency anaemia (IDA) and anaemia of inflammation (AI). The serum transferrin receptor (sTfR) may provide more information in this regard. The iron status of 561 pre-school children was determined and classified using the conventional measurements. The value of the concentration of sTfR, the ratio of sTfR (µg/ml) to LogSF (µg/l) (TfR-Index), and the Log of the ratio of sTfR (µg/l) to SF (µg/l) − (LogTfR:Fer ratio), in the classification of the iron status were determined by comparing their distributions across the classification of iron status. Although there were significant differences in sTfR and TfR-Index across the categories of iron status, there was considerable overlap. All subjects with iron deficiency had LogTfR:Fer ratio > 2·55, whereas in all subjects classified as AI it was < 2·55, thus clearly separating the two. The LogTfR:Fer ratio was not able to exclude IDA in the presence of inflammation. However, in cases of combined IDA and AI the LogTfR:Fer ratio was < 2·55 but increased to > 2·55 after resolution of the inflammation. This novel method of calculating the LogTfR:Fer ratio may provide a more precise classification of the iron status of children.
Iron deficiency is the most common nutritional deficiency worldwide affecting a quarter of the world's population (Walter, 1994). Laboratory measurements that are commonly used for the estimation of the iron status – serum iron, ferritin, transferrin saturation, mean cell volume and zinc protoporphyrin cannot clearly distinguish iron deficiency anaemia (IDA) from anaemia of inflammation (AI) (Burns et al, 1990; Guyatt et al, 1992; Cook et al, 1993). The identification of the stages of iron deficiency –[iron deficient erythropoiesis (IDE), iron deficient stores (IDS) and IDA] using these conventional laboratory tests is complex (Bainton & Finch, 1964; Skikne et al, 1990) and there is a need for simple laboratory methods to define this disorder. The serum ferritin concentration (SF), an indicator of body storage iron, identifies absent iron stores when the level is below 12 µg/l. However, values between 12 and 100 µg/l are difficult to interpret because inflammation, even in the presence of iron deficiency, causes elevation of SF (Blake et al, 1981; Burns et al, 1990; Worwood, 1997). Anaemic patients with an SF within this ‘grey’ area may still be iron deficient, particularly in the presence of inflammation (Worwood, 1997).
Circulating transferrin receptor (TfR) is a soluble form of the membrane receptor produced by proteolytic cleavage. Both the expression of TfR on the cell surface and the concentration of soluble TfR are inversely related to the level of intracellular iron (Baynes, 1996; Baynes & Cook, 1996). Serum transferrin receptor (sTfR) levels are thus a reliable indicator of functional iron deficiency and enhanced red cell production (Cook et al, 1993; Baynes, 1996; Baynes & Cook, 1996). Serum TfR levels are reported to be significantly elevated in IDA, IDE and IDS and are a reliable index of early tissue iron deficiency (Cook et al, 1994; Allen et al, 1998). The sTfR offers an advantage in assessing iron status because of its indicated ability to distinguish IDA from AI and because it can identify iron depletion and functional iron deficiency in patients with concurrent inflammation (Ferguson et al, 1992; Nielsen et al, 1994; Allen et al, 1998).
The combination of measurements of iron stores and of functional tissue iron as represented by the sTfR:SF ratio has potential advantages in distinguishing IDA from AI (Skikne et al, 1990; Allen et al, 1998). In this study, conventional measurements of iron status were used to classify children into different categories of iron status, and the values of sTfR and of the different methods of combining sTfR and SF as ratios (TfR-Index and LogTfR:Fer ratio) were compared across the categories.
Subjects and methods
Study subjects The study involved 561 South African children aged from 1 to 6 years in a rural area in the Northern Province. Approval to undertake the study was obtained from the University of the North Ethics Committee for Research on Human Subjects and informed consent was obtained from parents or guardians of all children included in the study.
Laboratory methods Venous blood was drawn into EDTA and plain vacutainer tubes. Full blood counts (FBCs) were carried out within 6 h of collection and sera were separated within 12 h of collection in separate cryotubes and stored at −70°C until needed.
Sample analysis Haemoglobin concentration (Hb) and red cell indices were measured using a SYSMEX K1000 autoanalyser (TOA Medical electronics, Kobe, Japan). Serum iron and total iron binding capacity (TIBC) were analysed on a Dimension ES Clinical Chemistry System (Dupoint, Wilmington, Delaware, USA) and percent transferrin saturation (TS) calculated. Serum TfR and SF were measured using an enzyme-linked immunosorbent assay (ELISA, Spectro-Ferritin and Ramco-TfR; Ramco Laboratories, Houston, Texas, USA) method. The results for sTfR are expressed as µg/ml and for SF as µg/l. In calculating the sTfR:Fer ratio, the sTfR values were converted to µg/l and the logarithm (base 10) of the ratio was used – LogTfR:Fer (µg/l:µg/l). The TfR-Index was calculated as defined by Punnonnen et al (1997) and Dimitru et al (2000)[i.e the ratio of sTfR to logarithm (base 10) of SF − sTfR/LogSF ratio] and the units were not converted –(µg/ml:µg/l). The manufacturer's normal range for sTfR in adults was 2·9–8·3 µg/ml.
Supplementation trial Fifty-four of the anaemic children in this study were included in a larger iron supplementation trial for a period of 3 months. The children were randomly assigned to two groups: the placebo group and the iron group. Each group had 27 children. Iron was given in a form of ferrous lactate solution [Ferro drop L, Warner – Lambert S.A (Pty) Ltd)], at a dose of 0·3–0·6 ml a day (equivalent to 1–2 mg iron per body mass). Full blood counts and iron status measurements were carried out before and after the supplementation trial.
Statistical analysis Standard statistical tests including anova for comparison between groups were performed using systat (version 7·00) software. Post hoc pairwise comparisons were made using the Bonferroni adjustment.
Subjects were classified into four categories of iron status (normal, IDS, IDE, IDA) on the basis of Hb, SF and TS percentage (Table I). Six per cent of subjects were anaemic (Hb < 11 g/dl) with low TS percentage (< 16%) and SF < 12 µg/l and were classified as IDA. Three per cent of subjects were anaemic with low TS% (< 16%) but had a SF > 50 µg/l. These were classified as AI. Thirteen per cent of subjects were not anaemic but had low TS percentage (< 16%) and SF < 12 µg/l and were classified as IDE, whereas 8% of subjects were not anaemic with transferrin saturation = 16% and serum SF < 12 µg/l and were classified as IDS. Almost 40% of cases were not classifiable into these conventional groups. Fifteen per cent were anaemic and hypoferraemic but had SF levels between 12 and 50 µg/l and were labelled as unclassified anaemia (UCA), the remaining 25%, labelled unclassifiable not anaemic (UCNA), had normal haemoglobin levels but had iron studies in the range of AI or UCA (Table I).
Table I. The mean (± SD) of sTfR) and LogTfR:Fer ratio in categories of iron status on the basis of haemoglobin concentration, SF and ST percentage.
Means with the same letter are not significantly different (P < 0·005).
AI, anaemia of inflammation; IDA, Iron deficiency anaemia; IDE, Iron deficiency erythropoiesis; IDS, iron deficient stores; UCA, unclassified anaemic and UCNA, unclassified non-anaemic.
As for AI or UCA
15 (2. 7%)
sTfR − mean (± SD)
TfR-Fer Index – mean (± SD)
LogTfR:Fer ratio − mean (± SD)
The sTfR was not able to distinguish between Normal, IDS, AI, UCA and UCNA (P > 0·05) (Table I and Fig 1). The mean sTfR was significantly higher in IDA and IDE compared with any other group (P > 0·001). Although the mean sTfR in AI was significantly lower than the IDA and IDE groups, it was not significantly different from the other groups (Table I). The mean TfR-Index in AI was significantly lower than in all iron deficiency groups (IDA, IDE and IDS) but was unable to distinguish between normal and AI (Table I). On the other hand, the mean LogTfR:Fer ratio was significantly lower in AI than any other group (Table I). Although the mean sTfR and TfR-Index in the iron deficient groups were significantly lower in the normal and AI groups, there was a considerable overlap of values (Figs 1 and 2). The sTfR values of many subjects lay within the quoted normal reference range (2·9–8·3 µg/ml) (Fig 1). A plot of the values of the Log TfR:Fer ratio shows that the Log TfR:Fer ratio was > 2·55 in all iron deficient groups, whereas all cases classified as AI had a ratio < 2·55 (Fig 3).
In the iron supplementation trial, which included 27 children in each of the iron and the placebo groups, the changes (Δ) in the parameters were calculated by subtracting the ‘before’ levels from the ‘after’ levels. Although a further drop in the serum ferritin and TS percentage was noted between the initial and final measurements of the placebo group, the fall was not significant (P > 0·05) (Table IIA). Significant improvements were noted in the Hb levels and all the iron status parameters of the children in the iron group (P < 0·05) (Table IIA). The mean (± SD) TfR levels of the iron group dropped significantly from 8·4 (3·8) µg/ml to 5·2 (1·7) µg/ml, whereas the mean (± SD) Log TfR:Fer ratio dropped from 2·7 (0·4) to 2·3(0·3).
At the end of the trial seven (25%) of the placebo group had haemoglobin levels in the normal range, whereas the other 20 (75%) were still anaemic. A comparison between the iron status of these groups was made. A significant increase in the Hb levels of the seven children (P < 0·01) was accompanied by a significant drop in their ferritin levels (P < 0·04) but no changes in the other iron status parameters. No differences were noted in the 20 children who were still anaemic (P > 0·05). The initial Log TfR:Fer ratios in the seven children with improved Hb levels was in the AI range (< 2·55) with the mean (SD) of 2·3 (0·2). Fifteen of the remaining 20 children had Log TfR:Fer ratios in the iron deficient range (> 2·55; Table II).
Using the Log TfR:Fer ratio value of 2·55 as a cut-off point, the children were divided into four groups: < 2·55 placebo and = 2·55 placebo groups, < 2·55 iron and = 2·55 iron groups (Table IIB). No significant changes were noted in the Δ mean of the Hb and the TS percentage of both the < 2·55 and > 2·55 placebo groups. Significant differences were noted between Δ mean of the serum ferritin, sTfR and Log TfR:Fer ratio between the placebo groups (P < 0·01). Iron treatment significantly improved all the measured parameters of both the < 2·55 and > 2·55 iron groups (P < 0·05), but no significant differences were noted between the changes in the Hb levels, TS percentage and serum ferritin levels between these groups (P > 0·1). However, the fall in sTfR and Log TfR:Fer ratio was significantly different, with a larger drop occurring in the = 2·55 iron group (P < 0·03) (Table IIB).
Considering the findings above, the LogTfR:Fer ratio and the haemoglobin concentration were used to reclassify the subjects into four iron status groups (Normal, Iron deficient (ID), IDA and AI (Table III). Table III shows the frequency of cases in each iron status group as defined by conventional criteria (rows) compared with the frequency when the LogTfR:Fer ratio was used as the defining criterion (columns). Over 70% of cases categorized as UCA with SF between 12 and 50 µg/l had a LogTfR:Fer ratio < 2·55 and were therefore classified as AI. The remainder was classified as IDA. Similarly, 23% of cases labelled as UCNA were iron deficient on this basis (Table 3).
Table III. Comparison of the conventional classification of iron status (rows) and a classification based on Log TfR: Fer ratio (columns).
AI, anaemia of inflammation; ID, iron deficiency; IDA, iron deficiency anaemia.
Anaemia of inflammation
Iron deficiency anaemia
Iron deficiency erythropoiesis
Iron deficient stores
Unclassified not anaemic
One of the major difficulties encountered in clinical medicine is to distinguish between IDA and the anaemia of inflammation, especially when they occur simultaneously (Burns et al, 1990; Cook et al, 1993). Although an SF of < 12 µg/l is diagnostic of iron deficiency its sensitivity is compromised in the presence of inflammation. Further investigations such as bone marrow examination or the determination of the response to iron treatment may be needed to solve this dilemma (Beck, 1990). The measurement of sTfR provides additional information in the assessment of iron status. Several studies have indicated that sTfR levels increase in IDA even in the presence of inflammatory diseases and that it is a reliable indicator of iron depletion (Cazzola & Beguin, 1992; Pettersson et al, 1994; Punnonen et al, 1994; Mast et al, 1998). Unlike serum iron, TS percentage and SF the sTfR level is not influenced by the acute phase response and it should be more reliable in the differential diagnosis of iron status (Cook et al, 1993; Mast et al, 1998; Shih, 1998; Tirkewitz et al, 1998). In addition to providing information on the level of intracellular iron, the sTfR is proportional to the total erythroid mass and reflects the rate of erythropoiesis (Huebers et al, 1990). In spite of its promise, several studies have indicated that the sTfR concentration adds little discriminatory power to SF (Pettersson et al, 1994;Mast et al, 1998).
The use of sTfR:Fer ratio is an improvement in specificity and sensitivity over the sTfR concentration alone (Punnonen et al, 1997; Cermak & Brabec, 1998; Dimitriou et al, 2000), but the method of calculating the ratio remains the subject of debate. The crude ratio of the values usually used in the assays (µg/ml for sTfR and µg/l for SF) is not better than sTfR alone (Punnonen et al, 1997). However, the ratio of TfR (µg/ml) and the log of the SF (TfR Index) proved a better indicator of iron depletion (Table I) (Punnonen et al, 1997). There are two problems with this approach. First, the units of measurement differ by three orders of magnitude (µg/ml vs µg/l). Second, in many instances the log of the SF values encountered in this situation is often close to unity and the TfR-Index therefore differs little from the sTfR alone. In this study, we have used the LogTfR:Fer after converting the values to SI units (µg/l). This ratio was significantly different in each of the recognized categories of iron status (except IDA and IDE) and was clearly different in normal, iron deficiency and AI (Table I, Fig 3). In contrast, neither the TfR alone nor the TfR-Index could be used to clearly distinguish between these groups (Table I, Figs 1 and 2).
A positive response of Hb levels to iron treatment in anaemic subjects is considered to be a clear indicator of pre-existing iron deficiency. In the treatment study, a significant improvement was noted in all anaemic children treated with iron, whereas no significant improvements were seen in anaemic children in the placebo group (P < 0·05). Using this criterion, all of the anaemic children treated with iron were iron deficient. However, 35% of these children had LogTfR:Fer ratios < 2·55. This appears to indicate that in this group the LogTfR:Fer ratio has a sensitivity of only 63% for diagnosing iron deficiency. It seems likely that this low sensitivity is caused by the presence of either acute or chronic inflammation combined with iron deficiency.
Evidence of this was seen in the placebo group in which anaemic children did not receive iron. In seven (25%) of these 27 children the Hb level rose into the normal range. These seven children all had Log TfR:Fer ratios < 2·55, and the improvement in the Hb level was accompanied by a fall in SF indicating a shift of iron from stores to haemoglobin. This suggests that the most likely cause of anaemia in these children was acute infection which resolved during the supplementation trial. Similar changes have been reported previously in studies of iron deficiency and infection in children (Jansson et al, 1984; Reeves et al, 1984; Olivares et al, 1989).
Previous workers have suggested that the TfR as a single additional measurement to the existing conventional methods does not provide more substantial information on iron status than SF (Pettersson et al, 1994; Fernandez-Rodriguez et al, 1999). This study confirms these observations, but indicates that the LogTfR:Fer ratio is a better criterion than either TfR alone or the so-called TfR-Index for defining the iron status of children. Using a cut-off level of LogTfR:Fer ratio > 2·55, non-anaemic children with marginal iron status who could not be identified using the conventional methods were identified as iron deficient. Similarly, the LogTfR:Fer ratio identified a higher number of anaemic children in either the IDA and AI groups than when the conventional methods were used. Although the ratio > 2·55 is specific for pure iron deficiency (IDA, IDE & IDS) the Log TfR:Fer ratio < 2·55 identifies both AI and combined IDA and IA. Therefore the ratio clearly distinguishes pure iron deficiency from AI and combined IDA–IA with a higher specificity than the conventional methods, but it still cannot distinguish between AI and combined IDA–IA. In cases of combined IDA–IA, the Log TfR:Fer ratio will be decreased and will rise to levels diagnostic of iron deficiency when the inflammation is cleared. Further studies are needed to confirm these findings and the possible use of the LogTfR: Fer ratio in the classification of the iron status of adults.
The help and co-operation of the community of the Dikgale village and their Chief is gratefully acknowledged. Funding was obtained from the South African Medical Research Council and Phizer (Parke-Davies; South Africa) provided the iron supplementation. Uta Schmidt provided technical assistance.