Soluble Transferrin Receptor during infancy and reference intervals for the Roche Cobas platform

Abstract Introduction Infant iron status assessments may be difficult to interpret due to infections. The soluble transferrin receptor (sTfR) has been suggested as a biomarker mainly unaffected by the acute phase response. Reference intervals reflecting dynamics of infant growth first year in life are not well established. Methods The sTfR and CRP concentrations were measured in samples from 451 term infants with the Roche Cobas platform in umbilical cord, at 48‐96 hours, 4 and 12 months. Reference values were constructed as the 2.5th and 97.5th percentiles. The relationship between CRP concentrations >1 mg/L and sTfR was tested by Kendall correlation. Results Reference intervals for girls and boys were 2.4‐9.5 mg/L at birth, 2.9‐8.4 mg/L at 48‐96 hours, 2.6‐5.7 mg/L at 4 months and 3.0‐6.3 mg/L at 12 months. No differences between sexes were observed except for at 4 months. sTfR did not covariate with CRP concentrations >1 mg/L except in 48‐96 hours samples. Conclusion This study reports reference intervals for sTfR from birth to 12 months of age in a large group of infants in a low‐risk area for iron deficiency. sTfR might add value to infant iron status diagnostics since no covariation with CRP was found at birth, at 4 months or at 12 months.


| INTRODUC TI ON
Iron has important physiologic roles in early life; thereby assessment of infant iron status is relevant to healthcare worldwide. Iron status biomarkers are more or less affected by interplay with the acute phase response. As a crucial nutrient for many microorganisms, including pathogens such as the malaria Plasmodium parasite, human biochemical responses are developed to keep iron unavailable to infecting microorganisms. 1 The soluble Transferrin Receptor (sTfR) has been suggested to be less influenced by the infectious and inflammatory status.
Extensively studied, mainly in adults, several studies have emphasized its advantages in distinguishing between anemia of inflammation and iron deficiency anemia (IDA). 2 It is generated by proteolytic cleavage of transmembrane dimeric glycoprotein transferrin receptor, as the red blood cell precursors mature; thus mainly shed from erythroblasts and reticulocytes. The relationship between tissue transferrin receptor and sTfR is reported to be constant, and sTfR by that reflects erythroid proliferation rate and iron turnover. 3 Whereas ferritin is a marker of iron stores, sTfR has been described to be a marker of iron needs, which means that these two biomarkers do not necessarily correlate. 4 Erythropoiesis as such is subject to large developmental changes subsequent to adaption to extrauterine life. Consequently, reference intervals need to be able to reflect the dynamics of infant growth.
Furthermore, frequent infections complicate infant iron status assessments, but use of sTfR as a marker of iron status in infants has been questioned. 5,6 Although to the best of our knowledge, there are in total 14 studies of sTfR reference intervals during infancy published, only two of these refer to one of the most widely used commercial assays; Roche Diagnostics. [6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] Standardization has not been widely established and conversion of reference intervals between platforms is not possible. 16 Age intervals vary across studies, and evaluation of published data is complicated. Moreover, previous studies have shown possible iron biomarker differences according to sex during infancy. 7,21 The aim of the present study was to calculate reference intervals, for the Roche Cobas sTfR assay, at four different time points in infancy and to investigate if partitioning according to sex is needed. A secondary aim was to describe sTfR in relation to increased C-reactive protein (CRP) in infants born in a geographic area with low risk of iron deficiency and a low burden of infections.

| Study population
This was a retrospective longitudinal cohort study. The infants were born in a low-risk area for iron deficiency, from uncomplicated pregnancies and uneventful perinatal circumstances and had a gestational age of 37 +0 -41 +6 weeks at delivery. The mothers were nonsmokers and healthy with uneventful pregnancies. Data were collected during 2008-2015 as part of studies assessing timing of umbilical cord clamping. [22][23][24][25] The first study population comprised of two studies performed at the county hospital of Halland, one study with children born in vaginal births (n = 400) and one study with children born by elective cesarean section (n = 64). These were approved by the regional research ethics committee at Lund Body weight at 12 months was within WHO Child growth standards 28 except for two girls and nine boys who had slightly higher weights than the WHO 99th percentile (12.7, 13.5 kg and 12.5-13.6 kg, respectively). Body lengths at 12 months were within WHO Child growth standards except for one girl (82 cm) and four boys (82-83 cm) slightly above the WHO 99th percentile.
Since a clinical decision limit excluding iron deficiencies for this age group has been debated, we chose to illustrate robustness of the population by presenting reference values excluding individuals with ferritin concentrations below two arbitrarily set cutoff points (20 and 30 µg/mL, respectively).

| Specimen collection and handling
Blood samples in studies performed at the hospital of Halland were taken at birth (umbilical cord blood), at 48-96 hours in conjunction with the metabolic screening, at 4 months and at 12 months. For the children born in Halland, blood was collected in tubes with serum separator (BD Vacutainer, Plymouth, UK). Blood samples in the study carried out at Karolinska University Hospital were taken at birth (umbilical cord blood) and at 4 months and blood was collected in serum tubes with serum separator (Sarstedt AG & Co, Nümbrecht, Germany).
Blood sampling (a maximum of 2.5 mL blood) was at 4 and 12 months performed after application of a local dermal analgesia with lidocaine 2.5% and prilocaine 2.5% (EMLA, AstraZeneca).

| Laboratory methodology
CRP, sTfR, and ferritin were analyzed using Cobas 6000 (Roche Diagnostics, Basel, Switzerland). Instruments were calibrated and run according to the manufacturer's instructions. The sTfR Tinaquant (a particle enhanced immunoturbidimetric assay) was according to the manufacturer standardized to an internal Roche reference material. Total imprecision was estimated by use of a commercial internal control material from SERO AS (Billingstad, Norway).

| Statistical methodology
Data at each sampling time point were assessed separately. Data were mathematically transformed by the method of Box-Cox.
Transformation was ruled in as successful if a) the visual inspection of the Q-Q-plot was approved and b) if the hypothesis test of Anderson-Darling passed with a significance level of 10%.
The 95% interpercentile reference interval was calculated for each group and a 90% confidence interval around each endpoint was estimated by using biweight quantile, transformed, reflected methodology with 500 bootstrap samples. Differences between time points and sexes were assessed as significant if 90% confidence intervals were nonoverlapping.

Relation between sTfR and CRP in samples with a CRP result
>1 mg/L was assessed with Kendall rank correlation as data were non-normally distributed.
Statistical analyzes were conducted using Analyse-it for Microsoft ExcEl 4.90.4 from Analyse-it Software Ltd.

| Reference intervals
Distribution of the original sTfR concentration data at the four sampling time points is shown by bee-swarm box plots in Figure 1. Lower and upper reference values with 90% confidence limits partitioned by age and the two suggested cutoffs for ferritin are shown in Table 1.

| Blood sampling from umbilical cord
The lower reference value of sTfR was slightly higher in boys than in girls and lower 90% confidence intervals were nonoverlapping.
Corresponding upper confidence interval pointed to no differences between sexes.

| Blood sampling at 48-96 hours
We found no significant differences in sTfR according to sex at 48-96 hours, neither at the lower nor at the upper reference levels.

| Association to CRP as a marker of the acute phase response
At 48-96 hours, a correlation between an increased CRP concentra-

| D ISCUSS I ON
The sTfR concentration has been suggested as an iron status biomarker in the pediatric population. There are many factors associated with normal growth that may influence interpretation of sTfR results in medical decision-making and reference intervals also con- Comparing data at 12 months, they reported the 5th percentile as well as the 97.5th percentile to be higher. 16 Differences may be due to study populations or bias between instruments/laboratories as well as statistical methodology used describing data.
We also compared the reference intervals in this study to A major strength of our study is the longitudinal design and data collection of blood samples from 451 healthy infants from uncomplicated pregnancies in an area with low risk for iron deficiency.
Parents declared subjective well-being of the child, but we cannot fully exclude the possibility of a child having some underlying condition. What criteria that with certainty could serve diagnosing infant iron deficiency is still a subject of discussion. Our study may have included some infants with iron deficiency. The reliability of our data was therefore tested by excluding infants with a corresponding ferritin <20 µg/L and <30 µg/L. The sTfR reference values at 4 and 12 months were not subject to significant changes by this partitioning. It is therefore unlikely that these sTfR reference values are affected by iron deficiency. Another concern is that the reference values could be affected by hematological disorders, such as thalassemia or hemochromatosis. However, prevalence of thalassemia is very low in northern Europe 35 and juvenile hemochromatosis is very rare. 36 Also, results could be affected by the timing of umbilical cord clamping. In this study, a minimum of 30 seconds was selected; based on the American College of Obstetricians and Gynecologists committee opinion, 26 but this time of clamping delay can be regarded as short as WHO recommendation is to delay cord clamping and cutting for 1-3 minutes for term infants. 37 A weakness of our study is that the approach used cannot establish decision limits, and the intention of our study is not that the 97.5th percentile should be used as such. To establish decision limits, studies need to be performed with a design based on clinical adverse outcomes which is not possible to perform ethically. Reference intervals do not imply a decision limit, but should be considered as a statistical model describing the distribution of a biomarker for a well-defined population. Also, the reference intervals calculated in this study are specific for the Roche Cobas assay and traceability to 1st WHO reference reagent preparation 07/202 is lacking. Another weakness is the few numbers of elevated CRP concentrations in the umbilical cord samples used to estimate the nonparametric correlations with sTfR.
In conclusion, this study present reference values for the iron status biomarker sTfR in a large cohort of presumably healthy infants measured on the Roche Cobas platform. The biomarker sTfR might add value to infant iron status diagnostics since no covariation with CRP was found at birth, at 4 months or at 12 months.