Hepcidin, Serum Iron, and Transferrin Saturation in Full-Term and Premature Infants during the First Month of Life: A State-of-the-Art Review of Existing Evidence in Humans

ABSTRACT Neonates regulate iron at birth and in early postnatal life. We reviewed literature from PubMed and Ovid Medline containing data on umbilical cord and venous blood concentrations of hepcidin and iron, and transferrin saturation (TSAT), in human neonates from 0 to 1 mo of age. Data from 59 studies were used to create reference ranges for hepcidin, iron, and TSAT for full-term-birth (FTB) neonates over the first month of life. In FTB neonates, venous hepcidin increases 100% over the first month of life (to reach 61.1 ng/mL; 95% CI: 20.1, 102.0 ng/mL) compared with umbilical cord blood (29.7 ng/mL; 95% CI: 21.1, 38.3 ng/mL). Cord blood has a high concentration of serum iron (28.4 μmol/L; 95% CI: 26.0, 31.1 μmol/L) and levels of TSAT (51.7%; 95% CI: 46.5%, 56.9%). After a short-lived immediate postnatal hypoferremia, iron and TSAT rebounded to approximately half the levels in the cord by the end of the first month. There were insufficient data to formulate reference ranges for preterm neonates.


Maternal control of fetal and early neonatal iron metabolism
Increases in maternal dietary iron uptake and placental iron transfer occur in the second and third trimesters (21,22), when maternal hepcidin decreases to trigger increased duodenal iron absorption (23), splenic macrophage iron recycling, and the release of maternal hepatic iron stores (24)(25)(26). The resulting increased circulating maternal iron is then freely available for transfer to the fetus. Factors that are thought to contribute to maternal hepcidin suppression in the second and third trimesters include maternal iron deficiency, erythropoiesis in the mother or fetus (25), estrogen (26), and progesterone receptor membrane component-1 (27). Conflicting evidence now exists as to whether pregnancy-induced plasma dilution may also play a role (15,28).

Fetal control of fetal and early neonatal iron metabolism
Eighty percent of all the iron transference from the mother to the fetus occurs in the last trimester (29). An illustration of the fetal demand for iron [amounting to 1.6-2.0 mg · kg −1 · d −1 (30)] is that umbilical cord FIGURE 1 Placental iron transfer between mother and fetus. Syncytiotrophoblasts in the placental villi take up Tf-bound iron from the maternal circulation by endocytosis via TFR1. Iron is released from TFR1 in acidified endosomes and transferred into the syncytiotrophoblast cytoplasm. Ferroportin transports iron out of placental syncytiotrophoblasts, and then ceruloplasmin, hephaestin, and zyklopen oxidize Fe 2+ to Fe 3+ , helping it pass through the endothelium to reach the fetal circulation. It is still unclear as to whether newly transported iron enters the fetal circulation as NTBI or bound to fetal Tf. Fetal-derived hepcidin is believed to regulate ferroportin expression on the fetal basal side of placental syncytiotrophoblasts (12,25). Maternal-derived hepcidin is believed to play a role in regulating TFR1 expression on the maternal side of placental syncytiotrophoblasts (31). Apo-Tf, unsaturated transferrin; CP, ceruloplasmin; DMT1, divalent metal transporter 1; fetal Tf, fetal-derived transferrin; Fe 2+ , ferrous iron; Fe 3+ , ferric iron; HEPH, hephaestin; NTBI, non-transferrin-bound iron; Tf, transferrin; TFR1, transferrin receptor 1; STEAP, six-transmembrane epithelial antigen of prostate; ZIP, zinc and iron related protein; ZP, zyklopen.

Placental control of fetal and early neonatal iron metabolism
The placenta may also independently regulate iron transfer to the fetus in some scenarios (58). A reduction of ferroportin expression on the apical fetal-facing membrane of placental syncytiotrophoblasts during maternal iron deficiency, in addition to increased expression of TFR1 on the maternal-facing side supports this hypothesis (28). Sangkhae et al. (28) propose that during maternal iron deficiency, iron is held in the placenta to ensure that its metabolic homeostasis is maintained. Placental protein synthesis and critical transfer mechanisms can then continue, ensuring the more detrimental condition of placental dysfunction does not occur. These findings were observed in murine and in vivo human trophoblast models, but not in respect to the human pregnancies analyzed (28).

Effects of infection on neonatal serum hepcidin concentrations
Intra-amniotic infections can cause an increase in fetal hepcidin (81). Multiple studies have documented an association of chorioamnionitis, perinatal acidosis, and neonatal sepsis with high umbilical cord hepcidin concentrations (81)(82)(83)(84)(85)(86). For example, an extremely high cord hepcidin concentration (437.6 ng/mL) was found in a neonate with confirmed Enterococcus faecalis early-onset sepsis (84). Similarly, very-lowbirth-weight, premature neonates with late-onset culture-confirmed sepsis exhibit elevated concentrations of hepcidin (83). Nevertheless, despite the well-documented regulatory pathways of infection and inflammation on iron regulation, it is important to note that multiple publications have shown a lack of correlation between hepcidin, IL-6, and C-reactive protein (CRP) in sick neonates (84,87). This is likely due to differences in the biochemical kinetics of these molecules. IL-6 concentrations spike very early in the course of perinatal infection, whereas the rise of CRP is delayed.

Standardizing hepcidin measurements
Multiple assays, including MS and immunochemistry ELISA methods, are available to quantify hepcidin in various body fluids (urine, serum, and plasma) (88). However, in the studies included in this state-of-theart review, none of these methods were calibrated using the same standards and, as a result, there were significant differences in hepcidin values between studies (89,90).
In 2016, van der Vorm et al. (89) harmonized many of the available hepcidin ELISA assays using native, lyophilized plasma with cryolyoprotectant as a commutable candidate reference material. Linear equations were formulated to standardize the hepcidin assays (89). These equations can now be used to conduct post hoc standardization of noncalibrated test results, aiding the retrospective comparison of data from previous publications. We have used these equations in this state-ofthe-art review to generate standardized hepcidin values (Supplemental Table 1). Standardized reference material, which was refined in 2019, is now available for purchase, allowing hepcidin measurements to be standardized in all laboratories (90).
To our knowledge, this is the first time that retrospective comparisons have been made between serum hepcidin concentrations in different studies, using post hoc standardized values to calculate weighted mean averages in umbilical cord and venous blood.
This state-of-the-art review contributes this comparative analysis and also offers an example for how other authors could approach retrospective comparisons of hepcidin concentrations from different studies.

Methods
In March 2019, we reviewed the literature searching 2 databases-PubMed and Ovid Medline-with no restrictions on language. The original search was for human studies only published during the date range of 1 January, 1975 to 1 December, 2019. Corresponding authors of extracted publications were not contacted. One individual carried out the inclusion/exclusion process of the retrieved studies, and there was no assessment of bias or the quality of studies as seen in a systematic review process. Table 1 displays the search strategy used. Figure 2 shows the flow diagram of the literature search. The search generated publications containing data on cord and venous concentrations of hepcidin and serum iron, and levels of transferrin saturation (TSAT), in the neonatal period. Studies that analyzed healthy neonates were included. Mean, median, or range of the gestational age of the study population was a requirement for inclusion. Neonates ≥37 wk at delivery were regarded as full-term-birth (FTB) neonates. Studies or study groups with a gestational age <37 wk were classed as premature [preterm-birth (PTB)] neonates. Retrieved publications had to report a mean time of bleed 0-720 h postdelivery to be analyzed. Mean (SD or 95% CI) or median (range, IQR, or 95% CI) data were extracted from the included publications. Studies reporting means (95% CIs) were included in the calculation of weighted means (95% CIs) and the associated Figures 3-5. Reference ranges for adults and children were presented for comparison (91,92). Many retrieved publications did not stratify results by birth weight; as a result, this variable was not recorded in Tables 2-7. Publications were not stratified by sample type (serum or plasma) owing to the overall lack of studies. If multiple publications on the same study population were retrieved, only 1 was included in the analysis.
The standardization of hepcidin values generated by different ELISA assays was performed using the slopes and intercepts from van der Vorm et al. (89)  To calculate the CI around the weighted mean, the weighted variance was calculated using the wtd.var function from the R package Hmisc. The SE derived from this weighted variance was then used to calculate the t statistic (i.e., weighted mean divided by weighted SE), from which the 95% CI was derived. GraphPad Prism version 8 (GraphPad Software) software was used to produce the graphical representation of the results.

Results
The initial search of 2 electronic databases for 3 different iron markers yielded 13,931 publications. After the exclusion of duplicated studies and selection criteria filtering, 20 publications were included in the analysis for hepcidin, 23 publications for TSAT, and 51 publications for serum iron. Many of these studies were found to contain information on multiple parameters of interest. Overall, we identified 59 publications containing data on hepcidin, serum iron, or TSAT in FTB neonates. Sixteen publications were found to contain data on PTB neonates. In publications detailing the effects of cord clamping interventions, all retrieved cord blood values were from groups that underwent 60 s of delayed cord clamping. This is consistent with current WHO policy (101). Cord blood weighted mean values are shown in Tables 2-7

Hypoferremia in FTB neonates
The weighted mean average for cord blood hepcidin was calculated using data from 11 studies. Almost all included studies reported a mean value between 11 and 41 ng/mL, apart from Kulik-Rechberger et al. (46). This study reported a much higher cord blood hepcidin value (67.9 ng/mL; 95% CI: 59.3, 76.5 ng/mL), as Supplemental Figure 1A shows. In addition, this study also recorded higher hepcidin concentrations in venous samples collected at 72 h (92.9 ng/mL; 95% CI: 83.  hepcidin production may provide a quick, comprehensive, and relatively long-lasting (0-3 d) hypoferremic response to aid protection during this vulnerable period (99). After the first few days, TSAT gradually increases as do serum iron concentrations, eventually reaching a plateau at ∼1 mo of age.

Lack of data on preterm neonates during the first 24 h
After analysis of the current literature, the extent of the role that hypoferremia plays in neonates with a gestational age <37 wk is still unclear. This is primarily due to the limited number of publications documenting hepcidin (n = 5), TSAT (n = 6), and serum iron (n = 13) in the first month of life in preterm neonates. The variability between the studies is Hemoglobin concentrations are provided to aid interpretation of neonatal iron status. 3 Hepcidin standardization was conducted using the linear equations documented in Supplemental Table 1. 4 Extracted SDs were converted to 95% CIs. 5 Medians ([IQRs] or 95% CIs or ranges) were not included in weighted means. 6 Values from Basu et al. (51) were not standardized because the study used the Hangzhou Eastbiopharm ELISA, which was not part of the van der Vorm et al. (89) analysis. 7 Median [IQR]. 8 Reference ranges for adults (male and female) and infants are displayed for comparison (92). Hemoglobin concentrations are provided to aid interpretation of neonatal iron status. 3 Hepcidin standardization was conducted using the linear equations documented in Supplemental Table 1. 4 Extracted SDs were converted to 95% CIs. 5 Medians ([IQRs] or 95% CIs) were not included in weighted means. 6 Reference ranges for adults (male and female) and infants are displayed for comparison (92).

CURRENT DEVELOPMENTS IN NUTRITION
vast and further complicated by the complex, intensive, and inconsistent care of premature neonates worldwide. Data analysis of the retrieved publications suggests that preterm neonates have lower cord hepcidin than full-term neonates, or infants and healthy adults. Weighted cord mean values are 250% higher in fullterm (29.7 ng/mL; 95% CI: 21.1, 38.3 ng/mL) neonates than in preterm (8.4 ng/mL; 95% CI: 2.0, 14.7 ng/mL) neonates. We speculate that this could be due to very early preterm neonates (<30 weeks of gestation) possessing circulatory monocytes with decreased surface expression of toll-like receptor 4 (TLR4), lower mRNA expression of TLR4, and reduced cytokine production (122). An effect on the production of IL-6 at delivery might then lead to a reduced ability to stimulate hepcidin expression as suggested in full-term infants.
Our analysis proposes that peripheral venous hepcidin values in preterm neonates increase to 44 ng/mL at 168 h. However, decreases in TSAT between the cord and venous samples are not observed (36.5%-45.6%). We propose that this is due to a lack of data on TSAT levels in preterm neonates over the first hours of life, potentially due to the complex ethical questions around bleeding preterm neonates so early in postnatal life. This results in the collection of skewed data, focusing only on later time points in the first month of life.

Limitations
The aim of this state-of-the-art review was to evaluate our current knowledge on neonatal iron homeostasis in preterm and full-term neonates. As a result of the dearth of publications detailing the parameters of interest during this period, our review has several limitations discussed below. First, we were unable to stratify by geographical location. Many studies do not stratify their study groups by gestational age (preterm: <37 wk, full-term: ≥37 wk). Subsequently, we have had to assign each study group or population by the mean gestational age. This will result in a reduction of any natural variation potentially caused by gestational age between the reviewed populations. This is also the case with respect to birth weight and hemoglobin concentration.
Similarly, the studies on preterm neonates are made up of multiple small sample size subgroups with different gestational ages. Owing to the lack of preterm studies, we have had to combine these study groups to formulate weighted means and figures. This in itself could distort the impact of gestational age on our results, because data from the very early preterm newborns are combined with those from the late preterm neonates.
The retrieval of gestational age was a crucial aspect of the search strategy; however, few studies documented the method used. There are large differences in the accuracy of different techniques (123).
Post hoc standardization of different hepcidin ELISA kits has, to our knowledge, never been completed before with retrospective data. However, care should be given to the accuracy of the standardized values, because standardization was only possible for DRG, Bachem, and Intrinsic Lifesciences ELISA test kits. Studies that used alternative methods (124) were not included in summary statistics.
An essential criterion of inclusion in this publication was that all neonatal data came from healthy newborns. However, documentation of labor practices (including mode of delivery) and postnatal care, along with postnatal medication, lack detail in the publications retrieved. Vaginal delivery is occasionally referred to as the method of delivery; TSAT, transferrin saturation; -, not determined because not applicable to the calculation of weighted mean hepcidin or standardized hepcidin values. 2 Hemoglobin concentrations are provided to aid interpretation of neonatal iron status. 3 Extracted SDs were converted to 95% CIs. 4 Medians ([IQRs] or 95% CIs) were not included in weighted means. 5 Median (minimum-maximum range). 6 Median [IQR]. 7 Reference ranges for adults and infants are taken from the NHANES, 1999-2000 (91). Hemoglobin concentrations are provided to aid interpretation of neonatal iron status. 3 Extracted SDs were converted to 95% CIs. 4 Medians ([IQRs] or 95% CIs) were not included in weighted means. 5 Median (minimum-maximum range). 6 AGA group of Hågå (128) can be identified in Supplemental Figure 2B. 7 SGA group of Hågå (128) can be identified in Supplemental Figure 2B. 8 Minimum-maximum range. 9 Reference ranges for adults and infants are taken from the NHANES, 1999-2000 (91).  Hemoglobin concentrations are provided to aid interpretation of neonatal iron status. 3 Extracted SDs were converted to 95% CIs. 4 Medians ([IQRs] or 95% CIs) were not included in weighted means. 5 Median (minimum-maximum range). 6 Umbilical cord vein and artery serum means were combined. 7 Median [IQR]. 8 Mean ± SD. 9 Mean (minimum-maximum range). 10 Reference ranges for adults and infants are taken from the NHANES, 1999-2000 (91). Hemoglobin concentrations are provided to aid interpretation of neonatal iron status. 3 Extracted SDs were converted to 95% CIs. 4 Medians ([IQRs] or 95% CIs) were not included in weighted means. 5 Median (minimum-maximum range). 6 AGA group of Hågå (128) can be identified in Supplemental Figure 3B. 7 SGA group of Hågå (128) can be identified in Supplemental Figure 3B. 8 Minimum-maximum range. 9 Ru et al. (120) can be identified in Supplemental Figure 3B. 10 Ru et al. (49) can be identified in Supplemental Figure 3B. 11 30-36 wk group of Sweet et al. (145) can be identified in Supplemental Figure 3B. 12 24-29 wk group of Sweet et al. (145) can be identified in Supplemental Figure 3B. 13 Mean (minimum-maximum range). 14 Reference ranges for adults and infants are taken from the NHANES, 1999-2000 (91).

CURRENT DEVELOPMENTS IN NUTRITION
however, the use of inflammation-inducing forceps, cesarean delivery, or vacuum delivery is not consistently reported in each publication.

Conclusion
Currently available data suggest that hepcidin and serum iron concentrations and TSAT levels for adults and infants are much lower than those found in cord blood and venous blood from neonates during the first month of life. We have strengthened the evidence that FTB neonates possess the ability to produce a hepcidin-mediated hypoferremic response postdelivery. Whether this mechanism is found in PTB neonates is still unclear. This is predominately due to the lack of studies on healthy preterm neonates during the first hours of life. If premature or low-birth-weight neonates are unable to mount a hypoferremic response, this could enhance their risk of early neonatal infections. Conversely, if the hypoferremic response is seen in both preterm and fullterm neonates, it will further support the hypothesis that regulation of iron distribution plays a fundamental role as an innate mechanism of protection against infection. In summary, serum hepcidin is likely triggered by the inflammatory effect of labor and delivery. We suggest that this intrinsic mechanism of protection protects newborns with immature immune systems as they transition from a semi-allogeneic, protected fetal setting to a microberich extrauterine environment (146,147). Hepcidin-induced hypoferremia then potentially provides a broad-action innate bacteriostatic action against invading micro-organisms, when physiological adaption to postnatal life is so critical for survival.