Erythrocyte Zinc Protoporphyrin Is Elevated With Prematurity and Fetal Hypoxemia
http://www.100md.com
《小儿科》
Department of Pediatrics, College of Medicine
Department of Biostatistics, College of Public Health, University of Iowa, Iowa City, Iowa
Department of Laboratory Medicine, University of Washington, Seattle, Washington
Steele Memorial Children's Research Center, University of Arizona College of Medicine, Arizona Health Sciences Center, Tucson, Arizona
ABSTRACT
Objective. To examine the utility of red blood cell (RBC) zinc protoporphyrin/heme ratio (ZnPP/H) as an indicator of fetal iron status, because unfavorable neurodevelopmental outcomes have been associated with poor iron status at birth, as indicated by low serum ferritin, and because few reliable indicators of fetal and early neonatal iron status exist.
Methods. Consecutively studied preterm and term fetuses at delivery included the following groups: (1) control nonhypoxic, (2) fetuses with intrauterine growth retardation (IUGR), and (3) fetuses of insulin-treated mothers (FDM). We hypothesized (1) that rapid growth velocity associated with an accelerated erythropoiesis among normal fetuses will lead to reduced iron delivery to a rapidly expanding RBC mass and higher umbilical cord blood RBC ZnPP/H and (2) that fetuses that are exposed to pathologic hypoxemia will experience an additional increase in erythropoiesis and higher cord ZnPP/H. ZnPP/H was determined on saline-washed cord blood erythrocytes by hematofluorometry and was examined for its relationship with clinical factors and cord blood laboratory measurements indicative of tissue oxygenation (plasma erythropoietin [EPO] and reticulocyte count) and iron status (plasma ferritin and erythrocyte indices). Statistical testing included 1-way analysis of variance, 2-way analysis of variance with covariates, simple linear regression, and multiple regression analysis.
Results. Among control group subjects, gestational age at birth was inversely correlated with RBC ZnPP/H and reticulocyte count and positively correlated with ferritin and EPO. Relative to control subjects, IUGR and FDM fetuses at specified gestational age groupings had higher ZnPP/H, lower plasma ferritin, and higher plasma EPO. Statistical modeling of the relationship between ZnPP/H and plasma ferritin among all study groups demonstrated significant impacts of gestational age, plasma EPO, maternal hypertension, and maternal smoking.
Conclusions. The inverse association of fetal ZnPP/H with gestational age at birth among control subjects is attributable to erythropoietic stimulation likely as a result of increasing growth velocity at the earliest gestational ages. The relatively higher ZnPP/H observed among fetuses in the IUGR and FDM groups likely is attributable to increased erythropoietic activity secondary to pathologic hypoxemia. Decreased placental iron transfer may also have limited iron availability and contributed to elevated ZnPP/H in the IUGR group. These data support the concept that increased erythropoietic activity and/or limited iron transport may place infants of diabetic mothers and infants with growth retardation at risk for developing systemic iron deficiency later in infancy and in early childhood.
Key Words: zinc protoporphyrin fetus hypoxemia iron status ferritin
Abbreviations: IUGR, intrauterine growth retardation FDM, fetuses of insulin-treated mothers RBC, red blood cell Hb, hemoglobin ZnPP/H, zinc protoporphyrin/heme ratio EPO, erythropoietin CRP, C-reactive protein MCV, mean corpuscular volume CH, erythrocyte Hb content CHCM, erythrocyte Hb concentration ANOVA, analysis of variance HBP, hypertension
Fetuses who experience chronic in utero hypoxemia are at risk for death or unfavorable neurodevelopmental outcomes. Potential contributors to this risk include hypoxic-ischemic injury and abnormal iron status.1,2 Examples include fetuses who manifest intrauterine growth retardation (IUGR) and those who are born to mothers with poorly controlled insulin-dependent diabetes (FDM). In response to chronic hypoxemia, red blood cell (RBC) production and hemoglobin (Hb) synthesis are accelerated as iron is preferentially directed toward Hb synthesis, becoming less available for incorporation as an essential co-factor for several enzymes critical for normal growth and development.3 An increasing body of evidence indicates that hypoxemia-induced events in early development result in decreased iron availability with an associated impairment of iron delivery to the developing brain and adverse neurodevelopmental consequences.4–6
Despite concern that systemic iron deficiency, ie, inadequate iron availability for meeting tissue needs, a readily treatable condition postnatally in infants, leads to significant adverse neurodevelopmental consequences, its detection during the neonatal period remains problematic. This is because several developmentally mediated biochemical and hematologic factors that are unique to fetuses and newborn infants confound the interpretation of traditional indicators of iron sufficiency in children and adults, eg, plasma transferrin receptor, plasma ferritin, and erythrocyte indices such as mean cell volume. Hence, additional indicators of iron sufficiency during the fetal and newborn periods are needed.
In marginal or preanemic iron deficiency, zinc is partially substituted for iron during heme synthesis, resulting in formation of zinc protoporphyrin (ZnPP) during RBC production. This metal substitution provides a very sensitive biochemical indicator of iron deficiency.7,8
Because of the utility of the RBC zinc protoporphyrin/heme ratio (ZnPP/H) as a screening tool for iron deficiency in children,9 we wanted to evaluate ZnPP/H as an indicator of iron status at birth. We hypothesized that (1) rapid growth velocity experienced by normal fetuses would increase erythropoiesis to the extent that iron availability becomes limiting, leading to increased ZnPP/H, especially in early gestation, when growth velocity is greatest, and (2) pathologic fetal hypoxemia superimposed on rapid growth, and decreased or limited maternal-to-fetal transplacental iron transport3, would further stimulate RBC production, further limiting iron availability and leading to even greater increases in ZnPP/H.
We tested our 2 hypotheses by examining the relationship of ZnPP/H in the 3 study groups with respect to relevant, clinically available laboratory measurements. The latter included (1) plasma erythropoietin (EPO), a direct indicator of hypoxemia; (2) reticulocyte count, a direct indicator of erythroid activity and an indirect indicator of tissue oxygenation; and (3) plasma ferritin, an indicator of tissue iron stores.9,10 To identify whether inflammation contributed to elevated plasma ferritin levels, we also measured C-reactive protein, a marker of inflammation. Finally, a multiple linear regression model was fitted to test for additional relationships of the clinical and laboratory factors with RBC ZnPP/H.
METHODS
Before beginning the study, we received institutional review board approval. Informed consent was not required because the study relied on residual blood obtained for clinical indications.
Study Subjects and Clinical Environment
Consecutive maternal and infant hospital records for the 3-month period from May to August 2001 were reviewed prospectively for maternal and infant demographic data needed to identify the control and high-risk study groups. Because the study was observational, therapeutic decisions were not altered by study findings. To address our hypothesis that gestational age would be inversely associated with ZnPP/H levels, we included as broad a gestational age spectrum as possible. Because of their limited numbers, all preterm fetuses were enrolled. We included fetuses with IUGR and FDMs as separate study groups because they are often delivered prematurely as a result of fetal and/or maternal complications, placing them at high risk for in utero hypoxemia. The remaining preterm fetuses were assigned to the control group.
Control Group Fetuses
Control group fetuses who were born at term included singleton or dichorionic-diamnionic twins who were born to women without factors that might adversely affect fetal oxygenation (Table 1). Birth weight z score, ie, birth weight standardized for gestational age, was included as a risk factor to avoid the inclusion of undergrown fetuses with IUGR or overgrown FDMs.11 The z scores were derived from the North American sea-level intrauterine growth curves of Arbuckle et al12 by subtracting the expected birth weight for gestational age and gender from the actual fetal weight at birth and then dividing by the SD at the gestational age at birth. Gestational age was determined from ultrasound, when done before 20 weeks, or from menstrual history, if reliable or, if not, from physical examination. Fifty-five term control group infants who were 38 weeks' gestational age were enrolled in the first 2 months, at which point data collection for this subgroup was discontinued, as it was deemed sufficient for analysis.
IUGR Group Fetuses
The IUGR group included fetuses with birth weight z scores that were <–1.75, adjusted for gender and gestational age. Study subjects in this group were permitted to exhibit any of the abnormal criteria listed in Table 1.
FDMs
The FDM group included fetuses who were born to mothers with diabetes that was treated with insulin during their pregnancy. To increase the sample size of the FDM group, we included 10 consecutively born FDMs who were delivered at the University of Arizona during the study period. As was the case with the IUGR group, FDM group subjects were permitted to exhibit any of the abnormal criteria listed in Table 1.
Collection and Handling of Blood Specimens
Umbilical cord whole-blood samples consisted of a mixture of arterial and venous blood collected in ethylenediaminetetraacetic acid as the anticoagulant (Becton Dickinson, Franklin Lakes, NJ) and stored at 4°C for up to 3 weeks before ZnPP/H analysis. Before storage, fresh (<3 days old) aliquots of cord blood were centrifuged, and plasma was frozen at –20°C for later analysis as described below.
Laboratory Analysis Procedures
Whole blood samples for ZnPP/H analysis were centrifuged, and the plasma was removed. The resultant RBC pellets were rinsed in a 10- to 20-fold excess of 0.9% NaCl and recentrifuged, and the supernatant was discarded. The washed RBCs were restored to their initial hematocrit by addition of normal saline. For each study subject, duplicate measurements on 30-μL saline-washed RBC were analyzed for ZnPP/H using the ProtoFluor-Z Hematofluorometer (Helena Laboratories, Inc, Beaumont, TX). Results are reported as a ratio of the micromole of ZnPP to the mole of heme.
Aliquots of fresh blood that was maintained at 4°C for <24 hours were analyzed by flow cytometry for reticulocyte count and combined erythrocyte and reticulocyte indices (ADVIA 120, Bayer Diagnostics, Inc, Tarrytown, NY), ie, erythrocyte mean corpuscular volume (MCV), Hb content (CH), Hb concentration (CHCM), and percentage of hypochromic RBCs. Plasma ferritin was determined using the Rainen Assay System Ferritin [125I] Radioimmunoassay Kit (Catalog no. NEA-078; DuPont, Billerica, MA). Plasma EPO was measured in triplicate using a double antibody radioimmunoassay procedure.13 CRP was measured using a high-sensitivity commercially available enzyme immunoassay kit (ALPCO Diagnostics, Windham, NH). The sensitivity of this assay is 0.05 mg/L, and intra- and interassay coefficients of variation are 6.0% and 11.6%, respectively.
Data Handling and Statistical Analysis
Statistical analysis was performed using SAS software program (Version 9.0; SAS Institute Inc, Cary, NC). Study variables that did not demonstrate normality were log-transformed. For analyses that required log-transformed data portrayed in bar graphs, back-transformed data were shown. Mean ZnPP/H, ferritin, and EPO were compared among the control, IUGR, and FDM groups using the 1-way analysis of variance (ANOVA). For variables that demonstrated significant F values testing for group effect, the Fisher protected least significant difference post hoc comparison was applied. When a sufficient number of study subjects was tested, multiple linear regression modeling was RBC ZnPP in human fetuses at birth performed to (1) determine whether maternal risk factors for fetal hypoxemia demonstrated significant effects on fetal ZnPP/H and (2) examine the relationship between ZnPP/H and the other study variables. The 1-way ANOVA was expanded to a 2-way ANOVA to include gestational age, categorized as <30 weeks, 30 to <35 weeks, and 35 weeks, to be able to compare study groups by gestational age. Because the study included infants with maternal factors (hypertension [HBP], chronic preexisting disease, tobacco use, or drug use) that possibly affect the level of the primary and secondary outcome laboratory variables, each of these factors was included as a covariate in the model. Maternal factors that showed a significant association with fetal iron delivery were included as covariates in the ANOVA and regression models. Test of mean contrast was used to test for selected comparisons of interest involving pair-wise mean comparisons between study groups within each gestational age group and to test for gestational age effect within each study group, with the P values for each set of comparisons adjusted using Bonferroni's method. Simple linear regression was used to investigate relationships among demographic and laboratory variables. 2 test was used in the comparison of categorical variables among the study groups. Descriptive statistics for the continuous variables are expressed as mean ± SEM; an value of P < .05 was considered statistically significant.
RESULTS
Demographics
The potential study population included 278 consecutive newborns who were born during the 3-month study period. Of these, 154 met the selection criteria for inclusion in 1 of the 3 defined study groups and had cord blood samples available for study. Fewer than 1% of infants who were born during this period failed to have cord blood samples available. Gestational age at birth, cord blood CHCM, and maternal age at delivery were similar among the 3 groups (Table 2). Compared with control subjects, the IUGR fetuses were lighter and the FDMs were heavier. None of the FDMs met IUGR group criteria. Twins were more common in the FDM group; C-section delivery was also more common in the FDM group. Of the 11 IUGR fetuses who were delivered by emergency C-section, 9 (82%) were performed for fetal reasons. Six of the 100 plasma CRP values exceeded 1.0 mg/dL; the 2 that exceeded 2.0 mg/dL were 2.4 and 6.9.
Correlation of Clinical and Laboratory Determinations in Control Fetuses
To identify the effect of development not confounded by risk factors, we tested only control group fetuses for the relationship of gestational age with ZnPP/H and other laboratory outcome variables. As indicated in Fig 1, ZnPP/H, plasma ferritin, plasma EPO, and reticulocyte count correlated significantly with gestational age at birth. Clots present in the cord blood sample tubes precluded flow cytometric analysis; therefore, only 28 samples could be analyzed for reticulocyte count. Despite this limitation, MCV and CH indices correlated significantly with gestational age (r = –0.47, P = .011, and r = –0.60, P = .0006, respectively), whereas CHCM and %Hypo RBCs did not (r = –0.27, P = .16, and r = –0.20, P = .32, respectively). Although control group ZnPP/H was significantly correlated with both reticulocyte count and plasma ferritin (Fig 2), no linear relationship was observed with EPO (r = 0.01, P = .96).
Comparisons and Correlations of Laboratory Variables Among Groups
When RBC ZnPP/H and important secondary laboratory outcome variables were compared among groups, group post hoc analysis indicated that the FDM group differed significantly from the control and IUGR groups for ZnPP/H, plasma ferritin, and plasma EPO but not for reticulocyte count (Fig 3). Although the control and IUGR groups demonstrated intermediate differences for ZnPP/H and plasma ferritin in the same direction observed for the control and FDM groups, these differences did not achieve statistical significance in the post hoc analyses.
Because of limited availability of flow cytometric data, RBC indices for premature fetuses could be analyzed meaningfully only for those who were >35 weeks' gestation (Fig 4). Mean MCV and the percentage of hypochromic RBCs were significantly greater in FDMs compared with the control group, and hypochromia was greater in the FDMs compared with the IUGR group. The 3 study groups did not differ significantly in mean CH and CHCM.
The correlation of laboratory parameters in the FDM and IUGR groups were in some ways similar to the control group fetuses but different in others. The FDM and IUGR groups both demonstrated significant inverse relationships of ZnPP/H and ferritin with higher correlation coefficients than the control group: r = –0.70, P = .002, n = 17; and r = –0.67, P < .0001, n = 36, respectively. Although a direct relationship was observed for ZnPP/H and EPO in both the FDM and IUGR groups, this reached statistical significance only for the IUGR group with its larger number of available subjects (r = 0.70, P = .002, n = 17, compared with r = 0.38, P = .04, n = 29 for the FDM group). The control group's significant direct association of ZnPP/H and reticulocyte count was not observed for either the FDM or the IUGR group: r = 0.41, P = .18, n = 12; and r = 0.17, P = .63, n = 10, respectively.
Comparisons of Laboratory Variables Among Groups by Gestational Age
When the 3 study groups were separated into 3 gestational age periods, significant group mean differences were observed for ZnPP/H, plasma ferritin, and plasma EPO. The control and IUGR groups demonstrated decreasing mean ZnPP/H ratios with increasing gestational age, with the IUGR group having significantly higher means for both the <30 weeks and 30 to 35 weeks intervals (Fig 5, A). In contrast, the FDM group had a mean ZnPP/H that increased with advancing gestation, such that the ZnPP/H at the >35 weeks interval was significantly higher compared with the other 2 groups.
In each of the 3 gestational age periods, plasma ferritin demonstrated greater relative variability and fewer among-group differences than ZnPP/H (Fig 5, B). The only significant differences were observed for the gestational interval >35 weeks, for which the mean ferritin level in the FDM group was lower relative to the other 2 groups.
The control and FDM groups demonstrated increasing mean plasma EPO levels with increasing gestational age, during which the mean EPO levels decreased among IUGR fetuses (Fig 5, C). The FDM group's mean EPO level was significantly higher than that in the control group at both 30 to 35 weeks and >35 weeks and higher than the IUGR group at >35 weeks.
Relationship Between ZnPP/H and Ferritin Using Modeled Covariates
In addressing our study's objective of evaluating ZnPP/H as an indicator of iron status at birth, the important and highly significant inverse relationship observed for cord blood ZnPP/H and plasma ferritin was examined using multiple linear regression analysis. The fitted regression model, which explained 64.2% (R2) of the variation in ZnPP/H, showed that in addition to plasma ferritin, ZnPP/H was significantly influenced by plasma EPO concentration and several clinical variables, including gestational age, maternal HBP, and maternal smoking (Fig 6). This model also identified a significant interaction between ferritin and maternal HBP (P = .0007), which indicated that the degree of association of ferritin with ZnPP/H varied with presence or absence of maternal HBP, and, likewise, the effect of maternal HBP on mean ZnPP/H was dependent on the ferritin level.
The multiple regression analysis showed a significant inverse relationship between ferritin and ZnPP/H with an estimated decrease of 29.1 ± 2.5% in mean ZnPP/H for a 10-fold increase in ferritin (P < .0001) among fetuses whose mothers did not manifest HBP. Among fetuses whose mothers manifested HBP, an even greater effect of ferritin was observed: a 10-fold increase in ferritin resulted in a 54.8 ± 2.0% drop in mean ZnPP/H (P < .0001).
The other laboratory measure that significantly affected ZnPP/H was EPO level. From the estimate of the EPO slope in the regression model, a 10-fold increase in EPO produced a 16.1 ± 0.7% (P = .007) increase in mean ZnPP/H. Although the significant effect of plasma EPO on the inverse relationship of plasma ferritin and ZnPP/H for all 3 study groups seems to contradict that suggested by Fig 1, A and C, for the control group alone, inclusion of the higher EPO values present from the 2 hypoxemic groups (see Fig 5, C) resulted in a small but significant upward deflection in the ZnPP/H versus ferritin relationship with increases in plasma EPO levels.
The clinical factors that influenced ZnPP/H were gestational age, maternal tobacco use, and maternal HBP. ZnPP/H decreased significantly with increasing gestational age, with mean ZnPP/H for fetuses who were >35 weeks being 47.5 ± 2.4% lower than for those who were <30 weeks' gestational age (P < .0001). Maternal tobacco use significantly increased mean ZnPP/H by 39 ± 4.6% (P < .0001). The effect of maternal hypertension on ZnPP/H significantly lessened with higher ferritin. When ferritin is low (eg, at 25th percentile, or 111 μg/L for this study), presence of maternal HBP significantly increased mean ZnPP/H by 17.6 ± 3.3% (P = .013). In contrast, when ferritin is high (eg, at 75th percentile, or 280 μg/L for this study), maternal HBP showed no significant effect on ZnPP/H (a decrease of 1.9 ± 3.4%; P = .813).
DISCUSSION
The present study's RBC ZnPP/H data at birth are consistent with and extend the work of others, demonstrating that mean ZnPP/H ratios in normal fetuses are higher than those in normal adults.10,14 More important, our data reinforce and extend the report of Juul et al14 that during the first week of life, ZnPP/H in normal fetuses is negatively associated with gestational age and is elevated further in those who are at high risk for in utero hypoxemia (ie, IUGR and FDM fetuses). Ours is the first study to demonstrate (1) that reticulocyte count in cord blood is positively associated with ZnPP/H and negatively associated with ferritin and (2) that the ZnPP/H–ferritin relationship at birth is affected by clinical conditions and laboratory variables indicative of fetal hypoxemia. Taken together, our findings indicate that ZnPP/H at birth reflects increased erythropoiesis accompanying normal growth and development; when chronic in utero hypoxemia is present, our observation of even higher fetal ZnPP/H levels raises the possibility that systemic iron availability to meet tissue needs, including the brain's, may have been limiting.4–6 The relationship of these findings to adverse neurodevelopmental outcome and the advantage that ZnPP/H may offer over the use of cord plasma ferritin15 have yet to be addressed.
Relationship of ZnPP/H to Iron Status in Normal Fetuses
Although the accumulated clinical and biochemical information indicates that the relationship between erythropoiesis and iron stores is closely related,8 the cumulative data of the present study support the speculation that high RBC ZnPP/H observed among normoxemic, iron-replete fetuses reflects the normal increase in erythropoiesis observed in fetuses relative to adults. This speculation is supported by data for iron-replete children and adults who experience increased erythropoiesis.7,16 That the control group data demonstrated a significant positive relationship between ZnPP/H and reticulocyte count supports the role of erythropoiesis leading to increased ZnPP; however, the negative association of ZnPP/H with plasma ferritin suggests that limited iron availability is a possible explanation. Because of limited reticulocyte data, it was not possible to determine statistically the dominant contributor to the increased ZnPP levels.
The normal fetal Hb levels and RBC indices observed among control group subjects further indicate that high RBC ZnPP/H does not necessarily represent a pathologic condition. We speculate that the greatest ZnPP/H levels were observed among the least mature control group fetuses because this group experienced the most rapid growth velocity and as such concomitantly experienced the greatest demand for tissue iron. Indeed, the normal rise that we and others have observed for fetal plasma ferritin with advancing gestational age may be the result of less intense erythropoiesis as gestation advances.17,18 Results of the present study do not rule out the possibility that nonspecific differences, ie, the concentration of fetal hemoglobin, in fetal RBCs as compared with the more adult-like RBCs produced later in infancy may lead to an artifactual increase in fetal ZnPP/H levels.
Response of RBC ZnPP/H to In Utero Stress and Hypoxemia
Ours is the first study to demonstrate that RBC ZnPP/H and the ZnPP/H–ferritin relationship at birth is perturbed by clinical and laboratory variables associated with in utero hypoxemia. Maternal HBP, maternal smoking, fetal hypoxemia (indicated by plasma EPO levels), and fetal immaturity each exerted an independent stimulatory effect that led to increased fetal ZnPP/H. Although the normal fetal increase in erythropoiesis relative to children and adults was the key stimulatory factor among control subjects, stress erythropoiesis, a vitally important homeostatic process from a survival standpoint, likely played an important role in the presence of fetal hypoxemia, as was observed in the IUGR and FDM groups. The normal homeostatic response of the fetus to chronic hypoxemia is to increase erythropoiesis for eventually increasing tissue oxygen delivery. Doing so places an increased demand on systemic iron stores to provide iron for Hb synthesis. Because under these conditions iron is preferentially directed toward Hb synthesis,5 tissue iron stores decrease, making less iron available for normal fetal growth and development. Increased RBC production coupled with limited or decreased maternal-to-fetal transplacental iron transport places the fetus at risk for inadequate iron availability to sustain essential metabolism.3 The evidence to date suggests that this sequence of events may contribute to serious life-long neurodevelopmental sequelae15 or death.
Relative to the control group subjects, fetuses with IUGR demonstrated increased ZnPP/H and plasma EPO with a tendency toward having low plasma ferritin levels. The higher ZnPP/H values that we observed are qualitatively consistent with the findings of Juul et al's14 in neonates during the first week of life. The latter group's higher mean ZnPP/H of 178 compared with the 78 μmol/mol heme observed in the present study was likely attributable to Juul et al's inclusion of earlier gestational age fetuses (33.9 vs 37.1 weeks). Although we did not attempt to determine why individual fetuses in the present study had IUGR, the underlying pathophysiology of this condition is most commonly attributable to utero-placental insufficiency. Autopsy studies in fetuses with IUGR indicate that their brain and liver iron content are decreased by 33% and 74%, respectively.19 In addition, reduced oxygen and nutrient transport by a typically small dysfunctional placenta results in fetuses' struggling to achieve normal growth while being in a chronically hypoxemic environment, as indicated by their markedly elevated plasma EPO levels.20–22 The inability of fetuses with IUGR to satisfy the already increased iron requirements of normal growth tends to increase erythropoiesis to a greater degree than normal. One result is even higher ZnPP/H levels than normal fetuses. Unfortunately, there were too few early-gestation fetuses with IUGR in the present study to permit an adequate assessment as to whether their erythropoiesis was increased. Nonetheless, the trend toward exceptionally high EPO levels among this early-gestation subgroup of fetuses with IUGR and that delivery was often necessitated on the basis of concern for fetal well-being suggest that their erythropoiesis may have been increased. As a result of their hypoxemic intrauterine environment, iron stores in fetuses with IUGR are frequently decreased at birth,23 predisposing these infants to early onset of iron deficiency,19 with some going on to manifest impaired neurodevelopmental function in infancy and childhood.15
Like the fetuses with IUGR, FDMs demonstrated increased ZnPP/H, plasma EPO, % hypochromic RBCs, and MCV relative to control fetuses, whereas plasma ferritin levels were decreased. Although statistically significant greater reticulocyte counts were not observed among FDM group subjects, whole-blood reticulocyte counts did not reflect the greater increase in growth and large blood volume experienced by this group. Taken together, this constellation of findings at birth among FDMs combined with those reported by others suggests a state of chronic fetal hypoxemia leading to the cascade of events beginning with increased EPO production and plasma EPO levels,1,24 leading to increased erythropoiesis and subsequent iron utilization, and ultimately resulting in diminished iron availability for Hb synthesis and decreased iron stores and brain iron content.13,24,25 In short, fetal hypoxemia is the result of accelerated metabolism resulting in increased fetal oxygen consumption in response to hyperglycemia-hyperinsulinemia brought about by diabetes-mediated hyperglycemia.26 Limited placental iron transport may also play a role in the lack of iron availability as indicated by increased ZnPP/H and low plasma ferritin.24,25 Recent psychomotor developmental data in FDMs indicate that resultant depletion of iron stores in fetal liver, heart, and brain as suggested by the present data and as verified by others in FDM autopsy studies may have neurodevelopmental consequences.27 This constellation of clinical features and increased number of stillbirths are especially true among macrosomic FDMs whose mothers' diabetes was poorly controlled.
Consistent with our findings, Juul et al14 reported that ZnPP/H is increased in the first week of life among FDMs. The considerably higher absolute FDM ZnPP/H reported by our group (191 vs 104 μmol/mol heme reported by Juul et al) was likely the result of earlier gestational age (34.2 vs 36.6 weeks) and poorer maternal diabetic control, because all of the FDMs in their study were admitted to the NICU, whereas ours were not. The ancillary laboratory and clinical data included in the present study and incorporated into the modeling analysis have allowed us more ably to suggest mechanistic explanations for our study's ZnPP/H findings.14
Study Limitations and Remaining Questions Regarding ZnPP/H in the Perinatal Period
A major limitation of our study is the lack of statistical power for identifying the relative roles of erythropoiesis and iron availability as contributors to the increased RBC ZnPP/H observed. This shortcoming was attributable primarily to the paucity of flow cytometric hematologic data, although the limited numbers of study subjects in some of the gestational age groupings as well as missing EPO data also played a role. In addition, the statistical modeling was complicated by the fact that the term control group did not exhibit the abnormal criteria present in the preterm control and the IUGR and FDM groups.
Longitudinal clinical and preclinical studies are needed for addressing key remaining questions regarding RBC ZnPP/H in the perinatal period: (1) What additional prenatal and postnatal factors, eg, RBC transfusion, oral iron therapy, parenteral iron therapy, EPO treatment, and events, affect ZnPP/H (2) What roles do ZnPP/H and ferritin play in predicting neurodevelopmental outcome, and, if both are associated, is one better than the other (3) Why are ZnPP/H measurements so variable from individual to individual Use of a recently described neonatal rat model for studying ZnPP/H28 offers an opportunity for directly measuring tissue iron in different organs but would require the development of analytic techniques for measuring plasma ferritin and plasma transferrin receptor.
CONCLUSIONS
On the basis of the present findings and those of Juul et al,14 we recommend during the immediate perinatal period that the use of RBC ZnPP/H data as an indicator of iron status be interpreted with reference to values derived from normal fetuses and neonates. The findings of our study indicate that elevated RBC ZnPP/H in cord blood at birth is associated with increased erythropoiesis compared with later life, not with functional or absolute iron deficiency. We base our conclusion on the observation that in the control group, umbilical cord plasma ferritin levels were in the ferritin range deemed desirable for adults who receive EPO treatment, yet this group had normal cord blood Hb levels and RBC iron indices. When viewed in the context of their laboratory and clinical findings, the elevated RBC ZnPP/H observed in fetuses at high risk for hypoxemia indicate functional iron deficiency. Whether they indicate true systemic iron deficiency with the potential for adverse neurodevelopment consequences remains to be established.
ACKNOWLEDGMENTS
This study was supported by grant RR00059 from the General Clinical Research Centers Program, National Center for Research Resources; National Institutes of Health Grant DK35816; and NIH P01 HL46925. The University of Arizona portion was supported by the American Heart Association, Arizona Affiliate AZGS-37-96 and Southwest Affiliate SW-GS-16-98 (Dr Kling).
We acknowledge the technical assistance and data collection contributions of Gretchen Cress, RN, Karen J. Johnson, RN, Erin McGuire, BS, Robert Schmidt, BS, Suzan J. Hays, Karen B. Lesser, MD, Carrie R. Daniel, BS, Ronald B. Schifman, MD, and Megan Dyer. Employees of the hospital blood bank are acknowledged for assistance with the storage of blood samples, and Mark A. Hart is acknowledged for secretarial help. Michael Georgieff, MD, provided helpful suggestions on data analysis and interpretation.
FOOTNOTES
Accepted Nov 23, 2004.
No conflict of interest declared.
REFERENCES
Schwartz R, Teramo KA. Effects of diabetic pregnancy on the fetus and newborn. Sem Perinatol. 2000;24 :120 –135
Gardosi J, Mul T, Mongelli M, Fagan D. Analysis of birthweight and gestational age in antepartum stillbirths. Br J Obstet Gynecol. 1998;105 :524 –530
Guiang SF, Georgieff MK. Fetal and neonatal iron metabolism. In: Polin RA, Fox WW, eds. Fetal and Neonatal Physiology. 2nd ed. Philadelphia, PA: WB Saunders Company; 1998:401 –410
Lozoff B. Perinatal iron deficiency and the developing brain. Pediatr Res. 2000;48 :137 –139
Rao R, Georgieff MK. Perinatal aspects of iron metabolism. Acta Paediatr Suppl. 2002;91 :124 –129
Sesma HW, Georgieff MK. The effect of adverse intrauterine and newborn environments on cognitive development: the experiences of premature delivery and diabetes during pregnancy. Dev Psychopathol. 2003;15 :991 –1015
Labbé RF, Vreman HJ, Stevenson DK. Zinc protoporphyrin: metabolite with a mission. Clin Chem. 1999;45 :260 –272
Labbé RF, Dewanji A. Iron assessment tests: transferrin receptor vis-a-vis zinc protoporphyrin. Clin Biochem. 2004;37 :165 –174
Harthoorn-Lasthuizen EJ, Lindemans J, Langenhuijsen MMAC. Erythrocyte zinc protoporphyrin testing in pregnancy. Acta Obstet Gynecol Scand. 2000;79 :660 –666
Griffin IJ, Reid MM, McCormick KP, Cooke RJ. Zinc protoporphyrin/haem ratio and plasma ferritin in preterm infants. Arch Dis Child Fetal Neonatal Ed. 2002;87 :F49 –F51
Arbuckle TE, Wilkins R, Sherman GJ. Birth weight percentiles by gestational age in Canada. Obstet Gynecol. 1993;81 :39 –48
Georgieff MK, Widness JA, Mills MM, Stonestreet BS. The effect of prolonged intrauterine hyperinsulinemia on iron utilization in fetal sheep. Pediatr Res. 1989;26 :467 –469
Juul SE, Zerzan JC, Strandjord TP, Woodrum DE. Zinc protoporphyrin/heme as an indicator of iron status in NICU patients. J Pediatr. 2003;142 :273 –278
Tamura T, Goldenberg RL, Hou J, et al. Cord serum ferritin concentrations and mental and psychomotor development of children at five years of age. J Pediatr. 2002;140 :165 –170
NCCLS. Erythrocyte Protoporphyrin Testing: Approved Guideline (NCCLS Document C42-A). Villanova, PA: National Committee on Clinical Laboratory Standards; 1996
Siimes AS, Siimes MA. Changes in the concentration of ferritin in the serum during fetal life in singletons and twins. Early Hum Dev. 1986;13 :47 –52
Sweet DG, Savage GA, Tubman R, Lappin TR, Halliday HL. Cord blood transferrin receptors to assess fetal iron status. Arch Dis Child Fetal Neonatal Ed. 2001;85 :F46 –F48
Georgieff MK. Fetal and neonatal iron metabolism. Semin Neonatal Nutr Metab. 2000;7 :1 –5
Ruth V, Autti-Ramo I, Granstrom ML, Korkcan M, Raivio KO. Prediction of perinatal brain damage by cord plasma vasopressin, erythropoietin, and hypoxanthine values. J Pediatr. 1988;113 :880 –885
Widness JA, Weiner CP, Giller RH, Kling PJ, Schmidt RL, Fomon SJ. Increasing plasma erythropoietin response to anemia during development. Pediatr Res. 1990;27 :153A
Teramo KA, Hiilesmaa VK, Schwartz R, Clemons GK, Widness JA. Amniotic fluid and cord plasma erythropoietin levels in pregnancies complicated by preeclampsia, pregnancy-induced hypertension and chronic hypertension. J Perinatal Med. 2004;32 :240 –247
Chockalingam UM, Murphy E, Ophoven JC, Georgief MK. Cord transferrin and ferritin values in newborn infants at risk for prenatal uteroplacental insufficiency and chronic hypoxia. J Pediatr. 1987;111 :283 –286
Georgieff MK, Landon MB, Mills MM, et al. Abnormal iron distribution in infants of diabetic mothers: spectrum and maternal antecedents. J Pediatr. 1990;117 :455 –461
Amarnath UM, Ophoven JJ, Mills MM, Murphy EL, Georgieff MK. The relationship between decreased iron stores, serum iron and neonatal hypoglycemia in large-for-date newborn infants. Acta Paediatr Scand. 1989;78 :538 –543
Madsen H. Fetal oxygenation in diabetic pregnancy. With special reference to maternal blood oxygen affinity and its effectors. Dan Med Bull. 1986;33 :64 –74
Siddappa AM, Georgieff MK, Wewerka S, Worwa C, Nelson CA, deRengnier RA. Iron deficiency alters auditory recognition memory in newborn infants of diabetic mothers. Pediatr Res. 2004;55 :1034 –1041
Dubuque SH, Dvorak B, Woodward SS, McCuskey RS, Kling PJ. Iron-deficient erythropoiesis in neonatal rats. Biol Neonate. 2002;81 :51 –57(David G. Lott, MD, M. Bri)
Department of Biostatistics, College of Public Health, University of Iowa, Iowa City, Iowa
Department of Laboratory Medicine, University of Washington, Seattle, Washington
Steele Memorial Children's Research Center, University of Arizona College of Medicine, Arizona Health Sciences Center, Tucson, Arizona
ABSTRACT
Objective. To examine the utility of red blood cell (RBC) zinc protoporphyrin/heme ratio (ZnPP/H) as an indicator of fetal iron status, because unfavorable neurodevelopmental outcomes have been associated with poor iron status at birth, as indicated by low serum ferritin, and because few reliable indicators of fetal and early neonatal iron status exist.
Methods. Consecutively studied preterm and term fetuses at delivery included the following groups: (1) control nonhypoxic, (2) fetuses with intrauterine growth retardation (IUGR), and (3) fetuses of insulin-treated mothers (FDM). We hypothesized (1) that rapid growth velocity associated with an accelerated erythropoiesis among normal fetuses will lead to reduced iron delivery to a rapidly expanding RBC mass and higher umbilical cord blood RBC ZnPP/H and (2) that fetuses that are exposed to pathologic hypoxemia will experience an additional increase in erythropoiesis and higher cord ZnPP/H. ZnPP/H was determined on saline-washed cord blood erythrocytes by hematofluorometry and was examined for its relationship with clinical factors and cord blood laboratory measurements indicative of tissue oxygenation (plasma erythropoietin [EPO] and reticulocyte count) and iron status (plasma ferritin and erythrocyte indices). Statistical testing included 1-way analysis of variance, 2-way analysis of variance with covariates, simple linear regression, and multiple regression analysis.
Results. Among control group subjects, gestational age at birth was inversely correlated with RBC ZnPP/H and reticulocyte count and positively correlated with ferritin and EPO. Relative to control subjects, IUGR and FDM fetuses at specified gestational age groupings had higher ZnPP/H, lower plasma ferritin, and higher plasma EPO. Statistical modeling of the relationship between ZnPP/H and plasma ferritin among all study groups demonstrated significant impacts of gestational age, plasma EPO, maternal hypertension, and maternal smoking.
Conclusions. The inverse association of fetal ZnPP/H with gestational age at birth among control subjects is attributable to erythropoietic stimulation likely as a result of increasing growth velocity at the earliest gestational ages. The relatively higher ZnPP/H observed among fetuses in the IUGR and FDM groups likely is attributable to increased erythropoietic activity secondary to pathologic hypoxemia. Decreased placental iron transfer may also have limited iron availability and contributed to elevated ZnPP/H in the IUGR group. These data support the concept that increased erythropoietic activity and/or limited iron transport may place infants of diabetic mothers and infants with growth retardation at risk for developing systemic iron deficiency later in infancy and in early childhood.
Key Words: zinc protoporphyrin fetus hypoxemia iron status ferritin
Abbreviations: IUGR, intrauterine growth retardation FDM, fetuses of insulin-treated mothers RBC, red blood cell Hb, hemoglobin ZnPP/H, zinc protoporphyrin/heme ratio EPO, erythropoietin CRP, C-reactive protein MCV, mean corpuscular volume CH, erythrocyte Hb content CHCM, erythrocyte Hb concentration ANOVA, analysis of variance HBP, hypertension
Fetuses who experience chronic in utero hypoxemia are at risk for death or unfavorable neurodevelopmental outcomes. Potential contributors to this risk include hypoxic-ischemic injury and abnormal iron status.1,2 Examples include fetuses who manifest intrauterine growth retardation (IUGR) and those who are born to mothers with poorly controlled insulin-dependent diabetes (FDM). In response to chronic hypoxemia, red blood cell (RBC) production and hemoglobin (Hb) synthesis are accelerated as iron is preferentially directed toward Hb synthesis, becoming less available for incorporation as an essential co-factor for several enzymes critical for normal growth and development.3 An increasing body of evidence indicates that hypoxemia-induced events in early development result in decreased iron availability with an associated impairment of iron delivery to the developing brain and adverse neurodevelopmental consequences.4–6
Despite concern that systemic iron deficiency, ie, inadequate iron availability for meeting tissue needs, a readily treatable condition postnatally in infants, leads to significant adverse neurodevelopmental consequences, its detection during the neonatal period remains problematic. This is because several developmentally mediated biochemical and hematologic factors that are unique to fetuses and newborn infants confound the interpretation of traditional indicators of iron sufficiency in children and adults, eg, plasma transferrin receptor, plasma ferritin, and erythrocyte indices such as mean cell volume. Hence, additional indicators of iron sufficiency during the fetal and newborn periods are needed.
In marginal or preanemic iron deficiency, zinc is partially substituted for iron during heme synthesis, resulting in formation of zinc protoporphyrin (ZnPP) during RBC production. This metal substitution provides a very sensitive biochemical indicator of iron deficiency.7,8
Because of the utility of the RBC zinc protoporphyrin/heme ratio (ZnPP/H) as a screening tool for iron deficiency in children,9 we wanted to evaluate ZnPP/H as an indicator of iron status at birth. We hypothesized that (1) rapid growth velocity experienced by normal fetuses would increase erythropoiesis to the extent that iron availability becomes limiting, leading to increased ZnPP/H, especially in early gestation, when growth velocity is greatest, and (2) pathologic fetal hypoxemia superimposed on rapid growth, and decreased or limited maternal-to-fetal transplacental iron transport3, would further stimulate RBC production, further limiting iron availability and leading to even greater increases in ZnPP/H.
We tested our 2 hypotheses by examining the relationship of ZnPP/H in the 3 study groups with respect to relevant, clinically available laboratory measurements. The latter included (1) plasma erythropoietin (EPO), a direct indicator of hypoxemia; (2) reticulocyte count, a direct indicator of erythroid activity and an indirect indicator of tissue oxygenation; and (3) plasma ferritin, an indicator of tissue iron stores.9,10 To identify whether inflammation contributed to elevated plasma ferritin levels, we also measured C-reactive protein, a marker of inflammation. Finally, a multiple linear regression model was fitted to test for additional relationships of the clinical and laboratory factors with RBC ZnPP/H.
METHODS
Before beginning the study, we received institutional review board approval. Informed consent was not required because the study relied on residual blood obtained for clinical indications.
Study Subjects and Clinical Environment
Consecutive maternal and infant hospital records for the 3-month period from May to August 2001 were reviewed prospectively for maternal and infant demographic data needed to identify the control and high-risk study groups. Because the study was observational, therapeutic decisions were not altered by study findings. To address our hypothesis that gestational age would be inversely associated with ZnPP/H levels, we included as broad a gestational age spectrum as possible. Because of their limited numbers, all preterm fetuses were enrolled. We included fetuses with IUGR and FDMs as separate study groups because they are often delivered prematurely as a result of fetal and/or maternal complications, placing them at high risk for in utero hypoxemia. The remaining preterm fetuses were assigned to the control group.
Control Group Fetuses
Control group fetuses who were born at term included singleton or dichorionic-diamnionic twins who were born to women without factors that might adversely affect fetal oxygenation (Table 1). Birth weight z score, ie, birth weight standardized for gestational age, was included as a risk factor to avoid the inclusion of undergrown fetuses with IUGR or overgrown FDMs.11 The z scores were derived from the North American sea-level intrauterine growth curves of Arbuckle et al12 by subtracting the expected birth weight for gestational age and gender from the actual fetal weight at birth and then dividing by the SD at the gestational age at birth. Gestational age was determined from ultrasound, when done before 20 weeks, or from menstrual history, if reliable or, if not, from physical examination. Fifty-five term control group infants who were 38 weeks' gestational age were enrolled in the first 2 months, at which point data collection for this subgroup was discontinued, as it was deemed sufficient for analysis.
IUGR Group Fetuses
The IUGR group included fetuses with birth weight z scores that were <–1.75, adjusted for gender and gestational age. Study subjects in this group were permitted to exhibit any of the abnormal criteria listed in Table 1.
FDMs
The FDM group included fetuses who were born to mothers with diabetes that was treated with insulin during their pregnancy. To increase the sample size of the FDM group, we included 10 consecutively born FDMs who were delivered at the University of Arizona during the study period. As was the case with the IUGR group, FDM group subjects were permitted to exhibit any of the abnormal criteria listed in Table 1.
Collection and Handling of Blood Specimens
Umbilical cord whole-blood samples consisted of a mixture of arterial and venous blood collected in ethylenediaminetetraacetic acid as the anticoagulant (Becton Dickinson, Franklin Lakes, NJ) and stored at 4°C for up to 3 weeks before ZnPP/H analysis. Before storage, fresh (<3 days old) aliquots of cord blood were centrifuged, and plasma was frozen at –20°C for later analysis as described below.
Laboratory Analysis Procedures
Whole blood samples for ZnPP/H analysis were centrifuged, and the plasma was removed. The resultant RBC pellets were rinsed in a 10- to 20-fold excess of 0.9% NaCl and recentrifuged, and the supernatant was discarded. The washed RBCs were restored to their initial hematocrit by addition of normal saline. For each study subject, duplicate measurements on 30-μL saline-washed RBC were analyzed for ZnPP/H using the ProtoFluor-Z Hematofluorometer (Helena Laboratories, Inc, Beaumont, TX). Results are reported as a ratio of the micromole of ZnPP to the mole of heme.
Aliquots of fresh blood that was maintained at 4°C for <24 hours were analyzed by flow cytometry for reticulocyte count and combined erythrocyte and reticulocyte indices (ADVIA 120, Bayer Diagnostics, Inc, Tarrytown, NY), ie, erythrocyte mean corpuscular volume (MCV), Hb content (CH), Hb concentration (CHCM), and percentage of hypochromic RBCs. Plasma ferritin was determined using the Rainen Assay System Ferritin [125I] Radioimmunoassay Kit (Catalog no. NEA-078; DuPont, Billerica, MA). Plasma EPO was measured in triplicate using a double antibody radioimmunoassay procedure.13 CRP was measured using a high-sensitivity commercially available enzyme immunoassay kit (ALPCO Diagnostics, Windham, NH). The sensitivity of this assay is 0.05 mg/L, and intra- and interassay coefficients of variation are 6.0% and 11.6%, respectively.
Data Handling and Statistical Analysis
Statistical analysis was performed using SAS software program (Version 9.0; SAS Institute Inc, Cary, NC). Study variables that did not demonstrate normality were log-transformed. For analyses that required log-transformed data portrayed in bar graphs, back-transformed data were shown. Mean ZnPP/H, ferritin, and EPO were compared among the control, IUGR, and FDM groups using the 1-way analysis of variance (ANOVA). For variables that demonstrated significant F values testing for group effect, the Fisher protected least significant difference post hoc comparison was applied. When a sufficient number of study subjects was tested, multiple linear regression modeling was RBC ZnPP in human fetuses at birth performed to (1) determine whether maternal risk factors for fetal hypoxemia demonstrated significant effects on fetal ZnPP/H and (2) examine the relationship between ZnPP/H and the other study variables. The 1-way ANOVA was expanded to a 2-way ANOVA to include gestational age, categorized as <30 weeks, 30 to <35 weeks, and 35 weeks, to be able to compare study groups by gestational age. Because the study included infants with maternal factors (hypertension [HBP], chronic preexisting disease, tobacco use, or drug use) that possibly affect the level of the primary and secondary outcome laboratory variables, each of these factors was included as a covariate in the model. Maternal factors that showed a significant association with fetal iron delivery were included as covariates in the ANOVA and regression models. Test of mean contrast was used to test for selected comparisons of interest involving pair-wise mean comparisons between study groups within each gestational age group and to test for gestational age effect within each study group, with the P values for each set of comparisons adjusted using Bonferroni's method. Simple linear regression was used to investigate relationships among demographic and laboratory variables. 2 test was used in the comparison of categorical variables among the study groups. Descriptive statistics for the continuous variables are expressed as mean ± SEM; an value of P < .05 was considered statistically significant.
RESULTS
Demographics
The potential study population included 278 consecutive newborns who were born during the 3-month study period. Of these, 154 met the selection criteria for inclusion in 1 of the 3 defined study groups and had cord blood samples available for study. Fewer than 1% of infants who were born during this period failed to have cord blood samples available. Gestational age at birth, cord blood CHCM, and maternal age at delivery were similar among the 3 groups (Table 2). Compared with control subjects, the IUGR fetuses were lighter and the FDMs were heavier. None of the FDMs met IUGR group criteria. Twins were more common in the FDM group; C-section delivery was also more common in the FDM group. Of the 11 IUGR fetuses who were delivered by emergency C-section, 9 (82%) were performed for fetal reasons. Six of the 100 plasma CRP values exceeded 1.0 mg/dL; the 2 that exceeded 2.0 mg/dL were 2.4 and 6.9.
Correlation of Clinical and Laboratory Determinations in Control Fetuses
To identify the effect of development not confounded by risk factors, we tested only control group fetuses for the relationship of gestational age with ZnPP/H and other laboratory outcome variables. As indicated in Fig 1, ZnPP/H, plasma ferritin, plasma EPO, and reticulocyte count correlated significantly with gestational age at birth. Clots present in the cord blood sample tubes precluded flow cytometric analysis; therefore, only 28 samples could be analyzed for reticulocyte count. Despite this limitation, MCV and CH indices correlated significantly with gestational age (r = –0.47, P = .011, and r = –0.60, P = .0006, respectively), whereas CHCM and %Hypo RBCs did not (r = –0.27, P = .16, and r = –0.20, P = .32, respectively). Although control group ZnPP/H was significantly correlated with both reticulocyte count and plasma ferritin (Fig 2), no linear relationship was observed with EPO (r = 0.01, P = .96).
Comparisons and Correlations of Laboratory Variables Among Groups
When RBC ZnPP/H and important secondary laboratory outcome variables were compared among groups, group post hoc analysis indicated that the FDM group differed significantly from the control and IUGR groups for ZnPP/H, plasma ferritin, and plasma EPO but not for reticulocyte count (Fig 3). Although the control and IUGR groups demonstrated intermediate differences for ZnPP/H and plasma ferritin in the same direction observed for the control and FDM groups, these differences did not achieve statistical significance in the post hoc analyses.
Because of limited availability of flow cytometric data, RBC indices for premature fetuses could be analyzed meaningfully only for those who were >35 weeks' gestation (Fig 4). Mean MCV and the percentage of hypochromic RBCs were significantly greater in FDMs compared with the control group, and hypochromia was greater in the FDMs compared with the IUGR group. The 3 study groups did not differ significantly in mean CH and CHCM.
The correlation of laboratory parameters in the FDM and IUGR groups were in some ways similar to the control group fetuses but different in others. The FDM and IUGR groups both demonstrated significant inverse relationships of ZnPP/H and ferritin with higher correlation coefficients than the control group: r = –0.70, P = .002, n = 17; and r = –0.67, P < .0001, n = 36, respectively. Although a direct relationship was observed for ZnPP/H and EPO in both the FDM and IUGR groups, this reached statistical significance only for the IUGR group with its larger number of available subjects (r = 0.70, P = .002, n = 17, compared with r = 0.38, P = .04, n = 29 for the FDM group). The control group's significant direct association of ZnPP/H and reticulocyte count was not observed for either the FDM or the IUGR group: r = 0.41, P = .18, n = 12; and r = 0.17, P = .63, n = 10, respectively.
Comparisons of Laboratory Variables Among Groups by Gestational Age
When the 3 study groups were separated into 3 gestational age periods, significant group mean differences were observed for ZnPP/H, plasma ferritin, and plasma EPO. The control and IUGR groups demonstrated decreasing mean ZnPP/H ratios with increasing gestational age, with the IUGR group having significantly higher means for both the <30 weeks and 30 to 35 weeks intervals (Fig 5, A). In contrast, the FDM group had a mean ZnPP/H that increased with advancing gestation, such that the ZnPP/H at the >35 weeks interval was significantly higher compared with the other 2 groups.
In each of the 3 gestational age periods, plasma ferritin demonstrated greater relative variability and fewer among-group differences than ZnPP/H (Fig 5, B). The only significant differences were observed for the gestational interval >35 weeks, for which the mean ferritin level in the FDM group was lower relative to the other 2 groups.
The control and FDM groups demonstrated increasing mean plasma EPO levels with increasing gestational age, during which the mean EPO levels decreased among IUGR fetuses (Fig 5, C). The FDM group's mean EPO level was significantly higher than that in the control group at both 30 to 35 weeks and >35 weeks and higher than the IUGR group at >35 weeks.
Relationship Between ZnPP/H and Ferritin Using Modeled Covariates
In addressing our study's objective of evaluating ZnPP/H as an indicator of iron status at birth, the important and highly significant inverse relationship observed for cord blood ZnPP/H and plasma ferritin was examined using multiple linear regression analysis. The fitted regression model, which explained 64.2% (R2) of the variation in ZnPP/H, showed that in addition to plasma ferritin, ZnPP/H was significantly influenced by plasma EPO concentration and several clinical variables, including gestational age, maternal HBP, and maternal smoking (Fig 6). This model also identified a significant interaction between ferritin and maternal HBP (P = .0007), which indicated that the degree of association of ferritin with ZnPP/H varied with presence or absence of maternal HBP, and, likewise, the effect of maternal HBP on mean ZnPP/H was dependent on the ferritin level.
The multiple regression analysis showed a significant inverse relationship between ferritin and ZnPP/H with an estimated decrease of 29.1 ± 2.5% in mean ZnPP/H for a 10-fold increase in ferritin (P < .0001) among fetuses whose mothers did not manifest HBP. Among fetuses whose mothers manifested HBP, an even greater effect of ferritin was observed: a 10-fold increase in ferritin resulted in a 54.8 ± 2.0% drop in mean ZnPP/H (P < .0001).
The other laboratory measure that significantly affected ZnPP/H was EPO level. From the estimate of the EPO slope in the regression model, a 10-fold increase in EPO produced a 16.1 ± 0.7% (P = .007) increase in mean ZnPP/H. Although the significant effect of plasma EPO on the inverse relationship of plasma ferritin and ZnPP/H for all 3 study groups seems to contradict that suggested by Fig 1, A and C, for the control group alone, inclusion of the higher EPO values present from the 2 hypoxemic groups (see Fig 5, C) resulted in a small but significant upward deflection in the ZnPP/H versus ferritin relationship with increases in plasma EPO levels.
The clinical factors that influenced ZnPP/H were gestational age, maternal tobacco use, and maternal HBP. ZnPP/H decreased significantly with increasing gestational age, with mean ZnPP/H for fetuses who were >35 weeks being 47.5 ± 2.4% lower than for those who were <30 weeks' gestational age (P < .0001). Maternal tobacco use significantly increased mean ZnPP/H by 39 ± 4.6% (P < .0001). The effect of maternal hypertension on ZnPP/H significantly lessened with higher ferritin. When ferritin is low (eg, at 25th percentile, or 111 μg/L for this study), presence of maternal HBP significantly increased mean ZnPP/H by 17.6 ± 3.3% (P = .013). In contrast, when ferritin is high (eg, at 75th percentile, or 280 μg/L for this study), maternal HBP showed no significant effect on ZnPP/H (a decrease of 1.9 ± 3.4%; P = .813).
DISCUSSION
The present study's RBC ZnPP/H data at birth are consistent with and extend the work of others, demonstrating that mean ZnPP/H ratios in normal fetuses are higher than those in normal adults.10,14 More important, our data reinforce and extend the report of Juul et al14 that during the first week of life, ZnPP/H in normal fetuses is negatively associated with gestational age and is elevated further in those who are at high risk for in utero hypoxemia (ie, IUGR and FDM fetuses). Ours is the first study to demonstrate (1) that reticulocyte count in cord blood is positively associated with ZnPP/H and negatively associated with ferritin and (2) that the ZnPP/H–ferritin relationship at birth is affected by clinical conditions and laboratory variables indicative of fetal hypoxemia. Taken together, our findings indicate that ZnPP/H at birth reflects increased erythropoiesis accompanying normal growth and development; when chronic in utero hypoxemia is present, our observation of even higher fetal ZnPP/H levels raises the possibility that systemic iron availability to meet tissue needs, including the brain's, may have been limiting.4–6 The relationship of these findings to adverse neurodevelopmental outcome and the advantage that ZnPP/H may offer over the use of cord plasma ferritin15 have yet to be addressed.
Relationship of ZnPP/H to Iron Status in Normal Fetuses
Although the accumulated clinical and biochemical information indicates that the relationship between erythropoiesis and iron stores is closely related,8 the cumulative data of the present study support the speculation that high RBC ZnPP/H observed among normoxemic, iron-replete fetuses reflects the normal increase in erythropoiesis observed in fetuses relative to adults. This speculation is supported by data for iron-replete children and adults who experience increased erythropoiesis.7,16 That the control group data demonstrated a significant positive relationship between ZnPP/H and reticulocyte count supports the role of erythropoiesis leading to increased ZnPP; however, the negative association of ZnPP/H with plasma ferritin suggests that limited iron availability is a possible explanation. Because of limited reticulocyte data, it was not possible to determine statistically the dominant contributor to the increased ZnPP levels.
The normal fetal Hb levels and RBC indices observed among control group subjects further indicate that high RBC ZnPP/H does not necessarily represent a pathologic condition. We speculate that the greatest ZnPP/H levels were observed among the least mature control group fetuses because this group experienced the most rapid growth velocity and as such concomitantly experienced the greatest demand for tissue iron. Indeed, the normal rise that we and others have observed for fetal plasma ferritin with advancing gestational age may be the result of less intense erythropoiesis as gestation advances.17,18 Results of the present study do not rule out the possibility that nonspecific differences, ie, the concentration of fetal hemoglobin, in fetal RBCs as compared with the more adult-like RBCs produced later in infancy may lead to an artifactual increase in fetal ZnPP/H levels.
Response of RBC ZnPP/H to In Utero Stress and Hypoxemia
Ours is the first study to demonstrate that RBC ZnPP/H and the ZnPP/H–ferritin relationship at birth is perturbed by clinical and laboratory variables associated with in utero hypoxemia. Maternal HBP, maternal smoking, fetal hypoxemia (indicated by plasma EPO levels), and fetal immaturity each exerted an independent stimulatory effect that led to increased fetal ZnPP/H. Although the normal fetal increase in erythropoiesis relative to children and adults was the key stimulatory factor among control subjects, stress erythropoiesis, a vitally important homeostatic process from a survival standpoint, likely played an important role in the presence of fetal hypoxemia, as was observed in the IUGR and FDM groups. The normal homeostatic response of the fetus to chronic hypoxemia is to increase erythropoiesis for eventually increasing tissue oxygen delivery. Doing so places an increased demand on systemic iron stores to provide iron for Hb synthesis. Because under these conditions iron is preferentially directed toward Hb synthesis,5 tissue iron stores decrease, making less iron available for normal fetal growth and development. Increased RBC production coupled with limited or decreased maternal-to-fetal transplacental iron transport places the fetus at risk for inadequate iron availability to sustain essential metabolism.3 The evidence to date suggests that this sequence of events may contribute to serious life-long neurodevelopmental sequelae15 or death.
Relative to the control group subjects, fetuses with IUGR demonstrated increased ZnPP/H and plasma EPO with a tendency toward having low plasma ferritin levels. The higher ZnPP/H values that we observed are qualitatively consistent with the findings of Juul et al's14 in neonates during the first week of life. The latter group's higher mean ZnPP/H of 178 compared with the 78 μmol/mol heme observed in the present study was likely attributable to Juul et al's inclusion of earlier gestational age fetuses (33.9 vs 37.1 weeks). Although we did not attempt to determine why individual fetuses in the present study had IUGR, the underlying pathophysiology of this condition is most commonly attributable to utero-placental insufficiency. Autopsy studies in fetuses with IUGR indicate that their brain and liver iron content are decreased by 33% and 74%, respectively.19 In addition, reduced oxygen and nutrient transport by a typically small dysfunctional placenta results in fetuses' struggling to achieve normal growth while being in a chronically hypoxemic environment, as indicated by their markedly elevated plasma EPO levels.20–22 The inability of fetuses with IUGR to satisfy the already increased iron requirements of normal growth tends to increase erythropoiesis to a greater degree than normal. One result is even higher ZnPP/H levels than normal fetuses. Unfortunately, there were too few early-gestation fetuses with IUGR in the present study to permit an adequate assessment as to whether their erythropoiesis was increased. Nonetheless, the trend toward exceptionally high EPO levels among this early-gestation subgroup of fetuses with IUGR and that delivery was often necessitated on the basis of concern for fetal well-being suggest that their erythropoiesis may have been increased. As a result of their hypoxemic intrauterine environment, iron stores in fetuses with IUGR are frequently decreased at birth,23 predisposing these infants to early onset of iron deficiency,19 with some going on to manifest impaired neurodevelopmental function in infancy and childhood.15
Like the fetuses with IUGR, FDMs demonstrated increased ZnPP/H, plasma EPO, % hypochromic RBCs, and MCV relative to control fetuses, whereas plasma ferritin levels were decreased. Although statistically significant greater reticulocyte counts were not observed among FDM group subjects, whole-blood reticulocyte counts did not reflect the greater increase in growth and large blood volume experienced by this group. Taken together, this constellation of findings at birth among FDMs combined with those reported by others suggests a state of chronic fetal hypoxemia leading to the cascade of events beginning with increased EPO production and plasma EPO levels,1,24 leading to increased erythropoiesis and subsequent iron utilization, and ultimately resulting in diminished iron availability for Hb synthesis and decreased iron stores and brain iron content.13,24,25 In short, fetal hypoxemia is the result of accelerated metabolism resulting in increased fetal oxygen consumption in response to hyperglycemia-hyperinsulinemia brought about by diabetes-mediated hyperglycemia.26 Limited placental iron transport may also play a role in the lack of iron availability as indicated by increased ZnPP/H and low plasma ferritin.24,25 Recent psychomotor developmental data in FDMs indicate that resultant depletion of iron stores in fetal liver, heart, and brain as suggested by the present data and as verified by others in FDM autopsy studies may have neurodevelopmental consequences.27 This constellation of clinical features and increased number of stillbirths are especially true among macrosomic FDMs whose mothers' diabetes was poorly controlled.
Consistent with our findings, Juul et al14 reported that ZnPP/H is increased in the first week of life among FDMs. The considerably higher absolute FDM ZnPP/H reported by our group (191 vs 104 μmol/mol heme reported by Juul et al) was likely the result of earlier gestational age (34.2 vs 36.6 weeks) and poorer maternal diabetic control, because all of the FDMs in their study were admitted to the NICU, whereas ours were not. The ancillary laboratory and clinical data included in the present study and incorporated into the modeling analysis have allowed us more ably to suggest mechanistic explanations for our study's ZnPP/H findings.14
Study Limitations and Remaining Questions Regarding ZnPP/H in the Perinatal Period
A major limitation of our study is the lack of statistical power for identifying the relative roles of erythropoiesis and iron availability as contributors to the increased RBC ZnPP/H observed. This shortcoming was attributable primarily to the paucity of flow cytometric hematologic data, although the limited numbers of study subjects in some of the gestational age groupings as well as missing EPO data also played a role. In addition, the statistical modeling was complicated by the fact that the term control group did not exhibit the abnormal criteria present in the preterm control and the IUGR and FDM groups.
Longitudinal clinical and preclinical studies are needed for addressing key remaining questions regarding RBC ZnPP/H in the perinatal period: (1) What additional prenatal and postnatal factors, eg, RBC transfusion, oral iron therapy, parenteral iron therapy, EPO treatment, and events, affect ZnPP/H (2) What roles do ZnPP/H and ferritin play in predicting neurodevelopmental outcome, and, if both are associated, is one better than the other (3) Why are ZnPP/H measurements so variable from individual to individual Use of a recently described neonatal rat model for studying ZnPP/H28 offers an opportunity for directly measuring tissue iron in different organs but would require the development of analytic techniques for measuring plasma ferritin and plasma transferrin receptor.
CONCLUSIONS
On the basis of the present findings and those of Juul et al,14 we recommend during the immediate perinatal period that the use of RBC ZnPP/H data as an indicator of iron status be interpreted with reference to values derived from normal fetuses and neonates. The findings of our study indicate that elevated RBC ZnPP/H in cord blood at birth is associated with increased erythropoiesis compared with later life, not with functional or absolute iron deficiency. We base our conclusion on the observation that in the control group, umbilical cord plasma ferritin levels were in the ferritin range deemed desirable for adults who receive EPO treatment, yet this group had normal cord blood Hb levels and RBC iron indices. When viewed in the context of their laboratory and clinical findings, the elevated RBC ZnPP/H observed in fetuses at high risk for hypoxemia indicate functional iron deficiency. Whether they indicate true systemic iron deficiency with the potential for adverse neurodevelopment consequences remains to be established.
ACKNOWLEDGMENTS
This study was supported by grant RR00059 from the General Clinical Research Centers Program, National Center for Research Resources; National Institutes of Health Grant DK35816; and NIH P01 HL46925. The University of Arizona portion was supported by the American Heart Association, Arizona Affiliate AZGS-37-96 and Southwest Affiliate SW-GS-16-98 (Dr Kling).
We acknowledge the technical assistance and data collection contributions of Gretchen Cress, RN, Karen J. Johnson, RN, Erin McGuire, BS, Robert Schmidt, BS, Suzan J. Hays, Karen B. Lesser, MD, Carrie R. Daniel, BS, Ronald B. Schifman, MD, and Megan Dyer. Employees of the hospital blood bank are acknowledged for assistance with the storage of blood samples, and Mark A. Hart is acknowledged for secretarial help. Michael Georgieff, MD, provided helpful suggestions on data analysis and interpretation.
FOOTNOTES
Accepted Nov 23, 2004.
No conflict of interest declared.
REFERENCES
Schwartz R, Teramo KA. Effects of diabetic pregnancy on the fetus and newborn. Sem Perinatol. 2000;24 :120 –135
Gardosi J, Mul T, Mongelli M, Fagan D. Analysis of birthweight and gestational age in antepartum stillbirths. Br J Obstet Gynecol. 1998;105 :524 –530
Guiang SF, Georgieff MK. Fetal and neonatal iron metabolism. In: Polin RA, Fox WW, eds. Fetal and Neonatal Physiology. 2nd ed. Philadelphia, PA: WB Saunders Company; 1998:401 –410
Lozoff B. Perinatal iron deficiency and the developing brain. Pediatr Res. 2000;48 :137 –139
Rao R, Georgieff MK. Perinatal aspects of iron metabolism. Acta Paediatr Suppl. 2002;91 :124 –129
Sesma HW, Georgieff MK. The effect of adverse intrauterine and newborn environments on cognitive development: the experiences of premature delivery and diabetes during pregnancy. Dev Psychopathol. 2003;15 :991 –1015
Labbé RF, Vreman HJ, Stevenson DK. Zinc protoporphyrin: metabolite with a mission. Clin Chem. 1999;45 :260 –272
Labbé RF, Dewanji A. Iron assessment tests: transferrin receptor vis-a-vis zinc protoporphyrin. Clin Biochem. 2004;37 :165 –174
Harthoorn-Lasthuizen EJ, Lindemans J, Langenhuijsen MMAC. Erythrocyte zinc protoporphyrin testing in pregnancy. Acta Obstet Gynecol Scand. 2000;79 :660 –666
Griffin IJ, Reid MM, McCormick KP, Cooke RJ. Zinc protoporphyrin/haem ratio and plasma ferritin in preterm infants. Arch Dis Child Fetal Neonatal Ed. 2002;87 :F49 –F51
Arbuckle TE, Wilkins R, Sherman GJ. Birth weight percentiles by gestational age in Canada. Obstet Gynecol. 1993;81 :39 –48
Georgieff MK, Widness JA, Mills MM, Stonestreet BS. The effect of prolonged intrauterine hyperinsulinemia on iron utilization in fetal sheep. Pediatr Res. 1989;26 :467 –469
Juul SE, Zerzan JC, Strandjord TP, Woodrum DE. Zinc protoporphyrin/heme as an indicator of iron status in NICU patients. J Pediatr. 2003;142 :273 –278
Tamura T, Goldenberg RL, Hou J, et al. Cord serum ferritin concentrations and mental and psychomotor development of children at five years of age. J Pediatr. 2002;140 :165 –170
NCCLS. Erythrocyte Protoporphyrin Testing: Approved Guideline (NCCLS Document C42-A). Villanova, PA: National Committee on Clinical Laboratory Standards; 1996
Siimes AS, Siimes MA. Changes in the concentration of ferritin in the serum during fetal life in singletons and twins. Early Hum Dev. 1986;13 :47 –52
Sweet DG, Savage GA, Tubman R, Lappin TR, Halliday HL. Cord blood transferrin receptors to assess fetal iron status. Arch Dis Child Fetal Neonatal Ed. 2001;85 :F46 –F48
Georgieff MK. Fetal and neonatal iron metabolism. Semin Neonatal Nutr Metab. 2000;7 :1 –5
Ruth V, Autti-Ramo I, Granstrom ML, Korkcan M, Raivio KO. Prediction of perinatal brain damage by cord plasma vasopressin, erythropoietin, and hypoxanthine values. J Pediatr. 1988;113 :880 –885
Widness JA, Weiner CP, Giller RH, Kling PJ, Schmidt RL, Fomon SJ. Increasing plasma erythropoietin response to anemia during development. Pediatr Res. 1990;27 :153A
Teramo KA, Hiilesmaa VK, Schwartz R, Clemons GK, Widness JA. Amniotic fluid and cord plasma erythropoietin levels in pregnancies complicated by preeclampsia, pregnancy-induced hypertension and chronic hypertension. J Perinatal Med. 2004;32 :240 –247
Chockalingam UM, Murphy E, Ophoven JC, Georgief MK. Cord transferrin and ferritin values in newborn infants at risk for prenatal uteroplacental insufficiency and chronic hypoxia. J Pediatr. 1987;111 :283 –286
Georgieff MK, Landon MB, Mills MM, et al. Abnormal iron distribution in infants of diabetic mothers: spectrum and maternal antecedents. J Pediatr. 1990;117 :455 –461
Amarnath UM, Ophoven JJ, Mills MM, Murphy EL, Georgieff MK. The relationship between decreased iron stores, serum iron and neonatal hypoglycemia in large-for-date newborn infants. Acta Paediatr Scand. 1989;78 :538 –543
Madsen H. Fetal oxygenation in diabetic pregnancy. With special reference to maternal blood oxygen affinity and its effectors. Dan Med Bull. 1986;33 :64 –74
Siddappa AM, Georgieff MK, Wewerka S, Worwa C, Nelson CA, deRengnier RA. Iron deficiency alters auditory recognition memory in newborn infants of diabetic mothers. Pediatr Res. 2004;55 :1034 –1041
Dubuque SH, Dvorak B, Woodward SS, McCuskey RS, Kling PJ. Iron-deficient erythropoiesis in neonatal rats. Biol Neonate. 2002;81 :51 –57(David G. Lott, MD, M. Bri)