Silva, Anadir M. MD1; Smith, Randi N.2; Lehmann, Christoph U. MD3; Johnson, Elizabeth A. MS4; Holcroft, Cynthia J. MD1; Graham, Ernest M. MD1
Chronic hypoxia is known to lead to medullary and extramedullary hematopoiesis through the renal production of erythropoietin. This leads to an increase in nucleated red blood cell counts in the neonate after a chronic hypoxic exposure. The primary stimulus of renal erythropoietin production is tissue hypoxia. Although the time interval between increased erythropoietin production and peak erythrocyte count is unknown, most evidence in animals and humans suggests that this response takes at least 24–48 hours and declines by 7 days.1,2 Based on the time necessary for tissue hypoxia to increase erythropoietin, and subsequently the nucleated red blood cell count, an acute intrapartum hypoxic event should not cause any significant increase, and when the nucleated red blood cell count is elevated at birth, this would suggest the presence of hypoxia of greater than 24 hours duration. Elevated nucleated red blood cell counts have been suggested as being associated with low umbilical arterial pH, intrauterine growth restriction (IUGR), perinatal brain damage, early-onset neonatal seizures, and cerebral palsy.3 Some investigators have proposed that following the changing nucleated red blood cell counts in neonatal blood may accurately identify the time before birth when brain-damaging ischemia and hypoxemia began.4 Our objective in this study was to estimate whether premature neonates with cerebral white matter injury diagnosed by head ultrasound within 6 weeks of birth have elevated nucleated red blood cell counts at birth, to estimate antenatal and intrapartum risk factors associated with increased counts, and to estimate whether these counts can be useful in identifying neonates at risk for the later development of brain injury.
PATIENTS AND METHODS
This was an institutional review board–approved case–control study of all infants born at a single tertiary care university hospital from November 1994 to October 2004 between 23 and 34 weeks of gestation. Infants born at 32 or fewer weeks of gestation had 3 head ultrasonograms, the first at 24–72 hours after birth, the second at 10–14 days, and the third at 6 weeks of gestation, to look specifically for cerebral white matter injury. Infants born between 32 and 34 weeks of gestation had a head ultrasonogram if it was thought warranted by the attending neonatologist based on the presence of neurologic symptoms, but not on a routine basis. Cases with ultrasound-diagnosed cerebral white matter injury characterized by periventricular leukomalacia (PVL) or ventricular dilatation due to white matter atrophy were identified and matched to the subsequent delivery by gestational age within 7 days without brain injury. Grade 3 and 4 intraventricular hemorrhage (IVH) is linked with an increased risk of neurologic morbidity, so in addition to not having white matter injury, the controls were also without grade 3 and 4 intraventricular hemorrhage. Neonates with chromosomal abnormalities or major congenital malformations were excluded. At our institution all neonates born at 34 or fewer weeks of gestation are admitted to the neonatal intensive care unit (NICU) immediately after birth, and at the time of their admission, have a cell blood count done with a Sysmex XE 2100 (Sysmex America Inc., Mundelein, IL), which includes nucleated red blood cell count per 100 white blood cells (WBCs). After excluding infants who did not have a nucleated red blood cell count drawn immediately on admission to the NICU, there were 176 cases that were matched to 176 controls
For this study we defined acute hypoxia as occurring during labor and chronic hypoxia as occurring before the onset of labor. Intraventricular hemorrhage was defined in the standard fashion, with grade 3 indicating hemorrhage with ventricular dilatation and grade 4 ventricular dilatation with parenchymal extension of hemorrhage. A course of antenatal steroids consisted of receiving 2 doses of 12-mg betamethasone given 24 hours apart. Oligohydramnios was defined as having an amniotic fluid index less than 5 cm. Intrauterine growth restriction was defined as a birth weight less than 10% for gestational age.5 The diagnosis of nonreassuring fetal heart rate (FHR) tracing was made by the physician attending delivery before performing a cesarean.
The nucleated red blood cell counts are reported here as nucleated red blood cells per 100 WBCs. A recent study that examined the nucleated red blood cell count in umbilical cord blood from 128 women found a close correlation between absolute nucleated red blood cells and nucleated red blood cells per 100 WBCs (r = 0.63).6 These investigators found that the automated hematology analyzer readings of nucleated red blood cell counts per 100 WBCs correlate well with readings by laboratory hematologists, which they considered to be the reference method. To minimize the effects of a few large nucleated red blood cell counts, a natural log transformation was used in the analysis. Normality of the natural log of the nucleated red blood cell counts was assessed using a normal probability plot. Summary statistics (median and 5th and 95th percentiles) for the nucleated red blood cell counts are presented on the original scale to maintain interpretability. Maternal and neonatal demographics and complications were compared across the cases and controls using paired t tests (continuous variables excluding gravity and parity), McNemar’s test (categorical variables), and the Wilcoxon matched-pair signed rank test (gravity and parity), with P < .05 considered significant. Variables with P < .10 in the univariate analyses were included in the multivariate analysis. Conditional multivariate logistic regression was used to identify variables associated with the presence of neonatal cerebral white matter injury. To determine the best neonatal nucleated red blood cell count to distinguish brain injured cases from neurologically normal controls, we dichotomized the neonatal nucleated red blood cell counts using various cut-points and constructed receiver operating characteristic (ROC) curves. The nucleated red blood cell cut-point with the maximal area under the ROC curve was used to calculate sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).
Secondary analyses assessed the relationship between nucleated red blood cell counts and risk factors of interest, such as severe fetal metabolic acidosis. Specifically, t tests were used to compare the natural log of the nucleated red blood cell counts among neonates with and without risk factors of interest. Subsequent multivariate linear regression was used to identify risk factors associated with significant changes in nucleated red blood cell counts. To estimate the relationship between the neonatal nucleated red blood cell count and metabolic acidosis, linear regression models of the natural log of the nucleated red blood cell count compared with umbilical artery pH and base excess were performed. Analysis was performed using Stata 7.0 (Stata Corporation, College Station, TX) and SPSS 12.0 (SPSS Inc., Chicago, IL).
When comparing brain injured cases with neurologically normal controls, there was no difference in maternal demographics, cesarean delivery, or exposure to antenatal steroids or magnesium (Table 1). There was a significant increase in multiple gestations among the cases (38 [21.6%] cases compared with 24 [13.6%] controls, P = .03). Even though the cases and controls were matched by gestational age within 7 days and had the same gestational age at delivery, the cases had a significantly smaller birth weight (mean ± standard deviation, 958 ± 306 g cases and 1,038 ± 381 g controls, P = .001) (Table 2). The cases had a higher incidence of Apgar scores less than 7 at 5 minutes, length of stay, chronic lung disease, and culture-positive neonatal sepsis. In addition to having cerebral white matter injury, 38 (21.6%) of the cases had grades 3 and 4 intraventricular hemorrhage. Of the 176 cases there were 9 neonatal deaths (5.2%), and of the 176 controls there were 19 neonatal deaths (10.8%). All of these infants had at least 1 head ultrasonogram, but only 5 of the cases and none of the controls had the 6-week head ultrasonogram. We did not exclude these infants, because we did not want to remove infants artificially who may have suffered from acute intrapartum hypoxia–ischemia and died soon after birth. One of the outcome measures we are interested in is the relationship between nucleated red blood cell counts and severe metabolic acidosis on an umbilical arterial blood gas. Removing cases and controls that died before 6 weeks of life would prevent us from being able to evaluate this relationship. The median, 5th percentile, and 95th percentile of the nucleated red blood cell count per 100 WBCs was significantly higher in the cases (22, 3 and 374 cases compared with 14, 1 and 312 controls, P = .02). There was no difference in hematocrit, WBC, lymphocyte, or platelet counts between the cases and controls.
Conditional multivariate logistic regression of variables from the univariate analyses with a P = .10 did not find a significant relationship between brain injury and nucleated red blood cell counts (odds ratio [OR] 1.15, 95% confidence interval [CI] 0.95–1.41, P = .16), but multiple gestation (OR 2.04, 95% CI 1.08–3.86, P = .03), birth weight (OR 1.00, 95% CI 1.00–1.00, P = .03), and intraventricular hemorrhage (OR 4.27, 95% CI 2.51–7.26, P < .001) were significantly associated with cerebral white matter injury. For the controls with a median nucleated red blood cell count of 14 per 100 WBCs and the cases with a median nucleated red blood cell count of 22 per 100 WBCs, ROC curves were used to find the cutoff point with the maximal ability to distinguish the 2 groups. Using a nucleated red blood cell count of 18 per 100 WBCs had an area of 0.58 under the ROC curve, with a sensitivity of 56.9%, specificity of 57.9%, PPV of 57.9%, and NPV of 56.9% to differentiate these 2 populations in this matched case–control sample.
In addition to identifying risk factors associated with neonatal cerebral white matter injury, we identified risk factors associated with changes in nucleated red blood cells. Specifically, we used t tests to compare the natural log of the nucleated red blood cell counts across neonates with and without risk factors of interest. Cesarean delivery, cesarean delivery for nonreassuring FHR tracings, preeclampsia, IUGR, oligohydramnios, 1- and 5-minute Apgar less than 7, umbilical arterial pH less than 7.0 or base excess less than −12 mM, and chronic lung disease were associated with statistically significant increases in nucleated red blood cell counts (Table 3). There was no difference in nucleated red blood cell counts between singleton and multiple gestations. Multivariate linear regression found cesarean delivery, IUGR, and 1-minute Apgar less than 7 to be associated with significant nucleated red blood cell increases (Table 4).
Last, linear regression found a statistically significant relationship between an increasing natural log of the nucleated red blood cell count and decreasing umbilical artery pH (R2 = 0.06, P < .001) and base excess (R2 = 0.07, P < .001).
If there is a 24–48 hour lag between the hypoxic insult and the increased production of nucleated red blood cells, then a high nucleated red blood cell count at birth in a child that later develops neurologic morbidity would indicate chronic hypoxia preexisting the onset of labor. One study that examined nucleated red blood cell counts as a marker of chronic hypoxia found that neonates with persistently nonreactive tracings from admission to delivery had higher nucleated red blood cell counts.7 Neither mild nor severe intrapartum fetal heart rate abnormalities were associated with a significant increase in nucleated red blood cell counts when the fetus was reactive at admission. They concluded that nucleated red blood cell counts identify the presence of chronic fetal asphyxia and that an acute intrapartum hypoxic event does not increase the nucleated red blood cell count. However, this study places great reliance on a nonreactive fetal heart rate tracing on admission to identify the chronically hypoxic fetus, and nonreassuring fetal heart rate patterns are quite imprecise in identifying these fetuses.8
Although the vast majority of the literature on the relationship between nucleated red blood cell counts and perinatal brain injury report nucleated red blood cell count per 100 WBCs, some have questioned the accuracy of this practice. One group compared neonatal nucleated red blood cell counts between 79 infants of diabetic mothers and 102 consecutively delivered controls and found that the infants of the diabetic mothers had significantly lower WBC counts with or without perinatal asphyxia.9 They concluded that the divergence in nucleated red blood cell counts from a control group can be inflated when results are expressed per 100 leukocytes when the cases have a relative leukopenia compared with the control group. In our study we did not find a difference in WBC count between cases and controls, so we did not calculate the absolute number of nucleated red blood cells. In addition to neonatal nucleated red blood cell counts, others have suggested that lymphocyte counts could help determine the time of neonatal neurologic injury. Lymphocyte counts and erythropoietin levels increase in the blood of adults made hypoxemic in high altitude chambers, probably from a generalized bone marrow stimulation.2 Neutrophil and lymphocyte counts are increased in the blood of hypoxemic subjects, but because many hypotonic neonates are later found to be hypotonic due to a bacterial infection and lymphocytosis is not a feature of acute bacterial infections, the lymphocyte count has been investigated as a marker for perinatal hypoxia rather than the total WBC count. Thrombocytopenia has also been investigated as a link to neonatal neurologic morbidity by way of antenatal thrombotic events that lead to increased placental impedance.10 In our study we did not find any difference in total WBC, lymphocyte, or platelet counts at the time of admission to the NICU between the brain injured cases and normal controls.
Multiple gestations are known to be at increased risk for cerebral palsy,11 but in this study we found no difference in nucleated red blood cell counts between singleton and multiple gestations. We chose to include both singletons and multiples in this study because if nucleated red blood cell counts are a marker of chronic hypoxia preexisting labor, they would apply equally to both, just as an umbilical arterial pH less than 7.0 and base excess less then −12 mM are used to identify acute hypoxia–ischemia sufficient to increase the risk of long-term neurologic morbidity for both singleton and multiple gestations.12
Approximately 70% of stillbirths exhibit growth restriction,13 and IUGR is associated with increased fetal plasma erythropoietin concentrations.14 A study of 441 infants with birth weights 499–1,751 g, of whom 94 developed intraventricular hemorrhage or PVL in the first week of life and had a complete blood count within 24 hours of birth that included a nucleated red blood cell count, did not find an elevated nucleated red blood cell count in the intraventricular hemorrhage and PVL group, but did find elevations in growth restricted fetuses.15 Including intraventricular hemorrhage and PVL in the study group may have contributed to clinical heterogeneity that, added to the small sample size, may not have had the power to detect a difference if it indeed exists. We chose to study cerebral white matter injury exclusively because of its susceptibility to hypoxia–ischemia; the vulnerability of its blood supply produces end zones in the white matter that are sites for the focal necroses of PVL and white matter atrophy producing ventriculomegaly.
Prematurity is known to be associated with increased nucleated red blood cell counts3; however, the brain-injured infants in this study were matched by gestational age within 7 days to ensure that any difference seen was not due to a difference in gestational age. Infection has been linked to increases in nucleated red blood cell counts. A study of preterm infants with clinical and histologic chorioamnionitis controlled for gestational age and birth weight percentile found that histologic chorioamnionitis was associated with elevated nucleated red blood cell counts.16 Some cytokines produced during an infection may act as hematopoietic stimulators. In our study we did not find a relation between clinical chorioamnionitis, histologic chorioamnionitis and funisitis, or culture-positive neonatal sepsis and nucleated red blood cell counts. Erythropoietin has been found to be significantly correlated with nucleated red blood cell counts; however, the relatively low R2 of 0.29 indicates that there are other mediators responsible for a majority of the increase.17 An umbilical arterial pH less than 7.0 or base excess less than −12 mM, indicators of acute intrapartum hypoxia–ischemia, were associated with significant nucleated red blood cell increases. Acute increases in nucleated red blood cell counts may be secondary to mobilization by endogenous cytokines such as interleukin-6, which is markedly increased in response to hypoxia.18 Epinephrine, an acute stress hormone, may modulate erythropoiesis and could stimulate rapid nucleated red blood cell increases.19 The increase in nucleated red blood cell and lymphocyte counts seen as early as 2 hours after acute fetal hypoxemia is certainly unrelated to erythropoiesis.20
In our study we found conditions associated with chronic in utero hypoxia–ischemia such as preeclampsia, IUGR, and oligohydramnios to be associated with significantly elevated nucleated red blood cell counts. Elevated nucleated red blood cell counts were also seen in neonates with acute hypoxia–ischemia, such as those with cesarean delivery for nonreassuring FHR tracings or an umbilical arterial pH less than 7.0 or base excess less than −12 mM. Because both acute and chronic hypoxia influence the nucleated red blood cell counts, whether drawn from umbilical or neonatal blood, this prevents the nucleated red blood cell count from being able to time the onset of neonatal brain injury. Although significant in the univariate analysis, the nucleated red blood cell count did not maintain significance in the multivariate analysis in predicting the presence of cerebral white matter injury. This shows that nucleated red blood cell counts do not have as strong an association with cerebral white matter injury as birth weight and intraventricular hemorrhage. The small R2 for the relationship of umbilical arterial pH and base excess with neonatal nucleated red blood cell counts shows that metabolic acidosis accounts for only 6–7% of the variability in nucleated red blood cell counts. Even though the cases had a significant increase in nucleated red blood cell counts, there was no difference in hematocrit suggesting that hypoxia may not be the dominant factor in producing these elevations.
In conclusion, although preterm neonates with cerebral white matter injury have significant increases in nucleated red blood cell counts at birth, a number of conditions suggestive of chronic hypoxia–ischemia, such as preeclampsia, IUGR, and oligohydramnios, are associated with elevated nucleated red blood cell counts, as is acute intrapartum hypoxia–ischemia, as manifested by an umbilical arterial pH less than 7.0 or base excess less than −12 mM, which may increase these counts independent of medullary or extramedullary hematopoiesis. Although the nucleated red blood cell count is introduced frequently in litigation involving allegations of intrapartum malpractice leading to brain injury, with defense experts claiming that high nucleated red blood cell counts show evidence of chronic hypoxia preceding the onset of labor and plaintiff experts claiming that high nucleated red blood cell counts are evidence of acute intrapartum hypoxia, because both acute and chronic hypoxia can increase the nucleated red blood cell counts, they will not be helpful in pinpointing the time of the hypoxic insult. Although the specificity, PPV, and NPV in this study are low, they are artificially elevated by not including all neonates born without injury, which shows that although nucleated red blood cell counts are a marker for neonatal cerebral white matter injury, they have poor predictive ability in identifying infants at birth who will later develop this form of injury.
1. Georgieff MK, Schmidt RL, Mills MM, Radmer WJ, Widness JA. Fetal iron and cytochrome c status after intrauterine hypoxemia and erythropoietin administration. Am J Physiol 1992;262:R485–91.
2. Siri WE, Van Dyke DC, Winchell HS, Pollycove M, Parker HG, Cleveland AS. Early erythropoietin, blood, and physiological responses to severe hypoxia in man. J Appl Physiol 1966;21:73–80.
3. Buonocore G, Perrone S, Gioia D, Gatti MG, Massafra C, Agosta R, et al. Nucleated red blood cell count at birth as an index of perinatal brain damage. Am J Obstet Gynecol 1999;181:1500–5.
4. Naeye RL, Localio AR. Determining the time before birth when ischemia and hypoxemia initiated cerebral palsy. Obstet Gynecol 1995;86:713–9.
5. Hadlock FP, Harrist RB, Sharman RS, Deter RL, Park SK. Estimation of fetal weight with the use of head, body, and femur measurements—a prospective study. Am J Obstet Gynecol 1985;151:333–7.
6. McCarthy JM, Capullari T, Spellacy WN. The correlation between automated hematology and manually read smears for the determination of nucleated red blood cells in umbilical cord blood. J Matern Fetal Neonatal Med 2005;17:199–201.
7. Korst LM, Phelan JP, Ahn MO, Martin GI. Nucleated red blood cells: an update on the marker for fetal asphyxia. Am J Obstet Gynecol 1996;175:843–6.
8. American College of Obstetricians and Gynecologists. Fetal heart rate patterns: monitoring, interpretation, and management. ACOG Technical Bulletin. Washington, DC: ACOG; 1995.
9. Green DW, Mimouni F. Nucleated erythrocytes in healthy infants and in infants of diabetic mothers. J Pediatr 1990;116:129–31.
10. Bernstein PS, Minior VK, Divon MY. Neonatal nucleated red blood cell counts in small-for-gestational age fetuses with abnormal umbilical artery Doppler studies. Am J Obstet Gynecol 1997;177:1079–84.
11. Pharoah PO, Cooke T. Cerebral palsy and multiple births. Arch Dis Child Fetal Neonatal Ed 1996;75:F174–7.
12. American College of Obstetricians and Gynecologists, American Academy of Pediatricians. Neonatal encephalopathy and cerebral palsy: defining the pathogenesis and pathophysiology. Washington DC: ACOG; 2003.
13. Morrison I, Menticoglou S, Manning FA, Harman CR, Cheang M. Comparison of antepartum tests and the relationship of multiple test results to perinatal outcome. J Matern Fetal Med 1994;3:75–83.
14. Snijders RJ, Abbas A, Melby O, Ireland RM, Nicolaides KH. Fetal plasma erythropoietin concentration in severe growth retardation. Am J Obstet Gynecol 1993;168:615–9.
15. Leikin E, Verma U, Klein S, Tejani N. Relationship between neonatal nucleated red blood cell counts and hypoxic-ischemic injury. Obstet Gynecol 1996;87:439–43.
16. Leikin E, Garry D, Visintainer P, Verma U, Tejani N. Correlation of neonatal nucleated red blood cell counts in preterm infants with histologic chorioamnionitis. Am J Obstet Gynecol 1997;177:27–30.
17. Ferber A, Fridel Z, Weissmann-Brenner A, Minior VK, Divon MY. Are elevated fetal nucleated red blood cell counts an indirect reflection of enhanced erythropoietin activity? Am J Obstet Gynecol 2004;190:1473–5.
18. Klausen T, Olsen NV, Poulsen TD, Richalet JP, Pedersen BK. Hypoxemia increases serum interleukin-6 in humans. Eur J Appl Physiol Occup Physiol 1997;76:480–2.
19. Mladenovic J, Adamson JW. Adrenergic modulation of erythropoiesis: in vitro studies of colony-forming cells in normal and polycythaemic man. Br J Haematol 1984;56:323–32.
20. Naeye RL, Lin HM. Determination of the timing of fetal brain damage from hypoxemia-ischemia. Am J Obstet Gynecol 2001;184:217–24.