Nucleated Red Blood Cells: Could They Be Indicator Markers of Illness Severity for Neonatal Intensive Care Unit Patients?

Background: We aimed to assess whether nucleated red blood cells (NRBCs) count could serve as a diagnostic and prognostic biomarker for morbidity and mortality in critically ill neonates. Methods: The association between NRBCs count and neonatal morbidity and mortality was evaluated in an observational cohort of critically ill neonates hospitalized in our neonatal intensive care unit over a period of 69 months. The discriminative ability of NRBCs count as diagnostic and prognostic biomarkers was evaluated by performing the Receiver Operating Characteristics (ROC) curve analysis. Results: Among 467 critically ill neonates included in the study, 45 (9.6%) of them experienced in-hospital mortality. No statistically significant difference was found with regards to NRBCs count between survivors and non-survivors, although the median value for NRBCs was sometimes higher for non-survivors. ROC curve analysis showed that NRBCs is a good discriminator marker for the diagnosis of perinatal hypoxia in neonates with area under the curve (AUC) [AUC 0.710; 95% confidence interval (CI), 0.660–0.759] and predominantly in preterm neonates (AUC 0.921 (95% CI, 0.0849–0.0993)) by using a cut-off value of ≥11.2%, with 80% sensitivity and 88.7% specificity. NRBCs also revealed significant prognostic power for mortality in septic neonates (AUC 0.760 (95% CI, 0.631–0.888)) and especially in preterms with sepsis (AUC 0.816 (95% CI, 0.681–0.951)), with cut-off value ≥ 1%, resulting in 81.6% sensitivity and 78.1% specificity. Conclusion: NRBCs count may be included among the early diagnostic and prognostic markers for sick neonates.


Introduction
Critically ill neonates constitute a fragile population with very special characteristics compared to children and adults; hence, early diagnosis and management of critical illness are of great importance for their short-term outcome and lifelong prognosis. Various scoring systems have Figure 1. Flowchart of study population. * 4 neonates with perinatal hypoxia were also included in this subgroup, as they suffered from late-onset sepsis. Abbreviation: NRBCs, nucleated red blood cell counts.
In all study neonates, the SNAPPE score and modified NEOMOD scoring system [33] were calculated on admission to the NICU (for preterm neonates with RDS or neonates with perinatal asphyxia/fetal distress) and on the onset of the disease (for neonates with suspected or confirmed sepsis).
In the first 6-12 h of life (for preterm neonates with RDS and those with perinatal asphyxia/fetal distress), prior to initiating antibiotic therapy-if required, blood specimens for culture and biochemical tests (serum glutamil oxaloacetic transaminase (SGOT), complete blood count, (CRP) were performed, while on the second day of life (DOL), further biochemical parameters (sodium, potassium, SGOT, serum glutamic pyruvic transaminase (SGPT), blood urea nitrogen (BUN), creatinine plasma level (Cr), bilirubin, albumin (ALB)) were measured in all neonates, as per our NICU protocol. As for neonates with suspected or confirmed sepsis, the aforementioned laboratory tests were performed and recorded at the onset of clinical deterioration [median day of life: 12, interquartile range (IQR) 7-20]. Chest radiograph was performed whenever clinically indicated.
Complete blood count was performed on a Sysmex XE-2100 analyzer (Roche, IL, USA). Peripheral blood smears prepared with Giemsa stain were examined, and band forms, NRBCs, Figure 1. Flowchart of study population. * 4 neonates with perinatal hypoxia were also included in this subgroup, as they suffered from late-onset sepsis. Abbreviation: NRBCs, nucleated red blood cell counts.
In all study neonates, the SNAPPE score and modified NEOMOD scoring system [33] were calculated on admission to the NICU (for preterm neonates with RDS or neonates with perinatal asphyxia/fetal distress) and on the onset of the disease (for neonates with suspected or confirmed sepsis).
In the first 6-12 h of life (for preterm neonates with RDS and those with perinatal asphyxia/fetal distress), prior to initiating antibiotic therapy-if required, blood specimens for culture and biochemical tests (serum glutamil oxaloacetic transaminase (SGOT), complete blood count, (CRP) were performed, while on the second day of life (DOL), further biochemical parameters (sodium, potassium, SGOT, serum glutamic pyruvic transaminase (SGPT), blood urea nitrogen (BUN), creatinine plasma level (Cr), bilirubin, albumin (ALB)) were measured in all neonates, as per our NICU protocol. As for neonates with suspected or confirmed sepsis, the aforementioned laboratory tests were performed and recorded at the onset of clinical deterioration [median day of life: 12, interquartile range (IQR) 7-20]. Chest radiograph was performed whenever clinically indicated.
Complete blood count was performed on a Sysmex XE-2100 analyzer (Roche, IL, USA). Peripheral blood smears prepared with Giemsa stain were examined, and band forms, NRBCs, myelocytes and Children 2020, 7, 197 4 of 10 metamyelocytes (evaluated as immature neutrophils in leukocyte formula), and Immature to Total Neutrophil Ratio (I/T ratio) were calculated.
NRBCs measurement, expressed as NRBC count per 100 White Blood Cells (WBC), was carried out both automatically by the formerly mentioned analyzer and also by manual correction so that the results would be accurate.

Statistical Analysis
Descriptive statistics of the baseline data are presented as mean ± standard deviation (SD), median (ranges), or as percentages when appropriate. The differences in categorized and continuous data were compared using the chi-square contingency test and the with Mann-Whitney U test. Time-to-event analysis was carried out in order to evaluate the variables independently associated with the in-hospital survival. Univariate Cox's proportional hazards analysis for relevant prognostic variables was performed. The hazards ratios or relative hazards derived from Cox's proportional hazards models are presented with 95% confidence intervals (CIs) and the respective p-values. A ratio higher than unity implies a higher probability of death than in the reference group.
NRBCs count was evaluated to predict mortality in the study group. Their discriminative ability as diagnostic and prognostic biomarkers was evaluated by performing the Receiver Operating Characteristics (ROC) curve analysis. The optimal cut-off thresholds were identified with the Youden index, and the sensitivity, specificity, positive and negative predictive values were also evaluated. All statistical tests were two-sided. Stata statistical software was used for all statistical analyses (Stata Corp., College Station, TX, USA). For all tests, a p-value lower than 0.05 indicates statistical significance.

Results
A total of 467 critically ill neonates were included in the study, of which 45 (9.6%) did not survive. The baseline characteristics of the neonates are presented in Table 1. Comorbidities recorded in study neonates are presented in Table 2.  Subsequently, we investigated the role of several parameters, and specifically, NRBCs, in the survival of the neonates (Table 3). No statistically significant difference between survivors and non-survivors regarding GA and BW was found. Almost all laboratory studied parameters had a significant role in the survival of the neonates except for neutrophils percentage, WBC, and NRBC. Although the median value for NRBCs was sometimes times higher for non-survivors, this finding did not allow for proving a statistical significance under the condition of p < 0.05, probably because of the great variability in NRBCs and the small number of cases. In addition, this analysis proved that all scoring systems (i.e., SNAPPE, modified NEOMOD, and TOLLNER scores) were significantly different between survivors and non-survivors.
The analysis of study population mortality (as a dependent variable) in relation to comorbidities revealed that all conditions had a significant role in survivability except for intrauterine growth retardation (IUGR) and NEC (Table 4). The performance of ROC curve analysis for the discriminative ability of NRBCs as diagnostic and prognostic markers in all study neonates is displayed in Table 5. Subsequently, study neonates were subclassified as term and preterm neonates. ROC curve analysis showed that NRBCs is a good discriminator marker for the diagnosis of perinatal hypoxia in preterm neonates (Figure 2), with AUC 0.921(95% CI, 0.0849-0.0993). Using a cut-off value of ≥11.2%, NRBCs resulted in 80% sensitivity and 88.7% specificity.
NRBCs also revealed significant prognostic power for mortality in preterm neonates with sepsis (AUC 0.816 (95% CI, 0.681-0.951)), with cut-off value ≥1%, resulting in 81.6% sensitivity and 78.1% specificity (Figure 3). This low cut-off value probably attributed to the day of measurement (median: 12 (IQR 7-20)). Contrarily, ROC curve analysis showed that NRBCs could not discriminate diagnosis or prognosis of sepsis or perinatal hypoxia between the term neonates of our study.

Discussion
In the present study, we evaluated the role of NRBCs count in the diagnosis and prognosis of morbidity and mortality in critically ill neonates. Our results revealed that NRBCs could serve, among others, as prognostic and diagnostic markers in this population, especially in preterm neonates.In healthy adults, NRBCs are not present in the peripheral blood but studies have reported Contrarily, ROC curve analysis showed that NRBCs could not discriminate diagnosis or prognosis of sepsis or perinatal hypoxia between the term neonates of our study.

Discussion
In the present study, we evaluated the role of NRBCs count in the diagnosis and prognosis of morbidity and mortality in critically ill neonates. Our results revealed that NRBCs could serve, among others, as prognostic and diagnostic markers in this population, especially in preterm neonates. In healthy adults, NRBCs are not present in the peripheral blood but studies have reported their detection in 10-30% of critically ill patients [13]. Inflammation, hypoxia, or massive hemorrhage seem responsible for the appearance of NRBCs in peripheral blood, as these situations increase erythropoietic pressure and result in failure of the spleen to remove these cells from the circulation. Purtle et al. [34] demonstrated that the presence of NRBCs is associated with a significant increase in the odds of post-discharge hospital mortality in critically ill adults.
Schaer et al. [15] found that NRBCs are not an independent risk factor for bad outcomes in pediatric intensive care; however, they may have a prognostic value in the first month of life, although their association with outcome is much less pronounced beyond the neonatal period.
The presence of NRBCs in the circulation at birth and during the first week of life is an expected and rather physiological finding. In healthy neonates, NRBCs disappear from the peripheral blood within a few weeks of life, but they can return to the circulation in several pathological situations and their elevated value has been often correlated with poor short outcome and risk factors for poor neurodevelopment in neonates [16,35].
In septic neonates, data show that excess NRBCs in peripheral blood may help with the diagnosis of sepsis and correlates with the mortality of these neonates [36]. In our study, NRBCs showed a good performance as a prognostic marker for mortality in neonates with sepsis, especially in preterms.
NRBCs could also serve as hematologic markers for placental dysfunction, hypoxemia, and asphyxia as hypoxemia triggers erythropoietin release, resulting in stimulation of red blood cells. NRBCs count elevation at birth or persistence is linked to adverse short-term and long-term outcomes of neonates with perinatal hypoxia [18]. Boskabadi et al. [17] found that NRBC count in neonates with birth asphyxia were significantly higher than healthy controls and associated with poorer short-term outcome. In accordance with these findings, the current study showed that NRBCs count could serve as a discriminator marker for early diagnosis of perinatal hypoxia with excellent performance in preterm neonates (AUC 0.921) by using a cut-off value of ≥11.2%, with 80% sensitivity and 88.7% specificity. In a more recent study, Boskabadi et al. [7] demonstrated that theNRBC count can be used as a prognostic marker for neonatal asphyxia, which in combination with the HIE grade could indicate high infant mortality, and complications of asphyxia. Furthermore, NRBCs were found to be very Children 2020, 7, 197 8 of 10 useful in differentiating the neonates with hypoxic ischemic encephalopathy, leading to appropriate management and better outcome of these newborns [37]. The role of NRBCs count in critically ill neonates is confirmed by Morton et al. [38], who reported that among critically ill neonates, NRBCs are associated with significantly elevated mortality risk. In our study, although the median value for NRBCs count was found to be sometimes times higher for non-survivors, this parameter did not have good discriminative ability for predicting mortality in all the study population, probably because of the great variability in NRBCs and the small number of cases.
This single-center study also has some limitations which arise from the variability in NRBCs count and the small number of subjects. Another limitation might be the evaluation of NRBCs count only at the disease onset and this perhaps weakens the strength of the study with regards to the prognostic role of NRBCs in mortality.
In conclusion, NRBCs count may be encompassed among the early diagnostic and prognostic markers for neonatal intensive care unit patients. Further studies are needed to assess trends in NRBC values for critically ill neonates.