Objective. We studied the frequency, onset, duration, and prognosis of neutropenia in a neonatal hospital population to define subgroups of neonates who might benefit from cytokine therapy.
Study Design. The study comprised of 2 parts: in a first retrospective study (I), clinical data of neonates with sepsis (n = 168) were analyzed; in a second retrospective and prospective study (II), clinical data of neonates with neutropenia (n = 131) were studied. In study I, the analysis focused on septic neonates with and without neutropenia, and in study II, on neutropenic neonates with and without primary infection. In the prospective part of study II, granulocyte colony-stimulating factor (G-CSF) plasma concentrations were analyzed in neutropenic neonates (n = 32).
Results. Thirty-eight percent of septic neonates were neutropenic. Neutropenia lasted <24 hours in 75% of these patients. It was recorded before or on the day of the clinical onset of sepsis in 87% of patients. The overall incidence of neutropenia was 8.1%. Seventy-two percent of these neutropenic episodes occurred in patients without infection at the time of diagnosis of neutropenia. In the latter patients, the risk of infection secondary to neutropenia was 9%, affecting only premature neonates. Neutropenic episodes without infection were of longer duration and were accompanied by lower G-CSF plasma concentrations than were episodes associated with infection. The percentage of neutropenic episodes primarily associated with infection was higher in VLBW neonates than in term neonates. Likewise, the risk of infection secondary to neutropenia (27%) and the mortality attributable to infection and neutropenia (28%) were significantly higher than in term newborns.
Conclusion. Considering the priming time for induction of neutrophilia, G-CSF therapy in neonates presenting with severe bacterial infection and neutropenia may be too late. In contrast, neutropenic very low birth weight neonates without primary infection might benefit from prophylactic G-CSF treatment. neonatal sepsis, neutropenia, granulocyte colony-stimulating factor.
- VLBW =
- very low birth weight •
- G-CSF =
- granulocyte colony-stimulating factor •
- CBC =
- complete blood cell count •
- WBC =
- white blood cell count •
- GBS =
- group B streptococci
Neutropenia is a frequently observed finding in neonates. It has been reported to be associated with an increased mortality in neonatal sepsis,1 ,2 and to be a risk factor for bacterial infections in very low birth weight (VLBW) neonates with a history of maternal preeclampsia.3 ,4 Neutropenia also occurs in healthy term or preterm neonates.5 In neonates with sepsis, neutropenia has been interpreted as an exhaustion of the storage and proliferative pools of the bone marrow.6Administration of recombinant granulocyte-colony stimulating factor (G-CSF) was shown to induce peripheral blood neutrophilia after 24 hours and bone marrow neutrophilia after 72 hours in neonates with presumed sepsis.7 However, whether G-CSF accelerates neutrophil production and decreases mortality and morbidity in septic neonates with neutropenia remains controversial.8–10
This study was initiated to evaluate the frequency of neonatal neutropenia and its relation to infection in a hospital population, and to characterize the clinical manifestations with emphasis on prognosis. We aimed to define subgroups of patients who might benefit from a prophylactic or therapeutic cytokine treatment. Two analyses were performed. In study I, data of neonates with sepsis were analyzed retrospectively with respect to neutropenia. In study II, data of neutropenic neonates were evaluated partly retrospectively and partly prospectively with respect to infection. It was intended to differentiate between neutropenia that was caused by infection, and neutropenia occurring independently of infection, and, thus, might be a potential risk factor for infection. In addition, in the prospective subgroup of part II of the study, G-CSF plasma concentrations in neutropenic neonates were measured.
The 2 study parts (I and II) were performed at the University Children's Hospital, Freiburg, a 145-bed pediatric tertiary care center, serving an urban and rural population of ∼400 000 inhabitants. Before entry into the prospective part of study II, parental informed consent was obtained for every patient. The local ethics committee has approved the whole study.
Study I: Retrospective Analysis of Neutropenia in Neonatal Sepsis
Medical records of all neonates admitted from January 1987 to December 1992 with the diagnosis of sepsis or suspected sepsis were analyzed (n = 168). Complete blood counts (CBCs) and differential cell counts at the time of diagnosis of sepsis and during the consecutive 8 days were evaluated. If the patient was neutropenic at the time of diagnosis of sepsis, previous CBCs were documented until 2 normal white blood counts (WBCs) were included. Similarly, if the patient was neutropenic at the end of the collecting period of 8 days, additional blood counts were evaluated until 2 consecutive measurements were outside the neutropenic range.
Study II: Retrospective and Prospective Analysis of Neonatal Neutropenia With Respect to Infection
Although in study I the charts of all neonates treated between 1987 to 1992 had to be reviewed for CBCs, establishment of a computer system in the hospital in 1991 provided easy access to all CBCs. Using this tool, all neonates admitted between September 1991 and May 1993 (21 months) were enrolled in a retrospective analysis, if they showed at least 1 neutropenic WBC during the first 28 days of life (n = 91). All WBCs during the time of neutropenia and 2 WBCs before and after the neutropenic period were recorded. In a prospective analysis, all neonates admitted between June 1993 and February 1994 (9 months) were enrolled, if they were neutropenic during the first 28 days of life (n = 40). In 4 eligible patients, parental consent was not obtained; they were excluded from the study. Beginning from the time of diagnosis of neutropenia, a WBC was obtained daily until 2 successive WBCs were recorded within the normal range. In patients with neutropenia of >8 days duration, CBCs were obtained at least weekly. In 32 of the 40 patients of the prospective study, G-CSF plasma concentrations were measured during neutropenia.
For study I and II, data analyzed included birth weight, gestational age, Apgar scores at 1, 5, and 10 minutes, primary site of infection, results of blood culture and swab cultures, antibiotic treatment, respiratory distress syndrome, and outcome. Maternal data analyzed included history of preeclampsia, fever, antibiotic treatment, time of rupture of membranes before delivery, and histopathology of placenta and cord.
Neutropenia was defined by modified reference ranges established by Manroe et al11 and Mouzinho et al12 (Fig 1). The cutoff points for neutropenia used in this study were well below the 95% confidence interval of the reference ranges for newborns with a birth weight >1500 g and ≤1500 g. The duration of neutropenia was defined as the time from the first to the last neutropenic WBC. If only 1 neutropenic WBC was obtained, the duration was set as 0 hour. In study II, neutropenia was defined as infection-associated, if an infection was diagnosed within 72 hours before or 24 hours after the diagnosis of neutropenia. These infections were called primary infections. Secondary infections were defined as infections diagnosed >24 hours after the diagnosis of neutropenia. Confirmed sepsis was defined as the clinical suspicion, which was confirmed by a positive blood culture, a positive cerebrospinal fluid culture, or a necrotizing enterocolitis diagnosed on radiograph. Suspected sepsis was defined as the clinical suspicion and at least 3 of the following criteria: 1) C-reactive protein >20 mg/L within 48 hours after onset of clinically suspected sepsis, 2) pneumonia diagnosed on radiograph or by microscopic or cultural evidence in tracheal aspirate, 3) gastric aspirate or urine with microscopic or cultural evidence of bacterial infection or positive latex agglutination test for urine for group B streptococci (GBS), 4) proportion of immature granulocytes (bands and less mature forms) to total neutrophils >.2 documented in any WBC within 48 hours after onset of clinically suspected sepsis, or 5) maternal fever or antibiotic treatment within 48 hours of delivery or premature rupture (>48 hours) of membranes before delivery or histopathology of amnionitis of placenta or cord. Early-onset sepsis was defined as the onset of sepsis during the first 72 hours of life, and late-onset sepsis was defined as diagnosis at the age of >72 hours. Pneumonia was diagnosed on radiograph or by clinical signs and evidence of bacteria in tracheal aspirate. Urogenital tract infections were defined by clinical symptoms, leukocyturia, and a positive urine culture of >104 bacteria/mL (midstream clean catch technique). Respiratory syncytial virus and rotavirus infections had to be confirmed by the presence of virus–antigen in secretions or stools. Death was considered to be related to infection if it occurred within 7 days after the diagnosis of infection and if an independent investigator judged it by medical records to be related. Preeclampsia was defined as hypertension (diastolic blood pressure >90 mm Hg) plus proteinuria and edema. The syndrome with hemolysis, elevated serum liver enzyme concentrations, and low platelet count was diagnosed when preeclampsia and hemolysis, elevated serum liver enzyme concentrations, and a low platelet count were present.
Blood was obtained by heelstick, venipuncture, or via an indwelling catheter. CBCs were performed electronically, and differential WBCs were performed microscopically on Pappenheim-stained blood films. All WBCs and differential counts were corrected for the presence of nucleated red blood cells. Plasma concentrations of G-CSF were measured by a double-sandwich enzyme immunoassay technique (Quantikine, R&D Systems, Europe, Abingdon, United Kingdom). Detection limit of the assay as indicated by the manufacturer was 10.9 pg/mL. Duplicate measurements were performed for each plasma sample. Frozen aliquots of plasma were thawed on ice at the time of analysis.
Discrete variables were compared by means of Fisher's exact test or χ2 test, as adequate. Continuous variables are presented as means ± standard deviation or medians with range, as adequate. Nonparametrically distributed data were compared using the Mann-Whitney U test. The statistical analysis was performed with Systat software (Systat, Inc, Evanston, IL). In part I of the study, the analysis was focused on a comparison of the 2 groups of sepsis with and without neutropenia, and in study II, on a comparison of the groups of neutropenia associated with primary versus secondary infection or no infection. Differences were considered to be statistically significant with P values <.05.
Study I: Retrospective Analysis of Neutropenia in Neonatal Sepsis
From 1987 to 1992, 4429 neonates had been admitted to the University Children's Hospital. In 168 neonates (3.8%), confirmed or suspected sepsis was diagnosed retrospectively (Table 1). When comparing the 73 neonates with confirmed sepsis with the 95 neonates with suspected sepsis, there was no significant difference in the clinical characteristics between the 2 groups except mortality, which was higher in neonates with confirmed sepsis. The frequency of neutropenia did not differ significantly between the 2 groups (Table 1). Therefore, we chose to present the analysis on the combined patient groups with confirmed and suspected sepsis. Early-onset sepsis was seen in 65%, and late-onset sepsis in 35% of the septic patients.
Frequency of Neutropenia in Neonatal Sepsis
In the 168 neonates with sepsis, a total of 869 WBCs were documented during the first 8 days after the diagnosis of sepsis, or—in case of persistent neutropenia—until 2 nonneutropenic WBCs had been obtained. In 63 of these newborns (38%) with sepsis or suspected sepsis, neutropenia was present. In early-onset sepsis, the frequency of neutropenia was 45% and significantly higher than in late-onset sepsis (24%; 50/110 vs 13/58; P = .004). Neutropenia occurred in all birth weight classes but was significantly associated with low gestational age (P = .005; Fig 2). The frequency of maternal preeclampsia did not differ between septic neonates with and without neutropenia (7/63 vs 6/105).
Mortality With Respect to Neutropenia
Applying our definition of neutropenia, modified from the reference values established by Manroe11 to neonates with birth weight >1500 g, and the reference range described by Mouzinho12 to VLBW neonates with birth weight ≤1500 g (Fig 1), the mortality attributable to sepsis did not differ significantly among neonates with and without neutropenia (12/63 vs 31/105). When we applied the reference values of Manroe8to all neonates including those with birth weight ≤1500 g—as authors of previous studies did4 ,5 ,9 ,13—the number of preterm neonates classified as neutropenic increased. Simultaneously, the mortality rate in the neutropenic group increased to 23% and was significantly higher than the rate of 8% in the nonneutropenic group (15/67 vs 7/101; P = .008). Mortality was associated with low gestational age and low birth weight (P < .001). Neutropenic patients who died from sepsis had a lower minimal neutrophil count than did neutropenic patients who survived (709/μL ± 680 vs 1489/μL ± 1090; P = .02).
Onset and Duration of Neutropenia in Septic Neonates
In 65% of the neonates (41/63) with sepsis and neutropenia, neutropenia was diagnosed at the day of the clinical onset of sepsis; in 13% (8/63), neutropenia developed within 3 days after the onset of sepsis; and in 22% (14/63), neutropenia was present before the clinical onset of sepsis (Fig 3). Although the range of the neutropenic period was 0 to 8 days, the median duration of neutropenia was 0 hour, defined as a single neutropenic WBC. In 75% of patients (47/63), the documented duration of neutropenia was <24 hours.
Neutropenia With Respect to Blood Culture Isolates
Sepsis was confirmed by a positive blood culture result in 30% (33/110) of the neonates with early-onset sepsis, compared with 69% (40/58) of patients with late-onset sepsis. GBS was the most frequently isolated pathogen in early-onset sepsis (17/33), Escherichia coli and Staphylococcus aureus in late-onset sepsis (11/40 each). There was no obvious correlation between the species of bacteria isolated and the incidence of neutropenia.
Study II: Retrospective and Prospective Analysis of Neonatal Neutropenia
The clinical characteristics of the 91 neonates in the retrospective and the 40 patients in the prospective part of study II did not differ significantly, except for a higher incidence of neonates with birth weight >1500 g in the retrospective group (24 vs 4;P = .04; Table 2). The analysis was performed separately on the study population of the retrospective and the prospective study. Because the results did not differ significantly, we present the following analysis on the combined data of the 131 patients of the retrospective and prospective part of study II.
Frequency of Neutropenia With Respect to the Onset of Infection
From September 1991 until March 1993, 1662 neonates were admitted to the hospital. The frequency of neutropenia during the study was 8.1% (135/1662). For 4 of these 135 patients, informed consent could not be obtained; therefore, they were not further evaluated.
At the time of diagnosis of neutropenia, 28% of the study population (37/131) had a primary infection, defined as an infection diagnosed within 72 hours before to 24 hours after the diagnosis of neutropenia. Of the 72% of neutropenic neonates (94/131) who did not have an infection at the time of diagnosis of neutropenia, 91% (86/94) remained free of infection. Only 9% (8/94) of the primarily not infected neutropenic neonates developed an infection during the later course of neutropenia. These infections were defined as secondary. The 8 children with secondary infection were preterm neonates with birth weight <2500 g, 4 of them ≤1500 g. One of these neonates died from the secondary infection. The frequency of maternal preeclampsia, a risk factor for neonatal neutropenia,3 ,4 was not significantly different in primarily infected neutropenic neonates (3/37), compared with the primarily not infected neutropenic neonates (13/94).
Neutropenia With Respect to Birth Weight and Onset of Infection
In neonates with neutropenia and birth weight ≤1500 g (n = 28), neutropenic episodes were accompanied by a primary infection in 46%, and by a secondary infection in 15% of cases. In contrast, in neonates with a birth weight of 1501 g to 2000 g (n = 44), only 16% and 4% of the neutropenic episodes were associated with a primary or secondary infection, respectively (P = .007; Fig 4). Because healthy full-term neonates with neutropenia are not admitted to the hospital, the percentage of neutropenic episodes associated with infection in neonates with birth weight >2000 g was higher than in those with birth a weight of 1501 g to 2000 g. Because of this selection, the correlation of low birth weight and the incidence of primary infection associated with neutropenia was significant for neonates with birth weight ≤2000 g (P = .004), but not for the entire study group.
Age at Diagnosis and Duration of Neutropenia
Seventy-seven percent of the neutropenic episodes were diagnosed during the first week of life. Neutropenic episodes associated with primary infection were significantly shorter than those associated with no or secondary infections (median: 0 vs 23 hours; range: 0–148 hours vs 0–1965 hours; P = .007). Neutropenia longer than 24 hours occurred in only 11% of the patients with primary infections, but in 40% of patients with no or secondary infections. Severity of neutropenia, represented by the lowest neutrophil count, did not differ significantly between the 2 groups.
Mortality Attributable to Infection
Five neonates died attributable to primary infections, and 1 died attributable to a secondary infection. All of these 6 neonates were born prematurely; 5 of them had birth weight ≤1500 g. The association of a high mortality and low birth weight or low gestational age was significant (P = .02 and P = .0091, respectively). None of the newborns >2500 g with infection and neutropenia died.
G-CSF Plasma Concentrations in Neutropenic Patients With Respect to Infection
In 32 of 40 patients of the prospective part of the study, G-CSF plasma concentrations were analyzed. Plasma had been collected at a median of 0 hours after diagnosis of neutropenia with a range 0 to 8 hours for the septic patients, and 0 to 142 hours for the nonseptic patients. For the group of septic patients, the first plasma sample after establishing the diagnosis of sepsis was analyzed. For the group of nonseptic patients, the first sample after onset of neutropenia was analyzed. When comparing G-CSF concentrations in different samples of neutropenic patients without sepsis during the period of neutropenia, no significant difference was observed (data not shown). In the group of septic patients, the period of neutropenia was very short. Therefore, as a rule, only 1 sample was available during neutropenia. The median age of the neonates at the time of collection was 58 hours. The time point of sample collection did not differ significantly between patients with and without infection. G-CSF plasma concentrations were significantly higher in neutropenic patients with primary infection (n = 10) than in patients with no or secondary infection (n = 22; median: 3461 pg/mL vs 52 pg/mL; P = .002; Fig 5).
In this study, like in others, neonatal sepsis proved to be an important clinical problem contributing significantly to morbidity and mortality (14% of neonates with confirmed and suspected sepsis). Neutropenia in neonatal sepsis was common, and in the same range as reported previously.13 Although several studies demonstrated a higher mortality in neutropenic than in nonneutropenic septic neonates,1 ,2 ,5 there was no significant difference in the mortality rate between the 2 groups in our study. The discrepancy is attributable to the different definitions of neutropenia; in this study, the physiologically lower neutrophil counts in VLBW neonates were considered. Because normal preterm VLBW neonates have reference ranges for neutrophils that differ significantly from that of mature neonates,12 ,14 ,15 we believe that birth weight-adjusted normal values should be applied when studying neutropenia and its complications.
In the majority of septic neonates, neutropenia was present at diagnosis of sepsis with a duration of <1 day in 75% of them. It has been suggested that the application of G-CSF in septic neonates will accelerate the recovery of neutrophil counts and improve prognosis. However, considering a priming time of 24 hours for induction of peripheral neutrophilia,7 one might speculate that G-CSF treatment at the day of diagnosis of sepsis and neutropenia will be too late. In fact, a recent placebo-controlled trial of G-CSF administration to neutropenic neonates with early-onset sepsis could not demonstrate significant differences in the neutrophil count between the 2 groups during the 3-day study.9 As might be expected from the natural course described in our study, the neutrophil count rose significantly by day 1 in both the G-CSF and placebo-treated groups. In another study with a retrospectively selected case–control group, G-CSF recipients reached significantly higher neutrophil counts only after 7 to 10 days of treatment.8 These observations are supported by data from the neonatal rat model demonstrating that G-CSF administration 6 to 18 hours after inoculation with GBS failed to improve survival, whereas 7 days of maternal G-CSF administration before delivery and neonatal inoculation did.16 ,17
Although neonates with sepsis with and without neutropenia have an increased mortality, two thirds of the neutropenic neonates did not have an infection at the time of diagnosis and remained free of infection. The severity of neutropenia in these patients did not differ from that observed in septic neonates, but its duration was significantly longer. In neutropenic neonates without infection at the time of diagnosis, the risk of a subsequent infection was only 8.5%. These so-called secondary infections were defined as infections occurring >24 hours after the onset of neutropenia. This definition was based on results of study I, in which only 6% of neutropenic episodes associated with sepsis had been diagnosed >24 hours before the clinical onset of sepsis. In particular, full-term neonates did not develop infections secondary to neutropenia, even when neutropenia was longstanding with 46, 47, 48, or 81 days duration. According to these data, G-CSF therapy in not infected neonates with birth weight >1500 g may not be warranted in terms of prevention of neutropenia-associated infection. In this study, perinatal events known to cause neonatal neutropenia (such as maternal preeclampsia, neonatal respiratory distress syndrome, perinatal asphyxia, or intraventricular hemorrhage5 ,13) were equally frequent in neutropenic neonates without infection as in those in which neutropenia was thought to be caused by the infection (data not shown).
In neutropenic neonates without infection, G-CSF plasma concentrations were significantly lower than in neonates with infection-associated neutropenia. In the latter group, G-CSF levels were elevated in the same range as previously reported.18–21 In some neonates, the G-CSF plasma concentration was >10 000 pg/mL, indicating that these levels may not be significantly increased after additional administration of G-CSF. Because only a limited number of samples were tested for G-CSF concentrations, these data have to be interpreted with caution.
Neutropenic VLBW neonates are a population at special risk. According to our results, only a minority of neutropenic VLBW newborns remained free of infection. It has been proposed that the well-known predisposition of premature neonates to bacterial infections is—at least in part—attributable to the physiologically low neutrophil counts observed in fetuses early in the third trimester.14Furthermore, it had been suggested that the low neutrophil count is caused by low G-CSF levels.18 Data on G-CSF serum levels in VLBW neonates are contradictory,18–20 and scarce in VLBW neonates with neutropenia. In this study, VLBW neonates with neutropenia had highly elevated G-CSF plasma concentrations. Two trials with small patient numbers indicated that G-CSF administration can increase neutrophil counts in low birth weight neonates with preeclampsia-associated neutropenia. Considering the relatively high incidence of secondary infections in neutropenic VLBW infants, further studies to evaluate the efficacy of prophylactic G-CSF therapy are warranted.
- Received March 8, 1999.
- Accepted October 18, 1999.
Reprint requests to (C.M.N.) University Children's Hospital, Mathildenstrasse 1, D-79106, Freiburg, Germany. E-mail:
- Gillian ER,
- Christensen RD,
- Suen Y,
- Ellis R,
- van de Ven C,
- Cairo SC
- ↵Kocherlakota P, La Gamma EF. Human granulocyte colony-stimulating factor may improve outcome attributable to neonatal sepsis complicated by neutropenia. Pediatrics. 1997;100(1). URL: http://www.pediatrics.org/cgi/content/full/100/1/e6
- Schibler KR,
- Osborne KA,
- Leung LY,
- Le TV, Baker SI, Thompson DD
- Mouzinho A,
- Rosenfeld CR,
- Sanchez PJ,
- Risser R
- ↵Davies NP, Buggins AGS, Snijders RJM, Jenkins E, Layton DM, Nicolaides KH. Blood leucocyte count in the human fetus. Arch Dis Child. 1992;:399–403
- Gessler P,
- Luders R,
- Konig S,
- Haas N,
- Lasch P,
- Kachel W
- Novales J,
- Salva AM,
- Modanlou HD,
- Medlock ES
- Cairo M,
- Plunkett J,
- Mauss D,
- van de Ven C
- Gessler P,
- Kirchmann N,
- Kientsch-Engel R,
- Haas N,
- Lasch P,
- Kachel W
- La Gamma EF, Alpan O, Kocherlakota P
- Copyright © 2000 American Academy of Pediatrics