|
|
| |||||||||||||||||||||||||||||||||||||||
Background Anemia of prematurity is characterized by low reticulocyte counts and inadequate erythropoietin response, for which many very-low-birth-weight infants receive multiple blood transfusions. We investigated whether early treatment of such infants with recombinant human erythropoietin would reduce their need for transfusions.
Methods We performed a controlled, blinded trial in 241 infants with very low birth weights at 12 centers in six European countries. When three days old, the infants were randomly assigned either to the epoetin group or to the control group. Those in the epoetin group received 250 IU of epoetin beta per kilogram of body weight subcutaneously three times a week from day 3 to day 42 (for a total of 17 doses); those in the control group did not receive this drug. Infants in both groups received oral iron (2 mg per day) from day 14 onward.
Results The control infants needed a mean of 1.25 transfusions each, as compared with 0.87 transfusion for epoetin-treated infants (P = 0.013). The median cumulative volume of blood transfused per kilogram per day was 0.41 ml in the control group (first quartile, 0 ml; third quartile, 0.8 ml) and 0.09 ml in the epoetin group (first quartile, 0 ml; third quartile, 0.8 ml) (P = 0.044). The rate of success, defined as an absence of need for transfusions and a hematocrit that never fell below 32 percent, was 4.1 percent in the control group and 27.5 percent in the epoetin group (P = 0.008). Epoetin was most beneficial in boys with birth weights of 1200 g or more and a base-line hematocrit of 48 percent or more. No toxic effects were observed in the epoetin group; as compared with the control group, the epoetin group had an increased incidence of septicemia (14 vs. 7 episodes, P not significant) and reduced weight gain (520 vs. 571 g, P = 0.02).
Conclusions Infants with very low birth weights have less need of transfusions if given epoetin beta during the first six weeks of life (250 IU per kilogram three times a week). We recommend early epoetin treatment for all such infants, but further studies of nutrition and iron supplementation during treatment are needed.
In vitro studies have indicated that recombinant human erythropoietin stimulates erythroid progenitors from preterm infants in a normal dose-response relation8,9. Pilot studies in small numbers of preterm infants suggested that low doses of this agent have some effect,10,11,12 but this finding was not confirmed in subsequent controlled trials6,13.
We undertook the present study to determine whether recombinant human erythropoietin, given as epoetin beta in a dose of 750 IU per kilogram of body weight per week, would prevent anemia and reduce the need for transfusion in infants with very low birth weights.
Methods
A randomized blinded trial was performed at 12 centers in six European countries, with the consent of appropriate ethics committees and the infants' parents. Eligible infants were randomly assigned to a control group or an epoetin-treated group on the third day of life, by means of numbered, sealed envelopes. Block randomization was performed at each center.
Primary End Point
Treatment was defined as successful if an infant had no need of transfusions and had a venous-blood hematocrit that never fell below 32 percent during the study period.
Calculation of Sample Size
To detect an increase in the success rate from 50 percent, as expected in controls,13 to 75 percent, with a power of 90 percent at a two-tailed level of significance of 5 percent, 85 infants would be required for each treatment group. The estimated rate of withdrawal during the study was 29 percent13. Consequently, 240 infants had to be enrolled.
Blinding
Most participating doctors were reluctant to administer repeated subcutaneous injections of a placebo to low-birth-weight infants. Therefore, two teams were formed at each center: treating physicians determined whether infants could be enrolled or withdrawn from the study, decided whether they should receive transfusions, and monitored them without knowing their treatment group; "dosing investigators" performed the randomization and administered the epoetin beta but were not involved in the infants' care. When treatment was to be given, a dosing investigator carrying a "black box" containing appropriate equipment visited each infant, administered the study medication, and placed adhesive strips on both thighs (of both epoetin recipients and controls), which remained there until the next visit. During this procedure, staff and parents had to leave. A treating physician or a dosing investigator assigned to an infant had to serve in that capacity as long as the infant was studied.
Criteria for Entry and Withdrawal
All infants with birth weights of 750 to 1499 g were considered for randomization. Infants with any of the following predefined criteria were excluded before randomization: hemolytic disease (11 infants), exchange transfusion (6), a venous-blood hematocrit above 59 percent (24), a leukocyte count above 60,000 per cubic millimeter (0), a platelet count above 750,000 per cubic millimeter (0), renal failure (10), systolic blood pressure above 90 mm Hg (1), cyanotic heart disease or a major congenital malformation requiring surgery (15), participation in another therapeutic trial (9), anticipated discharge before day 42 (113), a gestational age of more than 34 full weeks (10), or a lack of parental consent (155).
The criteria for withdrawal from the study after randomization included any of those listed above and the following: ventilation or a continuous positive airway pressure with a fraction of inspired oxygen above 0.40 after day 6, a severe reaction at the injection site, receipt of medication that might have bone marrow toxicity, a vertically transmitted infection, major surgery, discharge before 42 days, or withdrawal of parental consent.
Patients
From September 1991 to December 1992, 598 infants were admitted, of whom 354 were excluded from the study as described above. Three of the 244 infants who underwent randomization were also excluded: all data on 2 infants were lost, and treatment (epoetin) was inadvertently omitted in 1 infant, whose records were not completed. The remaining 241 infants were evaluated in an intention-to-treat analysis. Sixty-one infants had birth weights of 750 to 999 g, 86 weighed 1000 to 1249 g, and 94 weighed 1250 to 1499 g. One hundred thirty-six infants were born before 30 weeks of gestation.
After randomization, 61 infants were withdrawn for the reasons listed above. Three infants completed the study but underwent ventilation longer than the protocol allowed. Thus, 177 infants completed the study according to the protocol.
Administration of Epoetin Beta
Infants assigned to the epoetin group received 250 IU of epoetin beta per kilogram, injected subcutaneously into a thigh on Mondays, Wednesdays, and Fridays and begun as soon as possible after randomization (day 3 to 5). Treatment continued until day 40 to 42, for a total of 17 doses. Lyophilized epoetin beta (Boehringer-Mannheim, Mannheim, Germany) stored in vials containing 1000 IU was dissolved with sterile water so that the injected volume ranged from 0.25 to 0.55 ml.
Recommended Iron Treatment
Oral iron supplementation (2 mg per day) was started on day 14 in all infants. If the serum ferritin level fell below 100 ng per milliliter or signs of iron deficiency appeared (or both events occurred), the dose of iron was increased. Vitamin E supplementation was not part of the protocol.
Transfusions
Guidelines for transfusion were derived from the experience of the participating doctors. Infants who were receiving ventilation or who were less than two weeks old and had signs of anemia were given transfusions if their hematocrit fell below 40 percent, their hemoglobin concentration fell below 14 g per deciliter (8.7 mmol per liter), or blood samples totaling at least 9 ml per kilogram had been obtained from them since their previous transfusion. Spontaneously breathing infants more than two weeks old whose fraction of inspired oxygen was less than 0.40 were given transfusions if they had signs of anemia and their hematocrit fell below 32 percent and their hemoglobin concentration below 11 g per deciliter (6.8 mmol per liter); if they had no signs of anemia, the corresponding cutoff values were 27 percent and 9 g per deciliter (5.6 mmol per liter). The team of treating physicians decided the timing and volume of transfusions within these guidelines.
Monitoring and Blood Sampling
Delayed cord clamping was recommended14. The heart and respiratory rates, blood pressure, blood gas values, and renal and liver function were monitored (data not shown). Blood sampling was carried out on day 3, day 12 to 14, day 24 to 26, and day 40 to 42 as part of routine care. The volume of the blood samples was recorded. Cranial ultrasound scans were obtained at least on day 3 and from day 12 to day 14. Intraventricular hemorrhage was graded according to the classification of Papile et al.15. Ophthalmoscopic examination for retinopathy of prematurity16,17 was performed at least once up to four weeks after the estimated date of delivery.
Hematologic Measurements
The hematocrit was measured in venous blood and blood units after centrifugation. The volume of transfused red cells was determined from the volume of blood transfused and the blood-unit hematocrit. The leukocyte, erythrocyte, and platelet counts and the hemoglobin concentration were determined with automated blood counters. The neutrophil count was calculated from the number of leukocytes and the percentage of neutrophils in a blood smear. The reticulocyte count was "normalized" (corrected) to a hematocrit of 45 percent by multiplying the count by the actual hematocrit and dividing by 45.
Statistical Analysis
Kaplan-Meier estimations, which allow patients with censored data to be included in an analysis in an appropriate manner, were used to express the success rate and the rate of freedom from the need for transfusions in each treatment group. Log-rank tests were used to analyze the distributions of failure-free time and transfusion-free time since the start of treatment. Cox's proportional-hazards analysis was used to adjust the time-dependent treatment effects for prognostic factors found to be significant. Spearman's correlation coefficient was calculated for the volume of red cells transfused and blood sampled. The Mann-Whitney U test was carried out to compare the number of transfusions per infant, the cumulative volume of blood transfused, the cumulative blood loss, the absolute change between base-line and final values, and body weight between the two groups. Cox's proportional-hazards model was also used to analyze specific adverse events, taking into account censoring and predictors unrelated to treatment.
All P values are two-tailed. A P value below 0.05 was considered to indicate statistical significance. Analyses were carried out with the Statistical Analysis System (SAS Institute, Cary, N.C.).
Results
Table 1 shows the characteristics of both treatment groups at entry. There were no significant differences between the groups.
|
The median cumulative blood loss due to diagnostic tests was 0.74 ml per kilogram per day in the control group (first quartile, 0.51 ml; third quartile, 1.24 ml) and 0.83 ml in the epoetin group (0.57 ml and 1.45 ml, respectively) (P = 0.104). These losses were inversely related to birth weight: 0.59 ml per kilogram per day among infants weighing 1250 to 1499 g (first quartile, 0.43 ml; third quartile, 0.77 ml), 0.83 ml among those weighing 1000 to 1249 g (0.60 ml and 1.42 ml), and 1.27 ml among those weighing 750 to 999 g (0.95 ml and 1.70 ml). The median blood loss was 0.78 ml in boys and 0.75 ml in girls (P = 0.651).
At study entry, grade I or II intraventricular hemorrhage was found in 8 infants in the control group and 10 in the epoetin group, and grade III hemorrhage in 1 infant in the epoetin group.
Success Rate and Need for Transfusions
Seventeen infants in the control group and 28 in the epoetin group needed transfusions before they entered the study (P = 0.071). There was no significant difference between the groups in the need for transfusion and the success rates during the first two weeks of the study period (Figure 1 and Figure 2 and Table 2). Thereafter, the epoetin group had a reduced need for transfusion and a higher success rate (P<0.001 for each comparison). The most important predictors of the success of the treatment were birth weight and the base-line hematocrit (P<0.001 for each comparison). Moreover, boys responded better to epoetin beta than girls.
|
|
|
Changes in Hematologic Values
The changes in the hematocrit and reticulocyte count during the study are shown in Figure 3. The two treatment groups did not differ significantly at the end of the study in their leukocyte, neutrophil, and platelet counts or total plasma protein levels (data not shown).
|
There was no significant difference between the two groups in the number of infants withdrawn (Table 3). Of the 61 infants withdrawn, 15 had birth weights of 750 to 999 g, 25 weighed 1000 to 1249 g, and 21 weighed 1250 to 1499 g; 37 of these infants were born before 30 weeks of gestation, and 35 were boys. The infants withdrawn were similar in weight and gestational age to those who continued in the study.
|
|
The median cumulative dose of iron was 37 mg per kilogram in the control group (first quartile, 24 mg; third quartile, 54 mg) and 45 mg in the epoetin group (28 mg and 54 mg, respectively) (P = 0.219). The median serum ferritin concentration decreased during the study period by 16.5 ng per milliliter in the control group (first quartile, decrease of 97.0 ng; third quartile, increase of 45.3 ng) and by 73.0 ng per milliliter in the epoetin group (first quartile, decrease of 133.5 ng; third quartile, decrease of 22.5 ng) (P<0.001). No hemolysis due to iron supplementation was noted.
Weight Gain
Infants in the epoetin group gained weight more slowly than those in the control group (median gain, 520 vs. 571 g; P = 0.02 [first and third quartiles, 300 and 635 g vs. 361 and 750 g]).
Cost Effectiveness
The Zurich center calculated the cost effectiveness of treatment per infant. One vial containing 1000 IU of epoetin beta cost $25 (in U.S. dollars), one unit of red cells $80, transfusion equipment $35, treatment of hepatitis C (estimated incidence, 0.6 percent) $840, and treatment of infection with human immunodeficiency virus (estimated incidence, 0.005 percent) $7. With a mean of 1.25 transfusions per control infant and 0.87 transfusion per epoetin-treated infant, the calculated cost of treatment was $1,203 for a control infant and $1,262 for an epoetin-treated infant. However, the costs and rates of complications of transfusion vary from center to center and from country to country.
Discussion
Several studies of the treatment of small numbers of preterm infants with recombinant human erythropoietin have been published,6,10,11,12,18,19,20 usually reporting promising results. These papers have been difficult to compare because of differences in study design, criteria for entry, timing and dose of drug administration, and nutritional supplementation. Some studies were uncontrolled10,11,12.
We studied only infants with birth weights of less than 1500 g and gestational ages of less than 34 weeks so that we could exclude infants who were relatively mature but small for their gestational age, in whom the anemia of prematurity is not a major problem. We tried to prevent this anemia by beginning treatment early, as did Carnielli et al.18. Our approach therefore differed from that of Ohls and Christensen,19 who gave late treatment (from day 22 to day 70) for symptomatic anemia (hematocrit, 
30 percent), and that of Shannon et al.,6,20 who gave early-to-intermediate treatment (from day 4 to day 35) for anemia (hematocrit, 35 or 37 percent) to infants weighing less than 1250 g.
To exclude patients with severe iatrogenic anemia, we withdrew infants who required prolonged ventilation. This may explain why the need for transfusion in our control group was lower than the need reported by Strauss7 and Brown et al.21.
We have shown that epoetin beta reduces the need for transfusion in very-low-birth-weight infants, even though the infants treated with this agent received transfusions when their hematocrits were higher than those of the controls, and despite blood loss due to diagnostic tests. Boys whose birth weight was 1200 g or more and whose base-line hematocrit was 48 percent or higher benefited most from epoetin beta.
With the doses of epoetin beta and iron used in this study, we could not reduce the need for transfusion during the first two weeks of life. This result may have been due to clamping of the umbilical cord too early, high levels of blood loss, a delayed start of iron supplementation, or a delayed effect of epoetin beta. We therefore favor starting treatment with this drug early.
Although not statistically significant, the higher rate of transfusions in the epoetin group before entry may have biased the success rate. Epoetin beta may have averted only one donor exposure per infant, and its cost may not seem to favor prophylaxis with this agent. On the other hand, the hematocrits of the infants given the drug tended to be maintained nearer "normal" than those of the controls.
The decrease in serum ferritin levels in the epoetin group reflects active erythropoiesis. Although serum ferritin values do not provide complete information about iron storage, they were available at all participating centers. Our iron supplements might have been inadequate for optimal erythropoiesis. In pilot studies, Shannon et al. increased the daily intake of oral iron from 3 to 6 mg per kilogram6,20. Carnielli et al.18 administered 20 mg of iron per kilogram per week intravenously without complications.
Epoetin beta may affect other hematopoietic cells: at high concentrations, it has been found to decrease the formation of neutrophils from progenitors in vitro; however, no changes in blood neutrophils, monocytes, or lymphocytes were observed in weanling rats after injections of 2000 IU per kilogram22. In patients with the anemia of chronic renal failure, Dessypris et al.23 found that recombinant erythropoietin stimulated progenitors of erythrocytes, granulocytes, and platelets. All well-controlled studies conducted to date6,13,18,20 in preterm infants have found that this drug does not affect granulopoiesis. In the present study, we found no effects on platelets or granulocytes that could be attributed to the epoetin beta. Although not statistically significant, the higher incidence of septicemia in the epoetin group may be a cause for concern. Treatment with epoetin beta required 17 subcutaneous injections per infant, and it also depleted iron stores. Either factor might have increased the risks of infection, especially in very premature infants.
The infants we treated with epoetin beta gained less weight than the controls, like the infants studied by Shannon et al.20. Bechensteen et al.24 found similar growth rates among infants given recombinant erythropoietin and control infants, all of whom were fed extra protein. It is likely that the stimulation of erythropoiesis increases protein and caloric needs.
Recently, Emmerson et al.25 reported two cases of sudden infant death syndrome after treatment with recombinant human erythropoietin. In our trial, three infants died of this disorder: one infant had been treated with epoetin beta and completed the study, one had been assigned to this treatment but was withdrawn after receiving one dose because of prolonged ventilation, and one was a control. Neither the first trial of recombinant human erythropoietin for anemia of prematurity13 nor the present results suggest that the incidence of sudden infant death syndrome increases after this treatment.
Epoetin prevents the anemia of prematurity and effectively and safely reduces the need for transfusion. We regard the administration of epoetin beta as one tactic in a logical therapeutic strategy for managing anemia in very-low-birth-weight infants. This strategy should also include attempts to optimize these infants' blood count through placental transfusion at birth, minimize their blood losses due to diagnostic testing, and improve their nutrition to promote growth and hematopoiesis.
Supported by a grant (Sonderforschungsbereich 174/A9) from Deutsche Forschungsgemeinschaft and by Boehringer-Mannheim.
Source Information
From Universitatsklinikum Rudolf Virchow (R.F.M., M.O.) and Universitatsklinikum Steglitz (H.T.V.), Freie Universitat Berlin, Berlin, Germany; Boehringer-Mannheim, Mannheim, Germany (P.S.); Universitat Heidelberg, Heidelberg, Germany (O.L.); Universitat Zurich, Zurich, Switzerland (G.D.); Olgahospital, Stuttgart, Germany (G.H.); Royal Maternity Hospital, Belfast, United Kingdom (H.L.H.); Centre Hospitalier Universitaire Cochin Port-Royal, Paris (G.M.); Universitat Munster, Munster, Germany (G.J.); Universite Catholique de Louvain, Brussels, Belgium (G.V.); University Hospital Nijmegen, Nijmegen, the Netherlands (B.A.S.); Charite, Humboldt Universitat, Berlin (E.L.G.); Queen Mother's Hospital, University of Glasgow, Glasgow, United Kingdom (B.M.H.); and the Department of Hematology, University of Wales, Cardiff, United Kingdom (C.A.J.W.). The institutions and other coworkers of the European Multicentre Erythropoietin Study Group are listed in the Appendix.
Address reprint requests to Dr. Obladen at the Department of Neonatology, Universitatsklinikum Rudolf Virchow, Freie Universitat Berlin, Heubnerweg 6, D-14059 Berlin, Germany.
References
The following institutions and workers are members of the European Multicentre Erythropoietin Study Group.
Participating Centers (listed in order of the number of infants enrolled): Klinikum Rudolf Virchow, Freie Universitat Berlin, Berlin -- K. Abraham, C. Buhrer, and C. Domeyer; Universitat Heidelberg, Heidelberg -- J. Poschl; Universitat Zurich, Zurich -- P. Baeckert, A.M. Bucher, and H.U. Bucher; Olgahospital Stuttgart, Stuttgart -- M. Wagner, M. Kroll, and H. Gulde; Royal Maternity Hospital, Belfast -- M.G. O'Connor, F.A. Casey, and B.G. McClure; Universitatsklinikum Steglitz, Freie Universitat Berlin, Berlin -- M. Bartsch; Centre Hospitalier Universitaire Cochin Port-Royal, Paris -- C. Clamadieu, M. Monset-Couchard, and P. Mussat; Universitat Munster, Munster -- H. Rabe, E. Michel, and S. Lindner; Universite Catholique de Louvain, Brussels -- C. Debauche, D. Moulin, and S. Clement; University Hospital Nijmegen, Nijmegen -- L.A.A. Kollee; Charite, Humboldt Universitat, Berlin -- H. Weigel and S. Ledwon; and Queen Mother's Hospital, University of Glasgow, Glasgow -- G. Stewart.
Steering Committee: M. Obladen and R.F. Maier (Klinikum Rudolf Virchow, Freie Universitat Berlin, Berlin) and P. Scigalla (Boehringer-Mannheim, Mannheim, Germany).
Consultant in Hematology: C.A.J. Wardrop (University of Wales, Cardiff).
Consultant in Statistics: D. Messinger (Boehringer-Mannheim, Mannheim, Germany).
Data Management and Evaluation: W. Wierich, J. Bruning, and R. Vonk (AFB Comstat, Berlin).
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Recombinant Erythropoietin in Very-Low-Birth-Weight Infants
Beck D., Calame A., Masserey E., Carnielli V. P., Riol R. D., Montini G., van den Anker J. N., Montini G., Carnielli V. P., Maier R. F., Obladen M., Wardrop C. A.J.
Extract |
Full Text
N Engl J Med 1994;
331:676-678, Sep 8, 1994.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
