An association between infantile haemangiomas and erythropoietin treatment in preterm infants
- 1Division of Neonatology, University of Heidelberg Medical School, Heidelberg, Germany
- 2Department of Medical Biometry, University of Heidelberg, Heidelberg, Germany
- 3Department of Pediatric Hematology and Oncology, University Children's Hospital of Münster, Münster, Germany
- 4Department of Pediatrics, University Children's Hospital of Münster, Münster, Germany
- 5New York Obesity Nutrition Research Center, St. Luke's-Roosevelt Hospital Center, Columbia University, New York, New York, USA
- Correspondence to Jacqueline Bauer, Department of Pediatrics, University Children's Hospital of Münster, Albert-Schweitzer Str. 33, D-48149 Münster, Germany; jacqueline.bauer{at}ukmuenster.de
- Accepted 14 March 2011
- Published Online First 5 May 2011
Abstract
Background Infantile haemangiomas are benign vascular neoplasms that occur frequently in premature infants. The authors hypothesised that in addition to gestational age and birth weight, erythropoietin therapy may influence the incidence of these soft tissue tumours in preterm infants.
Methods 2563 infants born prematurely and admitted to the Division of Neonatology, University of Heidelberg Medical School were investigated in a retrospective analysis. Hospital charts for all infants were reviewed for clinical data. The primary endpoint was the percentage of infants who had received erythropoietin treatment and were diagnosed with a haemangioma.
Results Haemangiomas were diagnosed in 4.3% (n=110) of the 2563 preterm infants. These 110 infants had a median gestational age of 29 weeks (IQR 27–33 weeks) and the female:male ratio was 1.8:1. A higher incidence of haemangiomas (12–15%) was detected in premature infants with a lower gestational age (<31 weeks). Erythropoietin therapy was shown to be an independent risk factor after adjusting for all other known factors and oxygen therapy in multivariable analysis (HR 2.82, 95% CI 1.55 to 5.12). Subgroup analysis revealed that the effect was more pronounced in male than female infants (HR 3.61, 95% CI 1.52 to 8.57).
Conclusions This retrospective study demonstrates that erythropoietin treatment is associated with an increase in the incidence of these benign vascular tumours after adjusting for all other factors.
Introduction
Haemangiomas are the most common soft tissue tumours of infancy, recognised in 5–10% of children by 1 year of age.1,–,3 These benign vascular neoplasms consist of micro-vessels and endothelial cells that have the capacity for excessive proliferation but normally undergo eventual regression and involution.3 4 It has been suggested that growth factors affect the abnormal proliferation of endothelial cells in haemangiomas.4 Most infantile haemangiomas occur sporadically without a hereditary component, although a rare occurrence of familial vascular malformations with an autosomal dominant trait has been reported in the literature.5
Some authors have described a higher incidence in females, preterm infants, Caucasians and the newborns of mothers who have undergone chorionic villus sampling, although the reason for this higher frequency remains unclear.6,–,10 Two case reports have previously suggested a potential correlation between erythropoietin treatment and the development of haemangiomas in premature neonates.11 12 Erythropoietin is a well-known treatment for anaemia of prematurity,13 and stimulates erythropoiesis as well as angiogenesis.14 15 These effects might also influence the growth and possibly the development of infantile haemangiomas.
What is already known on this topic
-
▶ Haemangiomas are the most common soft tissue tumours of infancy.
-
▶ Lower gestational age and lower birth weight increase the risk of infantile haemangiomas.
What this study adds
-
▶ Erythropoietin treatment is associated with an increased incidence of these benign vascular tumours after adjusting for all other factors.
-
▶ Prospective trials to clarify the role of erythropoietin in the development or progression of haemangiomas are needed.
The aim of this study was to investigate the probable association between erythropoietin administration and the incidence of haemangioma in a large population of premature infants. We hypothesised that in addition to gestational age and birth weight, erythropoietin also increases the occurrence of these soft tissue tumours in premature infants.
Patients and methods
Study design
Premature infants (<37 gestational weeks) who were admitted to the Division of Neonatology, University of Heidelberg from 1999 to 2005 and survived to discharge, were enrolled in a retrospective cohort study. Hospital charts for all infants were hand reviewed for clinical data, the incidence of haemangiomas and treatment with erythropoietin for anaemia of prematurity. Gestational age was calculated from the last menstrual period and, if available, using early fetal ultrasound scans.
Haemangioma diagnosis
According to our usual clinical practice, haemangiomas were categorised using an international classification.6 Size, location and morphological subtypes were documented and in patients with ≥3 cutaneous haemangiomas, ultrasonography was performed to rule out haemangiomas in other organs.
Recombinant human erythropoietin treatment
Erythropoietin (NeoRecormon (epoetin beta); Roche Diagnostics, Mannheim, Germany) was administered subcutaneously or intravenously if haemoglobin was below 14 mg/dl and thrombocytes were <200 000 cells/µl from the fifth postnatal week for a period of 4 weeks. During the first 2 weeks of life, infants received red blood cell transfusions for anaemia of prematurity according to our guidelines for transfusion. The weekly erythropoietin dose was 750 IE/kg, divided into three single doses and adjusted weekly for body weight. Oral iron supplementation (2–4 mg/kg/day, Ferro Sanol; Schwarz Pharma, Mannheim, Germany) was given routinely to all infants. Erythropoietin treatment was prolonged up to 8 weeks if haemoglobin levels remained below 10 mg/dl and the reticulocyte count was low (<200 000 cells/µl). Administration was discontinued if thrombocytosis (>750 000 cells/nl) occurred.
Statistical analysis
Gestational age was categorised into three groups (22–26, 27–30 and 31–36 weeks) and birth weight was standardised by gestational age and grouped (0–25%, 25–75% and 75–100%).16
In addition to gestational age and standardised birth weight, the child's gender, oxygen therapy administration and erythropoietin treatment were investigated as potential independent factors influencing the development of haemangiomas. For analysis we fitted univariable and multivariable shared γ frailty models, with time from birth to occurrence of haemangioma or the date of discharge without observed haemangioma as the response variable. The inclusion of erythropoietin as a time-dependent covariate was chosen so as to model the different scenarios wherein erythropoietin administration may have taken place before or after the occurrence of the haemangioma. Furthermore, a frailty term was included in all models to account for multiple births. No model selection procedures were carried out; this was to ensure the reliability of the analysis of erythropoietin as the main factor of interest while adjusting for all other possible factors. We checked the assumption of proportional hazards in all models by plotting the scaled Schoenfeld residuals versus time and by testing for non-proportionality univariably for each factor as well as by performing a global test over all covariates as described by Grambsch.17 Further multivariable models including all main effects and one interaction term between erythropoietin and one of the other main factors at a time were carried out. In light of the known difference in the incidence of haemangioma between males and females, exploratory multivariable analyses were repeated in gender specific subgroups.
The significance level was fixed at α=0.05 and all tests of the HRs were carried out two-sided. Asymptotic 95% CIs for the HRs are given. The risk ratio of the incidence of haemangioma between all infants with and without erythropoietin was fixed as the main parameter of interest before the start of the evaluation. Hence only the result of the test concerning the effect of erythropoietin in the entire group of children is reliable. The p values of all other tests are to be interpreted with caution.
The analyses were carried out using SAS v 9.1 for Windows (SAS Institute, Cary, North Carolina, USA) or the statistical software R v 2.9.0 (R Foundation for Statistical Computing, Vienna, Austria).
Results
Prevalence
The study population comprised 1207 girls and 1356 boys, with a female:male ratio of 1:1.1 (table 1). There was one case of placenta praevia (1/110) in children with haemangiomas. In six of the 110 infants amniocentesis was performed with normal results, and there were no cases of chorionic villus sampling. Caesarean delivery was reported in all women.
Characteristics of the study population
Single or multiple haemangiomas were diagnosed in 4.3% (n=110) with a female:male ratio of 1.8:1. In four infants (3.6%) haemangiomas were present within the first 2 days after birth; in all other patients haemangiomas were recognised within the first 12 weeks of life (median 29 days, IQR 19–48 days after birth).
Location
The majority (62%) of the infants had a solitary haemangioma, 38 (34.5%) had two to five lesions, and four (3.6%) developed more than 10 lesions, which in three cases (2.7%) were associated with hepatic haemangiomas. Lesions were most commonly found on the trunk (52.6%), extremities (26.8%) and head and neck (20.6%).
Erythropoietin treatment and haemangiomas
The association between erythropoietin and haemangiomas is described in table 1. The median age for starting erythropoietin was the fifth week of life (IQR 3–7 weeks) and the median dose in each infant was 12 applications (IQR 10–15) of 250 IU/kg.
Possible risk factors for the development of haemangiomas were examined. Univariable analyses of all investigated factors revealed a significant effect of erythropoietin which increased the risk by a factor of 2.92, with female sex showing a 2.07-fold elevated risk compared to male gender (table 2). Multivariable analysis confirmed that these two factors were independently significant, while low birth weight showed a trend towards increased risk and oxygen therapy a trend towards decreased risk when adjusting for all other factors. The adjusted risk increase for erythropoietin treatment in this analysis is 2.82 (95% CI 1.55 to 5.12; p=0.0007).
Univariable and multivariable analysis of risk factors for the development of haemangioma
None of the tests for interaction between erythropoietin and other risk factors showed a significant result. However, exploratory subgroup analysis in gender specific groups showed that the magnitude of influence of the erythropoietin was different in males compared to females. As shown in table 3, males receiving erythropoietin have a 3.61-fold increased risk (95% CI 1.52 to 8.75; p=0.0037) of developing a haemangioma compared to untreated males, while in females the increase in risk is weaker (HR 2.22, 95% CI 1.02 to 4.82; p=0.045).
Multivariable analysis of risk factors in the subgroups of male and female infants
Early gestational age and low birth weight are both risk factors, but their influence is of similar magnitude and is not significant (tables 2 and 3).
Discussion
Our results confirm the hypothesis that gestational age and birth weight are not the only factors that influence the occurrence of infantile haemangiomas. Erythropoietin treatment seems to independently increase the incidence of haemangiomas, with exploratory subgroup analyses indicating that the effect is more pronounced in male than in female preterm infants. Haemangiomas were more common in females, with a female:male ratio of 1.8:1 in our study, consistent with previously published ratios ranging from 1.4:1 to 3:1.2 3
Amir et al8 investigated 973 preterm infants with birth weights of 500–2000 g during the first year of life and reported haemangiomas in 13%, with a female:male ratio of 1.4:1. Likewise, we found an increased incidence of haemangiomas in lower birthweight infants. Powell et al9 studied 615 preterm infants and found that the frequency of haemangiomas at 1 year of age was inversely related to gestational age at birth, being 8% for infants born after the 35th week, 11% for those born between the 30th and 35th weeks, and 19% for those born between the 25th and 29th weeks. Follow-up investigations were not available in our study, and so in the more mature infants (35–36 weeks) our analyses only reflect the first month of life, which might have caused the incidence of haemangiomas in this group to be slightly lower than expected. Drolet et al10 found that for every 500 g decrease in birth weight, the risk of haemangioma increased by 40%, and postulated that low birth weight was a stronger risk factor for haemangioma than gestational age. In our study no difference in the influence of these two factors was observed. In a recent prospective study it was found that infants with haemangiomas were more likely to be female, white non-Hispanic, premature and the product of multiple gestations (10.6%).18 In our population with haemangiomas, only 6% of the infants were the products of multiple births. A slightly higher incidence (17%) was detected in the infants below 30 weeks of gestation. No other perinatal variables were found in our cohort.
Immunohistochemical studies have demonstrated similarities between infantile haemangioma vasculature and placenta vasculature,19 with molecular profiling20 strongly suggesting a placental origin for infantile haemangioma. Additionally, a recent study has reported that haemangiomas are associated with the incidence and severity of retinopathy of prematurity and suggested that their occurrence and subsequent growth after birth may be a surrogate clinical marker for vasoproliferative retinopathy.21
Erythropoietin is used so as to avoid transfusions and/or treat anaemia in premature infants with low plasma levels of erythropoietin.22,–,24 Despite stimulating red blood cells, the administration of erythropoietin has not resulted in a clinically significant reduction in transfusions and or, consequently, in donor exposure in preterm infants.22,–,24 Several studies evaluating erythropoietin treatment reported limited adverse events, such as transient neutropenia and a possible increased risk of retinopathy of prematurity.13 23 24 Others suggested a protective effect of erythropoietin on the ischaemic retina.25 Recent investigations using various models of neonatal brain injury also favour early high-dose erythropoietin as a novel and effective pharmacological agent for neuroprotection at a time when the extremely immature brain is at highest risk of damage.26 No significant adverse events attributable to high doses of erythropoietin have been seen in the short-term in preterm infants. Nevertheless, extremely high concentrations of this agent may have adverse effects on neuronal cells and increase the risk of brain damage, particularly in combination with additional factors such as hypoxia and hyperoxia.26 27 The occurrence of haemangiomas was not considered in these studies.
Our investigation confirms an association between erythropoietin administration and the presence of haemangioma in premature infants. Erythropoietin was shown by multivariable analysis to be an independent prognostic factor, increasing the estimated risk by a factor of 2.82. Interestingly, in our study this effect was more pronounced in males receiving erythropoietin, as they had a 3.61-fold increased risk of developing a haemangioma compared to untreated males. One important limitation of our study is the retrospective collection of data. Nevertheless, the studied cohort was clearly defined before data collection, and routine erythropoietin administration and data documentation did not change during the study period.
Furthermore, prematurity-related disorders such as hypoxia and environmental factors during fetal growth could potentially cause an increase in the frequency of haemangiomas. It has been suggested that tissue ischaemia, a powerful stimulus for neovascularisation, is involved.25 The most common sites for haemangiomas established in this population support this assumption because the locations of the lesions are prone to ischaemia.28 On univariable and multivariable analyses we did not detect a statistically significant effect of oxygen therapy, although a trend for a decreasing effect was observed.
At present, accumulating information suggests that exogenous erythropoietin stimulates erythropoiesis and promotes angiogenesis.15 29 It is also possible that erythropoietin induces progression of malignant tumours either directly via erythropoietin receptors on cancer cells or indirectly through its angiogenic effects on tumour endothelial cells.30,–,32 A number of studies on erythropoiesis-stimulating agents have raised safety concerns regarding serious cardiovascular events and an increased risk of death from cancer.31,–,34 These agents might also act as growth factors for some tumours, speeding their progression and shortening survival. In cell lines derived from common paediatric tumours, the addition of exogenous erythropoietin increased the expression of antiapoptotic genes as well as the production and secretion of angiogenic growth factors.35 As erythropoietin expression is known to be induced by hypoxia, erythropoietin produced locally within hypoxic tumour tissue could stimulate tumour angiogenesis in a paracrine manner. The authors concluded that erythropoietin could thus play a putative oncogenic role in these tumours.35 Under physiological conditions, erythropoietin release in response to hypoxia is important in early embryogenesis by inducing vasculogenesis and angiogenesis. The prediction is not that erythropoietin increases the incidence of cancer, but rather that it stimulates progression and hence increases the detection of previously occult cancers. Addressing this concern will be difficult until the mechanism of erythropoietin stimulated tumour progression is better understood.
In summary, the present study demonstrates that erythropoietin and gender both independently influence the growth of haemangiomas. Furthermore, a trend was observed for an increase in risk by low birth weight and a decrease in risk by oxygen therapy. The absence of follow-up investigations after hospital discharge resulted in a very short observation period higher gestational age infants and most likely is responsible for the fact that the influence of low gestational age is not statistically significant in this study. We hypothesise that the angiogenic effects of erythropoietin may explain the haemangiomas in our study population. Because of the retrospective study design, it is not possible to conclude that the angiogenetic effects of erythropoietin have a definitive role in the development of haemangiomas in premature infants. Our results call for further research in prospective trials to clarify the role of erythropoietin in the development or progression of haemangiomas.
Footnotes
-
Competing interests None.
-
Provenance and peer review Not commissioned; externally peer reviewed.
