 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
ARTICLE |
a Department of Neonatology
b Laboratory of Psychoneuroimmunology
f Department of Obstetrics and Gynaecology, University Medical Center Utrecht, Utrecht, Netherlands
c Department of Neonatology, Isala Clinics Zwolle, Zwolle, Netherlands
d Department of Neonatology, University Medical Center Leiden, Leiden, Netherlands
e Department of Neonatology, VU University Medical Center, Amsterdam, Netherlands
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
PATIENTS AND METHODS. A total of 156 prematurely born children were included in this retrospective matched cohort study. Children treated with dexamethasone (n = 52) or hydrocortisone (n = 52) were matched for gestational age, birth weight, grade of infant respiratory distress syndrome, grade of periventricular or intraventricular hemorrhage, gender, and year of birth. A reference group of 52 children not treated with corticosteroids was included for comparison. Plasma adrenocorticotropic hormone and cortisol in response to a social stress task were determined. Cytokine production was analyzed after in vitro stimulation of whole-blood cultures.
RESULTS. The Trier Social Stress Test adapted for children induced an adrenocorticotropic hormone and cortisol response in all of the groups. The adrenocorticotropic hormone response was blunted in the dexamethasone group. The overall cortisol level was lower in the dexamethasone than in the hydrocortisone and reference group. Cortisol and adrenocorticotropic hormone in the hydrocortisone and reference groups were similar. The ratio of T-cell mitogen-induced interferon-
/interleukin-4 secretion was significantly higher in the dexamethasone group than in the hydrocortisone group. Interferon- production and the ratios of interferon-/interleukin-4 and interferon-/ interleukin-10 were significantly higher in the dexamethasone group than the reference group. However, production of these cytokines did not differ between the hydrocortisone and the reference groups.
CONCLUSION. Neonatal treatment of prematurely born children with dexamethasone but not with hydrocortisone resulted in long-lasting programming effects on hypothalamus-pituitary-adrenal axis and on the T-helper 1/T-helper 2 cytokine balance. Follow-up of these children is required to investigate long-term clinical consequences. We recommend that authors of previously performed randomized, controlled trials on neonatal glucocorticoid treatment include immune and neuroendocrine analyses in prolonged follow-up of these children.
Key Words: follow-up studies • glucocorticoids • immune response • neonatology • stress
Abbreviations: CLD—chronic lung disease • HPA—hypothalamus-pituitary-adrenal • Th—T helper • IFN—interferon • IL—interleukin • ACTH—adrenocorticotropic hormone • TSST-C—Trier Social Stress Test adapted for children • GA—gestational age • BW—birth weight • IRDS—infant respiratory distress syndrome • CS—cesarean section • MR—mineralo-glucocorticoid receptor • GR—glucocorticoid receptor
Neonatal glucocorticoid treatment, in particular dexamethasone, has been frequently and successfully used for the prevention of chronic lung disease (CLD) in prematurely born children. However, in humans, long-term adverse effects of neonatal dexamethasone therapy on cognitive and motor development have been described.1–5 In a double-blind randomized trial, it was shown that, at school age, ex-premature children treated neonatally with dexamethasone performed significantly poorer on cognition and motor development.5 Based on these data, neonatologists have been advised to limit the use of glucocorticoid to exceptional circumstances only and to lower the dose of glucocorticoid.6–8 However, in severe cases its use cannot be avoided.6,9 Moreover, worldwide, many prematurely born children have already been neonatally treated with relatively high doses of dexamethasone during the last 15 years. Therefore, it is mandatory to continue to investigate the possible long-term consequences of neonatal glucocorticoid treatment
The most widely used glucocorticoid for treatment of CLD is dexamethasone,10 although suggestions to use other glucocorticoids such as hydrocortisone have been made.10–14 In 2003, we reported the results of a retrospective study comparing the clinical efficacy and the long-term effects of neonatal treatment with hydrocortisone or dexamethasone on school performance.15 We did not observe differences in clinical efficacy (reduction of extra oxygen need and weaning from the ventilator) between hydrocortisone and dexamethasone. However, dexamethasone-treated children had a reduced school performance at 7 to 10 years of age as compared with hydrocortisone-treated children.15 More recently, we have confirmed these observations in an independent cohort, and we have described that dexamethasone-treated girls have more behavioral problems than hydrocortisone-treated girls or girls who had not been treated neonatally with glucocorticoid.16 Moreover, in the same cohort, motor skills of neonatally dexamethasone-treated children, as determined by using the movement assessment battery for children, were reduced compared with hydrocortisone-treated or untreated children.16
In our previous animal studies, we investigated effects of neonatal dexamethasone treatment on behavior, cardiovascular system, immune response, and neuroendocrine system in rats at adult age.17–23 In this rat model, neonatal treatment with dexamethasone not only resulted in long-lasting behavioral changes but also in a reduction of the hypothalamus-pituitary-adrenal (HPA) axis activity to novelty stress at adult age.21,22 In addition, the lipopolysaccharide-induced increase in corticosterone, the end product of the HPA axis, was reduced in adult rats treated with dexamethasone neonatally.18 Moreover, the T-helper (Th)1/Th2 cytokine balance of spleen cells of the rats was shifted toward a Th1-dominated phenotype.18 Th1 cytokines, for example, interferon (IFN)-, promote cellular immunity, whereas Th2 cytokines, such as interleukin (IL)-4 and IL-10, are known as anti-inflammatory cytokines and promote humoral immunity. A predominance in Th1 cytokine production has been associated with increased incidence of autoimmune diseases, whereas high Th2 cytokine levels are associated with allergy and asthma.24 Indeed, in our animal study on long-term effects of dexamethasone, the shift toward proinflammatory cytokines was associated with increased susceptibility for autoimmunity as determined in the animal model for multiple sclerosis, experimental autoimmune encephalomyelitis.18
The aim of our present study was to test the hypothesis that early neonatal glucocorticoid treatment also has long-lasting consequences for the cytokine balance and for the activity of the HPA axis in humans. In a retrospective cohort study, we included ex-premature children at the age of 7 to 10 years who had been treated neonatally with dexamethasone and compared them with a matched group of children treated with hydrocortisone and to a reference group of prematurely born children who had not been treated with glucocorticoids. The activity of the HPA axis was analyzed by determining plasma adrenocorticotropic hormone (ACTH) and cortisol levels in response to a psychosocial stress test, the Trier Social Stress Test adapted for children (TSST-C),25 which includes a public speech task followed by mental arithmetic in front of an audience. Possible effects of early neonatal treatment on the immune response were assessed by determining T-cell mitogen-induced and monocyte stimulation-induced cytokine production in a prestress blood sample.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Eligibility for inclusion in 1 of the study groups was as follows: surviving the neonatal period, availability to participate in the study protocol, neonatal cerebral ultrasound showing maximal grade 2 periventricular hemorrhage as classified according to Papile et al26, and absence of major congenital anomalies. Infants with periventricular leukomalacia, as defined by cerebral ultrasound at postdischarge follow-up, were also excluded. The hydrocortisone and dexamethasone treatment groups were composed as follows: the charts of all of the consecutively admitted preterm infants born after <32 completed weeks of pregnancy in the participating NICUs were systematically reviewed. From the NICU of the University Medical Center Utrecht, 131 eligible infants were treated with hydrocortisone during the defined time period. The dexamethasone group, recruited in a similar way from the other 3 NICUs, which used only dexamethasone, consisted of 198 eligible infants. Ultimately, we were able to reliably match 52 pairs from the hydrocortisone group and the dexamethasone group for gestational age (GA), birth weight (BW), gender, severity of infant respiratory distress syndrome (IRDS; scored as no, moderate, or severe respiratory distress syndrome according to clinical symptoms and the Giedeon classification),27 whether a minor intraventricular hemorrhage (grades 1–2) existed, the year of birth, and the time period of admission. The 52 selected hydrocortisone children in this study did not significantly differ from the 79 nonselected hydrocortisone children for GA, BW, gender, severity of IRDS, the incidence of periventricular or intraventricular hemorrhage (grades 1–2), or postnatal age at the start of glucocorticoid treatment. This was also true for the 52 children of the dexamethasone cohort versus the 146 nonselected dexamethasone-treated children. Importantly, the incidence of prenatal treatment with the glucocorticoid betamethasone did not differ between included and nonincluded hydrocortisone or dexamethasone children.
In a similar way, the same number of participants for a reference group was recruited from the 4 participating NICUs. The reference group consisted of prematurely born infants who had not been treated postnatally with glucocorticoids, did not have periventricular leukomalacia or major periventricular or intraventricular hemorrhage (grade 3 or more), and no other major complications during the neonatal period. However, it was impossible to match the reference group for the presence of IRDS or for gender (see also Table 1). The hydrocortisone and dexamethasone groups could be perfectly matched for the parameters indicated in Table 1.
|
The TSST-C has been described in detail by Buske-Kirschbaum et al.25 In brief, the test consists of a 45-minute relaxation period, after which the first blood sample is drawn. The child is then transferred to a conference room; on arrival in the room, the second blood sample is drawn. Just before performing the speech task in front of an audience, a third sample is taken. After a 5-minute speech and a subsequent 5 minutes of arithmetic (number subtraction), the fourth sample is drawn. After a 10-minute debriefing during which the child is praised for its excellent performance, the fifth blood sample is drawn. After transportation back to the relaxation room and another 45-minute relaxation video, the sixth and last poststress sample is obtained.
ACTH and Cortisol
Blood was collected in EDTA tubes, and plasma was separated by centrifugation and stored at –80°C. Cortisol was determined with a commercially available enzyme immunoassay (DSL, Sinsheim, Germany). Interassay and intra-assay coefficients of variance were <12% and <10%, respectively. Plasma ACTH was determined with automated immunoassays (Centauer, Bayer, Fernwald, Germany; intra-assay and interassay variability of <6 and <8%, respectively).
Immune Function
Heparinized venous blood was collected just before the stress test. IL-4, IL-10, and IFN- were measured in supernatants of stimulated whole-blood cultures as described previously.28 In short, 100 µL of whole blood (diluted 1:10 with Roswell Park Memorial Institute- 1640, Gibco, Grand Island, NY), 100 U/mL of penicillin, 100 µg/mL of streptomycin, and 2 mM of L-glutamine were stimulated for 72 hours with 50 µL of anti-CD2/CD28 monoclonal antibodies (CLB, Amsterdam, Netherlands; 37°C/5% CO2 in 96-well round-bottom plates).
Monocyte cytokine production was measured in cultures of whole blood 1:10 diluted with Roswell Park Memorial Institute-1640, supplemented with antibiotics and stimulated with lipopolysaccharide, (Escherichia coli 0127: B8, Sigma; final concentration: 2 ng/mL) at 37°/5% CO2 in 96-well flat-bottom plates for 18 hours. All of the supernatants were stored at –80°C. Cytokine levels in the culture supernatants were measured using standard ELISA kits (CLB).
Composition of the Peripheral Blood Cell Population
Circulating numbers of monocytes, granulocytes, T-cell subsets, B cells, and natural killer cells were assessed in whole blood by using dual color fluorescence analysis (monoclonal antibody labeled with either fluorescein or phycoerythrin) with a Becton Dickinson FACS Calibur flow cytometer (Becton Dickinson, Mountain View, CA): CD14+ (monocytes), CD3+ (total T cells), CD4+ (Th), CD8+ (suppressor/toxic T), CD19+ (B cells), and CD16/56+ (natural killer cells). Absolute numbers of cells were calculated on the basis of a total leukocyte blood count.
Statistics
Statistical analysis was performed by using SPSS 13.0 (SPSS Inc, Chicago, IL). For categorical values, 
2 tests were performed. For analysis of the effect of the stressor on HPA axis activity, repeated-measurement analysis of variance was used with posthoc Bonferroni tests. Cortisol data were log transformed before statistical analysis to obtain normal distribution. Area under the curve data were calculated by using the method described by Pruessner et al.29 Cytokine data were analyzed by analysis of variance using square root-transformed data and Bonferroni posthoc test. In tables and figures, untransformed values are presented. A P value of <.05 was considered statistically significant. Age at the day of testing, gender, BW, and IRDS score were used as covariates. We did not observe any effect of antenatal exposure to betamethasone, and correction for antenatal betamethasone exposure did not alter the results.
|
RESULTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
1 full course of antenatal betamethasone did not differ between groups (P = .84). At follow-up, the group of dexamethasone treated children was, on average, 4 to 5 months younger than the hydrocortisone (P < .01) and reference groups (P < .05). There were no other differences observed between the groups at follow-up (Table 2).
|
|
Activity of the HPA Axis: Cortisol
Exposure to the TSST-C induced a significant cortisol response in all of the groups (time effect, P < .0001; Fig 2B). There was a significant group effect (P = .01), but no time x group interaction. Posthoc analysis revealed that the plasma level of cortisol was significantly lower in the dexamethasone group than in the reference group (P = .04) or the hydrocortisone group (P = .02). Comparison of the area under the curve for cortisol during the entire period of measurement confirmed that there was a significantly lower output of cortisol in the dexamethasone group than in the other 2 groups (P < .05). The maximum stress-induced increase in cortisol from baseline level (plasma cortisol level at t = 20 minutes minus plasma cortisol level at t = –15 minutes) did not differ significantly between groups (P = .14). There were no differences in cortisol between the hydrocortisone group and the reference group. Correction for age, BW, gender, and IRDS grade did not alter the results.
Immune Response: Th1/Th2 Cytokine Pattern
As is shown in Table 3, the production of IFN- in the dexamethasone group was significantly higher than in the reference group (P < .05). We did not observe group differences in the capacity to produce IL-4 or IL-10 in response to stimulation with the T-cell mitogen (Table 3). Hence, there was a statistically significant group difference in the ratio between IFN- and IL-4 and in the ratio between IFN- and IL-10. The IFN-/IL-4 ratio was higher in the dexamethasone group than in the hydrocortisone group and the reference group. The ratio IFN-/IL-10 was higher in the dexamethasone group than in the reference group (P = .02). In contrast, however, we did not observe any differences in cytokine production between children in the hydrocortisone group and the reference group. We did not observe any group differences in the lipopolysaccharide-induced production of tumor necrosis factor-
and IL-6 (Table 3) or in the composition of the peripheral blood cell population (Table 4).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Neonatally dexamethasone-treated children showed a significantly lower cortisol level during and after exposure to a psychosocial stress test as compared with neonatally hydrocortisone-treated children and with children in the reference group. There was no difference between hydrocortisone-treated children and children in the reference group. Also, the overall level of ACTH tended to be lower in the dexamethasone group than in the hydrocortisone and reference groups. Moreover, the ACTH response to the stressor was blunted in the dexamethasone group. Although overall cortisol levels were lower in the dexamethasone-treated group, the stress-induced increase in cortisol over baseline was similar in all of the groups.
Our data indicate that HPA axis activity at school age is decreased in prematurely born children treated with dexamethasone in the early neonatal period. In previous animal studies, we and others have investigated the long-term effects of neonatal dexamethasone in a dose schedule that is comparable to the schedule used for CLD of prematurity on the HPA axis.18,22,30,31 In rats, neonatal dexamethasone treatment resulted in a blunted ACTH and cortisol response to novelty stress, to a conditioned fear paradigm, and to restraint, as well as to an inflammatory stressor at adult age (8 weeks).18,22,30,31 However, in the animal model we do not have indications for a reduced overall activity of the HPA axis. The diurnal rhythm of ACTH and cortisol was normal in rats treated with dexamethasone neonatally, and ACTH and corticosterone at baseline did not differ between groups.22,30 Moreover, the responses to infusion with ACTH or CRH were normal in dexamethasone-treated animals, but sensitivity to feedback inhibition by corticosterone was increased in the rats.22 Unfortunately, we do not have information on the diurnal rhythm of cortisol in the children or on the sensitivity for negative feedback. In future follow-up studies we will test glucocorticoid feedback capacity via a CRH-dexamethasone bolus administration. When comparing the results of the rat studies and the current results we have to take into account that the rats were relatively older than the children tested.
What are possible consequences of reduced HPA-axis activity for health and disease? Disturbances in HPA axis activity have been implicated in the development of learning disabilities and impaired coping with the environment.32 Although most studies suggest that increased rather than decreased glucocorticoid levels will impair learning, there is also evidence that hypocortisolism can modulate learning and memory.33 In this respect it is worth mentioning our recent studies showing that neonatal treatment of children with dexamethasone, but not with hydrocortisone, is associated with an increased need for special education at the age of 7 to 10 years.15,16 Moreover, in the same cohort of children, we have observed long-term behavioral consequences of neonatal dexamethasone treatment but not of hydrocortisone treatment; dexamethasone-treated girls displayed more behavioral problems, such as attention problems and social problems.16
From animal research, it is known that a low corticosterone level is associated with an increased sensitivity to infections and to autoimmunity.34 Indeed, we have described that neonatal treatment of rats with dexamethasone resulted in long-lasting changes in the HPA axis, as well as in increased production of proinflammatory Th1-type cytokines like IFN- at adult age.18 Interestingly, our present data suggest that, in humans, neonatal dexamethasone treatment can have long-lasting consequences for the cytokine balance as well. In line with the animal data, the Th1/Th2 (IFN-/IL-4) cytokine balance at the age of 7 to 10 years was significantly increased in prematurely born children treated with dexamethasone as compared with the hydrocortisone group. Moreover, IFN- production and hence, the IFN-/IL-4 and the IFN-/IL-10 ratios were higher in the dexamethasone group than in the reference group, whereas there were no differences between the hydrocortisone and reference groups on these parameters.
The possible clinical consequences of the altered Th1/Th2 balance remain to be determined. In rats, neonatal dexamethasone treatment resulted in a more severe course of experimental autoimmune encephalomyelitis, an animal model for multiple sclerosis, at adult age.18 For humans, the latter could mean that there is an increased risk for the development of autoimmune disease later in life after neonatal dexamethasone treatment. It is known, however, that most inflammatory autoimmune diseases do not develop until after puberty and, therefore, more prolonged follow-up studies are required.
The shift in immune balance toward Th1 cytokines may also lead to a reduced risk for the development of allergies and asthma.35 In our study sample, we did not observe a difference in the incidence of asthma and allergies between groups (data not shown), but more prolonged and detailed follow-up studies are required to confirm this observation.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
An intriguing finding in this long-term follow-up study was that we did not find the disturbances in the activity of the pituitary-adrenal axis or of the immune system in children treated during the neonatal period with hydrocortisone . It is known that 2 types of corticosterone receptors are involved in regulating the activity of the HPA axis: mineralo-glucocorticoid receptors (MRs) and glucocorticoid receptors (GRs). MRs have a high affinity for glucocorticoids and are thought to be responsible for maintaining baseline cortisol levels throughout the diurnal cycle. Feedback control of stress-induced elevations in cortisol is regulated predominantly by GR.37 In addition, it is known that dexamethasone has a high affinity for GR but does not bind MR. In contrast, hydrocortisone has a high affinity for MR and also binds GR.38 Furthermore, hydrocortisone, but not dexamethasone, binds to endogenous glucocorticoid binding globulin, which may have important consequences for biological availability and the capacity to penetrate the brain in vivo.37 The half-life of dexamethasone is thought to be 36 to 48 hours, whereas that of hydrocortisone is 8 to 12 hours.38 In human adults it has been shown that 50 times higher doses of hydrocortisone than dexamethasone are needed for suppression of HPA axis activity.39 The initial dose of hydrocortisone used to treat CLD of prematurity in the children included in our study was only 10 times higher than the dose of dexamethasone used. Although the acute clinical effects (lowering extra oxygen need and weaning from the ventilator) for dexamethasone and hydrocortisone were similar, it may well be that the biological effects on other target systems are markedly different because of the differences in receptor binding characteristics, biological availability, and half-life.15,40,41 The discrepancy cannot simply be explained by the fact that the dose of dexamethasone used for the treatment of CLD is too high, because it has been shown that a lower dose of dexamethasone has less beneficial effects on CLD.42
The present study also has limitations. First, the data obtained in this study were limited by the fact that we performed a retrospective matched cohort study and not a randomized prospective study. Another drawback is that the children from the dexamethasone and hydrocortisone groups were recruited from different centers, and we cannot exclude the possibility that other unknown differences between the centers that used dexamethasone and the center using hydrocortisone may have contributed to the observations, although the Dutch protocol concerning glucocorticoid use for CLD in preterm infants was precisely followed. The data presented in this article should, therefore, be taken with caution.
Another limitation is that a significant proportion of the children included in this study also had been exposed to antenatal glucocorticoid treatment. However, there seemed to be no differences between groups in the percentage of children that had received a full course of antenatal betamethasone (Table 1). Moreover, interestingly, in a separate analysis, we did not observe any effect of antenatal betamethasone exposure on any of the outcome parameters. Therefore, we conclude that the observed differences in stress response and immune activity were caused by neonatal rather than prenatal glucocorticoid exposure.
Our study raises important questions about the long-term effects of neonatal dexamethasone treatment for CLD in preterm infants. Because many children alive today received dexamethasone in the early neonatal period, it is imperative to gain additional insights about the possible health risks associated with this treatment by following these individuals to a later age. Finally, the results of the present study urge for a double-blind randomized, controlled trial on hydrocortisone versus placebo treatment for CLD of prematurity, because our data suggest that hydrocortisone may be a safer alternative for dexamethasone.
|
ACKNOWLEDGMENTS |
|---|
We are greatly indebted to all of the parents and children who volunteered to participate in this study. We thank Jitske Zijlstra and Mirjam Maas for excellent technical assistance, Dr C. Kirschbaum (University of Technology, Dresden, Germany) for analyzing cortisol, and Dr O. E. Janssen (Laboratory of Endocrinology, University of Essen, Essen, Germany) for determining ACTH.
|
FOOTNOTES |
|---|
Address correspondence to Cobi J. Heijnen, PhD, Laboratory of Psychoneuroimmunology, Office KC 03.068.0, University Medical Center Utrecht, Lundlaan 6, 3584 EA Utrecht, Netherlands. E-mail: c.heijnen{at}umcutrecht.nl
The authors have indicated they have no financial relationships relevant to this article to disclose.
| What's Known On This Subject In animals, neonatal dexamethasone treatment has long-lasting consequences for behavior, the immune system, and the neuroendocrine system. In humans, behavioral effects at school age after neonatal dexamethasone treatment have been described.
|
| What This Study Adds We describe for the first time immune and neuroendocrine consequences of neonatal dexamethasone but not hydrocortisone treatment at school age in a cohort of prematurely born children.
|
|
REFERENCES |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2008 American Academy of Pediatrics. All rights reserved. |
||
|
||
, reference group; 
, hydrocortisone; 
, dexamethasone. Data represent means and SDs.