Postnatal Corticosteroids to Prevent or Treat Chronic Lung Disease Following Preterm Birth

The discussion is focused on 4 areas: type of corticosteroid (DEX, hydrocortisone [HC], and other); intended purpose (prevention or treatment); dosing (low or high), and route of administration (systemic or inhaled). In general, the literature review was limited to randomized, prospective clinical trials and systematic reviews. When possible, odds ratios have been converted to either relative risk (RR) or relative benefit increase (RBI)^a to facilitate interpretation.

Of the various PCSs, DEX and HC have been studied the most extensively. Randomized control trials of these systemic CSs have been arbitrarily divided into early (prevention; started ≤7 days of age) and late (treatment; started >7 and <28 days of age) in most systematic reviews and meta-analyses published, and this statement follows this approach.^b Because dosing may alter the risk and benefit profile, a separate discussion of low versus high dosing has been added.

The most recent Cochrane review of the use of early PCS to prevent CLD concluded that the benefits of early DEX (successful extubation and survival without CLD at 36 weeks’ postmenstrual age [PMA]) may not outweigh adverse effects.⁹ There were no significant differences in mortality. However, there were higher rates of hyperglycemia, hypertension, hypertrophic cardiomyopathy, gastrointestinal bleeding or perforation, and growth failure. Long-term follow-up studies reported an increase in abnormal neurologic examinations and cerebral palsy, although the methodologic quality was inadequate in some studies with assessments limited to preschool age, and no study was powered to detect adverse long-term outcomes.

The most recent Cochrane review of the use of late PCS to treat CLD concluded that there was evidence of both benefit and harm of late DEX; that because of limitations in the evidence, it would be prudent to reserve the use of late PCS (including DEX) for infants who cannot be weaned from mechanical ventilation; and that, if DEX is used, the dose and duration should be minimized.¹⁰ This conclusion is similar to recent recommendations from a panel of European experts¹¹ as well as the Canadian Pediatric Society Fetus and Newborn Committee.¹² 

Since these 2 Cochrane reviews, there have been 2 RCTs of DEX in ventilator-dependent preterm infants at 10 to 21 days of age. The first study, called MINIDEX, compared a very low dose (50 µg/kg per day) of DEX versus a placebo with extubation as the primary outcome.¹³ This multicenter study planned to enroll 94 infants, but because of frequent open-label PCS use and a high rate of sepsis, recruitment proved difficult, and the study was halted. Among evaluable subjects, a higher proportion of infants in the DEX group were extubated by study day 7 (63% vs 33%), but no conclusions should be drawn from such a limited sample size.¹³ 

The second study compared a 42 day dosing regimen with a 9 day dosing regimen.¹⁴ Infants in the 9-day group could receive up to 2 additional 9 day courses if respiratory criteria were met; after 42 study days, open-label PCSs were permitted in both groups. Rates of survival without CLD and rates of rehospitalization for respiratory illness were similar. By age 7 years, 2 of 30 infants in the 42 day group were lost to follow-up, and 3 of 29 infants in the 9 day group had died. Of survivors who were evaluated, mean Wechsler Intelligence Scale scores were similar; however, significantly more infants in the 42 day group were in regular classrooms without individualized education plans (75% vs 38%, P < .05). The authors concluded that attempts to minimize PCS exposure may not be beneficial, although the small sample size, large number of exclusions (34% of eligible infants were excluded because of low 5 minute Apgar scores and/or birth weight below the 10th percentile), high percentage of multiple DEX courses in the 9 day group (33%), and frequent open-label use of PCSs (27%) among both groups suggest that this conclusion may not be generalizable.

Review of the literature for the 2010 AAP policy statement concluded that high daily doses (≥0.5 mg/kg) of DEX were linked to adverse neurodevelopmental outcomes, but because there had not been a linkage reported in low-dose trials, additional studies of lower-dose DEX were warranted. Authors of a 2017 Cochrane review analyzing RCTs that compared different dosing regimens concluded that recommendations on the optimal dosage and timing could not be made on the basis of the current level of evidence and suggested that if late PCSs are used, both dose and duration should be minimized.^9,10 There have been no new RCTs published since that Cochrane review.^c

Although DEX was the focus of most of the earlier studies, interest has turned to HC, given the numerous adverse effects associated with DEX use and a growing understanding of the physiologic differences between DEX, a synthetic glucocorticoid, and endogenous cortisol. DEX, unlike cortisol, has little mineralocorticoid effect but high potency and a long half-life; these features combine to increase the risk of deleterious effects on the developing nervous, cardiac, pulmonary, and gastrointestinal systems.^15,16 The developing brain has both mineralocorticoid and glucocorticoid receptors; exogenous glucocorticoids, like DEX, suppress endogenous cortisol secretion but do not substantially bind to mineralocorticoid receptors, leaving these receptors unoccupied for prolonged periods, which leads to neuronal apoptosis.^17,18 HC, which has both glucocorticoid and mineralocorticoid activity, acts more like endogenous cortisol and has been under intensive study in recent years as a potentially safer alternative to DEX.

The 2010 AAP policy statement concluded that existing data were insufficient to make any recommendation regarding the use of HC to prevent CLD, although there was evidence to suggest that giving HC for the first 2 weeks of life might increase survival without CLD, particularly for infants delivered in the setting of perinatal inflammation.⁸ A 2017 Cochrane review of early PCS to prevent CLD includes 3 RCTs published since the 2010 AAP policy statement.^9,10 Two were small feasibility trials in ELGANs with low blood pressure;^19,20 the third was the PREMILOC study, a large, multicenter randomized controlled trial (RCT) of low-dose HC in ELGANs for the first 10 postnatal days.²¹ The PREMILOC authors reported a statistically significant increase in survival without CLD at 36 weeks’ gestation (60% vs 51%; RBI, 0.22; 95% confidence interval [CI], 0.01 to 0.59; P = .04). However, in the subgroup of infants born at 24 to 25 weeks of gestation, the authors noted a statistically higher rate of late-onset sepsis in the HC group (40% vs 23%; RR, 1.70; 95% CI, 1.08 to 2.70; P = .02). Mortality was not significantly different between groups. Inclusion of these studies into the 2017 Cochrane review meta-analysis resulted in a small but statistically significant increase in survival without CLD at 36 weeks’ gestation (RBI, 0.04; 95% CI, 0.01 to 0.09), similar to that observed in trials using DEX (RBI, 0.08; 95% CI, 0.03 to 0.12). As with DEX, the authors concluded that the beneficial effects of early HC (either high or low dose) may not outweigh the adverse effects. In a predefined secondary analysis, the PREMILOC trial study authors found no predictive value of baseline cortisol levels with CLD. However, high cortisol levels early after birth were associated with an increased risk of severe intraventricular hemorrhage and spontaneous intestinal perforation in infants treated with hydrocortisone.²² 

In a separate meta-analysis of 4 RCTs designed specifically to test the efficacy of early HC for prophylaxis against relative adrenal insufficiency in preterm infants, the authors analyzed individual patient data from 982 study patients.²³ Similar to previous reports, they concluded that early, low-dose HC significantly improves rates of survival without CLD (RBI, 0.20, 95% CI, 0.06 to 0.35; P = .007) and survival to hospital discharge (RBI, 0.07; 95% CI, 0.01 to 0.12; P = .03). They also found significantly increased rates of spontaneous intestinal perforation (RR, 2.39; 95% CI, 1.32 to 4.22; P = .004) and late-onset sepsis (RR, 1.22; 95% CI, 1.01 to 1.43; P = .04), but there was no association of HC with spontaneous intestinal perforation in the absence of concurrent indomethacin treatment, and the observed difference in sepsis was not associated with increased mortality or other in-hospital adverse outcomes. The authors concluded that these adverse effects did not negate the overall benefit of HC therapy in this population.

Since those meta-analyses mentioned previously were conducted, there has been 1 newly reported RCT of HC in preterm infants. The STOP-BPD study compared early HC versus placebo, starting between 7 and 14 days of age, and continuing for 22 days.²⁴ The initial dose of HC was 5 mg/kg per day for 7 days, followed by a scheduled taper. The primary outcome, survival without CLD at 36 weeks’ PMA, occurred in 28% of the 372 study patients, and was not statistically different between groups (RBI, 0.11; 95% CI, 0.20 to 0.55; P = .52). However, there was a significant reduction in mortality by 36 weeks PMA in infants in the HC group (RR, 0.65; 95% CI, 0.43 to 1.00; P = .0498). These modest results may not significantly alter the conclusions from previous meta-analyses regarding the short-term risk and benefit profile of early HC in preterm infants. However, a preplanned analysis of long-term, 2-year corrected age outcomes for this study cohort may soon be available.

A recent meta-analysis examined the question of whether chorioamnionitis was an independent predictor of response to early HC.²⁵ The authors identified 3 RCTs published between 1999 and 2018 in which the presence of histologic chorioamnionitis was reported. In each study, enrollment was limited to ELGANs and/or infants with birth weight <1000 g. In the combined cohort of 407 infants exposed to chorioamnionitis, the rate of survival without CLD was significantly higher in the HC group (56% vs 42%; RBI, 0.32; 95% CI, 0.08 to 0.61; P = .0071) compared with 373 infants not exposed to chorioamnionitis (50% vs 47%; RBI, 0.06; 95% CI, −0.14 to 0.31; P = .58).

The 2010 AAP policy statement concluded that existing data were insufficient to make any recommendation regarding the use of HC to treat CLD.⁸ A 2017 Cochrane review included 1 additional pilot RCT, published since the 2010 AAP policy statement.^9,10 In that study, the authors were unable to discern any effect of a 1 week course of low-dose HC on pulmonary outcomes or regional brain volumes among 64 ventilator-dependent ELGANs.²⁶ Eight infants died in each study group, and there was no difference in the composite outcome of death or severe CLD, defined as oxygen and/or pressure support at 36 weeks’ PMA. The Cochrane review went on to note the limitations in the evidence to-date and concluded that it would be prudent to reserve the late use of PCS (including HC) for infants who cannot be weaned from mechanical ventilation and to minimize the dose and duration.

In an ongoing RCT study conducted by the Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network that enrolled 800 infants born at <30 weeks’ gestational age who remained intubated for >7 days and intubated at 7 to 14 days, HC facilitated extubation; however, survival without CLD was not significantly improved. Also, survival without moderate or severe neurodevelopmental impairment in the group receiving HC was similar to that in the placebo group.²⁷ 

There have been no RCTs directly comparing different doses of HC to prevent or treat CLD in preterm infants. Of all RCTs published to date, none have shown an increase in short-term adverse outcomes compared with placebo; this includes studies using low (≤2 mg/kg per day) and high (>2 mg/kg per day) doses of HC. This is reassuring, particularly in light of a recent observational study with more than 1427 preterm infants exposed to early HC in which the authors found an independent association between higher early HC doses and mortality.²⁸ 

To our knowledge, other systemic CSs commonly used in clinical practice have not been studied to prevent or treat CLD in preterm infants in RCTs.

To maximize pulmonary benefit while avoiding systemic adverse effects, there has been growing interest in the use of inhaled CSs to prevent or treat CLD in preterm infants.

The 2010 AAP policy statement concluded that existing data were insufficient to make any recommendation regarding the use of inhaled CS to treat CLD.⁸ A 2017 Cochrane Review of “early” (≤14 days of age) inhaled CS to prevent CLD included 2 new RCTs published since the 2010 AAP policy statement.^29,30 The authors noted increasing evidence for increased survival without CLD at 36 weeks’ PMA (RBI, 0.09; 95% CI, 0.01 to 0.33; P = .04), although the upper confidence limit around the number needed to treat to benefit 1 infant was infinity.

One new trial included in the 2017 Cochrane Review was the NEuroSIS trial, the largest RCT of inhaled steroids to date.³¹ This placebo-controlled trial enrolled 863 ELGANs; 437 received budesonide starting within 24 hours of age and continuing until they no longer required supplemental oxygen or positive pressure support or until they reached 32 weeks’ PMA. The authors found that the rates of survival without CLD at 36 weeks’ PMA were not statistically significant (60% vs 54%; RBI, 0.12; 95% CI, 0.01 to 0.26; P = .07). The risk of CLD at 36 weeks’ PMA was significantly reduced (RR, 0.74; 95% CI, 0.60 to 0.91; P = .004), but the authors noted that this advantage may have been gained at the expense of increased mortality (RR, 1.24; 95% CI, 0.91 to 1.69; P = .17).³¹ 

The second new RCT was a placebo-controlled trial that enrolled 211 ventilator-dependent ELGANs; infants in the inhaled group received fluticasone daily until extubation or 6 weeks of age.³² The overall rate of survival without supplemental oxygen need at discharge (the study definition of severe CLD) was 82% and was not significantly different between groups (RBI, 0.10; 95% CI, −0.03 to 0.25; P = .13). However, in preplanned subgroup analyses, the authors found a significantly higher rate of this composite outcome for treated infants in the 24- to 26-weeks’ gestational age strata (93% vs 83%; RBI, 0.12; 95% CI, 0.01 to 0.24; P = .03); this benefit appeared to be limited to infants in whom chorioamnionitis was confirmed by placental pathology (86% vs 59%; RBI, 0.44; 95% CI, 0.05 to 0.99; P = .02).³² 

Since 2017, there have been no new published RCTs of inhaled CSs to prevent CLD.^d

In a 2017 Cochrane review of late^e (≥7 days of age) inhaled CSs, the authors analyzed results from 8 RCTs that enrolled 232 preterm infants.³³ They found that inhaled CSs did not significantly increase the rate of survival without CLD at 36 weeks’ PMA.

Since 2017, there have been no new published RCTs of inhaled CSs to treat CLD.

The theoretical advantages of inhaled versus systemic PCSs would include a lower absolute dose, enhanced pulmonary benefit, and reduced systemic effects. However, a pair of 2017 Cochrane reviews that included RCTs that directly compared inhaled versus systemic PCSs found no evidence of differences in either effectiveness or adverse event profiles.^29,30 The authors concluded that inhaled CSs do not confer any net advantage over systemic CSs for the prevention or treatment of CLD in preterm infants.

A recent systematic review comparing pulmonary versus systemic delivery of PCSs included 2 RCTs of endotracheal instillation of CS using surfactant as a vehicle, in addition to the RCTs of inhaled CS included in the 2017 Cochrane review.³⁴ Studies of CS instilled with a vehicle had not previously been included in any systematic review. Meta-analysis of outcomes for 1716 preterm infants among 8 RCTs found that those who received direct pulmonary administration of CS had a significantly higher rate of survival without CLD at 36 weeks’ PMA (62% vs 52%; RBI 0.18; 95% CI, 0.09 to 0.29; P = .0001).

Clinical trials of inhaled CSs to prevent or treat CLD have used a variety of CSs, including beclomethasone, budesonide, fluticasone, flunisolide, and DEX, but there have been no RCTs directly comparing different inhaled CSs in preterm infants. In a recent systematic review limited only to 5 RCTs that used budesonide, the authors found marked heterogeneity between trials and could find no evidence that inhaled budesonide attenuates the severity of CLD.³⁵ 

Although cumulative data suggest that PCSs may increase survival without CLD among ELGANs, the interaction of various factors in determining the risk and benefit profile has made it difficult to determine the optimal approach. Type, dose, route, and duration of PCS all seem to be important modulators of response as well as the risk of both short-term and long-term adverse effects. Although one could not conduct an RCT or even a series of RCTs that adequately address all these parameters, it may be possible to employ novel analytical techniques to use the data we have accumulated thus far and gain new insights. One such technique is a Bayesian network meta-analysis, an extension of the classic pairwise meta-analysis that allows comparison of multiple interventions based on both head-to-head comparisons within trials and indirect comparisons across trials.³⁶ For example, if 1 set of trials compared A versus B and another set of trials compared B versus C, a Bayesian network analysis enables us to make comparisons between A and C, with B as the common comparator for both A and C. This technique has recently been applied to RCTs of PCSs to prevent or treat CLD in preterm infants. In 1 network meta-analysis that included 47 RCTs and 6747 participants, the authors found that DEX was more effective at reducing CLD than either HC or inhaled beclomethasone, and that early (≤7 days of age) initiation of high dose DEX (total dose >3.0 mg/kg) was associated with the most significant reductions in CLD.³⁷ Although the risk of cerebral palsy was not statistically significant between types of PCS, the combination of high dose and long-term DEX use was associated with an increased risk of adverse neurodevelopmental outcome.³⁷ In a more recent network meta-analysis that included 62 RCTs and 5559 participants, the authors parsed the data into 3 levels of PCS initiation (early, moderately early, and late) and total dose (low, medium, and high), and found that DEX provided moderately early (8–14 days of age) and at a medium total dose (2–4 mg/kg) was associated with the best overall reduction in mortality and/or CLD with hypertension as the only significantly increased adverse outcome.³⁸ 

Despite the limitations of network meta-analysis (eg, indirect comparisons must be considered observational evidence), such novel approaches may bring us closer to the optimal use of PCSs.

Because the risks associated with PCS use cannot be eliminated, patient selection is critical so as not to expose infants who are unlikely to benefit, ie, infants not likely to develop CLD. The importance of patient selection is underscored by an analysis first reported in 2005 and updated in 2014 in which the authors performed a weighted meta-regression between the risk differences for the combined outcome of death or cerebral palsy and the rate of CLD in controls from RCTs.^39,40 As in 2005, the authors found a strong negative relationship that favored PCS use if the incidence of CLD in the pretreatment population was >50%. With this information, one could apply clinical prediction models at an early postnatal age and treat only those infants at relatively high risk for CLD; however, this would still result in exposing vulnerable preterm infants who are not going to develop CLD, while also missing some infants who will go on to develop CLD. Ideally, a model that predicts CLD with an extremely high sensitivity and specificity at or soon after birth would maximize an individual infant’s risk and benefit profile. Although several models exist that use various demographic and clinical parameters, gestational age alone remains the strongest predictor within the first week of age.⁴¹ 

Initiating PCSs soon after birth has the advantage of interrupting the inflammatory cascade that leads to the development of CLD. Delaying treatment and using more selective criteria, such as continued need for respiratory support, has the advantage of capturing those infants most likely to develop CLD. The disadvantage to delayed intervention, however, is that infants with more severe CLD are less likely to benefit from PCS.⁴² 

Many RCTs of PCSs have not reported long-term outcomes, and those that have are limited to 18 to 24 months’ PMA. To better assess the impact of PCSs on preterm infant outcomes, longer-term assessments are needed; unfortunately, only a handful of RCTs have reported outcomes at school age (Table 1). In sum, these studies suggest that PCSs do not increase the risk of adverse outcomes at school age. Although preterm cohorts exposed to PCSs generally showed decreased growth, impaired lung function, and increased rates of neurodevelopmental impairment relative to term infants, they did not significantly differ from control cohorts for their respective RCTs. However, many studies were contaminated by relatively high rates of open-label PCS use; of the 3 placebo-controlled studies with low or no open-label PCS contamination, the 2 larger studies found increased rates of growth and/or neurodevelopmental impairment;^43,44 hence, the overall safety of PCS remains in question. A large, multicenter RCT by the Neonatal Research Network just completed 2-year follow-up, with plans to conduct school-aged outcomes assessment, which may provide data on nearly 800 infants, about the same as all previous studies combined.^f

1. PCSs, either to prevent or treat early CLD, may increase the rate of survival without severe CLD but carry significant short- and long-term risks.

2. Early (≤7 days) low-dose HC may prevent CLD or death in infants weighing less than 1000 g exposed to chorioamnionitis.

3. Late (>7 and <28 days of age) low-dose DEX may improve outcomes for preterm infants who remain on significant respiratory support but without evidence of severe CLD.

4. Inhaled CSs do not appear to offer any advantage over systemic corticosteroids and may be associated with increased mortality.

5. CSs may be more effective when delivered directly into the lung with surfactant as a vehicle, but data on long-term outcomes are lacking.

6. Data from RCTs are population based and may not apply to individual patients.

7. More data regarding long-term, school-aged outcomes are needed to guide the use of PCSs to prevent or treat CLD in preterm infant.

1. Routine use of PCSs cannot be recommended.

2. The decision to use PCSs to prevent or treat CLD should be individualized and made together with the parents, and the discussions should be documented in the patient’s medical record.

3. If a decision is made to administer a PCS, a low dose provided for a short, predefined duration (eg, extubation) is recommended. If the infant does not show a clinical response to the PCS within 72 hours of initiation, continued treatment is not recommended.

4. High-dose PCSs are not recommended to prevent or treat CLD in preterm infants.

5. Indomethacin should not be used concurrently with PCSs.

______________________________

• James J. Cummings, MD, MS, FAAP

• Arun K. Pramanik, MD, FAAP

• Eric Eichenwald, MD, Chairperson

• Charleta Guillory, MD

• Ivan Hand, MD

• Mark Hudak, MD

• David Kaufman, MD

• Camilia Martin, MD

• Ashley Lucke, MD

• Margaret Parker, MD

• Arun Pramanik, MD

• Kelly Wade, MD

• Timothy Jancelewicz, MD – AAP Section on Surgery

• Michael Narvey, MD – Canadian Pediatric Society

• Russell Miller, MD – American College of Obstetricians and Gynecologists

• RADM Wanda Barfield, MD, MPH – Centers for Disease Control and Prevention

• Lisa Grisham, APRN, NNP-BC – National Association of Neonatal Nurses

Jim Couto, MA

AAP

American Academy of Pediatrics

BPD

bronchopulmonary dysplasia

CI

confidence interval

CLD

chronic lung disease following preterm birth

CS

corticosteroid

DEX

dexamethasone

ELGAN

extremely low gestational age (<28 weeks) newborn

HC

hydrocortisone

PCS

postnatal corticosteroid

PMA

postmenstrual age

RBI

relative benefit increase

RCT

randomized controlled trial

RR

relative risk