Postnatal dexamethasone treatment for preterm infants at high risk for bronchopulmonary dysplasia is associated with improved regional brain volumes: a prospective cohort study Authors: Rahul Chandwani, Julia Kline, Mekibib Altaye, Nehal Parikh Published: 2025 ABSTRACT AbstractObjectiveTo evaluate the effect of low-dose postnatal dexamethasone therapy for bronchopulmonary dysplasia (BPD) prevention/treatment on MRI-derived regional brain volumes at term-equivalent age (TEA) and neurodevelopmental outcomes in a regional cohort of preterm infants.Study designWe prospectively recruited 392 preterm infants (≤32 weeks gestational age (GA)), who underwent structural MRI (3T Philips Ingenia) at TEA. We automatically segmented T2-weighted MRI scans using the Developing Human Connectome Project pipeline to derive a priori selected, two primary outcomes of interest: volumes of the cerebellum and subcortical grey matter. We estimated propensity scores for subjects with a logistic regression model and used weighted linear regression to determine the independent effects of dexamethasone on primary and two secondary outcomes: cortical surface area at TEA and motor scores at 2 years corrected age.ResultsOf 392 infants, 41 were treated with low cumulative dose dexamethasone (total 0.89 mg/kg) initiated at 36 days (median) of age for evolving BPD: 21 males; mean (SD) GA was 25.5 (1.6) weeks; postmenstrual age at MRI was 43.7 (1.2) weeks; and 33 had severe BPD. In multivariable linear regression, dexamethasone was significantly correlated with larger cerebellar (difference=0.510; 95% CI: 0.079 to 0.941) and subcortical grey matter volume (difference=0.138; 95% CI: 0.014 to 0.263). Dexamethasone was also positively correlated with motor scores (difference=5.220; 95% CI: 0.845 to 9.594).ConclusionLow-dose dexamethasone therapy after the first postnatal week for evolving/established BPD did not result in adverse macrostructural effects and may have a protective effect on motor development in preterm infants.Keywords: Neonatology, Neurology WHAT IS ALREADY KNOWN ON THIS TOPIC A greater risk of cerebral palsy and mortality has reduced the use of postnatal corticosteroid (PNC) therapy for bronchopulmonary dysplasia (BPD) prevention/treatment in at-risk infants. Short-term neurostructural measures to assess benefits and harms of PNC therapy on the developing brain around the time of birth are limited. A greater risk of cerebral palsy and mortality has reduced the use of postnatal corticosteroid (PNC) therapy for bronchopulmonary dysplasia (BPD) prevention/treatment in at-risk infants. Short-term neurostructural measures to assess benefits and harms of PNC therapy on the developing brain around the time of birth are limited. WHAT THIS STUDY ADDS In a prospective cohort of 392 preterm infants, low-dose dexamethasone initiated after the first postnatal week was independently associated with increased cerebellar and subcortical grey matter volume at term-equivalent age. Dexamethasone was also positively associated with motor outcomes at 2 years corrected age. In a prospective cohort of 392 preterm infants, low-dose dexamethasone initiated after the first postnatal week was independently associated with increased cerebellar and subcortical grey matter volume at term-equivalent age. Dexamethasone was also positively associated with motor outcomes at 2 years corrected age. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Our study demonstrates that low-dose dexamethasone does not result in adverse macrostructural effects and may be neuroprotective in the cerebellum/subcortical grey matter, suggesting the importance of considering timely PNC use to reduce the risk of neurodevelopmental sequelae of BPD. Our study demonstrates that low-dose dexamethasone does not result in adverse macrostructural effects and may be neuroprotective in the cerebellum/subcortical grey matter, suggesting the importance of considering timely PNC use to reduce the risk of neurodevelopmental sequelae of BPD. Introduction Infants born very preterm (VPT, <32 weeks gestational age (GA)) remain at high risk of mortality or neurodevelopmental morbidity from bronchopulmonary dysplasia (BPD).1 2 The incidence of BPD has been reported as either stable or increasing, likely due to improved survival of VPT infants and lack of sufficient therapies.3 Postnatal corticosteroids (PNCs) have been used for the prevention/treatment of BPD in at-risk infants, although the therapy is controversial because of possible adverse effects on the developing brain.4 A greater risk of cerebral palsy/mortality has admonished clinicians against the use of PNC for BPD treatment, especially when PNCs are given at high doses, within the first postnatal week, and/or to infants at low risk for developing BPD.5,8 The Dexamethasone: A Randomised Trial (DART) study found that low-dose dexamethasone, given after the first week of life, to ventilator-dependent VPT/extremely low birthweight infants shortened the duration of intubation without any increase in short-term complications or neurodevelopmental impairments/death at 2 years corrected age.9,11 Meta-analyses suggest that PNCs, when given to infants at high risk (ie, baseline risk of 50% or higher) for BPD/death and initiated after a week of age, still maintain an overall net benefit for treating BPD while being protective at reducing cerebral palsy/death.7 11 12 A majority of PNC studies were small pilot randomised trials and lacked school age neurodevelopmental follow-up or shorter-term neurostructural measures, thus unable to definitively demonstrate net benefit or harm. Previously, our group reported that postnatal dexamethasone (administered via the DART protocol) for severe BPD treatment was associated with a decreased risk for diffuse white matter abnormality (DWMA) on T2-weighted MRI, a known predictor of adverse neurodevelopmental outcomes.13 14 However, this study did not examine the effects of dexamethasone on other sensitive and objective measures of early brain injury/maturation and neurodevelopmental outcomes. Further evaluation is needed to understand the mechanisms of PNC benefits and harms and guide clinical recommendations of caring for preterm infants at high risk for BPD. Regional brain volumes are sensitive to the effects of PNC and directly correlate with neurodevelopmental outcomes.15,19 Prior studies have consistently demonstrated reduced volumes of the cerebellum and subcortical grey matter in PNC-treated infants at high risk for BPD.17,19 Thus, we considered cerebellar and subcortical grey matter volume as primary outcomes of interest in our study. Our objective was to evaluate the effects of low-dose dexamethasone therapy for BPD on brain volumes at term-equivalent age (TEA) and motor outcomes at 2 years corrected age in a regional cohort of preterm infants. To achieve this, we used a propensity scoring approach to account for selection bias and better isolate the independent effects of dexamethasone therapy on study outcomes. We hypothesised that low-dose dexamethasone therapy, initiated after a week of age, for preterm infants at high risk for BPD is associated with improved subcortical grey matter and cerebellar volumes, when compared with preterm infants at risk for BPD who did not receive dexamethasone therapy. Method Study design We prospectively enrolled a cohort of 392 preterm infants born at or below 32 weeks GA between September 2016 and November 2019, from five level III/IV neonatal intensive care units (NICUs) as part of the Cincinnati Infant Neurodevelopment Early Prediction Study (CINEPS). The NICUs included Cincinnati Children’s Hospital Medical Center (CCHMC), University of Cincinnati Medical Center, Good Samaritan Hospital, St. Elizabeth Healthcare and Kettering Medical Center. Infants were excluded if they had any chromosomal/congenital anomalies affecting the brain, spine or heart. Data acquisition All MRI data collection for the CINEPS cohort took place at the CCHMC. Infants underwent structural MRI during natural sleep on a 3T Philips Ingenia scanner with a 32-channel head coil. The scans were completed between 39 and 44 weeks postmenstrual age (PMA). To promote natural sleep, infants were: (1) fed 30 min prior to imaging, (2) fitted with silicone earplugs to dampen scanner noise (Instaputty, E.A.R. Inc, Boulder, Colorado, USA) and (3) swaddled in a blanket and vacuum immobilisation device (MedVac, CFI Medical Solutions, Fenton, Michigan, USA). We used the following T2-weighted MRI acquisition parameters: echo time 166 ms, repetition time 18 567 ms, voxel dimensions 1.0×1.0×1.0 mm3, flip angle 90° and scan time 3:43 min. Over 50 antenatal, intrapartum and neonatal clinical factors were collected prospectively.20 BPD was defined based on the type of respiratory support provided at 36 weeks PMA.21 Infants on room air did not have BPD, and for the remaining infants, BPD severity was classified as grade 1 if nasal cannula <2 L/min, grade 2 if nasal cannula >2 L/min or non-invasive positive pressure ventilation and grade 3 if mechanical ventilation was used.21 A total of 41 infants were treated with dexamethasone according to the DART protocol.9 No other dosing regimen was used. Motor aptitude of infants was determined by assessors trained to reliability in administering the Bayley Scales of Infant and Toddler Development-Third Edition (Bayley-III) standardised test.22 The Bayley-III Motor composite scale has a mean (SD) score of 100 (15) and range of 40–160, with higher scores indicating better gross and fine motor functioning. The cohort underwent Bayley-III testing between 22 and 26 months corrected age. Cognitive and language scores on the Bayley-III subscales were also obtained as part of this study. MRI data Processing To derive global and regional brain volumes, we processed all T2-weighted MRI data using the developing Human Connectome Project (dHCP) pipeline.23 24 The dHCP performs automatic cortical and subcortical tissue segmentation and volume estimation of 50 regions as established in the Gousias neonatal atlas.24 To summarise, the pipeline first performs bias correction and brain extraction of each T2-weighted image. The brain image is then segmented into tissue types (eg, cerebrospinal fluid, white matter, cortical and subcortical grey matter) using the Draw EM algorithm.23 White matter and pial surfaces are generated, enabling the estimation of cortical thickness. Simultaneously, surface inflation and spherical projection are used to calculate other metrics of cortical geometry/maturation.23 After processing our data, we visually inspected each segmented scan (figure 1) and excluded any scans with errors in tissue identification, improper segmentation or artefacts. Subjects who were unable to be processed due to severe brain injury and extreme ventriculomegaly were also excluded. We divided each regional brain volume by total brain volume to account for individual differences in infant head size. Our primary outcomes of interest were the relative (to total brain volume) volumes of the subcortical grey matter (left and right combined) and cerebellum.16,18 One secondary outcome—cortical surface area—was also derived from the pipeline.16 Segmentation from the Developing Human Connectome Project (dHCP) pipeline for an infant in our preterm cohort, shown in sagittal, coronal and axial views. Propensity scoring and statistical analysis In our prospective study, infants were not randomly assigned to dexamethasone therapy for BPD treatment.25 To account for selection bias and better understand the independent effects of dexamethasone on study outcomes, we used a propensity scoring approach, which balances the treated and untreated groups based on baseline demographic and clinical factors.25 The propensity score for each infant was determined as the predicted treatment probability from the logistic regression of dexamethasone with baseline characteristics as explanatory variables.26 We followed the Olmos and Govindasamy step-by-step guide to propensity score implementation.27 To optimise variables selected for generating propensity scores (ie, assessing degree of bias), we used an iterative process to ensure the minimum number of covariates/confounders remained significantly different between the treated and untreated groups after propensity score weighting. In linear regression analyses, we used the calculated propensity scores as weights to determine the independent effect of dexamethasone on brain volumes.27 In addition to known confounders, we included any remaining unbalanced baseline variables in each multivariable linear regression. We used a Shapiro-Wilk test to assess for normality, and a Mann-Whitney rank-sum test or χ2 test to compare baseline variables between the dexamethasone-treated and untreated subjects. Two-sided p values of <0.05 were considered statistically significant. Notably, we did not correct for multiple comparisons because of the preliminary nature of this study, having only two primary outcomes of interest.28 We also examined two secondary outcomes: cortical surface area on MRI at TEA and Bayley-III motor composite score.29 30 All statistical analyses were performed in R, using the following packages for regression modelling and causal inference methods: gbm, psych, rms and MatchIt (R Core Team 2020, https://www.R-project.org/).2731,34 Results Demographics Of 392 infants, 41 received dexamethasone between 32 and 40 weeks PMA for developing or established BPD: 21 males; mean (SD) GA 25.5 (1.6) weeks; birth weight 800 (229) g. Thirty-three (80.5%) preterm infants who received dexamethasone developed severe BPD.21 All 41 infants developed BPD and were treated with the low cumulative dose dexamethasone DART regimen (total 0.89 mg/kg administered orally/injected over 10 days) after the first postnatal week.9 Nine of these infants received a second course of DART, and one infant received three total courses. No other dexamethasone regimens were used. The median (range) age after birth for receipt of the first course of DART was 36 (10–113) days. Table 1 summarises the baseline characteristics of dexamethasone-treated and untreated infants. BPD developed in 35.3% of infants in the untreated group and 100% of infants in the treated group. Table 1Baseline characteristics of preterm infants who received postnatal corticosteroids for evolving or established bronchopulmonary dysplasia and infants who did notClinical variables*Postnatal steroid group(n=41)No steroid group(n=351)Difference in mean, median or proportion between groups (95% CI)P valueGestational age at birth, weeks*25.5 (1.6)29.7 (2.2)4.2 (3.5, 4.9) <0.001 Birth weight, grams*800 (229)1350 (433)550 (419, 685) <0.001 Male, N (%)21 (51.2%)189 (53.8%)2.6 (−13.5, 18.8)0.750Maternal antenatal steroids, N (%)37 (90.2%)324 (92.3%)2.1 (−7.4, 11.6)0.411Maternal magnesium therapy, N (%)33 (80.5%)296 (84.3%)3.8 (−8.9, 16.6)0.478Maternal hypertension, N (%)*13 (31.7%)141 (40.2%)8.5 (−6.7, 23.6)0.294Intubation at birth, N (%)*26 (63.4%)61 (17.4%)−46.0 (−61.3, −30.8) <0.001 Apgar <5 at 5 min*6.2 (2.2)7.7 (1.7)1.5 (0.80, 2.1) <0.001 Pulmonary haemorrhage, N (%)*4 (9.8%)2 (0.6%)−9.0 (−18.3, −0.1) <0.001 Caffeine therapy, N (%)*40 (97.6%)112 (31.9%)−65.7 (−58.9, −72.4) <0.001 Maternal breast milk diet, N (%)*40 (97.6%)333 (94.9%)−2.7 (−7.9, 2.6)0.448Parenteral nutrition duration, days *20.0 (21.0)11.0 (5.0)−9.0 (−22.0, −5.0) <0.001 Mechanical ventilation duration, days*46.0 (38.0)0.0 (2.0)−46 (−55, −40) <0.001 CPAP therapy duration, days*15.0 (30.0)36.0 (22.0)21.0 (11.0, 30.0) <0.001 Sepsis (culture positive), N (%)*15 (36.6%)24 (6.8%)−30.0 (−44.7, −14.8) <0.001 Any bronchopulmonary dysplasia, N (%)41 (100%)124 (35.3%)−64.7 (−69.7, −59.7) <0.001 Severe bronchopulmonary dysplasia, N (%)33 (80.5%)41 (11.7%)−68.8 (−81.4, −56.2) <0.001 Retinopathy of prematurity, N (%)35 (85.4%)99 (28.2%)−57.2 (−69.0, −45.4) <0.001 Postmenstrual age at MRI scan, weeks43.7 (1.2)42.6 (1.3)−1.1, (−1.5, −0.7) <0.001 Cincinnati Children’s NICU, N (%) *24 (58.5%)40 (11.4%)−47.1 (−62.6, −31.7) <0.001 All values are mean (SD), unless otherwise indicated. Median (IQR) is reported for parenteral nutrition, mechanical ventilation and CPAP therapy duration days. Italics indicate significant p values.*Variable was used in the calculation of propensity scores for each infant.CPAP, continuous positive airway pressure; NICU, neonatal intensive care units. Baseline characteristics of preterm infants who received postnatal corticosteroids for evolving or established bronchopulmonary dysplasia and infants who did not All values are mean (SD), unless otherwise indicated. Median (IQR) is reported for parenteral nutrition, mechanical ventilation and CPAP therapy duration days. Italics indicate significant p values. Variable was used in the calculation of propensity scores for each infant. CPAP, continuous positive airway pressure; NICU, neonatal intensive care units. Propensity score model We used the following a priori selected demographical and clinical factors (table 1) to calculate propensity scores in logistic regression (online supplemental table 1): GA, birth weight, maternal hypertension, Apgar scores, intubation, ventilation days, days on CPAP, caffeine therapy, pulmonary haemorrhage, sepsis, days of parenteral nutrition, maternal breast milk diet and clinical site. These variables were selected because of their known association with either treatment selection (ie, whether an infant received dexamethasone for BPD) or our primary and/or secondary outcome (brain volumes and motor function). Balancing of covariates between the treated/untreated groups (online supplemental table 2) was checked after propensity score weighting: GA, birth weight, maternal breast milk days, CPAP therapy duration, pulmonary haemorrhage, maternal hypertension and caffeine therapy remained significantly different between groups.27 These variables were considered for inclusion in the final regressions to control residual imbalance. Propensity score weighted linear regression analyses were performed for subcortical grey matter and cerebellar volumes as two primary outcomes using independent variables of dexamethasone, GA, PMA at MRI, severe BPD, sex, sepsis and birth weight. CPAP therapy duration, maternal hypertension, caffeine therapy and maternal breast milk diet were also included as predictors for residual imbalance. The same independent variables were also used in regression analysis for cortical surface area (secondary outcome). In weighted regression with Bayley-III motor scores as the other secondary outcome, we included dexamethasone, GA, high-risk socioeconomic status, severe BPD, sex, sepsis, birth weight, antenatal steroids, severe retinopathy of prematurity and global brain abnormality score. Maternal hypertension, caffeine therapy and maternal breast milk diet were imbalanced and also included in the model. Outcomes In our cohort, mean (SD) relative (to total brain) volumes of the cerebellum and subcortical grey matter were 7.4% (0.8%) and 7.4% (0.4%). For the two primary outcomes, in multiple linear regression analyses controlling for additional confounders, dexamethasone was significantly correlated with larger cerebellar volume (difference=0.510; p=0.021) and subcortical grey matter volume (difference=0.138; p=0.030) relative to total brain volume (figure 2A, B and table 2). Bayley-III motor composite scores were significantly lower (p<0.001) in the dexamethasone treated group compared with untreated infants (online supplemental table 3). However, as shown in figure 2C and table 3, dexamethasone therapy was significantly positively correlated with Bayley-III Motor (difference=5.220; p=0.020) composite scores in adjusted analyses. Lastly, dexamethasone therapy was not associated with cortical surface area (difference=1390 mm2; p=0.309) in multivariable weighted linear regression (figure 2D and online supplemental table 4). In addition, regression analyses demonstrating the relationship of dexamethasone with total brain volume, Bayley-III overall cognitive and language scores are shown in online supplemental tables 5–7, respectively. Association of dexamethasone and other clinical factors with primary and secondary outcomes of interest including the relative volume differences (%) for the cerebellum (A) and subcortical grey matter (B), differences in Bayley-III overall motor scores (C) and differences in cortical surface area (mm2) (D). BPD, bronchopulmonary dysplasia; SES, socioeconomic status; CPAP, continuous positive airway pressure. Table 2Weighted linear regression model of dexamethasone for BPD and other clinical factors associated with relative (to total brain tissue × 100%) volume of the cerebellum and subcortical grey matter on MRI in preterm infants at term-equivalent ageRelative volume of the cerebellum (%)Unadjusted analysisMultivariable analysis adjusted for clinical factors Coefficient (95% CI) P value Coefficient (95% CI) P valueDexamethasone for BPD0.935 (0.002, 1.87) 0.049 0.510 (0.079, 0.941) 0.021 Gestational age at birth0.534 (0.438, 0.630) <0.001 Postmenstrual age at MRI scan0.161 (0.057, 0.264) 0.002 Severe BPD−0.350 (−0.693, –6.70E–03) 0.046 Male sex0.778 (0.514, 1.04) <0.001 Sepsis (culture positive)−1.85 (–2.21, –1.48) <0.001 Birth weight−3.28E–04 (−8.64E–04, 2.08E–04)0.230CPAP therapy duration, days0.035 (0.028, 0.041) <0.001 Maternal hypertension0.520 (0.234, 0.806) <0.001 Caffeine therapy0.476 (0.082, 0.870) 0.018 Maternal breast milk diet−0.801 (–1.10, –0.505) <0.001 Relative volume of the subcortical grey matter (%)Unadjusted analysisMultivariable analysis adjusted for clinical factorsCoefficient(95% CI)P valueCoefficient(95% CI)P valueDexamethasone for BPD0.119 (–0.008, 0.246)0.0650.138 (0.014, 0.263) 0.030 Gestational age at birth0.060 (0.032, 0.088) <0.001 Postmenstrual age at MRI scan−0.015 (–0.045, 0.015)0.333Severe BPD−0.142 (–0.241, –0.043) 0.005 Male sex0.088 (0.012, 0.164) 0.023 Sepsis (culture positive)−0.047 (–0.152, 0.059)0.384Birth weight−3.47E–04 (−5.02E–04, –1.92E–04) <0.001 CPAP therapy duration, days1.36E–03 (−5.58E–04, 3.23E–03)0.166Maternal hypertension−0.025 (–0.107, 0.058)0.553Caffeine therapy−0.044 (–0.158, 0.069)0.443Maternal breast milk diet−0.034 (–0.120, 0.051)0.429Italics indicate significant p values.BPD, bronchopulmonary dysplasia. Weighted linear regression model of dexamethasone for BPD and other clinical factors associated with relative (to total brain tissue × 100%) volume of the cerebellum and subcortical grey matter on MRI in preterm infants at term-equivalent age Italics indicate significant p values. BPD, bronchopulmonary dysplasia. Table 3Weighted linear regression model of dexamethasone for BPD and clinical factors associated with Bayley-III overall motor scores in preterm infants at 2 years corrected ageBayley-III overall motor scoreUnadjusted analysisMultivariable analysis adjusted for clinical factorsCoefficient (95% CI)P valueCoefficient (95% CI)P valueDexamethasone for BPD6.754 (−0.191, 13.70)0.0575.220 (0.845, 9.594) 0.020 Gestational age at birth2.123 (0.975, 3.270) <0.001 High-risk SES−9.417 (−12.88, –5.959) <0.001 Severe BPD−4.452 (−8.201, –0.704) 0.020 Male sex6.208 (3.387, 9.029) <0.001 Sepsis (culture positive)−4.733 (−8.951, –0.515) 0.028 Birth weight−0.001 (−0.007, 0.005)0.691Antenatal steroids4.981 (−1.057, 11.02)0.106Severe retinopathy of prematurity−12.33 (−17.46, –7.196) <0.001 Global brain abnormality score−6.326 (−10.02, –2.637) <0.001 Maternal hypertension3.257 (0.176, 6.339) 0.038 Caffeine therapy4.401 (−0.014, 8.816)0.051Maternal breast milk diet−2.268 (−5.391, 0.855)0.154Italics indicate significant p values.BPD, bronchopulmonary dysplasia; SES, socioeconomic status. Weighted linear regression model of dexamethasone for BPD and clinical factors associated with Bayley-III overall motor scores in preterm infants at 2 years corrected age Italics indicate significant p values. BPD, bronchopulmonary dysplasia; SES, socioeconomic status. Discussion In this study, we determined the relationship between dexamethasone for BPD prevention/treatment and brain volumes at TEA in preterm infants. Using a propensity scoring method, we found that an average 10-day course of low-dose dexamethasone, initiated after the first postnatal week, was independently associated with increased cerebellar and subcortical grey matter volume for infants in our cohort. Low-dose dexamethasone had a comparable and inverse effect to the development of severe BPD on cerebellar (difference=0.510 vs −0.350) and subcortical grey matter (difference=0.138 vs −0.142) volumes, and was comparable to a 1-week increase in GA (difference=0.534) for cerebellar and more than 2-week increase in GA (difference=0.060) for subcortical grey matter volume. We also observed a positive association of dexamethasone with Bayley-III Motor scores at 2 years corrected age but no association with cortical surface area at TEA. Our findings suggest that dexamethasone, administered via the DART regimen, does not result in adverse macrostructural effects and may be neuroprotective for the cerebellum and subcortical grey matter. The cerebellum and subcortical grey matter structures such as paraventricular nuclei undergo rapid growth in the period of prematurity and are among the regions with the most abundant glucocorticoid receptors in the brain, potentially making them vulnerable to the effects of PNC.35 36 Smaller cerebellar and subcortical grey matter volumes at TEA have also been associated with worse motor outcomes in VPT infants.16 18 Prior studies investigating the neurostructural effects of PNC therapy have found decreased global and regional cerebral volumes in steroid-treated infants.17,19 Parikh et al (2007) observed reductions in cerebellum and subcortical grey matter volume for 11 extremely low birthweight infants who received a 7-day course of dexamethasone (cumulative dose 2.8 mg/kg) after 4 weeks of age, when compared with 30 untreated infants. Despite adjusting for group differences, the authors concluded that the deficits observed could represent a greater degree of illness from BPD in dexamethasone-treated infants, which may have persisted as residual confounding. In comparison, our study had a larger cohort size, followed the established DART protocol and used a propensity scoring method to minimise residual confounding and selection bias, which could contribute to differences found in primary outcomes. BPD delays brain maturation and is associated with white matter signal abnormalities and altered cortical maturation in at-risk infants.20 30 Our group previously examined perinatal clinical factors that contribute to the development of DWMA, a known predictor of neurodevelopmental deficits.20 One of the clinical factors analysed by Parikh et al was postnatal dexamethasone (DART protocol), which was associated in multivariable analyses with a decreased risk for DWMA when used in infants who develop severe BPD.20 In addition to examining the regimen’s effects on other sensitive and objective measures of early brain injury/maturation such as brain volumes, our current work also investigated its relationship with neurodevelopment, identifying a positive association between dexamethasone and motor development at 2 years corrected age. This is an important finding, as approximately half of the randomised controlled trials using dexamethasone after the first postnatal week for BPD prevention/treatment lack long-term follow-up of infants to evaluate risk of neurodevelopmental impairment and/or had a relatively small sample size.4 7 Douglas et al (2023) assessed 76 extremely preterm/extremely low birthweight infants who received PNC (dexamethasone or hydrocortisone) and found higher cumulative dose associated with higher odds of cerebral palsy and lower cognitive and motor developmental scores at 2 years.37However, these associations weakened after adjusting for GA, birth weight, sex and intraventricular haemorrhage. Other confounders (including BPD) were not controlled for in their models, which likely accounts for differences found between our studies.37 Confounding by indication is especially a major concern for observational studies of PNC for BPD prevention, considering that BPD is one of the top risk factors for motor and other neurodevelopmental impairments.21 37 38 A reluctance in prescribing PNC therapy, largely driven by concerns for increased risk of cerebral palsy and cognitive/language impairments, has contributed to the increased/stable incidence of BPD.4 However, it is important to balance this concern with the established fact that BPD is among the strongest risk factors for neurodevelopmental impairments.1 11 38 While initiating early therapy (<7 days) to prevent BPD is not recommended, most clinicians today use low-dose (<2 mg/kg), late-initiated (15–27 days) dexamethasone.10 Dexamethasone initiated moderately early (≥7 days)—the therapy received by infants in our cohort—does not raise concerns for long-term neurodevelopmental harm and could be beneficial in reducing the rates of BPD, death and extubation failure. This finding can be partially attributed to being able to more easily identify infants who are at high risk for BPD after the first postnatal week.4 12 A meta-regression has shown that infants with a baseline risk of 50% or higher for BPD may benefit the most from systemic dexamethasone to reduce the risk of cerebral palsy or death.11 Our study enrolled infants at very high risk of developing BPD, as demonstrated by the development of severe BPD in 80% of dexamethasone-treated infants. Though our study followed the established DART protocol (0.89 mg/kg, low-dose dexamethasone tapered over 10 days), it is common for infants to receive multiple PNC courses. High-dose (>4 mg/kg) dexamethasone has also been shown to lower the risk of neurodevelopmental impairment when compared with medium (2–4 mg/kg) or low-dose therapy.7 12 39 40 Collectively, our findings suggest the positive impact of low-dose dexamethasone prescribed after the first postnatal week, but will require additional studies to confirm improvement in long-term neurodevelopment. The strengths of our study include the objective assessment of brain volumes and adjustment for known predictors associated with brain and motor abnormality. Notably, infants who receive higher cumulative doses of dexamethasone often have a worse degree of illness and are already at greater risk of adverse outcomes. Randomised trials are the gold standard for assessing causal relationships, although it is not always possible to randomly allocate exposures, either due to known harms or lack of equipoise.9 In our observational study, we used a propensity scoring approach that aimed to account for residual confounding and selection bias. A large sample size is usually required to perform matching of subjects by propensity score, prior to comparing outcomes. Due to the insufficient overlap of propensity scores between dexamethasone-treated and untreated infants in our cohort, we were unable to perform 1:1 matching and directly compare outcomes. We addressed this by including significantly unbalanced variables as confounders in our final regression models.27 Another limitation is our lack of reporting on cognitive or behavioural outcomes because they cannot be evaluated robustly at 2 years corrected age. School age follow-up of our cohort is currently ongoing. Conclusion In summary, we demonstrated that dexamethasone for BPD prevention/treatment is associated with larger cerebellar and subcortical grey matter volumes at TEA and improved motor development at 2 years in a cohort of preterm infants. Given the lack of alternative therapies that exist to facilitate extubation, our study underscores the importance of considering dexamethasone after the first postnatal week to reduce the risk of adverse neurodevelopmental sequelae of BPD in at-risk infants. Supplementary material Acknowledgements We sincerely thank the Cincinnati Infant Neurodevelopment Early Prediction Study (CINEPS) Investigators; Principal investigator: Nehal A Parikh, DO, MS; Collaborators (in alphabetical order): Mekibib Altaye, PhD, Anita Arnsperger, RRT, Traci Beiersdorfer, RN BSN, Kaley Bridgewater, RT(MR) CNMT, Tanya Cahill, MD, Kim Cecil, PhD, Kent Dietrich, RT, Christen Distler, BSN RNC-NIC, Juanita Dudley, RN BSN, Brianne Georg, BS, Cathy Grisby, RN BSN CCRC, Lacey Haas, RT(MR) CNMT, Karen Harpster, PhD, OT/RL, Lili He, PhD, Scott K. Holland, PhD, V S Priyanka Illapani, MS, Kristin Kirker, CRC, Julia E Kline, PhD, Beth M Kline-Fath, MD, Hailong Li, PhD, Matt Lanier, RT(MR) RT(R), Stephanie L Merhar, MD MS, Greg Muthig, BS, Brenda B Poindexter, MD MS, David Russell, JD, Sara Stacey, BSN, Kari Tepe, BSN RNC-NIC, Leanne Tamm, PhD, Julia Thompson, RN BSN, Jean A Tkach, PhD, Hui Wang, PhD, Hui Wang, PhD, Jinghua Wang, PhD, Brynne Williams, RT(MR) CNMT, Kelsey Wineland, RT(MR) CNMT, Sandra Wuertz, RN BSN CCRP, Donna Wuest, AS and Weihong Yuan, PhD. Footnotes Funding: This work was supported by grants R01-NS094200 and R01-NS096037 from the National Institute of Neurological Disorders and Stroke (NINDS) and R01-HL164420 from the National Heart, Lung, and Blood Institute (NHLBI) of the National Institutes of Health. Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: The institutional review board at CCHMC approved our study locally (IRB# 2016-4483) and the four other NICUs (University of Cincinnati Medical Center, Good Samaritan Hospital, St. Elizabeth Healthcare and Kettering Medical Center) accepted this approval as each had a reciprocity agreement with CCHMC. Written informed consent was obtained from a parent or guardian of each infant prior to enrolment. Participants gave informed consent to participate in the study before taking part. Presented at: This work was selected for an oral presentation at the Pediatric Academic Societies Meeting in Denver, Colorado on 24 April 2022. Data availability statement Data are available upon reasonable request. References Associated Data Supplementary Materials Data Availability Statement Data are available upon reasonable request.