The association between bronchopulmonary dysplasia grade and risks of adverse neurodevelopmental outcomes among preterm infants born at less than 30 weeks of gestation

Abstract Background Bronchopulmonary dysplasia (BPD) is a multifactorial disease with neurodevelopmental implications. This study aims to quantify the risks of adverse neurodevelopmental outcomes for each BPD grade among preterm infants born at less than 30 weeks’ gestation. Methods We retrospectively studied infants who received care in our institution until at least 36 weeks postmenstrual age and had a formal neurodevelopmental assessment in our infant follow-up clinic using the Bayley Scales for Infant and Toddler Development (BSID). We assessed the association between BPD grade and adverse neurodevelopmental outcomes using descriptive statistics and regression models. Results Two hundred and fifty infants, including 89 (35.6%), 87 (34.8%), 65 (20.6%), and 9 (3.6%) with No BPD, Grade 1, Grade 2, and Grade 3 BPD, were included in the study. Small for gestational age, late pulmonary hypertension, dexamethasone administration, and adverse neurodevelopmental outcomes were more common as BPD grade increased. In a logistic regression analysis, Grades 2 and 3, but not Grade 1, BPD were associated with increased odds of a composite adverse neurodevelopmental outcome by 2.7 and 7.2 folds, respectively. A BSID domain-specific analysis showed that higher grades were associated with lower scores in the cognitive, gross motor, and fine motor domains. Conclusions Grades 2 and 3 BPD, but not Grade 1, correlate with risks of adverse neurodevelopmental outcomes at a grade-dependent manner in our single-center cohort retrospective study. Further validation using a multi-center large cohort is warranted.


Introduction
Bronchopulmonary dysplasia (BPD) is a disease commonly seen in preterm infants born at less than 32 weeks of gestation as a result of exposure to multiple antenatal and postnatal factors, causing impaired parenchymal and vascular growth, and forced early development of the immature lung [1]. Despite advances in perinatal and neonatal care, the incidence of BPD remains high, and it is a critical component of neonatal morbidity and mortality [2].
One reason that the incidence of BPD remains high compared to other neonatal morbidities may be the lack of a standardized definition that accurately captures the pathogenesis and disease severity in preterm infants [2]. The definition of BPD has evolved since its first description in the 1960s by Northway et al. who reported that a group of moderately preterm infants with surfactant deficiency survived with adverse outcomes from prolonged ventilation with high oxygen supply [3,4]. In 2000, the BPD definition was revised by the National Heart, Lung and Blood Institute (NHLBI), which used gestational age, oxygen supplementation levels, and the presence of positive pressure ventilation to classify BPD as none, mild, moderate, or severe [5]. However, with the advent of new respiratory modalities and improvements in clinical management, this definition quickly became outdated. In 2016, the definition of BPD was updated by the National Institute of Child Health and Human Development (NICHD). They proposed changing the severity grading to 1, 2, and 3, and accounting for newer noninvasive ventilation modes. They also required additional respiratory parameters and radiologic confirmations of lung disease [6]. The proposed grading system, albeit comprehensive, became cumbersome for clinical and research purposes. To resolve this, Jensen et al. improved the grading system to be more pragmatic by taking an evidence-based approach [7]. They showed that the optimal grading system of BPD would only include the mode of respiratory support at 36 weeks postmenstrual age (PMA). This grading system maintained the discretization of BPD severity as Grades 1, 2, and 3. The same group subsequently validated the use of this grading system to predict respiratory outcomes after NICU discharge [8]. With the grading system, infants who were previously categorized as severe BPD (requiring respiratory support with FiO2 ! 30% at 36 weeks PMA) now can be further subcategorized into Grade 1, 2, or 3. The BPD severity system based on the 2001 NICHD workshop has been correlated with neurodevelopmental outcomes. Survivors of prematurity with BPD are likely to have a higher magnitude of neurodevelopmental deficits, which manifest as low Bayley Developmental scores, cerebral palsy, deafness, and/or blindness [5,9]. Furthermore, the severity of BPD has been shown to serve as an independent risk factor for neurodevelopmental deficits. Moreover, the presence of persistent and severe BPD at term-equivalent age has been proposed to serve as a predictor of subsequent neurodevelopmental outcomes [10].
Although neurodevelopmental outcomes were included in the development of the Neonatal Research Network (NRN) grading system as a secondary outcome in order to resolve any equivalent results obtained from the concordance (c)-statistics using post-discharge mortality or respiratory morbidities as the outcome variable (the study primary outcome), it was not used to select the final definition of BPD grade. Although an association between BPD grade and neurodevelopmental outcomes was evident based on the c-statistics, the association was never quantified in terms of risk ratios based on the BPD grade. The objective of this study was to estimate the risks of adverse neurodevelopmental outcomes for each BPD grade in preterm infants born at less than 30 weeks of gestation.

Study participants
We conducted a retrospective study at the Loma Linda University Children's Hospital neonatal intensive care unit (LLUCH NICU), a level 4 regional referral unit, and at the Riverside University Health System Medical Center level 3 NICU [11]. Infants in both NICUs were cared for by the same medical team. Patients were included if they met the birth gestation criteria of less than 30 weeks between 2015 and 2020, had respiratory data available at 36 weeks PMA for BPD grading, and received a formal neurodevelopmental assessment using the Bayley Scales of Infants and Toddler Development (BSID) at the LLUCH infant follow-up clinic. BSID-Third Edition (Bayley III) was used until BSID-The Fourth Edition (Bayley IV) was introduced in 2019. We restricted the inclusion period to 2015 and later because documentation of neurodevelopmental assessment in the electronic healthcare records was not reliable prior to 2015. The study was approved by the Loma Linda University Human Research Protections program and the RUHS institutional review board office.

Data collection
Clinical data were extracted by one study member (A.N.) from the database of the electronic healthcare records. We extracted age, race/ethnicity, sex, gestational age (GA) at birth, birth weight, diagnosis list, medications, and respiratory support at 36 weeks PMA. A baby was considered to have received dexamethasone if the medication was administered for 10 days or more. Intraventricular hemorrhage (IVH) grades and necrotizing enterocolitis (NEC) stages were assigned based on the documentation of the international classification of diseases, 10 th edition clinical modification codes in the medical charts. Late pulmonary hypertension was defined based on an echocardiography assessment of pulmonary hypertension between 36 and 44 weeks PMA. If no echocardiography was performed, the baby was considered not having pulmonary hypertension. Small for gestational age (SGA) was defined as a birth weight z-score of < À1.28 (10 th percentile) on the 2013 Fenton sex-specific growth charts [12]. Maternal and perinatal data, including gestational diabetes mellitus (GDM), gestational hypertension (GHTN), delivery mode (vaginal vs. cesarean section), as well as BSID scaled scores, were manually extracted from NICU admission History & Physical notes and infant follow-up clinic notes, respectively. GHTN was defined as a maternal diagnosis of gestational hypertension, preeclampsia (with or without preexisting hypertension), or eclampsia. A mother was considered to have no GDM or GHTN if there was no documentation in the note. Extracted data were deidentified prior to further processing and analysis.

Definitions of BPD grade and adverse neurodevelopmental outcomes
The definition of BPD grades used in this study was based on a recent publication by Jensen et al. [7]. They developed a parsimonious grading system based on respiratory support modes at 36 weeks PMA. Grade 0 was assigned to infants without any mode of respiratory support. Grade 1 was assigned to infants receiving nasal cannula with a flow rate of 2 L per minute (LPM) or less. Grade 2 was assigned to infants receiving nasal cannula of more than 2 LPM flow rate or noninvasive positive pressure support. Grade 3 was assigned to infants receiving invasive support. The fraction of inspired oxygen provided and the duration of respiratory support in the first 28 days of life were not used in the grading definition.
We used the neurodevelopmental assessment report that was closest to 2 years of corrected age to define adverse neurodevelopmental outcome. The median age of neurodevelopmental assessment was 1.9 years. A composite adverse neurodevelopmental outcome was defined as a scaled score of less than 7 in any of the five domains (cognitive, receptive language, expressive language, fine motor, or gross motor), or a diagnosis of cerebral palsy regardless of type. For regression analysis, patients were excluded if scores in one or more domains were missing.

Statistical analysis
Descriptive statistics were performed for demographic comparisons. The median and interquartile range were used to represent ordinal variables (scaled scores). Ordinal variables (scaled scores) were presented as median and interquartile range. Continuous variables were presented as mean ± standard deviation. Chisquared tests were used for nominal variables. For ordinal variables, non-parametric one-way analysis of variance (ANOVA) and post-hoc pairwise analysis with Bonferroni correction were used. For continuous variables, a parametric one-way ANOVA was used for comparison. A p-value of < 0.05 is considered to reach the significance level.
Linear regression analyses were used to assess the association between BPD grade and scaled scores (treated as continuous variables) of each domain individually. Univariable and multivariable logistic regression analyses were performed to model the association between BPD grades and the composite adverse neurodevelopmental outcome. Odds ratio and 95% confidence interval (CI) were presented. All analyses were performed in R 4.1.1 using RStudio 1.4 [13,14].

Results
Between 2015 and 2020, 804 infants born at less than 30 weeks' gestation received care by the LLUCH NICU team; 721 infants had respiratory outcome data at 36 weeks PMA for BPD grade designation (Supplemental Figure 1). After chart review, we found that 188 infants did not follow up in our high-risk infant follow-up (HRIF) clinic. Among the remaining 533 infants, 283 did not have BSID scores for various reasons (Supplemental Table 1). Notably, out of 32 infants with Grade 3 BPD, 23 did not have BSID scores, including 18 (56%) that died after the first HRIF clinic visit. Overall, BSID assessment reports were available in 250 patients. These patients were included in the analysis.
Among the included patients, 54% (n ¼ 134) were male. The median birth GA was 26 weeks and 6 days, and 64.4% (n ¼ 161) of infants had BPD. The demographic characteristics for each BPD grade group are summarized in Table 1. The infants with BPD were born at a lower GA, weighed less at birth, and were more likely to be male and SGA. Antenatal exposures were comparable among groups. The percentage of infants with adverse composite neurodevelopmental outcomes increased with BPD grade, from 44% (39 out of 89) in infants with no BPD to 89% (8 out of 9) in infants with Grade 3 BPD. Furthermore, 81% (26 out of 32) infants diagnosed with Grade 3 BPD (n ¼ 32) either died after NICU discharge (n ¼ 18) or had adverse neurodevelopmental outcomes (n ¼ 8). The demographic and morbidity comparisons between the 250 included patients and the remaining 471 infants (188 without HRIF follow-up and 283 without BSID score data) were available in Supplemental Table 2.
As shown in Supplemental Figure 2, BSID scaled scores decreased as the BPD grade increased, with all domains reaching statistical significance. In a post-hoc pairwise analysis, scaled score differences between Grade 0 and Grade 1 did not reach statistical significance in all domains. However, the differences were significant between Grade 0 and Grade 2 in cognitive and fine motor domains. Furthermore, in cognitive, fine motor, and gross motor domains, Grade 0 and Grade 3 had significant statistical differences. In the expressive language domain, the differences were significant between Grade 0 and Grade 2 and between Grade 1 and Grade 2, with no significant differences noted in the receptive language domain. In this cohort, 56%, 46%, 25%, and 11% of infants in Grades 0, 1, 2, and 3 did not have pre-defined adverse neurodevelopmental outcomes at toddler age, respectively. In a univariable logistic analysis, when compared to BPD Grade 0, BPD Grade 2 (OR: 3.8, 95% CI: 1.9-7.9), BPD Grade 3 (OR: 10.3, 95% CI: 1.8-198.4), but not Grade 1 (OR: 1.5, 95% CI: 0.8-2.7), were found to significantly associated with an adverse composite neurodevelopmental outcome (Table 2). Additionally, birth gestation (OR: 0.7, 95% CI: 0.7-0.9) and dexamethasone exposure (OR: 2.6, 95% CI: 1.4-5.4), but not birth weight z-score, were also significantly associated with an adverse composite neurodevelopmental outcome. In a multivariable logistic analysis by including BPD grade, birth gestation, and dexamethasone exposure as independent variables and the composite neurodevelopmental outcome as the outcome variable, we found that, after adjusting for birth gestation and dexamethasone exposure, BPD Grade 2 and Grade 3 remained significantly associated with adverse neurodevelopmental outcomes, with increases in odds by 2.7 and 7.2-fold, respectively (Table 2). While birth gestation remained significantly associated with an adverse composite neurodevelopmental outcome after adjusting for other variables, the role of dexamethasone exposure became non-significant.
We then assessed the association between BPD grade and individual BSID scaled scores in a linear regression model adjusting for birth gestation and dexamethasone exposure. We found that Grade 2 and Grade 3 were each associated with lower scores in cognitive, fine motor, and gross motor domains,  whereas BPD grades were not associated with lower scores in the receptive and expressive language domains except for an estimated 2.4-point decrease in the receptive language domain when comparing Grade 3 BPD to no BPD (Supplemental Table 3).

Discussion
In this study, we reported the results of a series of analyses to assess the relationship between the NRN BPD grading system and toddler-age neurodevelopmental outcomes. The association between BPD and neurodevelopmental outcomes has been well established in the literature but the relative risks of adverse neurodevelopmental outcomes between each BPD grade has not been quantified [5]. In our single-center retrospective analysis, we found that Grade 2 and 3, but not Grade 1 BPD were significantly and independently associated with an adverse neurodevelopmental outcome.
Prior to the publication of the NRN grading system, studies were conducted to assess the relationship between prolonged invasive respiratory support needs and neurodevelopmental outcomes [9,15]. As part of the Extremely Low Gestational Age Newborns (ELGAN) study, Van Marter et al. assessed cerebral palsy outcome in infants who continued to require mechanical ventilatory support at 36 weeks PMA and found fourand six-fold increases in diparesis and quadriparesis at 24 months of age, respectively [15]. Another study found that infants mechanically ventilated for a longer period of time had worse mental and psychomotor developmental indices at 18 months of age compared to those infants who were mechanically ventilated for a shorter period of time [9]. Similarly, our results showed that preterm infants who were mechanically ventilated (Grade 3) had worse neurodevelopmental outcomes, although our findings may be biased due to post-discharge death and loss to follow-up. On the other hand, in our cohort, 24.3% of Grade 2 and 3 infants did not have the adverse composite neurodevelopmental outcomes. Further studies are merited to identify protective factors in this subpopulation.
Early studies suggested that dexamethasone use may be a stronger predictor of neurodevelopmental deficits when compared to BPD [16,17]. In our logistic analysis, dexamethasone exposure was significantly associated with adverse neurodevelopmental outcomes in a univariable model but became statistically insignificant after adjusting for birth gestation and BPD grade. This is supported by the findings from the 2017 Cochrane Review, which reported that adverse neurodevelopmental outcomes were not associated with dexamethasone use [18]. Our practice has a guideline for postnatal corticosteroid use in terms of indication and duration, but the decision to administer corticosteroids was made on an individual basis and was largely determined by the on-service neonatologist. There were no protocols on the timing and number of courses to be given to each individual patient. Such an approach provided an opportunity to adjust for the impact of corticosteroid use on adverse neurodevelopmental outcomes. Moreover, as our practice has a high prevalence of BPD, corticosteroid administration is unlikely to cause a detrimental impact on neurodevelopmental outcome or mortality, based on a recently revised meta-analysis by Doyle et al. [19].

Limitations
There were several limitations to our study. First, this was a single-center retrospective study. Second, only one-third of the infants with a BPD grade diagnosis had BSID score data, which may complicate the interpretation of the findings. However, the demographic and morbidity profiles between those with and without BSID data were rather comparable. Notably, as a majority of the infants died after 36 weeks PMA, our findings of higher risk of neurodevelopmental impairment in infants with Grade 3 BPD should be confirmed in larger populations. Lastly, socioeconomic status and maternal education, two well-established variables affecting neurodevelopmental outcomes of preterm infants, were not available for control in our analysis, making the findings of the differences in language outcomes of limited value [20][21][22].

Conclusion
The NRN BPD grading system replaces the old severity system to stratify infants based on post-discharge respiratory outcomes. The grading system also correlates with neurodevelopmental outcomes. Our study, albeit with limitations, further quantified the risks of adverse neurodevelopmental outcomes with increasing BPD grade. A larger cohort study is needed to verify our findings.

Disclosure statement
No potential conflict of interest was reported by the author(s).