The Complex Challenge of Defining Bronchopulmonary Dysplasia

Bronchopulmonary dysplasia (BPD), a condition that has been a focus of neonatal research for decades, is the most common chronic morbidity in extremely preterm infants.1 While the incidence of other neonatal complications has declined, the incidence of BPD has remained consistently high, affecting 30 to 60% of surviving extremely preterm infants.1,2 Evaluating BPD incidence over time is challenging because clinicians and clinical trialists base BPD diagnosis on management parameters, rather than specific structural, anatomic, or functional physiologic measures. This reliance on a management-based definition has led to the emergence of a number of proposed definitions, as reviewed in this issue of Pediatrics by Katz et al.3 Although development of a more robust, non-management-based definition would be ideal, the procedures necessary for such diagnosis, such as lung biopsy or pulmonary function tests,4 are either too invasive, not widely available, or cost-prohibitive.5 Thus, management-based definitions of BPD prevail.Identifying the "best" definition of BPD has been a long-standing point of contention in neonatology. One of the main drivers of the iterative attempts at a consensus BPD definition has been the overall goal of identifying a definition of BPD in the initial neonatal intensive care unit (NICU) hospitalization that improves the prediction of long-term respiratory outcomes in extremely preterm infants. Being able to predict long-term respiratory outcomes is critical, as it provides an opportunity to identify high-risk infants for timely interventions, allows clinicians to counsel parents about future expectations, and informs the development and assessment of clinical trials of interventions that affect these outcomes.6 Measuring these long-term outcomes to validate a BPD definition in the initial hospitalization can be time-consuming, cost-intensive, and difficult to conduct broadly, especially in outpatient settings, and even more so during childhood.6 However, because these NICU hospitalization management-based BPD definitions are associated with long-term respiratory and neurodevelopmental morbidities, BPD is frequently used as a surrogate predictor for these outcomes.7In their systematic review of 18 studies, Katz et al compared the prognostic accuracy of multiple BPD definitions on long-term outcomes in preterm infants with a mean gestational age under 30 weeks and a mean birth weight less than 1500 g.3 The authors used sensitivity, specificity, negative and positive predictive values, and c-statistics to assess predictive accuracy. Pulmonary outcomes of interest included respiratory hospitalizations and a group of categorical respiratory morbidities comprising asthma, wheezing, tracheostomy, and the use of supplemental oxygen. The authors also assessed associations between BPD and neurodevelopmental impairment (NDI) and late death. Late death was defined as that occurring after the diagnosis of BPD and before follow-up and was evaluated separately and as a composite outcome with respiratory outcomes or NDI. The systematic review revealed that none of the definitions compared was consistently superior to the others in predicting the outcomes of interest.This study highlights several areas of concern for clinicians and researchers interested in the consistent application of BPD diagnosis to inform long-term outcome prediction. The definitions reviewed by Katz et al incorporate the use of oxygen or respiratory support at varying time points of assessment.3 However, the need for supplemental oxygen or respiratory support is influenced by several factors, including clinicians' subjective assessment of respiratory status, oxygen saturation targets, immaturity of the central nervous system, the infant's growth trajectory, and use of medications such as diuretics and bronchodilators, which affect the respiratory system.8 In clinical practice, different definitions are applied based on local guidelines and preferences, and geographical location.9,10 Some definitions include the use of oxygen or respiratory support for at least 28 days; others require an assessment at 36- or 40-week postmenstrual age.10 In practical use, manually recording and counting days of support is burdensome and open to human error. Another issue to be considered is whether definitions established in one era should be applied to infants from a different era. Over time, our understanding of the pathophysiology of BPD has evolved, as has the availability of prevention therapeutics, such as antenatal steroids and surfactant, and the modalities of respiratory support at our disposal.4 In essence, the earlier definitions of BPD may reflect a different phenotype from "new" BPD, and comparing outcomes from different eras does not provide the most accurate information about contemporary surviving preterm infants.11Like the BPD definitions, the long-term outcomes that were assessed in the review by Katz et al differed considerably across the examined studies.3 In particular, the definitions of NDI were variable in all of the included studies. BPD has frequently been used as a predictor of neurodevelopmental outcomes in survivors, but as emphasized by Katz et al, ultimately these outcomes in preterm infants are influenced by numerous other factors and cannot be well predicted by BPD alone.3 For example, NICU medication management, including type and timing of corticosteroid exposure, affects the incidence of both BPD and NDI; understanding the potential clinical impact of such interventions underscores the need for a consistent definition.12 Given the challenges of predicting neurodevelopmental outcomes, prediction of long-term respiratory outcomes may be more achievable and has been the focus of definitions published in 2017 by the Canadian Neonatal Network and in 2019 by the Neonatal Research Network.13,14 As the field of neonatology looks to harmonize the definition of BPD, the authors also highlight the need for a core set of respiratory and neurodevelopmental outcomes that are consistently evaluated in clinical trials and follow-up studies. Using input from key stakeholders, clinically meaningful outcomes should be established that are patient-centered and impact the quality of life of both the patients and their families.15,16The potential benefits of an ideal BPD definition are clear, but how can we arrive at one? Unless neonatologists reach a consensus as a community, how will we accurately know that we are improving the care and outcomes of this vulnerable population? If the ideal definition of BPD were one that is developed based on its high predictive value of respiratory outcomes, should this definition be updated over time to evolve with changes in outcomes and neonatal care, including diagnostic tests and "big data" analyses? Can we agree on a definition that is sufficiently predictive of a consensus collection of long-term outcomes, is simple enough to verify, interpret, and apply consistently, is meaningful to families, and reflects the underlying pathophysiology of the lung disease? Or is our field fixated on something that is not fully achievable? We believe that neonatologists should, and can, agree on a universal definition of BPD and hold each other accountable for its consistent use across research endeavors and clinical metric reporting. While we may never arrive at a "perfect" management-based definition, consistency in the use of an agreed-on, validated definition will allow us at last to "compare apples to apples" and identify true changes and differences that may be associated with differing clinical practices. It is likely that such a definition will inevitably need updating because of the discovery and availability of new tools that can improve data assessment. However, with the consistent use of one definition at a time, the neonatal field will be a step closer to impacting the rate of long-term respiratory morbidity.Thank you to Dr P. Brian Smith and Dr C. Michael Cotten for their reviews of the manuscript.