Controversies about the definition of bronchopulmonary dysplasia at 50 years

Northway et al. 1 first described bronchopulmonary dysplasia (BPD) 50 years ago as a lung injury syndrome in preterm infants thought to be primarily caused by high levels of supplemental oxygen and airway pressure from mechanical ventilation. The disease has changed over the years because of improved respiratory support strategies that are used to care for progressively smaller and more immature infants. Presently, the pathophysiology of BPD in the very preterm infant is thought to be the simultaneous clash of the programmes for lung development, lung injury and lung repair over the weeks to months required to care for these infants. The definitions of BPD have evolved as the at-risk populations of infants and the care strategies have changed, as recently reviewed in the Journal by Hines et al. 2. In a letter to the Journal, Meyer et al. 3 propose modifications of the NIH Workshop definition for their ongoing trial of oral Vitamin A to prevent BPD 4. Their modifications are adaptations to newer care strategies such as high-flow nasal cannulas and very low flow high concentrations of oxygen that render infants either unclassifiable or misclassified as BPD. Unfortunately, these adaptations are band aids that do not solve the basic problems with the definitions, which are arbitrary classifications of infants with or without BPD based on their oxygen need at a predetermined time point. The inadequacy of current definitions of BPD is well recognized for their lack of precision that do not capture the severity of disease or have good predictive values for later pulmonary and neurodevelopmental outcomes. The root problem of most diagnostic criteria is that they use the clinical care variables of oxygen and pressure and not pathophysiology to define the disease. The elements that should be considered for a comprehensive definition are quite overwhelming as outlined in the Table 1 5. Multiple clinical variables contribute to the incidence and severity of BPD. Birthweight, gestational age and sex are potent biological variables that drive lung development, injury and repair pathways. Many extremely low-birthweight (ELBW) infants without BPD also have abnormal lung function at term and increased pulmonary morbidity for years 6. Thus, there are no 'normal' ELBW infants to calibrate the severity of BPD against. The disease is modulated by antenatal exposures (chorioamnionitis, antenatal steroids, maternal smoking) and foetal growth status as intrauterine growth restriction (IUGR) is a potent predictor of BPD. Then, early respiratory status (oxygen, airway pressure) and the amount and duration of respiratory support drive different phenotypes (described by oxygen exposure) towards BPD 7. Drug therapy (vitamin A, caffeine, postnatal steroids) can alter BPD incidence but may not change long-term outcomes. Present diagnostic criteria condense these complexities into simply oxygen and pressure exposure, at a single point in time, generally, 36 weeks postmenstrual age (PMA), on the continuum of disease progression. Other biological variables for which we have minimal translatable information could help us better define BPD. There are genetic contributors to the programmes of lung development, injury and repair that drive BPD that are just being identified 5. While multiple molecular biomarkers of inflammation, oxidant injury and growth factors associate with BPD, none are particularly useful for identifying clinical phenotypes that might be targeted for specific therapies. Identification of good biomarkers that would separate populations of infants at increased risk of BPD would be most helpful. A BPD definition such as the Shennan criteria that classifies infants at a single point of the clinical course of BPD at 36 weeks ignores the rich respiratory and clinical history of infants 8. Addition of quantification of oxygen exposure for >28 days is a modest increase in clinical information but minimally documents the chronicity of the respiratory course. With intense data collection for studies or perhaps with effective use of the electronic medical record, areas under the curves for oxygen and airway pressure exposures and different support modes could better define the degree of lung damage. Scoring systems using multiple clinical variables could further enrich the use of clinical information to generate BPD severity assessments. Recently, Isayama et al. 9 demonstrated that evaluation of BPD at weekly intervals from 34 to 44 weeks can change the prediction of outcomes at 18–21 months. After 36-week assessment, postconceptional age may better define infants at risk for subsequent adverse outcomes, but fewer infants will remain in the hospital for accurate assessments. A population that is missed by all current definitions is those infants that die of respiratory failure between two weeks and 36–40 weeks PMA. In the Prematurity and Respiratory Outcomes Program (PROP) study, 63 of 836 ELBW infants died prior to 36 weeks, and 71% were adjudicated to have died primarily from respiratory failure 10. These infants with 'early, lethal BPD' should be identified for innovative therapies such as stem cell treatments. A unified definition for BPD is further confounded by whom is interested in using the definition. Clinicians caring for premature infants would like to have reliable early predictors of whom will develop BPD to introduce early care strategies to decrease the risk. The clinician's interests are primarily focused on the weeks before diagnosis of BPD can be made. Epidemiologists would like a simple and robust definition to track across neonatal units and across time. Here, precision is perhaps less critical than consistency of diagnosis. Researchers of BPD would like a robust and nuanced diagnosis that could better refine clinical populations for more targeted therapy evaluations, that is definitions of different phenotypes of BPD that may have more consistent pathologies. Industry also wants a robust definition(s) of BPD that can be clearly scored despite variable care strategies across neonatal units and that can predict important longer-term outcomes. Finally, parents are primarily interested in survival and in the quality of life of their children. The reality is that clinical investigators have spoken with their feet using simple definitions of BPD. Of 47 randomized clinical trials of drugs for BPD, 34% did not identify the definition of BPD that was used. Of trials with definitions, 71% used only oxygen at 36 weeks with 6% refining that definition with a test of oxygen need 11. Oxygen requirement at 28 days was used for 32% of trials. Most recent clinical and epidemiology reports use one of the simple oxygen use definitions 2. Nevertheless, there are substantial efforts being made to develop better diagnostic criteria. The International Neonatal Consortium of industry, academic neonatologists and drug regulatory groups is drafting an analysis of inadequacies of the BPD diagnoses with suggestions as to how to develop better criteria. The National Institute of Child Health and Development held a workshop focused on BPD in the fall of 2016, and a report will also address BPD definitions. Dr. Judy Aschner is analysing the PROP cohort for developing a BPD definition that can predict respiratory outcomes at one year. There are likely other initiatives as well. More than one definition and diagnostic criteria may need to be developed to meet the requirements of the different users.