When bronchopulmonary dysplasia was first described by Northway in 1967 he didn’t try to produce a definition, his paper was a description of a small number of preterm survivors of high oxygen and positive pressure ventilation. He noted some years later the long term implications for pulmonary health of the survivors, during adolescence and as young adults, many had hyperinflation, and small airways dysfunction was common. Many had on-going respiratory symptoms.
Since then we have been trying to arrive at a way of predicting long term pulmonary health without waiting for those very prolonged outcomes. Andy Shennan came up with the idea of oxygen requirements at 36 weeks because that was a better predictor of post-discharge pulmonary problems than oxygen need at 28 days, which had become the rather arbitrary definition prior to his publication. Even with the first publication in 1988, the positive predictive value of O2 at 36 weeks for post-discharge respiratory problems was only 63% and the NPV was 90%. With changes in survival and the pattern of lung injury, the usefulness of oxygen at 36 weeks as a predictor of clinically important pulmonary injury has diminished.
A study by the Canadian Neonatal Network/Neonatal Follow-up Network examined the predictive ability of various combinations of oxygen therapy and other respiratory support among infants <29 weeks gestation for predicting what they called “serious respiratory morbidity” which was defined as either (1) 3 or more rehospitalizations after NICU discharge owing to respiratory problems (infectious or noninfectious); (2) having a tracheostomy; (3) using respiratory monitoring or support devices at home such as an apnea monitor or pulse oximeter; and (4) being on home oxygen or continuous positive airway pressure at the time of assessment between 18 and 21 months corrected age. (Isayama T, et al. Revisiting the Definition of Bronchopulmonary Dysplasia: Effect of Changing Panoply of Respiratory Support for Preterm Neonates. JAMA Pediatr. 2017;171(3):271-9.) At least 3 rehospitalizations was chosen because the 95th percentile of the number of readmissions owing to respiratory problems in this cohort was 2. Such serious morbidity occurred in 6% of the approximately 1500 babies included, and was not well predicted by oxygen requirements at 36 weeks, it was better predicted by either oxygen or respiratory support (which included high flow nasal cannulae at more than 1.5 litres per minute or other invasive or non-invasive support) and determining that at 40 weeks post-menstrual age was a better predictor than at 36 weeks. 24% of the babies in the cohort had oxygen or respiratory support at 40 weeks, of them 16% had “serious respiratory morbidity”, of those who did not have O2 or respiratory support at 40 weeks 2.4% had this severe morbidity.
Although this definition gives an adjusted odds ratio for serious respiratory morbidity of 6.1 (95% CI 3.4-11.0), which is better than a definition at 36 weeks, I am not sure how useful it is, with a PPV of 16% and a fairly high negative predictive value, a large majority of infants with this definition of BPD do not have ‘serious respiratory morbidity’.
The bigger problem maybe, is the definition of SRM (typing the whole phrase takes me too long!) being home on an apnea monitor doesn’t seem equivalent to me to having 3 hospitalisations, or still being on CPAP at 18 month follow-up. I think determining which definition of BPD, (a definition created by physicians), is best at predicting SRM, as defined by physicians, won’t necessarily help us a great deal. We should first determine which respiratory outcomes are most important to families, and then rank the clinically important outcomes (or rather ask parents to rank them), if we did that we could produce some sort of ordinal score of severity of preterm chronic lung disease. Then we could figure out if it is possible to predict those outcomes before they occur. We also should be very careful about introducing items which may be very strongly affected by different practice patterns, (such as home apnea or saturation monitoring), in some centers home use of such monitors is exceedingly rare, compared to others.
If I were to guess, I would think that families, during the first year or 2 or life, would be more disturbed by emergency room visits, and hospitalisations (especially PICU re-hospitalisations) and next by home oxygen, especially as the infant becomes more active, and next by daily use of respiratory medications. Tracheostomy would probably be way out at the top, but includes very few babies in Canada. But I don’t think my guesses are worth much, we should really ask parents, and other members of society.
Another recent publication from the NICHD network only examined possible variations of BPD definitions at 36 weeks, not at other post-menstrual ages. (Jensen EA, et al. The Diagnosis of Bronchopulmonary Dysplasia in Very Preterm Infants: An Evidence-Based Approach. Am J Respir Crit Care Med. 2019) They tested 18 different ways of analyzing different levels of respiratory support and FiO2 at 36 weeks for their accuracy in predicting post-discharge serious respiratory morbidity which was defined as death between 36 weeks and follow up or:
the occurrence of at least one of the following: tracheostomy placed any time prior to follow-up; continued hospitalization for respiratory reasons at or beyond 50 weeks PMA; use of supplemental oxygen, respiratory support, or respiratory monitoring (e.g. pulse oximeter, apnea monitor) at follow-up; or ≥2 re-hospitalizations for respiratory reasons prior to follow-up. Continued hospitalization at 50 weeks PMA is approximately 2 standard deviations above the mean age at discharge for extremely preterm infants included in Neonatal Research Network studies. Two or more re-hospitalizations represents the upper 75th percentile for re-hospitalization number among Network babies.
In this study, the best definition appeared to depend solely on the level of respiratory support at 36 weeks irrespective of the oxygen needs, using these diagnostic criteria, infants breathing in room air at 36 weeks PMA did not have BPD. Disease severity among the remaining infants was classified according to support: grade 1, nasal cannula at flow rates ≤2L/min; grade 2, nasal cannula at flow rates >2L/min or non-invasive positive airway pressure; and grade 3, invasive mechanical ventilation.
Using these definitions the percentage of babies with respiratory morbidity increased from 10% among infants without BPD, to 19% with grade 1, 35% with grade 2, and 77% with grade 3.
My comments about this study echo those for the CNN paper, the relative importance of these different aspects of SRM is variable, and an overall moderately good capacity to predict which infants may develop SRM is interesting, but would it be enough to be used as an interim outcome measure for clinical trials? Or as a marker of respiratory outcomes for quality control? In addition some of the measures are likely to be heavily influenced by practice patterns, some centers will use higher flow rates and lower oxygen concentrations, some will remove CPAP earlier than others, or extubate with higher ventilatory requirements.
One of the reasons for bringing this up now is a publication of a study aimed at reducing lung injury in the very preterm infant. Davis JM, et al. The role of recombinant human CC10 in the prevention of chronic pulmonary insufficiency of prematurity. Pediatr Res. 2019. I think this is the second small preliminary trial of Clara Cell Protein in very preterm babies, the previous one having 22 babies total (7 controls, 7 with low dose and 7 with higher dose Clara Cell Protein, called CC10 in the new study). It is a recombinant version of a protein produced by human Clara cells, which has several potentially beneficial effects, including reduction of lung inflammation. In the previous trial there was a reduction of signs of lung inflammation with a single intratracheal dose, that study was underpowered for clinical outcomes, and didn’t really show any evidence of clinical benefit. The new trial included 44 babies in all, of 24 to 29 weeks and intubated for RDS. They were included in 1 of 2 sub-studies, the first randomized kids to low dose (1.25 mg/kg) or placebo, the second to higher dose (5 mg/kg) or placebo, given as a single intratracheal dose within 4 hours of surfactant, and at less than 24 hours of age.
The primary outcome of the study was survival without what they called CPIP, chronic pulmonary insufficiency of prematurity, which was defined as presence of any one of the following : (1) evidence of respiratory symptoms (e.g., coughing and wheezing) or use of respiratory medications by parental diaries or pulmonary questionnaires (CPIP-SS), (2) one or more re-hospitalizations for respiratory causes (CPIP-RH), (3) administration of respiratory medications (including oxygen) (CPIP-RM), and (4) at least one non-routine medical visit for respiratory causes (CPIP-DV).
These sound like things that are likely to be of some importance to families; the primary outcome was analyzed as a dichotomous, yes/no outcome, and in addition they recorded the number of criteria for CPIP that were satisfied.
These criteria are clearly much less severe than the outcomes of the 2 other studies discussed above, and almost all of the babies had CPIP, with no difference between placebo and CC10 groups, at either dose. Overall only 8 of the 44 babies survived without CPIP. About half of the babies in the trial did not have BPD by the usual criterion of oxygen needs at 36 weeks, and most of the babies who did not have BPD still had CPIP.
What should the outcome of interest be for studies trying to reduce lung injury in the very preterm infant? Is an interim outcome such as BPD, by any of these definitions, of sufficient predictive capacity to use it in place of clinically important pulmonary dysfunction in the first years of life? I think we have to be very careful that, even if we could come up with criteria that predict chronic pulmonary symptoms among the general preterm population, an intervention that acutely reduces the chances of satisfying those criteria will not necessarily improve long term pulmonary function. Post-natal steroids, for example, by reducing inflammation, reduce the number of babies who satisfy definitions of BPD which are based on oxygen needs, but there is no evidence that they improve long term pulmonary function.
In order to answer the question of the usefulness of “BPD” as a surrogate outcome for improvements in pulmonary health in clinical trials, Anna Maria Hibbs and her colleagues have performed a systematic review of large trials of BPD prevention that also published data on long term respiratory health outcomes Corwin BK, et al. Bronchopulmonary dysplasia appropriateness as a surrogate marker for long-term pulmonary outcomes: A Systematic review. J Neonatal Perinatal Med. 2018;11(2):121-30. The review found 5 trials, the DINO trial from Melbourne of DHA for preventing developmental impairment, which also looked at lung injury, the NO-CLD trial, the early iNO trial which disappointingly has no obvious acronym, SUPPORT, and the SOD the lungs trial (that is actually my own personal acronym for the trial).
All the studies reported BPD, using a variety of definitions, and all had some measures of longer term pulmonary health: all reported healthcare utilization, (hospitalizations, and visits to the doctor or the emergency room); 3 reported respiratory illness, (e.g. asthma); 4 reported respiratory medication use (e.g. bronchodilators, steroids, diuretics, oxygen); one study reported 1 year mortality.
A consistent relationship between BPD and subsequent markers of respiratory morbidity was not seen in the trials studied. Only the NO CLD trial found a significant decrease in rates of BPD. This study also saw significant decreases in the use of respiratory medications including oxygen, but in no other outcome measures. The SOD study also found a significant decrease in respiratory illness requiring asthma medications, but failed to find a significant decrease in BPD. Similarly, the SUPPORT study failed to find a significant decrease in BPD, but did see significant decreases in measures of healthcare utilization and respiratory illness.
Not only therefore is BPD, using any definition, a very limited way of describing severity of early pulmonary injury, it is clearly of very low value as a surrogate for clinical trials aimed at improving pulmonary health.
This brings into even sharper relief the concerns about using BPD as part of the combined outcome of BPD or death. Mortality matters, BPD, not so much.
Chronic pulmonary dysfunction and respiratory illness in early life are important adverse outcomes for families, finding better ways to describe, predict, and prevent them is a priority.