Intermittent Hypoxemia and Bronchopulmonary Dysplasia: Manifestations of Immature Respiratory Control and the Preterm Lung

Bronchopulmonary dysplasia (BPD), first reported by Northway and colleagues (1), continues to be redefined in its diagnosis, causalities, and therapies (2). BPD is a heterogeneous chronic condition of parenchymal, airway, and cardiovascular pathology after preterm birth that has strong associations with adverse childhood morbidities (3). It is a disease both caused by and defined by the degree of respiratory therapy needed to prevent hypoxemia and death during the first weeks of life. New data indicate that unfavorable responses to these therapies, as measured by an increase in intermittent hypoxemic events, may contribute to or are associated with the diagnosis of severe BPD at 36 weeks’ postmenstrual age (4).

Intermittent hypoxemia (IH) is an emerging descriptor of the physiologic manifestations of immature respiratory control and the developmentally preterm lung. In this issue of the Journal, Jensen and colleagues (pp. 1192–1199) describe, in a multinational cohort of more than 1,000 preterm newborns, associations in the first weeks of life between both prolonged IH events (oxygen saturation [Sp_O2] < 80% for >1 min as measured by pulse oximetry) and cumulative hypoxemia (% time with Sp_O2 < 80%) with future diagnosis of severe BPD (4). This work confirms reported associations between IH and BPD from smaller single-center cohorts (5, 6) and that IH events in extremely preterm newborns increase in frequency during the first weeks of life, even in newborns in whom BPD was not diagnosed. Furthermore, this study confirms associations between severe BPD and late death or adverse neurodevelopment suggesting that early and prolonged IH may contribute to the known risk of poor developmental outcomes in infants with severe BPD.

A range of IH definition criteria have been described in the neonatal research literature including a fall in Sp_O2 to <80–85% with or without accompanying duration criteria. Is a threshold of <80%, as used by Jensen and colleagues and others, too low considering American Academy of Pediatrics recommendations for Sp_O2 targets of 90–95% (4)? Or appropriate to reduce supplemental oxygen exposure? The jury is still out. Minimum duration criteria are also important in identifying “shorter” events that do not require intervention. Jensen and colleagues report on IH >1 minute in duration based on their previous data showing that shorter IH events were not associated with poor outcomes in this cohort (7). Previous single-site studies have confirmed an association between longer IH duration and BPD, although neither study identified a specific duration cutoff (5, 6). The authors did not report a maximum duration threshold, which is important to distinguish intermittent from sustained hypoxemia that can have differing physiological effects (8). Lastly, Raffay and colleagues looked at the time interval between subsequent IH events and found that BPD infants had more event clusters occurring <1 minute and 1–20 minutes apart with no differences observed in isolated events (>20 min apart) (6). Taken as a whole, these studies suggest that not all IH events may be detrimental and future trials should focus on identifying high risk patterns of IH.

An important contributor that can influence identification of high-risk patterns is the pulse oximeter averaging time. With no current standard, 8 or 10 seconds is the most commonly used setting in the neonatal ICU to reduce nuisance alarms. Longer averaging times can result in short, closely clustered IH being combined into fewer but longer events (9). Thus, with the 16-second averaging time used by Jensen and colleagues, interpretation of IH <1 minute in duration as nonhazardous should be made with caution.

Pulse oximetry may be an important noninvasive tool that can identify infants at risk for poor outcomes. Jensen and colleagues note that IH frequency, more so than time spent in hypoxemia, predicted BPD risk, suggesting current-generation oximeter Sp_O2 histograms could be improved. More sophisticated unbiased modeling of large data sets could give insight into which IH parameters are important. One such effort to investigate the entire Sp_O2 waveform is the Pre-Vent multicenter trial of more than 500 extremely preterm patients from five contributing centers (10). Investigators will use the highly comparative time-series analysis developed by Fulcher and colleagues (11) to identify a range of models that best describe the longitudinal profile of cardiorespiratory waveform patterns during the first few months of life. The goals of this collaborative are to delineate mechanisms of immature ventilatory control in preterm infants and to identify predictive markers and ascertain their contributions to respiratory morbidities, including but not limited to traditional BPD definitions as extremely preterm newborns are likely to display some BPD phenotype in terms of lung development and longer-term outcomes (2). The results from this trial could inform the next generation of advanced pulse oximeter monitoring.

Complicating interpretation of observational IH data is the concomitant use of supplemental oxygen to treat hypoxemia and maintain Sp_O2 targets. Of note, higher Sp_O2 targets are associated with decreased IH events (12) but also increased supplemental oxygen use at 36 weeks’ postmenstrual age (13). As noted by the authors, determining association versus causation from observational studies and confounding by indication remains a challenge. What can reasonably be speculated is that the most at-risk infants for BPD likely have the worst respiratory control, worst lung disease, and most IH events and are on the highest Fi_O2. Looking to translational animal studies (14, 15) exposure to intermittent hypoxia alone does not result in a BPD phenotype; however, when superimposed with hyperoxia, the lung pathology occurs in excess of sustained hyperoxia alone. Thus, the hyperoxic overshoot after an IH event may be the villain to battle.

Classifying pathologic versus expected patterns of IH will be critical to understanding and predicting risk for poor respiratory outcomes, particularly as automated supplemental oxygen titration enters the clinical realm (16, 17).