Article Text

Achieved oxygen saturations and risk for bronchopulmonary dysplasia with pulmonary hypertension in preterm infants
  1. Samuel J Gentle1,
  2. Avinash Singh2,
  3. Colm P Travers1,
  4. Arie Nakhmani2,
  5. Waldemar A Carlo1,
  6. Namasivayam Ambalavanan1
  1. 1Department of Pediatrics, The University of Alabama at Birmingham, Birmingham, Alabama, USA
  2. 2Department of Electrical and Computer Engineering, The University of Alabama at Birmingham, Birmingham, Alabama, USA
  1. Correspondence to Dr Samuel J Gentle, Department of Pediatrics, The University of Alabama at Birmingham, Birmingham, Alabama, USA; sjgentle{at}uabmc.edu

Abstract

Objective Characterisation of oxygen saturation (SpO2)-related predictors that correspond with both bronchopulmonary dysplasia-associated pulmonary hypertension (BPD-PH) development and survival status in infants with BPD-PH may improve patient outcomes. This investigation assessed whether (1) infants with BPD-PH compared with infants with BPD alone, and (2) BPD-PH non-survivors compared with BPD-PH survivors would (a) achieve lower SpO2 distributions, (b) have a higher fraction of inspired oxygen (FiO2) exposure and (c) have a higher oxygen saturation index (OSI).

Design Case–control study between infants with BPD-PH (cases) and BPD alone (controls) and by survival status within cases.

Setting Single-centre study in the USA.

Patients Infants born at <29 weeks’ gestation and on respiratory support at 36 weeks’ postmenstrual age.

Exposures FiO2 exposure, SpO2 distributions and OSI were analysed over the week preceding BPD-PH diagnosis.

Main outcomes and measures BPD-PH, BPD alone and survival status in infants with BPD-PH.

Results 40 infants with BPD-PH were compared with 40 infants with BPD alone. Infants who developed BPD-PH achieved lower SpO2 compared with infants with BPD (p<0.001), were exposed to a higher FiO2 (0.50 vs 0.34; p=0.02) and had a higher OSI (4.3 vs 2.6; p=0.03). Compared with survivors, infants with BPD-PH who died achieved a lower SpO2 (p<0.001) and were exposed to a higher FiO2 (0.70 vs 0.42; p=0.049).

Conclusions SpO2-related predictors differed between infants with BPD-PH and BPD alone and among infants with BPD-PH by survival status. The OSI may provide a non-invasive predictor for BPD-PH in preterm infants.

  • Neonatology
  • Respiratory Medicine
  • Intensive Care Units, Neonatal

Data availability statement

Data are available upon reasonable request. The data that support the findings of this study are available upon reasonable request to the corresponding author (SG).

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

WHAT IS ALREADY KNOWN ON THIS TOPIC

  • Bronchopulmonary dysplasia-associated pulmonary hypertension (BPD-PH) increases children’s risk for adverse outcomes into adulthood; however, physicians are currently limited in their ability to predict which children will develop disease.

WHAT THIS STUDY ADDS

  • Achieved oxygen saturations (SpO2) and the oxygen saturation index (OSI; a parameter which relates respiratory support to SpO2) helped identify infants with BPD-PH. Infants with BPD-PH who died achieved a lower SpO2 compared with survivors.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

  • SpO2 targeting may be a modifiable exposure that reduces infants’ risk for the development of BPD-PH. In addition, the OSI may be used as a non-invasive, continuous predictor for BPD-PH.

Introduction

Bronchopulmonary dysplasia (BPD) impacts ~50% of infants born extremely preterm,1 and it is now known that up to 40% of these infants will have BPD-associated pulmonary hypertension (BPD-PH).2 BPD-PH, the most severe endotype of BPD, has been reported to culminate in death in up to 48% of infants before the age of 2.3 Therefore, identification of exposures and predictors for the development of BPD-PH are critical to improve outcomes in this vulnerable population.

Evidence from adults indicates that chronic exposure to high altitudes4 5 and chronic hypoxia from lung disease6 7 cause pulmonary vascular remodelling, right ventricular hypertrophy and PH. In lambs with PH, targeting pulse oximetry saturations (SpO2) of 85–89% results in reduced pulmonary perfusion and higher pulmonary vascular resistance compared with higher SpO2 targets.8 9 It is unknown whether achieved SpO2 is associated with preterm infants’ risk for PH. While an SpO2 target of 91–95%, as compared with 85–89%, increased the need for supplemental oxygen at 36 weeks’ postmenstrual age (PMA) in a meta-analysis of the previous five large, randomised trials in preterm infants, none of these trials reported the outcome of PH.10–12

As PH may limit gas exchange, previous investigators13 14 have characterised PH severity using the relationship between oxygen administration and oxygen exchange using the oxygen saturation index (OSI) defined using the mean airway pressure (MAP), fraction of inspired oxygen (FiO2) and SpO2 (OSI=MAP×FiO2×100/SpO2).15 We hypothesised that infants with BPD-PH (compared with infants with BPD alone) and non-survivors with BPD-PH (compared with survivors with BPD-PH) would spend a higher proportion of time (on average) with lower SpO2 distributions and a higher ratio of oxygen delivery to diffusion as estimated by the OSI.

Methods

Patient population

This was a single-centre, observational investigation of data prospectively collected between 2018 and 2023 at the University of Alabama at Birmingham (UAB). High-fidelity, cardiorespiratory data were available for a discovery cohort of infants born between 2018 and 2020 as a secondary analysis of infants from UAB enrolled in the Prematurity-Related Ventilatory Control (PreVENT) study.16 From this cohort, we included infants who were born at <29 weeks’ gestation and on any respiratory support at 36 weeks’ PMA with use of the grade BPD definition.17 Infants were excluded if they had major congenital anomalies or genetic syndromes. At UAB, infants meeting these inclusion criteria are systematically screened for PH by echocardiography starting at a month of age, and then once a month until discharge. Infants were characterised as having BPD-PH if echocardiographic evidence of PH was demonstrated on any echocardiogram performed during hospitalisation using previously reported criteria.18

Comparison groups for analyses were (1) infants with echocardiographic BPD-PH and infants with BPD alone and (2) within infants with BPD-PH by survival status up until discharge irrespective of postnatal age. Infants were matched by gestational age at birth (±1 week). The cardiorespiratory parameter analyses were constrained to the postnatal week preceding the first instance of echocardiographic evidence in infants with BPD-PH and, in infants with BPD alone, the week preceding a postnatal age-matched echocardiogram demonstrating absence of PH.

The predictive utility of the OSI was evaluated in this discovery cohort, which was further evaluated in a validation cohort born between 2021 and April 2023. As OSI values are reliant on MAP values for calculations, these analyses were limited to infants on continuous positive airway pressure, non-invasive ventilation or invasive ventilation.

Parameter comparisons from discovery cohort

In infants with any grade of BPD,17 distributions of SpO2 were recorded using Philips IntelliVue MP70 or MP50 monitors with Nellcor pulse oximetry sensors (Nellcor, Mansfield, Massachusetts) with an averaging time of 8 s. Data were collected and stored using the BedMaster system (Excel Medical Electronics, Jupiter, Florida, USA) which collected data as .STP files, which were converted to HDF5 files and later analysed using MATLAB (MathWorks, Natick, Massachusetts, USA) in-house code.

In infants with grade 2 BPD receiving positive airway pressure and grade 3 BPD,17 predictive parameters considered included (1) FiO2, (2) the ratio of SpO2 to FiO2 and (3) OSI over the course of the postnatal week preceding echocardiographic PH. While this constrained the population analysed, infants with severe forms of BPD are at highest risk for long-term neurodevelopmental impairment.17 Median data, derived from hourly data during this week, were collected from the electronic medical record with concurrent abstraction of SpO2, MAP and FiO2 for OSI calculations. In infants on continuous positive airway pressure, the positive end expiratory pressure value was used as a surrogate for MAP. SpO2 targets were also compared between groups to ensure differences in SpO2 distributions were not impacted by targeting.

Parameter comparisons in the validation cohort

The ability of predictive parameters to accurately classify infants with BPD-PH was appraised using a receiver operating characteristic curve-informed cut-off value. These values were used to evaluate the sensitivity, specificity, negative predictive values and positive predictive values in the validation cohort.

Second, predictors with the greatest predictive utility, based on the predictor with the highest area under the curve (AUC), were further evaluated over the entirety of the hospitalisation, in conjunction with data from the discovery cohort, to determine whether predictive parameters could be functionally used for risk stratification within a specific postnatal time period either by (a) postnatal week or (b) PMA.

Power analysis

To our knowledge, SpO2 distributions have not previously been compared between infants with BPD alone and those with BPD-PH. We have previously reported the duration of severe hypoxaemia between infants with BPD alone and those with BPD-PH in which a similar duration of time was spent <80% and <70% in both groups.19 However, the distribution of SpO2 has not been reported. Regarding the OSI, previous studies in infants with congenital diaphragmatic hernia (CDH) have reported an AUC of 0.71 for the detection of PH.20 Using this AUC, an overall prevalence in our dataset of 50% (based on aforementioned matching approach), level of shrinkage of 0.90 and an expected Cox-Snell R-squared of 0.12 (based on the c-statistic and prevalence), we anticipated the need for 70 infants for model derivation and an additional 70 infants for model validation.

Statistical analysis

Distributions of achieved SpO2 were compared using an Anderson-Darling test. Receiver operating characteristics (ROC) analysis identified optimal thresholds for BPD-PH accuracy. A Fisher’s exact test was used for analyses of categorical variables between comparison groups. Following tests of normality, the appropriate parametric or non-parametric test was used for continuous variables. Two-way analysis of variance was used for longitudinal OSI comparisons. Analyses were performed with SPSS V.28 and GraphPad Prism V.8.2.1 with a p<0.05 considered significant.

Results

80 infants (40 with BPD-PH gestational age matched to 40 infants with BPD alone) were available from the UAB PreVENT cohort for parameter comparisons for the discovery cohort. The median gestational age and birth weight were 24.4 (IQR 23–26) vs 24.6 (IQR 23–26) weeks’ gestation (p=0.65) and 589 (IQR 529–660) vs 648 (IQR 581–851) g (p=0.004) in infants with BPD alone and BPD-PH, respectively. The respective proportion of infants with grade 1, 2 and 3 BPD among infants with BPD-PH was 40%, 33% and 23% (two infants died prior to 36 weeks’ PMA), respectively, as compared with 55%, 30% and 15% among infants with BPD alone (p=0.46). We previously reported demographic and clinical characteristics in this population,17 with additional characteristics detailed in online supplemental table 1.

BPD-PH versus BPD alone

In the week preceding echocardiographic diagnosis, infants who developed BPD-PH achieved a lower SpO2 compared with infants with BPD alone (figure 1; p=0.0005) despite the similar SpO2 alarm limits (online supplemental table 2). Representative tracings from infants with BPD alone and BPD-PH are available in online supplemental figure 1. Compared with infants with BPD alone, infants with BPD-PH were exposed to a higher FiO2 (0.50 vs 0.34; p=0.02), had a higher OSI (4.26 vs 2.64; p=0.03) and had a similar SpO2:FiO2 ratio (2.2 vs 2.9; p=0.09) in the week prior to echocardiographic identification of BPD-PH (figure 2A–C).

Figure 1

Characterisation of achieved oxygen saturations (SpO2) in the week preceding initial echocardiographic diagnosis of bronchopulmonary dysplasia-associated pulmonary hypertension (BPD-PH) as compared with saturations in infants with BPD without PH (BPD alone). The distribution of SpO2 differed between infants with and without BPD-PH (p<0.001).

Figure 2

Differences in respiratory parameters between infants with bronchopulmonary dysplasia (BPD) alone and BPD-associated pulmonary hypertension (BPD-PH) (A–C) and among infants with BPD-PH by survival status (D–F). Parameters of comparison include fraction of inspired oxygen (FiO2), oxygen saturation index (OSI) and ratio of oxygen saturation (SpO2) and FiO2. *P<0.05.

In comparing these parameters, OSI had the highest AUC of the ROC curve (0.70), followed by FiO2 (0.51) and SpO2:FiO2 ratio (0.48) (figure 3). The optimal cut-off value for OSI was determined to be 2.79, which was used to classify infants in the validation cohort. The validation cohort consisted of 87 infants (44 with BPD-PH and 45 with BPD alone). The median gestational age in infants with BPD-PH was 24.4 weeks (IQR 24–25), and 24.0 weeks (IQR 23–25) in infants with BPD alone. From this validation cohort, there were 26 infants with BPD-PH and 13 infants with BPD alone exposed to positive airway pressure for OSI performance appraisal. Test sensitivity was 92% (95% CI 75% to 99%) and specificity was 46% (95% CI 19% to 75%), with a positive predictive value of 77% and a negative predictive value of 75%.

Figure 3

Receiver operating characteristic curves for oxygen saturation index (OSI), fraction of inspired oxygen (FiO2) and oxygen saturation (SpO2):FiO2 ratio. AUC, area under the curve; BPD, bronchopulmonary dysplasia; BPD-PH, BPD-associated pulmonary hypertension.

In longitudinal OSI comparisons, there was increasing discriminatory utility and separation in OSI values between infants with BPD alone and BPD-PH beyond 16 postnatal weeks and 36 weeks’ PMA (figure 4A,B).

Figure 4

Oxygen saturation index (OSI) over hospitalisation by postnatal week (A) and postmenstrual age (B). Data include infants from both the discovery and validation cohorts. Tables beneath panels indicate the number of unique patients per time period available for comparisons. *P<0.05. BPD, bronchopulmonary dysplasia; BPD-PH, BPD-associated pulmonary hypertension.

BPD-PH survivors versus BPD-PH non-survivors

When achieved SpO2 was compared within infants with BPD-PH by survival status, infants with BPD-PH who died achieved lower SpO2 compared with survivors (figure 5; p=0.0005). Within infants with BPD-PH, non-survivors were exposed to a higher FiO2 (p=0.049) and had a similar OSI and SpO2:FiO2 ratio compared with survivors (figure 2D–F).

Figure 5

Characterisation of achieved oxygen saturations (SpO2) in the week preceding initial echocardiographic diagnosis of bronchopulmonary dysplasia-associated pulmonary hypertension (BPD-PH) by survival status. The distribution of SpO2 differed between non-survivors and survivors with BPD-PH (p<0.001).

Discussion

In this prospective observational study, infants with BPD-PH achieved lower SpO2 distributions compared with infants with BPD alone. In evaluating related predictive parameters to differentiate infants with BPD-PH from infants with BPD alone, infants with BPD-PH had a higher OSI in the week preceding echocardiographic diagnosis with progressive separation in OSI values after 36 weeks’ PMA. Whereas non-survivors with BPD-PH achieved a lower SpO2 distribution compared with survivors, OSI was not predictive of survival.

Given that infants that developed BPD-PH achieved lower SpO2 compared with infants with BPD alone, it is plausible that higher SpO2 targets may reduce risk for BPD-PH development. The American Thoracic Society and the American Heart Association recommend targeting an SpO2 of 92–95% in infants with established BPD-PH21; however, there is currently limited evidence for what SpO2 targeting strategy reduces risk for BPD-PH. In both the Surfactant, Positive Pressure, and Oxygenation Randomized Trial10 and the Neonatal Oxygenation Prospective Meta-analysis,22 progressive divergence in survival between randomisation groups occurred with increasing postnatal age starting about a month after randomisation, indicating that infants in the low SpO2 group are at increased risk for late death. While PH was not routinely assessed in these trials, the time course of the progressive divergence is consistent with the natural history of PH in extremely preterm infants.2 Evidence from the present study also identified differences in achieved SpO2 distributions in infants with BPD-PH by survival status, which may further support SpO2 targets higher than currently recommended.21 Collectively, these data support the need for additional investigations evaluating the impact of SpO2 targeting on both the development of BPD-PH and survival in infants with established BPD-PH.

Data from animal studies provide mechanistic associations between SpO2 targets and pulmonary vascular resistance. In a model of term gestation lambs, pulmonary blood flow was assessed following asphyxia and randomisation to three SpO2 targets: 85–89%, 90–94% and 95–99%. Lambs in the 95–99% SpO2 target group had the highest pulmonary blood flow and lowest pulmonary vascular resistance.9 However, as higher SpO2 targets may increase retinopathy of prematurity (ROP) risk,22 the benefits of higher SpO2 targets must be considered in context to infants’ competing risk for ROP.

In addition to different SpO2 distributions between infants with and without PH, infants with BPD-PH also had a higher OSI compared with infants with BPD. The OSI has been used in other clinical contexts to continuously measure the presence and severity of PH. In a cohort of infants (n=42) with CDH, an OSI ≥12.5 was associated with a high risk for PH, whereas an OSI ≥22 was associated both with extracorporeal membrane oxygenation and mortality.23 Animal models have also substantiated the relationship between OSI and increasing pulmonary vascular resistance.24 Receiver operating curve analysis in the present study identified an optimal OSI cut-off value of ~2.8 in discriminating between infants with BPD-PH from infants with BPD alone. Compared with infants with CDH, the predictive utility of this metric may be more limited given that infants at risk for BPD-PH may not be on a mode of respiratory support from which an OSI can be calculated. In our study, 44% of the discovery cohort and 55% of the validation cohort were on a mode of respiratory support without an estimated MAP precluding an OSI calculation. However, given that a higher proportion of infants with more severe endotypes of BPD develop BPD-PH,2 the OSI may still have clinical value in infants at highest risk for BPD-PH. While FiO2 differed between infants with BPD-PH and BPD alone (consistent with previous literature18 25) and has the potential for more generalised applicability, the predictive utility, as estimated by the AUC, was only 0.51 compared with 0.70 using OSI. The magnitude of an individual patient’s FiO2 may therefore be non-specific to the BPD-PH and may also correspond to other BPD phenotypes.26

In longitudinal analyses not accounting for the timing of echocardiographic BPD-PH identification, the timing of OSI separation between infants with BPD alone and BPD-PH occurred at around 36 weeks’ PMA. This postnatal age of separation is consistent with the natural history of BPD-PH development reported in observational studies.2 In considering plausible explanations for this timing of separation, we speculate that this may be driven by progressive intracardiac or extracardiac shunting in infants with BPD-PH, which, by potentially increasing the FiO2 to SpO2 ratio over time, would mathematically increase the OSI. Other changes in clinical management, such as subsequent increases in SpO2 targets, would likely have a limited effect given the anticipated increase in both FiO2 and SpO2. Use of the OSI during hospitalisation in infants with grade 2 and 3 BPD may therefore be used as a continuous biomarker to identify the development of BPD-PH. Moreover, the predictive utility of OSI, as assessed in the validation cohort, conferred a sensitivity of 92%, though the specificity of OSI was only 46% for which echocardiographic confirmation is warranted to further establish the presence or absence of BPD-PH.

There are many strengths with this analysis of SpO2 achievement and related predictive parameters in infants with and without BPD-PH. The high-fidelity, prospective collection of cardiorespiratory parameters in a large cohort of infants with BPD-PH enabled a comparison of both SpO2 achievement both between infants with and without BPD-PH and within infants with BPD-PH by survival status. Validation of predictive parameters identified within the discovery cohort and identification of the postnatal age of OSI separation further supports the application of OSI measurements in clinical practice. This observational study was within a single centre for which further evaluation is warranted. Observations within the discovery cohort related to achieved SpO2, FiO2 and OSI were also limited to the week preceding echocardiographic diagnosis in infants with BPD-PH. Observational differences between infants with BPD alone and BPD-PH are limited to associations. Whether SpO2 targeting has preventative and therapeutic utility for infants with BPD-PH needs further evaluation in randomised studies of differential SpO2 targets.

Conclusion

In this observational cohort study comparing SpO2 achievement between infants with BPD-PH and BPD alone, SpO2 achievement differed between comparison groups and within infants with BPD-PH by survival status. The OSI, which relates oxygen administration to oxygen exchange, differentiated infants with BPD-PH from those with BPD alone with high sensitivity but low specificity with progressive discrimination following 36 weeks’ PMA. Further evidence to support the influence of SpO2 targeting and the predictive utility of OSI in infants either at risk for or with established BPD-PH are warranted.

Data availability statement

Data are available upon reasonable request. The data that support the findings of this study are available upon reasonable request to the corresponding author (SG).

Ethics statements

Patient consent for publication

Ethics approval

The institutional review board provided approval prior to all analyses.

References

Supplementary materials

Footnotes

  • Contributors SG is guarantor and helped with manuscript conceptualisation, data curation, formal analysis, original draft preparation, review and editing, and provided final approval of the version to be published. AS and AN contributed to interpretation of data for the work, statistical analysis, manuscript review and editing, and provided final approval of the version to be published. CPT contributed to original draft preparation, review and editing, and provided final approval of the version to be published. WAC and NA contributed to conceptualisation, interpretation of data for the work, original draft preparation, review and editing, and provided final approval of the version to be published. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

  • Funding NIH/NHLBI (U01 HL133536, U01 HL133536-05S1), National Heart, Lung, and Blood Institute (K23HL157618), American Heart Association (24CDA1275188), LDCC (U01 HL133708).

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.