Base excess serves as a mediator in the hemoglobin-intensive care unit stay length relationship in neonatal respiratory distress syndrome

Introduction

Neonatal respiratory distress syndrome (NRDS) is a common and critical pulmonary disease in newborn infants (1). It is primarily caused by a type II alveolar surfactant deficiency, leading to progressive alveolar collapse (2). NRDS worsens inspiratory difficulty, with patients often exhibiting symptoms such as tachypnea, grunting during respiration, and nasal flaring (3). These signs appear at birth or shortly thereafter and worsen within 48–72 hours (4). NRDS is usually accompanied by diffuse alveolar inflammation or interstitial pulmonary edema syndrome, which can lead to chronic lung disease, respiratory failure, and in severe cases, shock or death (5,6). This makes NRDS a common reason for admission to the neonatal intensive care unit (ICU), accounting for 30–40% of neonatal hospitalizations (7,8).

Certain factors increase the risk of NRDS in newborn infants. Except for low birth weight, asphyxia, or cesarean section, prematurity is a significant risk factor for NRDS (9). It is reported that only 7% of near-term infants develop NRDS, whereas the incidence rises to 30% in preterm infants with a gestational age (GA) of less than 30 weeks and 60% for those under 28 weeks (10,11). This may be due to the immature lung development in preterm infants, making them more susceptible to lung injury (12). To reduce the incidence and mortality of NRDS, the World Health Organization recommends the use of prenatal corticosteroids and tocolytics for women at risk of preterm labor; and continuous positive airway pressure and surfactant administration for newborn infants (13). Despite these efforts, NRDS remains a significant public health issue. Current treatments for NRDS are limited to supportive measures, such as respiratory support, caffeine therapy, and electrolyte management (4). There is a lack of understanding of the etiology and pathogenesis of NRDS, along with an absence of effective identification tools (4,14). Therefore, there is an urgent need to identify modifiable factors, study the pathogenesis of NRDS, and develop new therapeutic targets.

This was a comprehensive study using the MIMIC-III database to identify and analyze key clinical factors related to the length of ICU stay in newborn infants with NRDS. This study aimed to develop more effective management protocols and interventions by elucidating these factors, ultimately improving the prognosis and care of newborn infants suffering from NRDS. We present this article in accordance with the TRIPOD reporting checklist (available at https://tp.amegroups.com/article/view/10.21037/tp-2025-559/rc).

Methods

Data source and study population

This retrospective observational cohort collected data from the MIMIC-III database (https://physionet.org/content/mimiciii/1.4/), which is large and freely accessible, with 7,870 neonates. Since neonates were not directly involved, this study did not receive approval from our hospital. The database included de-identified medical records for patients admitted to the ICU at the Beth Israel Deaconess Medical Center (Boston, Massachusetts, USA). Therefore, the individual consent was also waived. The authors have completed the Collaborative Institutional Training Initiative (CITI) course and passed both the “Conflicts of Interest” and “Data or Specimens Only Research” exams to access the database. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Inclusion criteria: (I) NRDS was defined according to the ninth edition of the International Classification of Disease (ICD-9) codes (15), yielding 1,315 neonates with NRDS; (II) length of ICU stay ≥1 day (n=1,241); (III) missing values were less than 5% for all variables. The missing values were imputed using the random forest imputation method. Only the first admission into the ICU was included if the patient was admitted to the ICU multiple times. There were 1,241 samples included in the final study, with 518 females and 723 males. The flowchart of patient selection is shown in Figure S1.

Data extraction

By referencing the previous research on NRDS, we extracted clinical data, including demographics, vital signs, and laboratory tests, using structured query language from the MIMIC-III database. Demographics: age, height, birth weight, and GA; vital signs: partial pressure of arterial oxygen/fraction of inspired oxygen (pO₂/FiO₂), and heart rate; laboratory results: hematocrit, lymphocyte, neutrophil, red blood cell (RBC), red cell distribution width (RDW), potential of hydrogen (pH), partial pressure of carbon dioxide (pCO₂), urine output, platelet, white blood cell (WBC), anion gap (AG), hemoglobin, glucose, base excess (BE), and bilirubin; comorbidities were identified based on the ICD-9 codes: acute kidney injury (AKI), jaundice, acidosis, anemia, pneumonia, apnea, sepsis, intraventricular hemorrhage (IVH), and congenital malformation. Treatment: dopamine use, mechanical ventilation (MV) use, surfactant administration, and methasone use. The first monitoring record for the variables within the first 24 hours after ICU admission was collected. Only the first admission into the ICU was included if the patient was admitted to the ICU multiple times. The in-time to the ICU and out-time were acquired. The length of ICU stay was calculated and served as the study outcome.

Statistical analysis

All statistical analyses were performed using SPSS software (version 23.0) and R software (version 4.4.1). Baseline characteristics of the females and males among NRDS patients were exhibited. Continuous variables with skewed distribution were expressed as median [interquartile range] and the comparisons between the two groups were conducted using the Mann-Whitney U test. Categorical characteristics were represented by count (percent) and the Chi-squared test was adopted to assess the differences between the groups. Then, the univariable generalized linear model (GLM) was performed to select the parameters related to the length of ICU stay. After lasso regression to reduce the overfitting of variables, feature importance analysis was performed using three machine-learning algorithms, including XG Boost, ridge regression, and random forest, to determine the factors closely associated with the length of ICU stay.

A generalized additive model (GAM) was adopted to evaluate the nonlinear relationship between independent and dependent variables (both were continuous variables). Besides, threshold effect analysis was conducted to identify the infection point of the relationship between independent variables and the length of ICU stay. Subsequently, a multivariable GLM model was applied to identify independent predictors for the length of ICU stay. Age, height, birth weight, heart rate, dopamine, jaundice, acidosis, anemia, pneumonia, apnea, and AKI were adjusted. Sensitivity analyses were conducted to evaluate the robustness of the association of hemoglobin with length of ICU stay in the following subgroups: excluding sepsis/IVH/congenital malformation (n=960); with MV treatment (n=1,171); with surfactant administration (n=927); without methasone use (n=1,183); with complete GA data (n=373). The Pearson correlation test was used for correlation analysis. The mediation analysis evaluated the mediating effect and potential regulatory pathway. P<0.05 was considered statistically significant.

Results

Baseline characteristics of the participants

A total of 1,241 individuals (518 females) diagnosed with NRDS were enrolled. The study population (median age, 13 hours) had an average height of 46.00 cm with an average weight of 1.545 kg. Among them, 121 samples had AKI (9.750%), 1,117 had jaundice (90.008%), 98 had acidosis (7.897%), 302 had anemia (24.335%), 64 had pneumonia (5.157%), and 650 had apnea (52.377%). A total of 166 (13.376%) and 1,171 (97.910%) individuals received dopamine and MV treatment, respectively. The distribution of AKI, anemia, and apnea in the females was significantly different from the males (P<0.05). Males were prone to be taller and heavier than females (P<0.05). There were statistical differences in the levels of RBC, platelet, and hemoglobin between the two groups (P<0.05); however, no remarkable differences were found in the hematocrit, lymphocytes, neutrophils, RDW, pH, pCO_2, urine output, WBC, AG, glucose, BE, and bilirubin (all P>0.05). More detailed information on the baseline characteristics is presented in Table 1.

Identification of key variables related to length of ICU stay

We first conducted an univariable analysis to obtain factors associated with the length of ICU stay. As shown in Table 2, except for gender, pO₂/FiO₂, pH, pCO₂, and MV use, the remaining variables were all statistically significant. To further reduce the feature numbers and avoid overfitting, we performed a lasso regression analysis that selected 18 features, including AKI, anemia, weight, height, bilirubin, BE, glucose, hemoglobin, AG, WBC, platelet, urine output, RDW, neutrophils, lymphocytes, hematocrit, heart rate, and age (Figure 1A). Three machine-learning algorithms were employed to rank the importance of the 18 features and select the top 10 features for the following intersection analysis (Figure 1B). The Venn plot showed 4 common features among the XG Boost, ridge regression, and random forest methods: height, weight, hemoglobin, and BE (Figure 1C).

Figure 1 Selection of the key variables related to length of intensive care unit stay. (A) Lasso regression obtained 18 variables. (B) Three machine learning methods to select 10 of 18 features according to their importance. (C) Four common features of the three methods revealed by the Venn diagram. AKI, acute kidney injury; WBC, white blood cell; RDW, red cell distribution width.

Nonlinear association of hemoglobin and BE with length of ICU stay

Since height and weight were uncontrollable factors, we next focused on exploring the value of hemoglobin and BE in NRDS. To analyze the dose-response relationship between hemoglobin, BE, and length of the ICU stay, GAM analysis was initially carried out, and exhibited that hemoglobin was nonlinearly linked to the length of ICU stay (P for nonlinear <0.001) (Figure 2A). A similar nonlinear relationship was observed between BE and the study outcome (P for nonlinear <0.001) (Figure 2B).

Figure 2 Generalized additive model analysis. The nonlinear association of (A) hemoglobin and (B) base excess with the length of intensive care unit stay. ICU, intensive care unit.

Due to the nonlinear association, we employed threshold effect analysis to obtain the vital turning point. As exhibited in Table 3, a significant negative relationship existed for hemoglobin level ≥12.1 g/dL and the length of ICU stay (β=−3.7, 95% CI: −4.6 to −2.7); however, this relationship became insignificant when hemoglobin level was <12.1 g/dL (β=0.1, 95% CI: −3.2 to 3.3). In addition, there was no significant relationship between the BE level <−8.3 mmol/L and the length of ICU stay (β=0.5, 95% CI: −0.3 to 1.4), but a negative relationship existed when the BE level was ≥−8.3 mmol/L (β=−2.9, 95% CI: −3.6 to −2.1) (Table 4).

Hemoglobin independently predicted shorter length of ICU stay

After demonstrating the association of hemoglobin and BE with the length of ICU stay, we then investigated their independent roles using multivariable GLM analysis (Table 5). Hemoglobin level as a continuous variable was an independent predictor for shorter length of ICU stay (β=−0.503, 95% CI: −0.958 to −0.049) (P<0.05). When hemoglobin level was divided into <12.1 g/dL group and ≥12.1 g/dL group, the patients in the ≥12.1 g/dL group had a favorable prognosis (β=−7.242, 95% CI: −11.539 to −2.944) (P <0.05). In contrast, BE as a continuous and categorical variable (grouped by the inflection point) could not independently predict the length of ICU stay (β=−0.133, 95% CI: −0.389 to 0.124) (β=−2.008, 95% CI: −5.832 to 1.815). These results highlighted the central role of hemoglobin in NRDS. BE was significantly connected with the length of ICU stay but did not serve as an independent factor.

Sensitivity analysis

We performed GLM analysis in different subgroups to examine the robustness of the association between hemoglobin levels and ICU stay. These analyses consistently revealed a significant inverse relationship between hemoglobin and ICU stay across all subgroups: excluding sepsis/IVH/congenital malformation (n=960, β=−2.344, 95% CI: −3.174 to −1.513, P<0.001); with MV treatment (n=1,171, β=−3.060, 95% CI: −3.929 to −2.191, P<0.001); with surfactant administration (n=927, β=−3.203, 95% CI: −4.222 to −2.185, P<0.001); and without methasone use (n=1,183, β=−2.768, 95% CI: −4.124 to −1.058, P<0.001) (Table S1).

In addition, we analyzed the association of hemoglobin levels with length of ICU stay by incorporating GA as the covariate in a subset of the cohort (n=373). As shown in Table S2, hemoglobin still had an inverse relationship with the ICU stay (P<0.05). The persistent significance and direction of the estimates confirmed the central role of hemoglobin levels in the length of ICU stay among neonates with NRDS.

Association of hemoglobin with length of ICU stay mediated by BE

Hemoglobin and BE were identified as core controllable variables in NRDS prognosis, but only hemoglobin had a remarkable relationship with the length of ICU stay independent of other covariates. These triggered us to explore the potential involvement of BE in the association of hemoglobin with the length of ICU stay. Through the correlation test, we found a positive connection between hemoglobin and BE with cor. =0.24 (P<0.05) (Figure 3A). In the mediation model, hemoglobin (X) served as a predictor for the length of ICU stay (Y), and BE (M) was a mediator (Figure 3B). The direct association of X with Y was significant (P<0.05). The indirect relationship between X and Y through M was also significant, revealing a significant mediation effect (P<0.05). Of note, this was a partial mediation as the variance of the direct path (X to Y) did not reduce to zero after the mediator was added. Altogether, hemoglobin elevation was an independent indicator for predicting the length of ICU stay. Moreover, their intimate relationship was partially mediated by BE.

Figure 3 Association of hemoglobin with the length of ICU stay mediated by base excess. (A) Positive correlation between hemoglobin and base excess. (B) The relationship between hemoglobin and the length of ICU stay was partially mediated by base excess. ICU, intensive care unit.

Discussion

NRDS is one of the most common causes of respiratory failure and death in newborns (5). This study investigated the risk factors for the length of ICU stay for newborns diagnosed with NRDS. Using machine learning algorithms, we identified that height, weight, hemoglobin, and BE were significantly associated with the length of ICU stay for NRDS patients. These factors are crucial in predicting NRDS, consistent with previous studies that have found a correlation between height, weight, and NRDS (16,17). Since height and weight are uncontrollable factors, our research focused on the specific association of hemoglobin and BE with NRDS. The results indicated that hemoglobin levels (≥12.1 g/dL) and BE (≥−8.3 mmol/L) within a certain threshold were negatively correlated with the length of ICU stay.

To ensure blood flow to critical organs such as the heart and brain, the oxygen supply to the lung, intestine, and other issues is reduced; as one of the most fragile organs, the lung is easily damaged (18). Consistent with our findings, previous studies have emphasized a significant association between lower hemoglobin levels and the development of NRDS (11). For instance, one study found that 19.9% of infants clinically diagnosed with NRDS had hemoglobin levels below 15 mg/dL (19). Moreover, in cases of acute respiratory distress syndrome, each 1 g/dL increase in hemoglobin was associated with a 0.066-fold reduction in mortality risk (20). We propose that the central role of hemoglobin lies in its fundamental capacity for oxygen transport (21). In NRDS, surfactant deficiency leads to generalized atelectasis, ventilation-perfusion mismatch, hypoxemia, and respiratory acidosis (22). Against this background of impaired pulmonary oxygenation, reduced hemoglobin levels further compromise systemic oxygen delivery. This exacerbates tissue hypoxia, particularly in metabolically active organs, prompting a shift toward anaerobic glycolysis. Consequently, lactate accumulates, driving metabolic acidosis and reducing BE—a phenomenon observed in premature calves with respiratory distress syndrome (23). Blood pH and BE levels are important for detecting respiratory acidosis (24). Our study extends current knowledge by not only revealing that hemoglobin levels and BE are negatively correlated with the ICU stay duration of NRDS patients, but also revealing that hemoglobin levels can independently predict ICU stay duration, whereas BE cannot. Importantly, we identified a positive correlation between hemoglobin and BE, consistent with reports that respiratory dysfunction often disrupts acid-base balance (25).

A key novel finding is that BE partially mediates the relationship between hemoglobin and ICU stay duration. This suggests a plausible pathophysiological pathway: low hemoglobin intensifies hypoxia, which promotes anaerobic metabolism and lactate production, thereby lowering BE (17,26,27). Subsequently, acidotic stress may amplify inflammatory responses, including leukocyte recruitment and elevated IL-6 and TNF-α, ultimately causing tissue injury (Figure 4) (28). Previous studies have demonstrated that increasing hemoglobin-oxygen affinity ameliorated hypoxia tolerance (29-31). Therefore, we speculate that optimizing hemoglobin within an appropriate range may help attenuate acid-base imbalance and potentially improve outcomes in NRDS.

Figure 4 Role of hemoglobin and base excess in neonatal respiratory distress syndrome. Hemoglobin decrease leads to reduced oxygen release to other tissues, causing the body to be hypoxia. This further induces anaerobic metabolism, accumulating lactate and disrupting the body’s acid-base balance. Ultimately, this leads to leukocyte recruitment and inflammation. IL-6, interleukin-6; TNF-α, tumor necrosis factor-alpha.

The results of this study have significant clinical implications. Hemoglobin level emerges as a robust, readily accessible biomarker for risk stratification, with the identified threshold of 12.1 g/dL providing a potential clinical reference for predicting a prolonged ICU course in neonates with NRDS. Monitoring and managing hemoglobin levels in neonates with NRDS might be an effective strategy to improve outcomes. Ensuring adequate hemoglobin levels might reduce ICU stay duration, thus decreasing healthcare costs and improving overall prognosis. Moreover, the partial mediation effect of BE illuminates a plausible physiological pathway linking impaired oxygen transport to metabolic acidosis, thereby underscoring the value of integrated assessment of both hematological and acid-base parameters in clinical monitoring. However, several limitations must be acknowledged. The observational nature of the study precludes the establishment of causality. Future research should focus on prospective studies and randomized controlled trials to validate these findings. Exploring interventions aimed at optimizing hemoglobin levels and understanding the precise mechanisms by which BE mediates the hemoglobin-ICU stay relationship would be beneficial.

Conclusions

In conclusion, our study identifies hemoglobin level as a robust and independent biomarker for the length of ICU stay in neonates with NRDS. The relationship between hemoglobin and ICU stay is partially mediated by BE, highlighting the interconnected roles of oxygen transport and acid-base balance in neonatal outcomes. These findings suggest that targeting hemoglobin levels might be valuable in managing NRDS, ultimately improving patient outcomes and reducing ICU stays.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tp.amegroups.com/article/view/10.21037/tp-2025-559/rc

Peer Review File: Available at https://tp.amegroups.com/article/view/10.21037/tp-2025-559/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tp.amegroups.com/article/view/10.21037/tp-2025-559/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Song X. The diagnostic and prognostic value of serum angiopoietin-like 4 level in neonatal respiratory distress syndrome. Ann Clin Biochem 2025;62:293-302. [Crossref] [PubMed]
2. Lin W, Ruan J, Liu Z, et al. Exploring the diagnostic value of ultrasound radiomics for neonatal respiratory distress syndrome. BMC Pediatr 2024;24:215. [Crossref] [PubMed]
3. Duangkum C, Pattamathamakul S, Chaiyarach S, et al. Predictors of delivery at 48 hours or more in pregnant women with preterm prelabor rupture of membrane: A retrospective cohort study. Eur J Obstet Gynecol Reprod Biol X 2025;26:100393. [Crossref] [PubMed]
4. Yadav S, Lee B. Neonatal Respiratory Distress Syndrome. 2025.
5. Bakri AH, Hassan MH, Abd-Elbaseer KA, et al. Association of macrophage migration-inhibitory factor gene and growth differentiation factor 15 gene polymorphisms and their circulating levels with respiratory distress syndrome among preterm neonates. Clin Exp Pediatr 2025;68:680-9. [Crossref] [PubMed]
6. De Luca D. Respiratory distress syndrome in preterm neonates in the era of precision medicine: A modern critical care-based approach. Pediatr Neonatol 2021;62:S3-9. [Crossref] [PubMed]
7. Shi ZN, Zhang X, Du CY, et al. Effects of pulmonary surfactant combined with noninvasive positive pressure ventilation in neonates with respiratory distress syndrome. World J Clin Cases 2024;12:5366-73. [Crossref] [PubMed]
8. Legesse BT, Abera NM, Alemu TG, et al. Incidence and predictors of mortality among neonates with respiratory distress syndrome admitted at West Oromia Referral Hospitals, Ethiopia, 2022. Multi-centred institution based retrospective follow-up study. PLoS One 2023;18:e0289050. [Crossref] [PubMed]
9. Bao LY, Dao XY, Du K. Progress in the Application of Lung Ultrasound for the Evaluation of Neonates with Respiratory Distress Syndrome. J Multidiscip Healthc 2024;17:1-9. [Crossref] [PubMed]
10. Xiong Y, Zang X, Xie T, et al. Additional Adverse Perinatal Outcomes With No Effect on Neonatal Mortality and Birth Defects in Pregnancies Conceived by Assisted Reproductive Technology. Front Pediatr 2022;10:809259. [Crossref] [PubMed]
11. Lei Y, Qiu X, Zhou R. Construction and evaluation of neonatal respiratory failure risk prediction model for neonatal respiratory distress syndrome. BMC Pulm Med 2024;24:8. [Crossref] [PubMed]
12. Shabil M, Gaidhane S, Ballal S, et al. Maternal COVID-19 infection and risk of respiratory distress syndrome among newborns: a systematic review and meta-analysis. BMC Infect Dis 2024;24:1318. [Crossref] [PubMed]
13. WHO Recommendations on Interventions to Improve Preterm Birth Outcomes. WHO Guidelines Approved by the Guidelines Review Committee. Geneva: World Health Organization; 2015.
14. Yang J, He Y, Ai Q, et al. Lung-Gut Microbiota and Tryptophan Metabolites Changes in Neonatal Acute Respiratory Distress Syndrome. J Inflamm Res 2024;17:3013-29. [Crossref] [PubMed]
15. De Coster C, Quan H, Finlayson A, et al. Identifying priorities in methodological research using ICD-9-CM and ICD-10 administrative data: report from an international consortium. BMC Health Serv Res 2006;6:77. [Crossref] [PubMed]
16. Wang Y, Yuan H, Jiang R, et al. Effects of soluble dietary fiber on glycolipid metabolism in gestational diabetes mellitus: study protocol for a randomized controlled clinical trial. Trials 2025;26:349. [Crossref] [PubMed]
17. Cheng K, Zhu H, Zhou Z, et al. Value of brain tissue oxygen saturation in neonatal respiratory distress syndrome: a clinical study. Eur J Transl Myol 2024;34:11863. [Crossref] [PubMed]
18. Di Martino E, Ambikan A, Ramsköld D, et al. Inflammatory, metabolic, and sex-dependent gene-regulatory dynamics of microglia and macrophages in neonatal hippocampus after hypoxia-ischemia. iScience 2024;27:109346. [Crossref] [PubMed]
19. Tigabu Kebede Z, Matebe YH, Demisse AG, et al. Hematologic Profiles of Ethiopian Preterm Infants With Clinical Diagnoses of Early-Onset Sepsis, Perinatal Asphyxia, and Respiratory Distress Syndrome. Glob Pediatr Health 2020;7:2333794X20960264.
20. Wang Z, Xing L, Cui H, et al. A Nomogram for Predicting the Mortality of Patients with Acute Respiratory Distress Syndrome. J Healthc Eng 2022;2022:5940900. [Crossref] [PubMed]
21. Obeagu EI, Igwe MC, Obeagu GU. Oxidative stress's impact on red blood cells: Unveiling implications for health and disease. Medicine (Baltimore) 2024;103:e37360. [Crossref] [PubMed]
22. Peppiatt I, Retrouvey M, Conran RM. Educational Case: Neonatal respiratory distress syndrome. Acad Pathol 2024;11:100115. [Crossref] [PubMed]
23. Ider M, Naseri A, Ok M, et al. Biomarkers in premature calves with and without respiratory distress syndrome. J Vet Intern Med 2021;35:2524-33. [Crossref] [PubMed]
24. Martínez-Barbero G, García-Mesa Y, Cobo R, et al. Acid-Sensing Ion Channels' Immunoreactivity in Nerve Profiles and Glomus Cells of the Human Carotid Body. Int J Mol Sci 2023;24:17161. [Crossref] [PubMed]
25. Castro D, Patil SM, Zubair M, et al. Arterial Blood Gas. In: StatPearls. Treasure Island (FL): StatPearls Publishing; 2024.
26. Liu H, Li J, Guo J, et al. A prediction nomogram for neonatal acute respiratory distress syndrome in late-preterm infants and full-term infants: A retrospective study. EClinicalMedicine 2022;50:101523. [Crossref] [PubMed]
27. Olofsson P. Umbilical cord pH, blood gases, and lactate at birth: normal values, interpretation, and clinical utility. Am J Obstet Gynecol 2023;228:S1222-40. [Crossref] [PubMed]
28. Putz ND, Shaver CM, Dufu K, et al. GBT1118, a compound that increases the oxygen affinity of hemoglobin, improves survival in murine hypoxic acute lung injury. J Appl Physiol (1985) 1985;124:899-905. [Crossref] [PubMed]
29. Dufu K, Yalcin O, Ao-Ieong ESY, et al. GBT1118, a potent allosteric modifier of hemoglobin O(2) affinity, increases tolerance to severe hypoxia in mice. Am J Physiol Heart Circ Physiol 2017;313:H381-91. [Crossref] [PubMed]
30. Geng X, Dufu K, Hutchaleelaha A, et al. Increased hemoglobin-oxygen affinity ameliorates bleomycin-induced hypoxemia and pulmonary fibrosis. Physiol Rep 2016;4:e12965. [Crossref] [PubMed]
31. Dufu K, Williams AT, Muller CR, et al. Increased hemoglobin affinity for oxygen with GBT1118 improves hypoxia tolerance in sickle cell mice. Am J Physiol Heart Circ Physiol 2021;321:H400-11. [Crossref] [PubMed]