Effect of low-to-moderate hyperoxia on lung injury in preclinical animal models: a systematic review and meta-analysis

Background Extensive animal investigation informed clinical practice regarding the harmful effects of high fractional inspired oxygen concentrations (FiO 2 s > 0.60). Since questions persist whether lower but still supraphysiologic FiO 2 ≤ 0.60 and > 0.21 (FiO 2 ≤ 0.60/ > 0.21) are also harmful with inflammatory lung injury in patients, we performed a systematic review examining this question in animal models. Methods Studies retrieved from systematic literature searches of three databases, that compared the effects of exposure to FiO 2 ≤ 0.60/ > 0.21 vs. FiO 2 = 0.21 for ≥ 24 h in adult in vivo animal models including an inflammatory challenge or not were analyzed. Survival, body weight and/or lung injury measures were included in meta-analysis if reported in ≥ 3 studies. Results More than 600 retrieved reports investigated only FiO 2 s > 0.60 and were not analyzed. Ten studies with an inflammatory challenge (6 infectious and 4 noninfectious) and 14 studies without, investigated FiO 2 s ≤ 0.60/ > 0.21 and were analyzed separately. In seven studies with an inflammatory challenge, compared to FiO 2 = 0.21, FiO 2 ≤ 0.60/ > 0.21 had consistent effects across animal types on the overall odds ratio of survival (95%CI) that was on the side of harm but not significant [0.68 (0.38,1.23), p = 0.21; I 2 = 0%, p = 0.57]. However, oxygen exposure times were only 1d in 4 studies and 2–4d in another. In a trend approaching significance, FiO 2 ≤ 0.60/ > 0.21 with an inflammatory challenge consistently increased the standardized mean difference (95%CI) (SMD) in lung weights [0.47 (− 0.07,1.00), p = 0.09; I 2 = 0%, p = 0.50; n = 4 studies] but had inconsistent effects on lung lavage protein concentrations ( n = 3), lung pathology scores ( n = 4) and/or arterial oxygenation ( n = 4) ( I 2 ≥ 43%, p ≤ 0.17). Studies without an inflammatory challenge had consistent effects on lung lavage protein concentration ( n = 3) SMDs on the side of being increased that was not significant [0.43 (− 0.23,1.09), p = 0.20; I 2 = 0%, p = 0.40] but had inconsistent effects on body and lung weights ( n = 6 and 8 studies, respectively) ( I 2 ≥ 71%, p < 0.01). Quality of evidence for studies was weak. Interpretation Limited animal studies have investigated FiO 2 ≤ 0.60/ > 0.21 with clinically relevant models and endpoints but suggest even these lower FiO 2 s may be injurious. Given the influence animal studies examining FiO 2 > 0.60 have had on clinical practice, additional ones investigating FiO 2 ≤ 0.60/ > 0.21 appear warranted, particularly in pneumonia models.