Plasma proteomic biomarkers of respiratory failure severity and mechanical ventilation duration among critically ill children

Background: Pediatric acute respiratory failure is associated with elevated mortality and morbidity. Prognostic indicators of poor clinical trajectory are limited to invasive mechanical ventilation (IMV) parameters such as oxygenation index (OI).

Aims: Characterize the plasma proteome of children undergoing IMV via discovery proteomics and assess candidate markers for extended IMV duration and elevated OI.

Methods: We performed a single-center, observational cohort study of critically ill children <18 years of age undergoing IMV for acute respiratory failure from 2018 to 2020. Blood was collected within 24- and 48-72 hours following endotracheal intubation in sodium citrate tubes and processed to platelet-poor plasma. Tandem Mass Tag (TMT)-labeled liquid chromatography mass spectrometry (LC-MS) was conducted. Protein abundance data analyses included normalization, log2-transformation, removal of proteins with >30% missing values, z-scoring, imputation of missing values from normal distribution, and normalization to remove batch effects. The Mass Dynamics platform was utilized for proteomic analyses. Cohorts were defined and compared by IMV duration (long duration [upper quartile, >16 days] versus short duration [lower quartile, <4 days]) and ARDS severity (high [OI>16] versus mild [OI<8]) to identify proteins of interest (POIs).

Results: Eighty-one samples from 48 subjects were analyzed, with 759 total proteins identified. Compared to those with a short IMV duration, those with a long IMV duration had the following POIs noted <24 hours of IMV: YWHAZ, TKT, ALDOA, SERPINB1, TALDO1, and RNPEP with higher abundance and CDH13, ACAN, MMRN2, ICAM2, CTSA, IGLV1-47, and IGHV3-43 with lower abundance (Fig. 1A,B). Of those, SERPINB1, TALDO1, and ICAM2 were differentially abundant at both <24- and 48-72 hours of IMV. Compared to those with mild ARDS, those with severe ARDS had the following POIs noted <24 hours of IMV: TKT, LTA4H, ICAM1, PIGR, RNH1, INHBE, HSPA5, CD163 with higher abundance and KIT, TFRC, KRT77, SELENOP, APOM, and TTR with lower abundance (Fig. 1D,E). Of those, LTA4H, ICAM1, PIGR, and KIT were differentially abundant at both <24- and 48-72 hours of IMV. Univariate logistic regression analysis (yielding odds ratios with 95% confidence intervals and p-values for severe ARDS (OI>16) and long duration of IMV (>16 days), respectively) identified SERPINB1 (5.61 (1.05-29.85), p = 0.043; 7.0 (1.28-38.28), p = 0.025), TALDO1 (2.73 (1.04-7.16), p = 0.04; 3.48 (1.18-10.24), p = 0.023), TKT (3.61 (1.37-9.49), p = 0.009; 2.97 (1.19-7.37), p = 0.019), RNH1 (6.95 (1.5-32.11), p = 0.013; 3.96 (1.08-14.5), p = 0.038), and HSPA5 (9.48 (1.43-62.79), p = 0.02; 11.42 (1.59-81.72), p = 0.015) as proteins of interest with differential abundance at 24 hours (Fig. 1C,F).

Conclusions: Using a TMT-LC-MS approach, we identified 5 POIs that after further validation may represent putative prognostic indicators of ARDS severity and poor clinical trajectory.