Supplemental Digital Content is available in the text.
Keywords: biomarkers, cardiopulmonary bypass, congenital cardiac, lung injury, pediatrics, proteomics
OBJECTIVES:
Cardiopulmonary bypass triggers systemic inflammation, resulting in lung injury, and frequently leads to prolonged mechanical ventilation. Biomarkers of systemic inflammation are required to predict the risk of such complications. We hypothesize that specific serum proteins can be used as biomarkers to predict the severity of lung injury following cardiac surgery.
DESIGN:
Retrospective chart review study.
SETTING:
Clinical variables were collected and used in conjuncture with unbiased proteomic analysis using mass spectrometry that was performed on frozen plasma samples from a study group (patients with mechanical ventilation > 48 hr post surgery) and a control group (patients with mechanical ventilation < 48 hr post surgery).
SUBJECTS:
Subjects included were infants who underwent cardiac surgery with similar complexity (Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery 3 or 4) using cardiopulmonary bypass. Patients in both groups were matched for their weight, age, and duration of cardiopulmonary bypass.
INTERVENTION:
None.
MEASUREMENTS AND MAIN RESULTS:
Four-hundred eighty-three proteins were identified (99% minimum confidence and two peptides minimum, protein false discovery rate 0.1%) on proteomic analysis of four control and four study patients at precardiopulmonary bypass, 0, and 48 hours postcardiopulmonary bypass samples. Thirty-six of 178 proteins were significantly different (≥ 1.5-fold; p < 0.05) at precardiopulmonary bypass (top increased: tenascin; top decreased: tetranectin), 18 of 140 proteins at 0 hour (top increased: hemoglobin beta; top decreased: C8 beta), and 25 of 166 proteins at 48 hours post surgery (top increased: proteoglycan 4; top decreased: galectin-3–binding protein). The top pathway involved cytoskeleton remodeling. Other pathways involved immune response and blood coagulation. Proteoglycan 4 was validated by enzyme-linked immunosorbent assay in a different set of samples (n = 20/group; mean ± sd: 128 ± 67 vs 195 ± 160 ng/mL) (p = 0.037).
CONCLUSIONS:
Multiple proteomic biomarkers were associated with worse respiratory outcomes. Precardiopulmonary bypass biomarkers might indicate risk factors (e.g., abnormalities of coagulation), whereas those identified at 0 hour and post cardiopulmonary bypass may reflect mechanisms of ongoing pathobiology.
Lung injury is an expected complication following cardiac surgery using cardiopulmonary bypass (CPB) (1–3). Extracorporeal circulation during CPB triggers a systemic inflammatory response that results in pulmonary endothelial cell swelling and the infiltration of several proteolytic proteins and enzymes in the alveolar and interstitial tissues (4–7). Post-CPB lung injury manifestations vary in terms of severity and may significantly impact clinical management. Post-CPB lung injury can initially present with hypoxemia and abnormal ventilation and may occasionally lead to acute respiratory distress syndrome (ARDS) and acute respiratory failure, requiring invasive mechanical ventilation (8). Hammermeister et al (9) reported that up to 20% of patients who had cardiac surgery using CPB required mechanical ventilation for longer than 48 hours post surgery. In a recent report using the Pediatric Cardiac Critical Care Consortium data, 25% of pediatric patients with cardiac surgery required mechanical ventilation for longer than 44 hours (10). Furthermore, more prolonged ventilation has been associated with ARDS, hospital-acquired infections, sepsis, and mortality (11).
Identifying specific biological molecules as a diagnostic or prognostic biomarker is critical to earlier detection of diagnosis and prognosis of diseases or adverse effects. Several biomarkers have been explored, including proteins that are variable, diverse, and directly related to cellular pathology (12, 13). Unbiased mass spectrometry (MS) is an evolving technology that enables proteomic analysis at a large scale. All detectable proteins in a sample can be identified and quantified using targeted data analysis (14, 15).
We hypothesized that pediatric patients who underwent CPB surgery might have variable serum levels of specific proteins that correlate with clinical outcomes. These proteins can be used as biomarkers that may have a prognostic value of prolonged mechanical ventilation after cardiac surgery. In our study, we performed unbiased proteomic analysis on a limited number of plasma samples. Proteomic analysis identified several biomarker candidates associated with prolonged mechanical ventilation, and we validated one of the top candidates as a proof-of-concept.
MATERIALS AND METHODS
Patient Selection
At Children’s of Alabama, under separate institutional review board approval, all patients younger than 8 years who are having cardiac surgery are approached for participation in the congenital heart center biorepository. From this repository, subjects were selected. This study was reviewed and approved by the University of Alabama at Birmingham’s Institutional Review Board (IRB) approval number IRB-300003366. Informed consent was waived. Of these subjects, children younger than 1 year old requiring cardiac surgery with Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery (STAT) (16) categorizations 3 and 4 were selected for our study. Patients who required mechanical ventilation before surgery were excluded. We had a total of 458 patients who met our selection criteria over 7-year period (January 2012 to May 2019). The previous literature suggested the length of mechanical ventilation following CPB to be less than 44 hours in 75% of the cases (10). Our blood sampling was done before and at 0 and 48 hours after CPB.
Consequently, we selected the 48 hours cutoff of mechanical ventilation to classify our subjects into two groups. The study group included patients who required invasive mechanical ventilation for longer than 48 hours post surgery. The control group included patients who needed either no mechanical ventilation or mechanical ventilation for less than 48 hours post surgery. We had a pool of 40 patients in each group who were matched based on age (± 3 mo), weight (± 0.5 kg), STAT categories, and CPB time (± 20 min) length.
Plasma Samples
Blood samples before and after CPB were collected by bedside nurses in heparinized tubes. Subsequently, blood plasma was isolated by centrifugation and then stored in the heart center biorepository –80°C freezer. Plasma samples were studied at three time points, pre CPB and 0 and 48 hours post CPB.
The Discovery Phase
We randomly selected four subjects from the control group and four patients from the study group. For each subject, we obtained the pre-CPB and 48 hours post-CPB plasma samples and did the unbiased proteomic analysis. The characteristics of this group of patients are presented in Table 1.
TABLE 1.
Patient Characteristics of the Discovery Phase
| Variable | Control (n = 4) | Study (n = 4) |
|---|---|---|
| Age (d) | 34 ± 48 | 31 ± 53 |
| Weight (kg) | 3.9 ± 0.8 | 3.4 ± 0.9 |
| Gestational age (wk) | 38.5 ± 1.29 | 38 ± 1.4 |
| Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery risk of mortality category | 3.8 ± 0.4 | 3.8 ± 0.4 |
| Cardiopulmonary bypass time (min) | 91 ± 14 | 108 ± 19 |
| Length of mechanical ventilation (hr) | 23.7 ± 17 | 160 ± 141 |
| ICU LOS (d) | 20.75 ± 14 | 28 ± 16 |
| Hospital LOS (d) | 25.5 ± 13 | 30 ± 15 |
LOS = length of stay.
Data presented as mean ± sd.
Proteomics Analysis.
To improve proteomic coverage, each sample was individually depleted of the most abundant proteins using a depletion spin column kit as per the manufacturer’s instructions (Thermo Scientific Pierce Protein Biology, catalog number: A36369, Abundant Protein Depletion Spin Columns; depletes: albumin, immunoglobulin G, a1-acid glycoprotein, a1-antitrypsin, a2-macroglobulin, apolipoprotein A-I, apolipoprotein A-II, fibrinogen, haptoglobin, immunoglobulin A, immunoglobulin M, transferrin). The flow-through fractions obtained after the immunodepletion step were collected, exchanged into 100 mM triethylammonium-bicarbonate, and concentrated to approximately 20–30 μL using molecular weight cutoff centrifugal filter devices (Amicon Ultra—0.5 mL 3k, Millipore, Billerica, MA). The protein fractions were quantified; 40 µg of protein per sample were reduced with dithiothreitol and denatured at 70°C for 35 minutes before loading onto 10% bis-tris protein gels and separated. The gels were stained overnight with colloidal coomassie for visualization purposes, the entire gel lane was cut into 6-molecular weight fractions, and each plug was equilibrated in 100 mM ammonium bicarbonate and digested overnight with Trypsin Gold, MS Grade (Promega, Cat. number V5280, WI) following manufacturer’s instruction. Peptide extracts were reconstituted in 0.1% formic acid/double distilled water at 0.1 µg/µL. MS was carried out, and the data were processed, searched, filtered, grouped, and quantified, as previously reported in detail (17). Following protein identification and relative quantification by normalized spectral counting (NSC), the most statistically significant changed proteins from each pairwise comparison were further analyzed using gene ontology assignments and pathway analysis (carried out in full detail as previously reported in Ludwig et al [17], under section 2.5 LLC-ESI-MS2 and Protein IDs for GeLC).
Statistical and Multivariate Analysis
Nonparametric statistical analysis was performed for the proteomic data generated between each pairwise comparison, t test (single tail, unequal variance, cutoff of p < 0.05). For protein abundance ratios determined with NSCs, we set a 1.5-fold change as the threshold for significance, determined empirically by analyzing the inner-quartile data from the control experiments using ln-ln plots, where the Pearson’s correlation coefficient (R) is 0.98, and greater than 95–99% of the normalized intensities fell between the set fold change. In each case, both tests (t test and fold change) have to pass to be considered significant. All multivariate analysis, including 2D hierarchical clustering analysis HeatMaps, principal component analysis plots, etc., were carried out using Qlucore Omics Explorer (Qlucore, Lund, Sweden).
Systems Analysis.
Gene ontology (GO) assignments and pathway analysis were carried out using MetaCore (GeneGO, St. Joseph, MI). Interactions identified within MetaCore were manually correlated using full-text articles. Detailed algorithms have been described previously (18, 19).
The Validation Phase
Proteoglycan 4 Enzyme-Linked Immunosorbent Assay Analysis.
We selected serum samples from 20 patients in the control group and 20 patients in the study group to be included in our validation phase. Samples were analyzed for proteoglycan 4 concentration by enzyme-linked immunosorbent assay (ELISA) (Cusabio Proteoglycan 4 ELISA kit, Wuhan, P.R. China, Catalog Number. CSB-E14124h). Samples were diluted 10-fold and analyzed following the manufacturer’s protocol.
Statistical Analysis.
Tests were applied at two time points: right before surgery (pre CPB) and 48 hours post surgery (post CPB). Descriptive analyses (means, sds, medians, and interquartile ranges) of the study subject’s demographics and characteristics were performed. One-way analysis of variance (ANOVA) described the differences among groups as appropriate. Groups means were compared using all pairwise multiple comparison procedures (Holm-Sidak method).
RESULTS
Results of the Proteomic Analysis
1) Before CPB: Table 2 shows the comparisons of the mean levels of the proteins before CPB. Zinc-alpha-2-glycoprotein, alpha-2-Heremans Schmid-glycoprotein, angiotensinogen, and beta-2-glycoprotein 1 were approximately two-fold higher in the study than in the control group. Vitamin K–dependent protein S, coagulation factor V, apolipoprotein A-I, and tetranectin were lower in the study group compared with the control group. Proteoglycan 4 levels trended higher, although not statistically significant, in the study group (p = 0.1).
2) 0 hour after CPB: Table 3 shows the comparisons of the mean levels of the proteins after CPB at 0 hour. Tyrosine-protein kinase was increased three-fold in the study group. Angiotensinogen, transmembrane protease serine 13, thyroxine-binding globulin, zinc-alpha-2-glycoprotein, and alpha-antichymotrypsin were around 1.5–1.8-fold higher in the study group, whereas corticosteroid-binding globulin and complement component C8 beta chain were two- to four-fold decreased.
3) 48 hours post CPB: Table 4 shows the comparisons of the mean levels of the proteins 48 hours after CPB. Proteoglycan 4, keratin type I, and keratin type II were 2.6-fold higher in the study group, whereas selenoprotein P and galectin-3–binding protein were lower than the control group.
TABLE 2.
The Top Hit Proteins of the Proteomic Analysis of the Precardiopulmonary Bypass Plasma Samples
| Protein Name Pre Cardiopulmonary Bypass, Normal Spectrum Counts | Control | Study | Statistical Analysis | ||
|---|---|---|---|---|---|
| Mean ± sd | Mean ± sd | Significance Analysis of Microarrays | t test | Folds | |
| Tenascin | 4.36 ± 2.60 | 19.23 ± 18.59 | 0.70 | 0.1038 | 4.4 |
| Corticosteroid-binding globulin | 8.29 ± 6.08 | 24.05 ± 19.62 | 0.61 | 0.1041 | 2.9 |
| Complement factor I | 6.78 ± 5.54 | 19.06 ± 11.17 | 0.74 | 0.0569 | 2.8 |
| Coagulation factor XIII A chain | 1.65 ± 0.69 | 3.94 ± 2.30 | 0.77 | 0.1112 | 2.4 |
| Gamma-glutamyl hydrolase | 3.69 ± 1.23 | 7.86 ± 3.76 | 0.84 | 0.0548 | 2.1 |
| Zinc-alpha-2-glycoprotein | 17.35 ± 11.22 | 35.54 ± 5.68 | 1.08 | 0.0196 | 2.0 |
| Alpha-2-Heremans Schmid-glycoprotein | 31.73 ± 6.15 | 62.92 ± 16.42 | 1.38 | 0.0127 | 2.0 |
| L-selectin | 1.66 ± 0.19 | 3.28 ± 1.32 | 1.07 | 0.0828 | 2.0 |
| Transthyretin | 43.05 ± 21.72 | 84.32 ± 44.17 | 0.63 | 0.0814 | 2.0 |
| Angiotensinogen | 35.24 ± 15.12 | 68.83 ± 21.24 | 0.92 | 0.0230 | 2.0 |
| Beta-2-glycoprotein 1 | 14.63 ± 9.36 | 28.05 ± 3.28 | 1.06 | 0.0289 | 1.9 |
| Fibrinogen beta chain | 42.88 ± 41.65 | 81.88 ± 19.17 | 0.64 | 0.0803 | 1.9 |
| Thyroxine-binding globulin | 15.82 ± 6.26 | 30.14 ± 14.22 | 0.70 | 0.0684 | 1.9 |
| Plexin domain-containing protein 2 | 2.31 ± 0.62 | 4.10 ± 1.61 | 0.80 | 0.0552 | 1.8 |
| Carboxypeptidase N catalytic chain | 7.52 ± 2.07 | 13.27 ± 5.02 | 0.81 | 0.0880 | 1.8 |
| Plasma kallikrein | 7.51 ± 3.72 | 12.77 ± 4.19 | 0.67 | 0.0548 | 1.7 |
| Plasma protease C1 inhibitor | 65.3 ± 39.42 | 107.05 ± 21.03 | 0.69 | 0.0630 | 1.6 |
| Pregnancy zone protein | 57.58 ± 21.39 | 86.64 ± 4.65 | 1.12 | 0.0686 | 1.5 |
| Ficolin-3 | 11.12 ± 3.42 | 6.75 ± 2.79 | 0.70 | 0.0482 | –1.6 |
| Ig heavy chain V-III region BRO (extra-cellular) | 5.29 ± 2.21 | 2.96 ± 1.34 | 0.65 | 0.0736 | –1.8 |
| Vitamin K–dependent protein S | 8.71 ± 2.09 | 4.57 ± 1.38 | 1.19 | 0.0101 | –1.9 |
| Ig kappa chain V-I region AG (plasma membrane) | 8.88 ± 3.76 | 4.37 ± 1.69 | 0.83 | 0.0455 | –2.0 |
| Beta-2-microglobulin | 7.66 ± 4.36 | 3.73 ± 1.59 | 0.66 | 0.0860 | –2.1 |
| Coagulation factor V | 21.07 ± 6.63 | 9.88 ± 5.39 | 0.93 | 0.0388 | –2.1 |
| Fibulin-1 | 4.50 ± 2.89 | 2.07 ± 0.98 | 0.63 | 0.1403 | –2.2 |
| Ig kappa chain V-I region WEA | 5.48 ± 2.97 | 2.46 ± 1.49 | 0.68 | 0.0684 | –2.2 |
| Apolipoprotein A-I | 156.29 ± 47.47 | 67.33 ± 52.49 | 0.89 | 0.0230 | –2.3 |
| Cadherin-5 | 5.02 ± 3.40 | 1.91 ± 0.71 | 0.76 | 0.0814 | –2.6 |
| Ig kappa chain V-IV region Len | 6.47 ± 4.34 | 2.37 ± 1.27 | 0.73 | 0.0773 | –2.7 |
| Keratin, type II cytoskeletal 6A | 31.34 ± 18.05 | 10.41 ± 8.06 | 0.80 | 0.0859 | –3.0 |
| Ig kappa chain V-I region EU | 4.90 ± 4.31 | 1.53 ± 0.58 | 0.69 | 0.1079 | –3.2 |
| Keratin, type I cytoskeletal 16 | 30.97 ± 21.8 | 9.36 ± 6.89 | 0.75 | 0.0703 | –3.3 |
| Apolipoprotein D | 11.42 ± 8.16 | 3.38 ± 1.10 | 0.87 | 0.0714 | –3.4 |
| Apolipoprotein A-II | 9.35 ± 6.5 | 2.62 ± 1.20 | 0.87 | 0.0643 | –3.6 |
| Tetranectin | 10.22 ± 5.02 | 1.97 ± 1.11 | 1.34 | 0.0215 | –5.2 |
Ig = immunoglobulin.
Top hits for the “precardiopulmonary bypass” specimens were generated by applying the stats and fold change to the normalized spectral counts as indicated in the methods provided (significance analysis of microarrays > 0.6 or p < 0.05, with fold < or > 1.5).
TABLE 3.
The Top Hit Proteins of the Proteomic Analysis of the 0 Hour Postcardiopulmonary Bypass Plasma Samples
| Protein Name 0 hr, Normal Spectrum Counts | Control | Study | Statistical Analysis | ||
|---|---|---|---|---|---|
| Mean ± sd | Mean ± sd | Significance Analysis of Microarrays | t test | Folds | |
| Hemoglobin subunit beta | 7.27 ± 1.84 | 44.08 ± 49.11 | 0.72 | 0.1154 | 6.1 |
| Tyrosine-protein kinase hematopoietic cell kinase | 1.44 ± 1.62 | 4.45 ± 2.0 | 0.83 | 0.0300 | 3.1 |
| Ig kappa chain V-IV region Len | 2.70 ± 2.63 | 7.12 ± 4.79 | 0.60 | 0.0855 | 2.6 |
| Keratin, type II cytoskeletal 6A | 11.42 ± 9.35 | 24.69 ± 12.02 | 0.62 | 0.0676 | 2.2 |
| Transthyretin | 45.42 ± 36.14 | 95.33 ± 46.13 | 0.61 | 0.0711 | 2.1 |
| Plasma kallikrein | 5.86 ± 3.77 | 10.99 ± 4.10 | 0.65 | 0.0577 | 1.9 |
| Angiotensinogen | 17.01 ± 6.49 | 31.49 ± 5.16 | 1.24 | 0.0070 | 1.9 |
| Transmembrane protease serine 13 | 5.23 ± 2.27 | 9.38 ± 3.27 | 0.75 | 0.0438 | 1.8 |
| Keratin, type II cytoskeletal 5 | 17.29 ± 10.97 | 29.25 ± 6.91 | 0.67 | 0.0620 | 1.7 |
| Ig kappa chain V-I region WEA | 2.08 ± 0.33 | 3.48 ± 1.5 | 0.77 | 0.0783 | 1.7 |
| Thyroxine-binding globulin | 8.09 ± 1.94 | 13.31 ± 3.22 | 1.01 | 0.0198 | 1.6 |
| Zinc-alpha-2-glycoprotein | 15.22 ± 4.34 | 23.8 ± 5.63 | 0.86 | 0.0274 | 1.6 |
| Alpha-1-antichymotrypsin | 55.5 ± 11.06 | 86.44 ± 7.28 | 1.69 | 0.0025 | 1.6 |
| Ig gamma-2 chain C region | 46.71 ± 19.47 | 29.69 ± 7.48 | 0.63 | 0.0902 | –1.6 |
| Corticosteroid-binding globulin | 12.65 ± 1.87 | 7.11 ± 3.5 | 1.03 | 0.0210 | –1.8 |
| Ig gamma-3 chain C region | 51.12 ± 24.14 | 25.84 ± 6.01 | 0.84 | 0.0625 | –2.0 |
| Fibrinogen beta chain | 46.35 ± 29.99 | 22.17 ± 10.81 | 0.59 | 0.1041 | –2.1 |
| Complement component C8 beta chain | 7.97 ± 4.5 | 1.82 ± 1.51 | 1.02 | 0.0330 | –4.4 |
Ig = immunoglobulin.
Top hits for the “0 hr” specimens were generated by applying the stats and fold change to the normalized spectral counts as indicated in the methods provided (significance analysis of microarrays > 0.6 or p < 0.05, with fold < or > 1.5).
TABLE 4.
Top Hit Proteins of the Proteomic Analysis of the 48 Hours Postcardiopulmonary Bypass Plasma Samples
| Protein Name Post, Normal Spectrum Counts | Control | Study | Statistical Analysis | ||
|---|---|---|---|---|---|
| Mean ± sd | Mean ± sd | Significance Analysis of Microarrays | t test | Folds | |
| Proteoglycan 4 | 4.62 ± 2.44 | 12.33 ± 4.33 | 1.14 | 0.0144 | 2.7 |
| Keratin, type I cytoskeletal 14 | 7.78 ± 6.78 | 20.36 ± 7.16 | 0.90 | 0.0218 | 2.6 |
| Keratin, type II cytoskeletal 5 | 8.16 ± 7.74 | 21.18 ± 7.51 | 0.85 | 0.0261 | 2.6 |
| Tenascin | 7.59 ± 7.44 | 17.35 ± 8.37 | 0.62 | 0.0663 | 2.3 |
| Attractin | 6.97 ± 4.84 | 13.66 ± 2.4 | 0.93 | 0.0314 | 2.0 |
| Thyroxine-binding globulin | 9.23 ± 6.56 | 17.48 ± 4.93 | 0.72 | 0.0475 | 1.9 |
| Angiotensinogen | 32.35 ± 5.11 | 58.18 ± 18.44 | 1.10 | 0.0316 | 1.8 |
| Cholinesterase | 5.15 ± 3.78 | 8.73 ± 2.18 | 0.60 | 0.0821 | 1.7 |
| Leucine-rich alpha-2-glycoprotein | 18.11 ± 10.0 | 30.59 ± 8.3 | 0.68 | 0.0524 | 1.7 |
| Alpha-2-Heremans Schmid-glycoprotein | 23.14 ± 6.72 | 37.96 ± 13.16 | 0.75 | 0.0539 | 1.6 |
| Transforming growth factor-beta–induced protein ig-h3 | 3.56 ± 1.13 | 5.85 ± 2.12 | 0.70 | 0.0609 | 1.6 |
| Vasorin | 3.92 ± 2.22 | 6.40 ± 1.08 | 0.75 | 0.0541 | 1.6 |
| Zinc-alpha-2-glycoprotein | 20.33 ± 7.35 | 33.0 ± 1.69 | 1.40 | 0.0189 | 1.6 |
| Alpha-1-antichymotrypsin | 143.84 ± 26.2 | 221.76 ± 79.84 | 0.73 | 0.0722 | 1.5 |
| Pigment epithelium-derived factor | 7.97 ± 2.51 | 12.23 ± 1.02 | 1.21 | 0.0175 | 1.5 |
| Beta-2-glycoprotein 1 | 14.93 ± 5.05 | 22.75 ± 5.37 | 0.75 | 0.0392 | 1.5 |
| Epididymis luminal protein 214 | 97.37 ± 20.63 | 55.02 ± 35.33 | 0.76 | 0.0476 | -1.8 |
| Trypsin-3 | 6.11 ± 3.19 | 3.33 ± 1.23 | 0.63 | 0.0904 | -1.8 |
| Ig gamma-4 chain C region | 48.18 ± 18.44 | 23.44 ± 20.3 | 0.64 | 0.0608 | -2.1 |
| Ig lambda chain V-I region HA | 2.93 ± 1.79 | 1.23 ± 0.83 | 0.65 | 0.0783 | -2.4 |
| Apolipoprotein A-II | 10.83 ± 5.66 | 4.19 ± 2.62 | 0.80 | 0.0482 | -2.6 |
| Selenoprotein P | 3.07 ± 0.71 | 1.14 ± 0.93 | 1.18 | 0.0091 | -2.7 |
| Immunoglobulin J chain | 5.09 ± 2.87 | 1.82 ± 1.34 | 0.77 | 0.0524 | -2.8 |
| Galectin-3–binding protein | 5.54 ± 2.34 | 1.95 ± 1.38 | 0.97 | 0.0235 | -2.8 |
Ig = immunoglobulin.
Top hits for the “post” specimens were generated by applying the stats and fold change to the normalized spectral counts as indicated in the methods provided (significance analysis of microarrays > 0.6 or p < 0.05, with fold < or > 1.5).
Pathways Analysis.
Pathway analysis is used to identify highly associated and well-characterized molecular physiologic cellular processes. This approach was applied to the most significantly changed proteins stemming from the combined 0 and 48 hours post-CPB samples. There were six significant albeit loosely associated pathways that include cytoskeleton remodeling as the most robust (illustrated in Fig. S3, http://links.lww.com/CCX/A779) due to changes in keratin isoforms. In addition, cell adhesion, blood coagulation, and immune response were also highlighted as potentially interesting pathways in Table S1 (http://links.lww.com/CCX/A780). However, due to the lack of proteins associated with any one pathway, we also included “Network” analysis, which pulls in a much broader array of biological associations that do not necessarily belong to any single pathway.
Network Analysis.
Network analysis is used to identify any associated and yet well-characterized GO-related molecular or biological processes that may not be occurring within a specific cellular mechanism or even within a particular cell type, and for this reason, it is better suited for blood pool–derived specimens. This approach was applied to the most significantly changed proteins stemming from the combined 0- and 48-hour post-CPB study plasma specimens. There were seven significantly associated networks (Table S2, http://links.lww.com/CCX/A781) that included number 2 cytokine signaling and number 6 regulation of blood coagulation (illustrated in Fig. S1, http://links.lww.com/CCX/A777 and Fig. S2, http://links.lww.com/CCX/A778) due to 27 and 11 significant protein changes (also called “seed nodes”), respectively. The total node number is always higher than the seed nodes, which stem from the top hit protein list identified from our datasets due to the fact that second-degree associations are also common, and these additional nodes will connect the dots between proteins that are not directly but somewhat indirectly connected by a separate protein variable. Of note, for our illustrations and to gain a higher level of confidence, we only explore those networks that contain a high degree of connectivity without the need for more than a few second-degree nodes required for connectivity. As a whole, proteins identified within these datasets appear to point toward response to an endogenous hormone, proliferation, cytokine-mediated response, exocytosis/secretion, blood coagulation, and complement activation. Focusing on any one of these areas of apparent response will likely yield additional information regarding the pathology of the disease process. Table S3 (http://links.lww.com/CCX/A782) shows the main GO processes of the discovered proteins in the combined 0- and 48-hour post-CPB study plasma specimens. Again, the GO processes involve cytoskeletal remodeling, cell adhesion, blood coagulation, and immune response processes.
Results of the Validation Phase
Patient Characteristics.
There were no statistical differences among the groups in terms of age, weight, or STAT category. The mean length of mechanical ventilation (hr) was 30 hours for the control group and 110 hours for the study group (p < 0.01). Although longer in the study group, differences in ICU and hospital length of stay did not reach statistical significance (p = 0.08, p = 0.07, respectively) (Table 5).
TABLE 5.
Patient Characteristics of the Validation Phase of the Study
| Variable | Control (n = 20) | Study (n = 20) | p |
|---|---|---|---|
| Age (d) | 56 ± 49 | 32.2 ± 45 | 0.12 |
| Weight (kg) | 4.06 ± 1.13 | 3.56 ± 1.09 | 0.16 |
| Gestational age (wk) | 37.6 ± 1.7 | 37.35 ± 1.8 | 0.6 |
| Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery risk of mortality category | 3.8 ± 0.4 | 3.8 ± 0.4 | 1.0 |
| Cardiopulmonary bypass time (min) | 97 ± 32 | 101 ± 27 | 0.7 |
| Length of mechanical ventilation (hr) | 29.72 ± 16 | 109.74 ± 93 | < 0.01 |
| ICU LOS (d) | 17.45 ± 10 | 54.62 ± 93 | 0.08 |
| Hospital LOS (d) | 20.9 ± 10 | 59.35 ± 92 | 0.07 |
LOS = length of stay.
Data presented as mean ± sd.
Proteoglycan 4 Plasma Levels.
No differences in proteoglycan 4 levels at the pre-CPB time point were seen (p = 0.48). The study group (128 ± 67 ng/mL) had lower levels of proteoglycan 4 compared with the control group (195 ± 160 ng/mL) in the 48-hour post-CPB sampling (p = 0.03) (Table 6).
TABLE 6.
Descriptive Analysis of the Control and Study Proteoglycan 4 Plasma Concentration Pre and Post Cardiopulmonary Bypass
| Timepoint | Mean ± sd | se | Median (Quartile 1, Quartile 3) |
|---|---|---|---|
| Pre-CPB study | 146 ± 82 | 18 | 131 (71, 225) |
| Post-CPB study | 128 ± 67 | 15 | 127 (74, 172) |
| Pre-CPB control | 168 ± 49 | 11 | 156 (136, 203) |
| Post-CPB control | 195 ± 160 | 36 | 147 (97, 250) |
| Comparison | Difference of Means | t | p |
| Pre study vs post study | 17.680 | 0.566 | 0.573 |
| Pre control vs post control | 26.400 | 0.845 | 0.401 |
| Pre study vs pre control | 22.275 | 0.713 | 0.478 |
| Post study vs post control | 66.355 | 2.125 | 0.037 |
CPB = cardiopulmonary bypass.
Comparison of the proteoglycan 4 concentrations (ng/mL) among and within the groups.
DISCUSSION
Lung dysfunction is one of the most common consequences after cardiac surgery (1–3). The clinical significance of pulmonary dysfunction can be variable, but it can lead to prolonged mechanical ventilation support for a significant percentage of patients following CPB (8, 9). Our study did not aim to identify patients with lung dysfunction. Instead, we assumed that all patients have some degree of lung dysfunction after cardiac surgery. Some patients have lung complications requiring prolonged mechanical ventilation. Given the clinical importance of postoperative lung injury, we aimed to discover serum biomarkers that can provide a prognosis of the severity of lung injury and suggest potential therapeutic targets for future study.
To identify biomarkers for the lung injury following CPB, we obtained unbiased proteomic analysis before and after CPB in a limited number of patients and identified a large number of proteins that are involved in multiple overlapped cellular processes. Consequently, we performed pathway analysis to determine the molecular and physiologic processes that these proteins are involved with, revealing six significant pathways.
The cytoskeleton remodeling pathway was one of the most significant pathways as multiple keratin isoforms were increased in the study samples (Fig. S3, http://links.lww.com/CCX/A779). Other discovered pathways involved blood coagulation, cell adhesions, and immune response, which were further investigated by using network analysis. Based on significant protein changes, both cytokine signaling and regulation of blood coagulation were among the most important networks.
As a result of the detailed molecular and biological processes analysis, the identified proteins are involved in multiple pathophysiologic processes that could be responsible for the lung disease and coagulopathy that are seen after cardiac surgery. Validating some of these proteins could lead to a better understanding of the disease process and could open the door for future diagnostic and therapeutic targets.
Proteoglycan 4 is one of the proteoglycans found in many different organs, including the lungs. Proteoglycan 4 is found to have an essential role in the inflammatory and innate immunologic response in the lung and be part of the cytoskeletal structure of the extracellular matrix, cellular plasma membrane, and intracellular lung structures (20). Recent evidence suggests the important role of proteoglycan 4 in regulating the inflammatory response by binding to some surface receptors in multiple organs, including the lungs (21–26). Proteoglycan levels and structures are altered significantly in response to lung injury related to inflammation or infection, as exemplified in ARDS, asthma, and chronic obstructive pulmonary disease (27–30). Our initial proteomic analysis showed the difference in proteoglycan levels before and after CPB for the study group compared with the control. In the validation study, we confirmed that lower plasma levels of proteoglycan 4 were associated with mechanical ventilation longer than 2 days for pediatric patients following CPB. The prospect of a biomarker, or a panel of biomarkers, that would predict lung injury would be very useful for postoperative clinical management. It could provide a potential avenue for future therapies to mitigate lung injury associated with CPB.
CONCLUSIONS
A limitation of this study is the small sample size. In addition, several other factors, such as hemodynamic instability, can impact mechanical ventilation usage after cardiac surgery. We did not factor the indication for mechanical ventilation in our classification of the included patients into the study versus control. Further complicating our study is the variable definition of prolonged ventilation after congenital heart surgery that has been previously described (31). To address this, we carefully matched the patient characteristics, including the STAT category, during both the discovery and validation phases, as demonstrated in Tables 3 and 5. We assumed that patients with a matched length of CPB time, weight, age, and surgical complexity have lung dysfunction as the indication for mechanical ventilation. On the other hand, mechanical ventilation is a cause of lung injury by itself, and this cannot be ruled out that as a reason for alterations of proteoglycan 4 levels.
Our proteomic analysis discovered multiple proteins involved in cytoskeleton remodeling, blood coagulation, and complement and immune response activation. These findings indicate multiple pathways and possible therapeutic targets that can be further investigated. The validation phase of the study is the first to identify the potential prognostic value of plasma levels of proteoglycan 4 in lung injury and prolonged mechanical ventilation following cardiac surgery using CPB. The precise role and behavior of proteoglycan 4 in lung injury after cardiac surgery will be investigated in future studies. Future studies evaluating proteoglycan 4 and other biomarkers may have significant diagnostic and possible therapeutic implications.
Supplementary Material
Footnotes
Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website (http://journals.lww.com/ccejournal).
Supported, in part, by the National Institutes of Health R01 HL129907, and the proteoglycan 4 enzyme-linked immunosorbent assay testing was supported, in part, by Section of Cardiac Critical Care internal funds.
The authors have disclosed that they do not have any conflicts of interest.
REFERENCES
- 1.Clark SC. Lung injury after cardiopulmonary bypass. Perfusion. 2006; 21:225–228 [DOI] [PubMed] [Google Scholar]
- 2.Kirklin JK. Prospects for understanding and eliminating the deleterious effects of cardiopulmonary bypass. Ann Thorac Surg. 1991; 51:529–531 [DOI] [PubMed] [Google Scholar]
- 3.Ng CS, Wan S, Yim AP, et al. Pulmonary dysfunction after cardiac surgery. Chest. 2002; 121:1269–1277 [DOI] [PubMed] [Google Scholar]
- 4.Westaby S. Organ dysfunction after cardiopulmonary bypass. A systemic inflammatory reaction initiated by the extracorporeal circuit. Intensive Care Med. 1987; 13:89–95 [DOI] [PubMed] [Google Scholar]
- 5.Apostolakis E, Filos KS, Koletsis E, et al. Lung dysfunction following cardiopulmonary bypass. J Card Surg. 2010; 25:47–55, 2010 [DOI] [PubMed] [Google Scholar]
- 6.Vohra HA, Whistance R, Modi A, et al. The inflammatory response to miniaturised extracorporeal circulation: A review of the literature. Mediators Inflamm. 2009; 2009:707042. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Giacinto O, Satriano U, Nenna A, et al. Inflammatory response and endothelial dysfunction following cardiopulmonary bypass: Pathophysiology and pharmacological targets. Recent Pat Inflamm Allergy Drug Discov. 2019; 13:158–173 [DOI] [PubMed] [Google Scholar]
- 8.Huffmyer JL, Groves DS. Pulmonary complications of cardiopulmonary bypass. Best Pract Res Clin Anaesthesiol. 2015; 29:163–175 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Hammermeister KE, Burchfiel C, Johnson R, et al. Identification of patients at greatest risk for developing major complications at cardiac surgery. Circulation. 1990; 84:IV380–389 [PubMed] [Google Scholar]
- 10.Gaies M, Tabbutt S, Schwartz SM, et al. Clinical epidemiology of extubation failure in the pediatric cardiac ICU: A report from the pediatric cardiac critical care consortium. Pediatr Crit Care Med. 2015; 16:837–845 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Monteverde E, Fernández A, Poterala R, et al. Characterization of pediatric patients receiving prolonged mechanical ventilation. Pediatr Crit Care Med. 2011; 12:e287–e291 [DOI] [PubMed] [Google Scholar]
- 12.Li XJ, Yi EC, Kemp CJ, et al. A software suite for the generation and comparison of peptide arrays from sets of data collected by liquid chromatography-mass spectrometry. Mol Cell Proteomics. 2005; 4:1328–1340 [DOI] [PubMed] [Google Scholar]
- 13.Nesvizhskii AI, Aebersold R. Interpretation of shotgun proteomic data: The protein inference problem. Mol Cell Proteomics. 2005; 4:1419–1440 [DOI] [PubMed] [Google Scholar]
- 14.Domon B, Aebersold R. Options and considerations when selecting a quantitative proteomics strategy. Nat Biotechnol. 2010; 28:710–721 [DOI] [PubMed] [Google Scholar]
- 15.Gillet LC, Navarro P, Tate S, et al. Targeted data extraction of the MS/MS spectra generated by data-independent acquisition: A new concept for consistent and accurate proteome analysis. Mol Cell Proteomics. 2012; 11:O111.016717. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Jacobs ML, O’Brien SM, Jacobs JP, et al. An empirically based tool for analyzing morbidity associated with operations for congenital heart disease. J Thorac Cardiovasc Surg. 2013; 145:1046–1057. e1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Ludwig MR, Kojima K, Bowersock GJ, et al. Surveying the serologic proteome in a tissue-specific Kras(G12D) knockin mouse model of pancreatic cancer. Proteomics. 2016; 16:516–531 [DOI] [PubMed] [Google Scholar]
- 18.Bhatia VN, Perlman DH, Costello CE, et al. Software tool for researching annotations of proteins: Open-source protein annotation software with data visualization. Anal Chem. 2009; 81:9819–9823 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Ekins S, Bugrim A, Brovold L, et al. Algorithms for network analysis in systems-ADME/Tox using the MetaCore and MetaDrug platforms. Xenobiotica. 2006; 36:877–901 [DOI] [PubMed] [Google Scholar]
- 20.Frevert C, Wight TN. Matrix proteoglycans.: The Encyclopedia of Respiratory Medicine. Laurent GJ. (Eds). London, United Kingdom, Elsevier, 2006, pp 184–187 [Google Scholar]
- 21.Das N, Schmidt TA, Krawetz RJ, et al. Proteoglycan 4: From mere lubricant to regulator of tissue homeostasis and inflammation: Does proteoglycan 4 have the ability to buffer the inflammatory response? Bioessays. 2019; 41:e1800166. [DOI] [PubMed] [Google Scholar]
- 22.Iqbal SM, Leonard C, Regmi SC, et al. Lubricin/proteoglycan 4 binds to and regulates the activity of Toll-like receptors in vitro. Sci Rep. 2016; 6:18910. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Alquraini A, Jamal M, Zhang L, et al. The autocrine role of proteoglycan-4 (PRG4) in modulating osteoarthritic synoviocyte proliferation and expression of matrix degrading enzymes. Arthritis Res Ther. 2017; 19:89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Alquraini A, Garguilo S, D’Souza G, et al. The interaction of lubricin/proteoglycan 4 (PRG4) with toll-like receptors 2 and 4: An anti-inflammatory role of PRG4 in synovial fluid. Arthritis Res Ther. 2015; 17:353. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Nahon JE, Hoekstra M, Havik SR, et al. Proteoglycan 4 regulates macrophage function without altering atherosclerotic lesion formation in a murine bone marrow-specific deletion model. Atherosclerosis. 2018; 274:120–127 [DOI] [PubMed] [Google Scholar]
- 26.Parish CR. The role of heparan sulphate in inflammation. Nat Rev Immunol. 2006; 6:633–643 [DOI] [PubMed] [Google Scholar]
- 27.Bensadoun ES, Burke AK, Hogg JC, et al. Proteoglycan deposition in pulmonary fibrosis. Am J Respir Crit Care Med. 1996; 154:1819–1828 [DOI] [PubMed] [Google Scholar]
- 28.Merrilees MJ, Ching PS, Beaumont B, et al. Changes in elastin, elastin binding protein and versican in alveoli in chronic obstructive pulmonary disease. Respir Res. 2008; 9:41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Araujo BB, Dolhnikoff M, Silva LF, et al. Extracellular matrix components and regulators in the airway smooth muscle in asthma. Eur Respir J. 2008; 32:61–69 [DOI] [PubMed] [Google Scholar]
- 30.Lee KY, Chuang HC, Chen TT, et al. Proteoglycan 4 is a diagnostic biomarker for COPD. Int J Chron Obstruct Pulmon Dis. 2015; 10:1999–2007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Rose L, McGinlay M, Amin R, et al. Variation in definition of prolonged mechanical ventilation. Respir Care. 2017; 62:1324–1332 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.