LIPS and PaO2/FiO2 Combined Plasma Biomarkers Predict Onset of Acute Respiratory Distress Syndrome in Patients of High Risks in SICU: A Prospective Exploratory Study

2024-09-29

作者Ziyuan Shen, Zhongnan Yin, Senhao Wei, Zhukai Cong, Feng Zhao, Hua Zhang, Xi Zhu
来自Mediators of Inflammation
DOI: doi.org/10.1155/2024/4936265
 
摘要
Objective. To explore and validate the value of clinical parameters combined with plasma biomarkers for predicting acute respiratory distress syndrome (ARDS) in patients of high risks in the surgical intensive care unit (SICU). Materials and Methods. We conducted a prospective, observational study from January 2020 to December 2023, which enrolled 263 patients of high risks in the SICU of Peking University Third Hospital consecutively; they were classified as ARDS and non-ARDS according to whether ARDS occurred after enrollment. Collected clinical characteristics and blood samples within 24 hr of admission to SICU. Blood samples from the first day to the seventh day of SICU were collected from patients without ARDS, and patients with ARDS were collected until 1 day after ARDS onset, forming data based on time series. ELISA and CBA were used to measure plasma biomarkers. Endpoint of the study was the onset of ARDS. Cox proportional hazard regression analysis was used to find independent risk factors of the onset of ARDS, then constructed a nomogram and tested its goodness-of-fit. Results. About 84 of 263 patients ended with ARDS. Univariate analysis found 15 risk factors showed differences between ARDS and non-ARDS, namely, interleukin 6, interleukin 8 (IL-8), angiopoietin Ⅱ, LIPS, APACHEⅡ, SOFA, PaO2/FiO2, age, sex, shock, sepsis, acute abdomen, pulmonary contusion, pneumonia, hepatic dysfunction. We included factors with p<0.2 in multivariate analysis and showed LIPS, PaO2/FiO2, IL-8, and receptor for advanced glycation end-products (RAGE) of the first day were independent risk factors for ARDS in SICU, a model combining them was good in predicting ARDS (C-index was 0.864 in total patients of high risks). The median of the C-index was 0.865, showed by fivefold cross-validation in the train cohort or validation cohort. The calibration curve shows an agreement between the probability of predicting ARDS and the actual probability of occurrence. Decision curve analysis indicated that the model had clinical use value. We constructed a nomogram that had the ability to predict ARDS in patients of high risks in SICU. Conclusions. LIPS, PaO2/FiO2, plasma IL-8, and RAGE of the first day were independent risk factors of the onset of ARDS. The predictive ability for ARDS can be greatly improved when combining clinical parameters and plasma biomarkers.