The included transcriptome data were downloaded from gene expression omnibus (GEO) databases (http://www.ncbi.nlm.nih.gov/geo/) (Barrett et al., 2013 (link)). Only peripheral blood samples collected within 24 h of diagnosis or ICU admission were included. The RNA sequencing data of 91 adult samples (including 19 septic shock, 20 sepsis, 12 uncomplicated infection and 40 healthy controls) in the GSE154918 dataset, which were pre-processed using the DESeq2 package by the contributors (Love et al., 2014 (link); Herwanto et al., 2021 (link)), were used as discovery dataset to explore genes, modules and mechanisms associated with septic shock. Additionally, the array data and survival information of 479 adult sepsis samples with a 28-day cumulative death rate about 23.80% in the GSE65682 dataset were read in R language to determine the prognostic significance of interested genes in sepsis patients. The gene expression profiles of GSE65682 were background-subtracted and normalized by a robust multi-array average algorithm using the affy package. The row count matrix of 345 adult sepsis samples including 52 dead and 293 survival samples in the GSE185263 dataset was downloaded to validate survival significance of the hub gene.
Clinical blood laboratory examinations data of sepsis and septic shock patients were extracted from the MIMIC-IV (version 2.0) database in the physionet (https://physionet.org/content/mimiciv/2.0/) for the further validation (Goldberger et al., 2000 (link); Johnson et al., 2022 (link)). One of the authors who has finished the required Collaborative Institutional Training Initiative examination (Certification number 53459610 for Zhao) can access the database. The adult ICU stay samples meeting the sepsis-3 definition at the first day of ICU admission were included (Singer et al., 2016 (link)). The patients’ parameters including absolute neutrophil count, absolute CD3 count (i.e., T cell count), absolute CD4 count and absolute CD8 count from blood specimens and survival data were extracted for further analysis. Specifically, we extracted the max values of neutrophil counts of each ICU stay within 6 h before ICU admission and 24 h after; while the chart time requirements of the other three items were limited to 6 h before ICU admission and 48 h after, concerning their more time costs waiting for the reports. In our study, the data about neutrophil counts of 8250 ICU stays containing 40.5% septic shock samples and with a 28-day cumulative mortality rate (CMR) about 22.3% were extracted. However, among them only 69 had the time-limited data about CD3 counts and 68 had desirable CD4 counts and CD8 counts due to their less clinical applications. More details were shown in Supplementary Table S1. The code used for data extraction can be available on GitHub (https://github.com/MIT-LCP/mimic-iv).
Free full text: Click here