BioProject

To stratify hydrocortisone application in septic shock, we investigated an immune sub-study of the CORTICUS trial (Sprung et.al 2008, NEJM) employing machine learning to a panel of 120 parameters of 84 patients (n=24 non-survived, n=60 survived, 28 days) with special emphasis on potentially disadvantageous corticosteroids effects in the context of sepsis including clinical parameters, organ failure scores, lymphocyte counts and plasma protein concentrations of cytokines. We identified the ratio of IFNγ/IL10 in serum before randomization to serve as a valid biomarker for treatment stratification. This was validated with cytokine serum levels of patients (n=49) from the SISPCT study (Bloos et.al 2016, JAMA) and the early arm (n=20) of a hydrocortisone cross-over study (Keh et.al. 2003, Am J Respir Crit Care Med). Integrating these three studies, we yielded an odds ratio of 2.1 and a 95% confidence interval of 0.99-4.52 (P=0.03).In vitro assays revealed IFNγ/Il10 to reflect the burden or severity of systemic infection. Severity was evidencedbyserum levels of these cy-tokines in the patients with septic shock we observed, and also in patients with less severe sepsis. Elucidating the molecular regulation of leukocytes during treatment and placebo by a transcriptomics analysis pointed to induced recovery of immune cell function due to hydro-cortisone treatment, particularly in the predicted hydrocortisone responders. IFNγ/IL10 is a molecular marker with high potential for support of hydrocortisone therapy decision. IFNγ/IL10 is reasonable for this as it reflects the burden and recovery of the immune system which seems to be indicative for corticosteroids treatment of septic shock. Overall design: Paired samples of 47 septic shock patients from the CORTICUS-trial (ClinicalTrials.gov number, NCT00147004) treated with either Placebo or Hydrocortisone (HC) were included. Whole blood from patients was collected in the morning (pre) and 24 hours (post) later. Total cellular RNA from circulating leukocytes was isolated using PaxGene Blood RNA Kit (PreAnalytix) for transcriptome profiling. The first analysis compared the pre-treatment samples of 28 patients with IFNg/IL10 high versus 19 patients with IFNg/IL10 low. Genes with standard deviation or average expression above the 25% quantile were tested for significant up- or downregulation using two sided t-tests. In the second analysis, the effect of HC on 7 IFNg/IL10 low-ratio patients (benefitting from HC therapy) was observed by comparing gene expression 24h post start of the treatment versus pre-treatment (0h). Differentially expressed genes were obtained using two sided t-tests followed by multiple testing correction (Benjamini-Hochberg, FDR < 0.1). A similar comparison was made with 12 low-ratio patients treated with placebo. Genes distinctly up or down regu-lated in patients of the HC arm compared to the placebo arm were selected for gene set enrichment analysis.