Tjitske van Engelen

107 Host response in patients with ICU-acquired pneumonia Causative pathogens Causative pathogens were determined post hoc based on isolation of a respiratory pathogen from any lower respiratory tract specimen (including both clinical and study surveillance cultures) or blood culture in the 3 days before and after the day of pneumonia diagnosis [2, 3]. Sample collection Ethylenediaminetetraacetic acid (EDTA) anticoagulated blood samples were centrifuged at the local site at 1500 g for 15 minutes (preferably at 4°C), after which plasma was transferred into plasma storage tubes and stored at -70°C as soon as possible and within 3 hours after sample collection at the ICU. Samples were regularly transferred from the local sites to the central laboratory at the University of Antwerp, Belgium, and further transferred in four batches to the Amsterdam UMC biobank, the Netherlands, all on dry ice. Normal reference values were obtained in plasma collected from 19 age and sex matched subjects after having provided written informed consent (part of the ELDERBIOME study, clinicaltrials.gov identifier NCT02928367). Assays We measured 19 host response biomarkers on a BioPlex-machine (Bio-Plex 200, Bio-Rad, Hercules, CA, USA) using a custom-made Luminex multiplex assay (R&D Systems Inc, Minneapolis, MN, USA). All samples were diluted 1:2 in assay buffer and all biomarkers were in one assay. We categorized these biomarkers into four pathophysiological domains: interleukin (IL)-6, IL-8, IL-10, and IL-1 receptor antagonist (IL-1RA)(reflecting cytokine release); matrix metalloproteinase (MMP)-8, soluble triggering receptor expressed on myeloid cells (sTREM)-1, soluble cluster of differentiation (sCD)163, soluble receptor for advanced glycation endproducts (sRAGE), tenascin-C and procalcitonin (reflecting systemic inflammation); sE-selectin, soluble vascular cell adhesion protein (sVCAM)-1, fractalkine, syndecan-1, soluble thrombomodulin, angiopoietin-1, and angiopoietin-2 (reflecting endothelial activation and function); soluble tissue factor and D-dimer (reflecting coagulation activation). Samples were randomly distributed over assay plates stratified per subgroup of interest (i.e. case baseline, case event, case day 7, control baseline, control day 7, and healthy subjects), such that the proportion of each subgroup on each plate was the same, and the central tendency and spread of each biomarker on each plate was expected to be the same. We removed individual analytes if fewer than 25 beads were measured for that analyte in that sample. Batch effects (non-biological and non-random variation in biomarker levels) between plates were corrected using multiple bridging samples present on each plate in duplicate. We fitted linear regression models per analyte to obtain scaling factors for each plate relative to the first plate of the measurement day. Biomarker measurements below the limit of quantification were imputed as half the lower limit of quantification. This was necessary for 418/1796 [23.3%] of IL-10 measurements, but only 9/32328 (0.03%) of all other biomarker measurements. We accepted extrapolated biomarker values up to ten times the upper limit of quantification. Measurements higher than ten times the upper limit of quantification (in 115/34124 [0.34%] total data points) were set to the upper limit of quantification. 5

RkJQdWJsaXNoZXIy MTk4NDMw