Transcriptional instability during evolving sepsis may limit biomarker based risk stratification

Abstract

Background: Sepsis causes extensive morbidity and mortality in children worldwide. Prompt recognition and timely treatment of sepsis is critical in reducing morbidity and mortality. Genomic approaches are used to discover novel pathways, therapeutic targets and biomarkers. These may facilitate diagnosis and risk stratification to tailor treatment strategies.

Objective: To investigate the temporal gene expression during the evolution of sepsis induced multi-organ failure in response to a single organism, Neisseria meningitidis, in previously healthy children.

Method: RNA was extracted from serial blood samples (6 time points over 48 hours from presentation) from five critically ill children with meningococcal sepsis. Extracted RNA was hybridized to Affymetrix arrays. The RNA underwent strict quality control and standardized quantitation. Gene expression results were analyzed using GeneSpring software and Ingenuity Pathway Analysis.

Result: A marked variability in differential gene expression was observed between time points and between patients revealing dynamic expression changes during the evolution of sepsis. While there was evidence of time-dependent changes in expected gene networks including those involving immune responses and inflammatory pathways, temporal variation was also evident in specific "biomarkers" that have been proposed for diagnostic and risk stratification functions. The extent and nature of this variability was not readily explained by clinical phenotype.

Conclusion: This is the first study of its kind detailing extensive expression changes in children during the evolution of sepsis. This highlights a limitation of static or single time point biomarker estimation. Serial estimations or more comprehensive network approaches may be required to optimize risk stratification in complex, time-critical conditions such as evolving sepsis.

Conflict of interest statement

Competing Interests: The authors have declared that no competing interests exist.

Figures

Figure 1. Gene expression changes over time.

(a) Plot of 28,869 genes with normalized intensity values against time for the 5 patients, from GeneSpring 11.5 analysis (see Methods). Colors of plots are representative of genes' change in expression level compared to the corresponding gene's expression at 0 hours for patient 1. Blue is for upregulated expression, yellow is for no change in expression, and red is for downregulated expression compared to patient 1 at time 0. (b) Numbers of genes up- and down-regulated over time for each of the 5 patients, compared to each patient's gene expression at 0 hours. The top panel indicates numbers of up-regulated genes, the bottom panel indicates numbers of down-regulated genes, for patients 1 (•), 2 (▪), 3 (Δ), 4 (x), and 5 (*).

Figure 2. Temporal variation in gene expression changes.

Numbers of genes that change at different time points as compared to their baseline gene expression level are shown. Expression levels were analyzed by pairwise comparisons between each of 5 time points (4, 8, 12, 24, and 48 hours) compared to each patient's corresponding 0 hour gene expression profiles, and expression levels of greater than 2-fold changes were noted. This was performed for all 5 patients (except in patient 4 where the 4-hour sample had degraded RNA), 24 pairwise comparisons in total. Numbers in the sections of the Venn diagram correspond to numbers of genes that were up- or down-regulated by ≥2-fold for at least 1 in the 5 patients; numbers in the outer sections correspond to the numbers of genes uniquely regulated at the corresponding time points, and the number in the middle corresponds to the number of genes differentially regulated at all time points during the time-course.

Figure 3. Ingenuity Pathway Analysis network.

Example of a gene network derived from genes uniquely regulated at 48 hours, comprising 3 merged networks consisting of 444 genes (red, increased expression; green, decreased expression) as derived from Ingenuity Pathway Analysis. Selected regions of the network are highlighted, consisting of genes involved in: (a) Antimicrobial response, Inflammatory response, Cell-To-Cell Signaling and Interaction (“MX1 genes”, network score 23); (b) Cell Death, Cell Signaling, DNA Replication, Recombination and Repair (“CASP genes”, network score 33); (c) Cellular Development, Hematological System Development and Function, Hematopoiesis (“RB1 genes”, network score 37).

Figure 4. Expression variability of proposed biomarkers for sepsis.

The gene expression changes of a selection of proposed biomarkers are shown, to demonstrate their variability over the time-course. Starting from (a) in clockwise direction: (a) GZMB and (b) CCL4 are up- or down-regulated compared to 0 hours at all time points (“common”); (c) IL8 is up- or down-regulated compared to 0 hours at 8, 12, 24, and 48 hours; (d) MMP8 is up- or down-regulated compared to 0 hours at 12, 24, and 48 hours; (e) CCL3 is up- or down-regulated compared to 0 hours at 24 and 48 hours; (f) SULF2 is up- or down-regulated compared to 0 hours at 48 hours only. Up- and down-regulation is filtered at ≥2-fold change compared to 0 hours.