Insights Into a "Negative" ICU Trial Derived From Gene Expression Profiling. Issue 12 (December 2019)

• Record Type: Journal Article
• Authors: Hoekstra, Mary; Maslove, David M.; Veldhoen, Richard A.; Marshall, John C.; Muscedere, John
• Is Part Of: Critical care medicine. Volume 47:Issue 12(2019)
• Publication Date: 2019-12
• Subjects: critical care -- intensive care -- lactoferrin -- nosocomial infection -- randomized controlled trial -- transcriptomics; Critical care medicine -- Periodicals; Soins intensifs -- Périodiques; 616.028
• DOI: 10.1097/CCM.0000000000003693
• Languages: English
• ISSNs: 0090-3493

Abstract

Abstract : Objectives: Randomized controlled trials in the ICU often fail to show differences in endpoints between groups. We sought to explore reasons for this at a molecular level by analyzing transcriptomic data from a recent negative trial. Our objectives were to determine if randomization successfully balanced transcriptomic features between groups, to assess transcriptomic heterogeneity among the study subjects included, and to determine if the study drug had any effect at the gene expression level. Design: Bioinformatics analysis of transcriptomic and clinical data collected in the course of a randomized controlled trial. Setting: Tertiary academic mixed medical-surgical ICU. Patients: Adult, critically ill patients expected to require invasive mechanical ventilation more than 48 hours. Interventions: Lactoferrin or placebo delivered enterally and via an oral swab for up to 28 days. Measurements and Main Results: We found no major imbalances in transcriptomic features between groups. Unsupervised analysis did not reveal distinct clusters among patients at the time of enrollment. There were marked differences in gene expression between early and later time points. Patients in the lactoferrin group showed changes in the expression of genes associated with immune pathways known to be associated with lactoferrin. Conclusions: In this clinical trial, transcriptomic data provided a useful complement to clinical data, suggesting that the reasons for the negative result were less likely related to the biological efficacy of the study drug, and may instead have been related to poor sensitivity of the clinical outcomes. In larger studies, transcriptomics may also prove useful in predicting response to treatment. Abstract : Supplemental Digital Content is available in the text.