Extremes of Interferon-Stimulated Gene Expression Associate with Worse Outcomes in the Acute Respiratory Distress Syndrome

Abstract

Acute Respiratory Distress Syndrome (ARDS) severity may be influenced by heterogeneity of neutrophil activation. Interferon-stimulated genes (ISG) are a broad gene family induced by Type I interferons, often as a response to viral infections, which evokes extensive immunomodulation. We tested the hypothesis that over- or under-expression of immunomodulatory ISG by neutrophils is associated with worse clinical outcomes in patients with ARDS. Genome-wide transcriptional profiles of circulating neutrophils isolated from patients with sepsis-induced ARDS (n = 31) and healthy controls (n = 19) were used to characterize ISG expression. Hierarchical clustering of expression identified 3 distinct subject groups with Low, Mid and High ISG expression. ISG accounting for the greatest variability in expression were identified (MX1, IFIT1, and ISG15) and used to analyze a prospective cohort at the Colorado ARDS Network site. One hundred twenty ARDS patients from four urban hospitals were enrolled within 72 hours of initiation of mechanical ventilation. Circulating neutrophils were isolated from patients and expression of ISG determined by PCR. Samples were stratified by standard deviation from the mean into High (n = 21), Mid, (n = 82) or Low (n = 17) ISG expression. Clinical outcomes were compared between patients with High or Low ISG expression to those with Mid-range expression. At enrollment, there were no differences in age, gender, co-existing medical conditions, or type of physiologic injury between cohorts. After adjusting for age, race, gender and BMI, patients with either High or Low ISG expression had significantly worse clinical outcomes than those in the Mid for number of 28-day ventilator- and ICU-free days (P = 0.0006 and 0.0004), as well as 90-day mortality and 90-day home with unassisted breathing (P = 0.02 and 0.004). These findings suggest extremes of ISG expression by circulating neutrophils from ARDS patients recovered early in the syndrome are associated with poorer clinical outcomes.

Conflict of interest statement

The authors have declared that no competing interests exist.

Figures

Fig 1. Transcriptome analysis to characterize ISG expression of ARDS patient and healthy control neutrophils.

(A) Log2-transformed expression of 31 unique transcripts within the Type I Interferon-mediated Signaling Pathway Cluster from neutrophils isolated from sepsis-induced ARDS patients (n = 31) and healthy volunteers (n = 19). ISG expression was ordered by hierarchical clustering (Euclidean distance with complete linkage), and contains only genes determined to significantly change between subjects. Three major clusters of subjects (columns) are broadly grouped as High ISG expression (left), Mid (middle) and Low (right). Subject groupings are represented by yellow (healthy) and blue (ARDS) blocks at the profile base. Only ARDS patients were contained within the Low ISG expression subject cluster. When these genes were ranked for extent of variance of expression using relative size of the standard deviation between subjects, the genes with the largest variance were identified as MX1, ISG15, IFIT1, and IFIT3 (identified by shading). (B) A Principal Component Analysis identified the same 4 genes that comprised nearly all of the variability between subjects in Panel A.

Fig 2. Classification of ARDS patients based on neutrophil ISG expression.

The mean of the transformed expression values of neutrophil MX1, IFIT1, and ISG15 is plotted for ARDS patients (triangle, n = 120) or healthy subjects (circle, n = 40). Subjects whose ISG expression was greater or less than one standard deviation from the mean were designated as High (red) or Low (blue) ISG expressers, respectively. ISG expression in healthy subjects overlapped only with the High and Mid cohorts of ARDS patients.

Fig 3. Kaplan-Meier Analysis of 90-Day Discharge to Home with Unassisted Breathing and Survival.

(A) Proportion of patients confirmed to be discharged to home with unassisted breathing within 90 days of study enrollment. Overall log-rank P-value comparing groups was 0.009. After adjusting for age, race, gender and BMI, rate of discharge to home was significantly lower for both the High ISG (red line, P = 0.006) and Low ISG (blue line, P = 0.004; p-values from Cox PH model) expressing cohorts when compared to the Mid (black line). (B) Proportion of patients confirmed to be dead within 90 days of study enrollment. Overall log-rank P-value comparing groups was 0.01. After adjusting for age, race, gender and BMI, rate of death was significantly worse for both the High ISG (red line, P = 0.02) and Low ISG (blue line, P = 0.02; p-values from Cox PH model) expressing cohorts when compared to the Mid (black line).

Fig 4. Association of ISG expression with confirmed viral infections and circulating IFNα.

(A) The presence of a viral infection, as determined either through available medical history or within the course of clinical care, did not reach significance in the cohort with High ISG expression (5 of 21) compared to the Mid (8 of 82) and Low (1 of 17) cohorts (Fisher’s exact test P = 0.23). (B) Detection of circulating IFNα tended to occur more frequently within the High ISG cohort (11 of 21) compared to the Mid (20 of 76) and Low (5 of 15) cohorts, but this difference did not reach significance (Fisher’s exact test P = 0.08). (C) Levels of circulating IFNα trended higher within the High ISG cohort (n = 21) compared to the Mid (n = 76), (Kruskal-Wallis overall P-value = 0.03; High vs. Mid pairwise P = 0.009). n = 15 for the Low cohort. Plot depicts range (minimum≈0, maximum = upper whisker), 1st to 3rd quartile (box) and median (line).