An IL-18-centered inflammatory network as a biomarker for cerebral white matter injury

Abstract

Chronic systemic sterile inflammation is implicated in the pathogenesis of cerebrovascular disease and white matter injury. Non-invasive blood markers for risk stratification and dissection of inflammatory molecular substrates in vivo are lacking. We sought to identify whether an interconnected network of inflammatory biomarkers centered on IL-18 and all previously associated with white matter lesions could detect overt and antecedent white matter changes in two populations at risk for cerebral small vessel disease. In a cohort of 167 older adults (mean age: 76, SD 7.1, 83 females) that completed a cognitive battery, physical examination, and blood draw in parallel with MR imaging including DTI, we measured cerebral white matter hyperintensities (WMH) and free water (FW). Concurrently, serum levels of a biologic network of inflammation molecules including MPO, GDF-15, RAGE, ST2, IL-18, and MCP-1 were measured. The ability of a log-transformed population mean-adjusted inflammatory composite score (ICS) to associate with MR variables was demonstrated in an age and total intracranial volume adjusted model. In this cohort, ICS was significantly associated with WMH (β = 0.222, p = 0.013), FW (β = 0.3, p = 0.01), and with the number of vascular risk factor diagnoses (r = 0.36, p<0.001). In a second cohort of 131 subjects presenting for the evaluation of acute neurologic deficits concerning for stroke, we used serum levels of 11 inflammatory biomarkers in an unbiased principal component analysis which identified a single factor significantly associated with WMH. This single factor was strongly correlated with the six component ICS identified in the first cohort and was associated with WMH in a generalized linear regression model adjusted for age and gender (p = 0.027) but not acute stroke. A network of inflammatory molecules driven by IL-18 is associated with overt and antecedent white matter injury resulting from cerebrovascular disease and may be a promising peripheral biomarker for vascular white matter injury.

Fig 1. Imaging and fluid analysis workflows in the MarkVCID and ASPIRE cohorts.

Workflow diagram of the MarkVCID cohort of 167 subjects that underwent detailed cognitive evaluations, MRI including DTI, and serum collections (left). Workflow diagram of the ASPIRE cohort of 202 subjects presenting with acute neurologic symptoms who underwent MRI and concurrent serum collection (right).

Fig 2. ICS components form an inflammatory network.

STRING database query of the six ICS component analytes reveals a biologically interconnected network centered on IL-18 and highly related to inflammation (p-value for protein interactions = 0.00022). ICS component analytes are shown as colored nodes (bold) while first level interacting proteins are shown as white nodes. Line width reflects the strength of data support.

Fig 3. ICS correlates with MRI measures of cerebrovascular injury.

Heatmap of z-scores for each of the individual analytes composing the ICS for each included subject in the MarkVCID cohort ordered left to right by ICS (average z-score of each analyte) (A). Scatter plot and regression line of logWMH vs. ICS (n = 110) (B). Scatter plot and regression line of free water vs. ICS (n = 49) (C). Red dashed lines indicate 95% confidence intervals.

Fig 4. Association of ICS with overt and antecedent white matter injury.

Average intensity maps of free water (FW) and frequency maps of white matter hyperintensities (WMH) of groups with low (upper) and high (middle) ICS groups dichotomized around median ICS. Lower panels illustrate voxel differences in FW and WMH between low and high ICS groups.

Fig 5. ICS increases with vascular risk factors.

Mean ICS in groups with one or more vascular risk factor diagnoses (red) vs. those with less vascular risk factor diagnoses (black). All group comparisons were statistically significant at adjusted p<0.008 except between those with 6 vascular risk factor diagnoses (n = 2).

Fig 6. ICS is associated with white matter injury in those at risk for stroke.

Scree plot of principal components analysis of data from ASPIRE cohort serum biomarker panel with two main factors (Factor 1 and Factor 2) driving variance (A). Scatter plot of Factor 1 values versus ICS in this cohort demonstrating a significant correlation (r = 0.94) (B). Scatter plot and regression line of modified Fazekas score and ICS for individual subjects (C). Red dashed lines indicate 95% confidence intervals. Representative T2/FLAIR MR images of ASPIRE subjects with low (left) or high (right) ICS scores (D).