Chronic cerebral infarctions and white matter lesions link to long-term survival after a first ischemic event: A cohort study

Background and purpose: To investigate the association of different phenotypes, count, and locations of chronic covert brain infarctions (CBI) with long-term mortality in patients with first-ever manifest acute ischemic stroke (AIS) or transient ischemic attack (TIA). Additionally, to analyze their potential interaction with white matter hyperintensities (WMH) and predictive value in addition to established mortality scores.

Methods: Single-center cohort study including consecutive patients with first-ever AIS or TIA with available MRI imaging from January 2015 to December 2017. Blinded raters adjudicated CBI phenotypes and WMH (age-related white matter changes score) according to established definitions. We compared Cox regression models including prespecified established predictors of mortality using Harrell's C and likelihood ratio tests.

Results: A total of 2236 patients (median [interquartile range] age: 71 [59-80] years, 43% female, National Institutes of Health Stroke Scale: 2 [1-6], median follow-up: 1436 days, 21% death during follow-up) were included. Increasing WMH (per point adjusted Hazard Ratio [aHR] = 1.29 [1.14-1.45]), but not CBI (aHR = 1.21 [0.99-1.49]), were independently associated with mortality. Neither CBI phenotype, count, nor location was associated with mortality and there was no multiplicative interaction between CBI and WMH (p > .1). As compared to patients without CBI or WMH, patients with moderate or severe WMH and additional CBI had the highest hazards of death (aHR = 1.62 [1.23-2.13]). The Cox regression model including CBI and WMH had a small but significant increment in Harrell's C when compared to the model including 14 clinical variables (0.831 vs. 0.827, p < .001).

Discussion: WMH represent a strong surrogate biomarker of long-term mortality in first-ever manifest AIS or TIA patients. CBI phenotypes, count, and location seem less relevant. Incorporation of CBI and WMH slightly improves predictive capacity of established risk scores.