Cerebral inflammation is an underlying mechanism of early death in Alzheimer’s disease: a 13-year cause-specific multivariate mortality study

INTRODUCTION: Although Alzheimer’s disease (AD) is associated with early death, its life expectancy differs greatly between patients. A better understanding of this heterogeneity may reveal important disease mechanisms underlying the malignancy of AD. The aim of this study was to examine the relation between AD pathologies and early death in AD caused by dementia. METHODS: At a memory clinic, 247 referred consecutive patients with AD were monitored during 12.6 ± 1.6 years. Multivariate Cox regression analyses were performed with baseline measures of amyloid beta (Aβ) pathology (APOE genotype, cerebrospinal fluid (CSF) Aβ42) tau pathology (CSF phosphorylated tau and total tau), cerebrovascular pathology (white-matter lesions and CSF/serum albumin ratio), neuroinflammatory pathology (CSF soluble vascular cell adhesion molecule-1, sVCAM-1), frontal, temporal, and central brain atrophies, global cognition, sex, and age. Comorbidities and medications also were analyzed. All continuous variables were transformed to z scores to compare hazard ratios (HRs) and 95% confidence intervals (CIs). RESULTS: At follow-up, 89% of the patients had died. The mean survival time was 6.4 ± 3.0 years. The AD pathology that independently predicted an early death caused by dementia was cerebral inflammation (sVCAM-1; HR, 1.32; 95% CI, 1.07–1.64). Other independent predictors were lower global cognition (HR, 0.51; 95% CI, 0.43–0.61), frontal atrophy (HR, 1.38; 95% CI, 1.12–1.70), and medial temporal atrophy (HR, 1.23; 95% CI, 1.02–1.49). When examining death caused by dementia and related causes (vascular diseases and infections), age (HR, 1.23; 95% CI, 1.04–1.46) and cerebrovascular pathology (white-matter lesions: HR, 1.17; 95% CI, 1.01–1.36; and CSF/serum albumin ratio: HR, 1.16; 95% CI, 1.001–1.34) were also significant risk factors in addition to the previous variables. No comorbidity or medication was significant in the specific-cause models. CONCLUSIONS: This is the first study to link neuroinflammation independently to early death in AD and, hence, a rapidly progressing disease. Frontal and medial temporal atrophies and low cognition were also significant predictors. These are probably downstream biomarkers that reflect neuronal degeneration and late-stage disease. Our results suggest that inflammation, and not amyloid or tau pathology, is an independent underlying mechanism in the malignancy of AD.