Molecular biomarkers for vascular cognitive impairment and dementia

As disease-specific interventions for dementia are being developed, the ability to identify the underlying pathology and dementia subtypes is increasingly important. Vascular cognitive impairment and dementia (VCID) is the second most common cause of dementia after Alzheimer disease, but progress in identifying molecular biomarkers for accurate diagnosis of VCID has been relatively limited. In this Review, we examine the roles of large and small vessel disease in VCID, considering the underlying pathophysiological processes that lead to vascular brain injury, including atherosclerosis, arteriolosclerosis, ischaemic injury, haemorrhage, hypoperfusion, endothelial dysfunction, blood–brain barrier breakdown, inflammation, oxidative stress, hypoxia, and neuronal and glial degeneration. We consider the key molecules in these processes, including proteins and peptides, metabolites, lipids and circulating RNA, and consider their potential as molecular biomarkers alone and in combination. We also discuss the challenges in translating the promise of these biomarkers into clinical application. Vascular cognitive impairment and dementia is the second most common cause of dementia after Alzheimer disease. In this Review, the authors examine the potential of key molecules in the pathophysiology as biomarkers of vascular cognitive impairment and dementia and consider the challenges of clinical translation. Key points VCID has multiple underlying pathologies that can contribute to cognitive impairment. Identification of molecular biomarkers that can differentiate VCID from healthy ageing and Alzheimer disease remains challenging. Multiple molecular biomarkers have been associated with VCID, but none has yet been translated into clinical application. The heterogeneity and complexity of VCID means that use of multiple biomarkers in combination tends to be necessary. Promising biomarkers related to various pathophysiological pathways could be combined into panels to optimize sensitivity and specificity; machine learning could be useful for constructing these panels.