The term mild cognitive impairment (MCI) broadly defines an intermediate state of cognitive decline, predominately linked to impaired memory function, which is thought to be predictive of dementia, primarily AD. Various definitions have been proposed in the literature, each with differences in focus (for example, age-associated change versus pathological decline) and diagnostic criteria (for example, memory versus non-memory impairment) 1011121314. As a possible tool for identifying individuals at increased risk of dementia, MCI is an important concept. Indeed, in clinical samples, individuals with a case diagnosis of the anmestic subtype of MCI (A-MCI) 151617 have been found to progress to dementia at a rate of 10% to 15% per year compared with a progression rate of only 1% to 2% in normal controls 1819. In contrast, in the general population, the positive predictive validity of A-MCI is poor 19. Many incident dementia cases are found to be excluded from an A-MCI case diagnosis, and of persons with A-MCI, many remain stable or revert to normal cognitive status at follow-up 10. These findings are consistent across all MCI definitions (for example, amnestic versus non-amnestic and single-versus multiple-domain MCI) 19.

No MCI criteria can be recommended for population screening of individuals at high dementia risk 20. This raises a number of questions regarding the precision and utility of current diagnostic criteria and what the best indicators of dementia risk are and whether these are being captured in current diagnostic methods. Poor predictability possibly results from limitations in case findings due to a lack of clinical judgement and inflexibility in operationalisation of criteria when a diagnosis of MCI is made outside the clinical setting. However, it has been suggested that MCI predictability may be improved through consideration of the underlying pathogenesis of cognitive decline 21. Subclassification of MCI with and without co-morbid vascular disease may therefore be important for discriminating individuals at high versus low dementia risk in the general population.

Although the fact is not always explicitly stated in MCI-defining criteria, an MCI case diagnosis is usually made following the exclusion of individuals with psychiatric and vascular co-morbidity 101222. As such, the association between vascular disease and MCI is not widely studied. Where vascular disease has been evaluated in the context of MCI, no clear pattern between vascular disease and incident MCI or between vascular disease, MCI and risk of dementia progression has been found. An increased risk of incident MCI has been associated with elevated midlife blood pressure 23, elevated total cholesterol level 23, history of coronary artery bypass grafting 24, stroke 25 and midlife hypertension 21 in some, but not other 26, studies. Only atrial fibrillation and low folate levels have been associated with an increased risk of dementia progression from MCI 27. Inconsistencies in conclusions possibly result from differences in patient sources (for example, clinic-versus population-based), type of impairment (for example, narrow [A-MCI] versus global measures of cognitive impairment as captured in the term 'cognitive impairment, no dementia' [CIND]), definitions of disease and measurement of vascular factors over varying time frames and with instruments of different sensitivities (for example, subjective report versus objective measure). To better identify the link between vascular disease and cognitive impairment, the term vascular cognitive impairment (VCI) was introduced 92829303132.

VCI refers to cognitive decline attributable to vascular disease. However, unlike MCI (which is a narrow term capturing a pre-clinical form of dementia), VCI encompasses individuals affected with any degree of cognitive decline caused by or associated with vascular disease and its risk factors. As such, the level of impairment in VCI ranges in severity from mild to vascular dementia (VaD) or mixed VaD, in which cerebrovascular and AD pathologies co-occur. Calls for more specific staging have recently led to further subclassification of VCI to capture vascular disease-related impairment not fulfilling criteria for dementia. This stage is defined using the term VCIND and has been further subdivided to include specific terms for pre-clinical impairment, including vascular MCI (V-MCI), MCI of vascular type, pre-clinical VCI, vascular pre-dementia MCI and mild VCI (M-VCI/MCI-vas) 29. Each is analogous to the concept of MCI in terms of stage. However, similar to MCI, VCIND and its component states lack consistent standardised indicators (for example, symptoms) and a unique case definition.

Whether within VCIND there is a state predictive of dementia is largely unknown. Commonly, VCIND is not seen as a pre-clinical dementia state. Where longitudinal outcome across the spectrum of VCI has been investigated, progression is not always clinical (that is, decline/dementia), with many cases improving or remaining stable at follow-up 9333435. Predictive ability may depend on the nature of the vascular disturbance in addition to methodological factors (for example, case description, sample and nature of cognitive impairment). As with MCI, terminology and diagnostic criteria for VCIND have not been harmonised, making cross-study comparison of disease outcomes difficult. Yet unlike MCI, in which algorithms have been created for each different classification 10, no such algorithm yet exists for VCIND and its many possible subtypes.

Neuropsychological studies have identified attentional-executive deficits and psychomotor slowing, with relatively preserved language and recognition memory in individuals with vascular disease 4748. However, not all studies agree on the importance of each cognitive domain and no single deficit or pattern of deficits as yet accurately signals an underlying vascular cause 49. This is not unexpected given the multiplicity of aetiologies for vascular disease and the fact that the pattern and extent of cognitive deficits would likely reflect not only disease type, but also its severity 5051. For example, cognitive impairment as a consequence of stroke would likely depend on not only timing and the anatomical location of the stroke, but also the laterality, severity and extent of the lesion. Furthermore, impairments in attention and executive and motor function are not necessarily restricted to vascular causes of dementia and have also been associated with Lewy body dementia 52 and fronto-temporal dementia 53.

Where VCIND has been followed longitudinally, cognitive impairment associated with memory (free and cued recall) and category fluency was found to predict risk of incident dementia 54. These findings suggest that the pattern of impairment in VCIND conforms more to AD than to VaD. Indeed, almost half of the cases progressed to AD or mixed AD at 5 years of follow-up. However, whether these findings extend across the different vascular causes of VCIND (that is, stroke-versus hypertension-related cognitive impairment) and other cognitive domains (that is, perception and motor performance) remains to be tested.

With regard to MCI, no consistent cognitive profile exists across the many different case definitions. The focus of MCI is predominantly impaired memory, but deficits in other cognitive domains will also be observed when, by definition, they are also included 55. Subdivisions between different cognitive subtypes of MCI (for example, amnestic versus non-amnestic and single-versus multi-domain) have implications for inference about aetiology and outcomes. Indeed, while A-MCI 151656 is thought to be a precursor of AD, non-amnestic subtypes of MCI have been found to identify individuals at high risk of both AD and VaD 57.

Where the cognitive profile of individuals with MCI and co-morbid vascular disease has been compared with that of individuals with MCI and no vascular disease, group differences have been reported in some 5859, but not all 60, studies. Where differences have been observed, the MCI vascular group shows more extensive cognitive impairment primarily in speed, attention and executive function 58, consistent with the general pattern of cognitive difficulties resulting from vascular disease alone 29. In other studies, vascular pathology in MCI (for example, white matter lesions) has been associated with decreased risk of progression of cognitive pathology and a more stable cognitive change profile 6162, although this is not always replicated 63. These findings suggest a complex relationship between vascular disease and cognitive progression in MCI which might relate to the specific vascular disease factor. Of importance is determining whether the presence of vascular disease accelerates or intensifies the cognitive disturbance in all cases of MCI and which factors might mediate this relationship (for example, age and reserve). The specific type of cognitive impairment associated with vascular disease needs to be defined and measures that are sensitive, specific and appropriate for longitudinal and observational assessment of cognition in the context of vascular disease (that is, memory versus non-memory domains) need to be identified in order to facilitate the development of diagnostic criteria for cognitive decline in the presence (VCIND) versus the absence (MCI) of vascular disease.

Neuroimaging in VCIND shows a pattern of vascular lesions that are similar to, but less severe than, changes observed in VaD 64. Pathology includes evidence of leukoaraiosis and white matter infarction 286566, with mild hippocampal and enthorhinal cortex atrophy relative to the level seen in MCI/AD 64. In contrast, neuroimaging in MCI generally shows a pattern of changes similar to that observed in AD, namely temporal and hippocampal atrophy, reduction in whole-brain glucose metabolism and white matter degeneration, including hyperintensities and white matter lesions identified using diffusion tensor imaging 6768697071727374. Development of dementia from MCI has been associated with hippocampal volume changes 697576777879, medial temporal lobe atrophy 80 and metabolic alteration 7881.

Although neuroimaging studies of VCIND and MCI suggest different pathological processes, findings are not always consistent and such changes are imperfect predictors of disease. Considerable intra-individual variation exists, and overall, neuroimaging-identified abnormalities correlate poorly with cognitive profile 43. Indeed, such changes have also been observed in individuals who do not exhibit cognitive deficits, raising questions about the uniqueness of findings 82. Inconsistency in results possibly arises due to differences in the use of subjective visual rating scales to assess the extent of pathology across groups, regional focus of disease (global versus focal), in addition to differences in enrolment criteria and the type and severity of vascular disease across imaging cohorts. Indeed, different vascular disease factors are associated with varying types and levels of pathology: hypertension has been associated with reduced cerebral blood flow 83 and an increased risk of periventricular white matter lesions 8485; lower arterial oxygen saturation and chronic obstructive pulmonary disease have been associated with cerebral white matter lesions, but not lacunar infarcts 86; diabetes has been associated with cortical and hippocampal atrophy 8788, white matter lesions 89 and lacunar infarcts 89; and current smoking status, diabetes and hypertension are associated with both neurodegenerative (that is, decreased brain volume) and vascular (that is, lacunar infarcts and white matter lesions) changes 90.

The severity and type of lesions required for a diagnosis of MCI and VCIND remain controversial. Vascular disease and its risks are associated with brain changes but the clinical relevance of such changes in the prediction of cognitive decline and dementia progression remains uncertain. Isolating unique disease effects from the effects of ageing and other risk factors (that is, genetic susceptibility) will be important in determining cellular/molecular/functional vulnerability as a consequence of vascular disease as well as establishing with accuracy those changes that distinguish who will and will not develop cognitive decline and subsequent dementia.

The neuropathological profile of MCI has been derived mainly from a relatively small number of studies with MCI defined predominately using the A-MCI subtype. Compared with highly selected controls, MCI cases generally show an increase in neurofibrillary tangle pathology in memory-related cortical regions, including the entorhinal cortex, fusiform gyrus and temporal pole 9192. These changes are thought to represent one of the earliest pathological substrates of this condition 93 and have been taken to suggest that many MCI cases are early or prodromal AD 9495. Vascular pathology has also been reported in MCI such that the neuropathology of some cases includes features associated with both AD and VaD 96. However, there is considerable heterogeneity in findings and not all individuals with MCI at death or those who progress from MCI to dementia have been reported to show any particular neuropathology 97.

Rather, they have been indistinguishable from control groups. Inconsistency in outcome may arise from differences in study population (for example, specialised clinic versus population), age group differences (that is, young-old versus old-old), operational definition of MCI and neuropathological criteria (for example, Khachaturian, Consortium to Establish a Registry for Alzheimer's Disease, or National Institute on Aging-Reagan).

Whether there is a consistent neuropathological profile across the spectrum of vascular causes and severity levels of VCI is unknown but seems unlikely. Indeed, VCI is a multi-factor disorder related to a wide variety of lesions and causes and as such the pathological profile, similarly to the psychological and radiological profiles, would be expected to be heterogeneous. In autopsy studies, an increased prevalence of cerebral vascular pathology has been found in individuals with stroke 98, diabetes mellitus 99100, angina (with co-morbid dementia) 101102 and hypertension 103. Pathological features have included large- and small-vessel disease, gliosis, microvascular brain damage (severe cribri-form change), white matter damage, microinfarction and haemorrhage 104. Cardiovascular risk factors have also been associated with AD-like neuropathological lesion formation in some, but not all, studies, with the extent of pathology typically being more severe in APOE (apolipoprotein E) e4 carriers 100103105106107. In contrast, in other cases, an inverse association between vascular disease and the extent of cerebral degenerative pathology has been found 101. The profile of pathology across the different vascular disease factors is heterogeneous and the significance of such changes in the development of cognitive impairment is not known. Furthermore, neuropathological associations appear to be risk factor-specific and population-specific, being absent when vascular disease is assessed using composite cerebrovascular index scores and in non-Caucasian populations 108109.

Across the spectrum of age-associated brain changes, no neuropathological profile yet exists that reliably distinguishes impairment of different severity levels and causes. In the general population, currently identifiable pathological features have not been found to correlate well with observed clinical and cognitive profiles: many non-demented healthy controls also show evidence of pathological brain changes associated with both AD and VaD 43110111. Techniques that better characterise the impact of vascular disease on brain structure and more sensitive measures for accurately staging cognitive status which incorporate known risk factors are needed for diagnostic differentiation between an at-risk and a not at-risk brain. However, as with AD, expecting neuropathology to be a gold standard at any given age for the diagnosis of VCI is an oversimplification 112.