There is a complex relationship between cerebrovascular disease (CVD) and cognition. Many studies investigate the relationship between one subgroup of CVD (i.e., clinical/imaging evidence of prior ischemic stroke, imaging evidence of small vessel disease, etc.) and diagnoses of cognitive impairment/dementia or scores on cognitive assessments. As different measures of CVD burden are independently associated with impaired cognition, it is reasonable to hypothesize that these abnormalities could have an additive negative impact on cognition in patients who have multiple types of CVD. Few studies have evaluated the combined effects of different types of CVD pathology, representing a significant gap in the understanding of vascular cognitive impairment given that many patients have more than one type of CVD. Xu et al. sought to investigate the relationship between different CVD markers and both global and domain-specific cognition as assessed by neurocognitive testing.

The authors conducted a case-control study of patients 50 years of age or older recruited from two memory clinics in Singapore who were screened by their performance on a locally validated neuropsychological battery (Vascular Dementia Battery, VDB). Eligible subjects were those who scored 1.5 or more standard deviations below education-adjusted norms. Controls recruited from the community and had no impairment on the Vascular Dementia Battery. A total of 399 subjects were categorized into one of five groups:

1) Non-vascular Cognitive Impairment-No Dementia (CIND), defined as impairment in at least one domain on the VDB that did not meet DSM-IV criteria of dementia who had no documented history of stroke or TIA (n=100)

2) Vascular CIND, who met criteria of CIND as above but had a positive history of stroke or TIA (n=62)
3) AD, as per the NINCDS-ADRDA criteria (n=116)
4) VaD, as per the NINCDS-AIREN criteria (n=27)
5) Controls who had no cognitive impairment on the VDB (n=94)

The participants had a 3T MRI, with markers of CVD categorized as binary variables recorded according to the presence or absence of multiple cortical infarcts, intracranial stenosis, multiple lacunar infarctions, moderate/severe white matter hyperintensities (WMH) and multiple microbleeds. Each subject’s cognitive status was assessed using a modified protocol based on the NINDS-CSN Vascular Cognitive Impairment battery, assessing working memory, executive function, language, visuomotor speed, visuospatial function and recall and recognition. 


Correlations between CVD markers were reported as phi coefficients, and independent associations between individual CVD markers and global and domain specific performance were assessed using general linear models, controlling for age, gender, education, ethnicity and other vascular risk factors. Analyses of variance were employed to relate different CVD burden scores with global and domain-based cognitive performance.

Moderate to severe white matter hyperintensities were independently associated with lower global cognition (p<0.001) as well as poorer performance on all individual cognitive domains. Multiple microbleeds were associated with poor performance on visuomotor speed (p<0.008), visuoconstruction (p<0.008), executive function (p<0.008) and language (p<0.008). No other single CVD marker was associated with cognitive impairment, but the presence of 2 or more of any of the CVD measures was associated with poorer global cognitive performance (p<0.01), independent of WMH. Based on these findings, the authors generated a weighted CVD burden score with two points awarded for the presence of moderate/severe WMH and one point when at least 2 of the other CVD markers were present. Subjects with a CVD burden index of 2-3 had poorer global and domain specific performance as compared to those with a CVD burden index score of 0-1. Subjects with a CVD burden score of 3 showed particularly poor relative performance in visuoconstruction.

The conclusion that WMH have a significant effect on global cognition spanning multiple domains is consistent with previous studies, theoretically due to the diffuse pattern of involvement seen in moderate/severe disease causing disruption between cortical/subcortical connections in multiple areas. Interestingly, the two CVD markers with the strongest associations (i.e., WMH and microbleeds) are both pathologic sequelae of Cerebral Amyloid Angiopathy, which is linked to cognitive impairment in both the presence and absence of concomitant AD pathology. Given this, it is also important to consider that both of these markers represent sequelae related to a diffuse process, in contrast to the focal nature of cortical and lacunar infarction. With an understanding of the regional anatomic basis of the different domains of cognition, future studies investigating the effects of cortical and lacunar infarction on specific cognitive domains could potentially benefit by grouping subjects based on the location of infarction.

This report represents an important initial step in recognizing and assessing the cumulative heterogeneous effects of varying types of cerebrovascular pathology on cognition. External validation including samples drawn from multiple centers, larger sample sizes, appropriate age and education matched controls and evaluation of changes over time is essential to assess the usefulness of the proposed CVD burden index.