Within a limited range of vascular burden, we found no evidence of an effect of vascular risk on the longitudinal change of CSF Aβ42
, FDG uptake in typical AD-related ROIs, and MRI hippocampal atrophy during cognitive decline. CSF Aβ42
is an amyloid-specific marker for AD, and its decline appears to be faster early in the disease course11
; however, none of these vascular proxy measures was associated with CSF Aβ42
either cross-sectionally or longitudinally. Likewise, MRI hippocampal volume serves as a sensitive surrogate marker for AD pathology,18
but neither cardiovascular risk profile nor WMH was clearly associated with hippocampal volume or its rate of atrophy. Although there was a suggestion of a relationship between WMH and longitudinal hippocampal atrophy in the NC group, the lack of a relationship between WMH change and hippocampal change argues against its significance. Our findings are consistent with previous studies that cerebrovascular burden, at least of mild degree, and AD pathology may be 2 independent factors contributing to dementia.10,19–21
Vascular factors not only increase the risk of AD but also predict dementia progression in AD.22
Their contribution is considered additive in augmenting the expression of AD.23
Animal models have shown increased amyloid deposition after stroke, and white matter injury may predispose to the neurodegenerative effects of Aβ.9
Although the neurovascular hypotheses concerning Aβ clearance seem plausible,24
little evidence from human studies suggests that vascular risks are also amyloidogenic.25
Even though amyloid deposition can be enhanced by circulatory defects, vascular effects may be easily overwhelmed once AD pathology becomes advanced.20
However, it is possible that vascular risks play an initiating role, and, therefore, observation of their influence on amyloid accumulation may require longer study periods.
The significant association between higher baseline WMH volume and lower FDG uptake in the AD group may reflect the fact that vascular burden contributes to reduced synaptic activity in patients with AD. However, the rate of glucose hypometabolism was not influenced by vascular risks in MCI or AD, arguing against the interaction between vascular risks and AD pathologic progression. We used the average normalized intensity from 5 FDG ROIs including bilateral temporal and angular gyri and posterior cingulate gyrus, which were commonly affected in AD. Although an earlier study demonstrated that WMH was associated with frontoparietal metabolism in FDG-PET rather than our ROIs,26
the underlying neuropathology for these regions is still undetermined. FDG-PET generally represents synaptic activity,27
and ROIs may be sensitive to AD pathologic changes. However, FDG ROIs may be not specific enough to exclude vascular or other pathologic contributions. Our finding in the NC group that a greater amount of WMH was associated with faster decline of FDG ROI uptake is in line with this notion. These cognitively normal participants are not necessarily going to develop AD, and they may be considerably different from participants with MCI and AD in ADNI. Therefore, the significant associations between FDG uptake in PET and WMH volume in our study may reflect the general effect of vascular burden on synaptic activity rather than AD-specific pathology.
Age has long been considered a prominent predictor of WMH severity,28
and here with repeated measures we have further shown that after adjustment for age, WMH volume increased significantly over time and the rate of increase appeared to be faster for participants with cognitive impairment. Although vascular risks such as hypertension are important factors for WMH severity,29
we cannot exclude the contribution from other pathologic conditions. A voxel-based morphometric MRI study has shown that gray matter reduction is correlated with the increase in WMH volume30
; however, the temporal relationship between these findings is unknown. People with MCI or AD may have accelerated gray matter or cortical atrophy and thus develop faster WMH progression. Without further investigation of the underlying neuropathology of WMH, we cannot be sure that WMH change simply represents the progression of vascular burden.
4 is a strong genetic risk factor for AD and predictive of cognitive decline.31
As a key component in the transport of cholesterol and lipid, APOE
4 plays a role in both coronary risk and cerebral amyloid deposition.32,33
We did not find any association between APOE
4 and cardiovascular risk score or WMH. The relationship between AD biomarkers and vascular burden was not affected by the presence of APOE
4 either. Conversely, APOE
4 has been shown to be associated with lower baseline CSF Aβ42
, FDG uptake, and accelerated hippocampal atrophy in an earlier ADNI study.11
4 seems to contribute to AD via the amyloid pathway more than through increasing vascular burden.
Greater cardiovascular risks and WMH volume were, as expected, associated with worse cognitive performance, particularly executive function in the MCI group, consistent with previous studies, which showed that WMH was more associated with psychomotor speed or executive function.26,34
We also found that although cognitive performance on MMSE and ADAS-cog, typical measures for AD, was not influenced by changes in WMH volume, the decline in executive function significantly correlated with the increase of WMH volume in the NC group. This finding suggested that vascular burden and AD-type pathology target different cognitive domains in aging and dementia.
There are several limitations in our study. First, vascular burden in the study was generally low because enrollment criteria excluded people with Hachinski Ischemic Scale scores of 4 or greater. As a result, the range of vascular burden in ADNI was narrow, and our findings cannot be generalized to populations with more than mild vascular risks. Furthermore, our length of observation ranged from 2 to 5 years. The influence of mild vascular burden may not be detectable within the period of observation. Nevertheless, this limitation provides us a unique opportunity to evaluate mild vascular effects that are uncomplicated by comorbidities, motor and sensory alterations, and other signs and symptoms that might increase measurement error, especially for cognition. Another limitation is that we do not have pathologic data showing the concordance between these AD-type biomarker changes and the severity of amyloid deposition. How specific these AD biomarkers are to detect the amyloid pathologic cascade at each cognitive stage in contrast with vascular progression is not known. Cardiovascular risk profile was derived from hypertension, diabetes, and other factors measured only once at baseline, and many of them may not have been quantified with adequate precision. Although the cardiovascular risk score was normally distributed, it predicts future cardiovascular events but is not necessarily reflective of the underlying vascular pathology. Outcomes could also have been affected by treatment of hypertension or other vascular risk factors, and the degree of medical control was not considered in our analyses. These treatments presumably change over time, depending on the previous outcome and also affecting the follow-up predictor as well as outcome. Handling of these time-varying confounders is beyond the scope of ADNI data, and we are not sure how the lack of handling of these confounders would affect our results.
The unique strength of the study is its longitudinal setting and repeated measurement to capture the dynamics of AD biomarkers and WMHs. Time-varying predictors and outcome are rarely available in population studies; therefore, the assumption that an age effect is uniform across different participants is not as necessary as in cross-sectional studies. In addition, the majority of participants had 3 or more repeated measures, allowing us to evaluate not only the single difference between 2 time points but also the variance of change. With more than 3 or 4 repeated measures, the regression toward the mean effect can be further minimized.
The longitudinal changes in AD biomarkers were not modified by mild vascular risks during cognitive decline in ADNI. There is no evidence that cerebral amyloid deposition is affected by mild vascular burden. Vascular contribution to AD dementia is probably additive although not necessarily independent of the amyloid pathway.