Blood Pressure and Cerebral White Matter Share Common Genetic Factors in Mexican-Americans

Background and Purpose

Elevated blood pressure (BP) is a well-known risk factor for atrophy of cerebral white matter (WM) and this can lead to cognitive decline and disability in the elderly¹. BP and the integrity of cerebral WM are under strong genetic control, with up to 80% of individual variance explained by genetic factors^2–8. We hypothesized that the genetic factors responsible for elevation in BP were also responsible for decline in white WM integrity. This hypothesis was tested in a well-characterized population Mexican-Americans randomly selected from large extended families⁹. WM integrity was gauged as fractional anisotropy (FA) using Diffusion Tensor Imaging (DTI), a fully quantitative technique that is capable of ascertaining subtle decline in WM integrity¹⁰.

First, we investigated if shared genetic variance between BP and FA measurements was consistent with the pulse-wave encephalopathy (PWE) theory, as previously suggested^{5, 6}. PWE posits the direct and indirect biological effects to explain the regional pattern of BP-related decline in the integrity of cerebral WM^11–13. The direct PWE effect suggests that an increase in arterial pulse pressure (PP) can lead to gliosis of periventricular WM, due to mechanical damages associated with increased amplitude of CSF movement⁵. The indirect PWE effect suggests that an increase in systolic BP (SBP) can lead to focal gliosis in subcortical WM due to stenosis and loss of permeability of small cerebral blood vessels^{13, 14}. Next, we investigated if shared genetic variance between BP and FA was influenced by difference in WM development rates¹⁵. Prior studies suggested a connection between the rates of cerebral development and the genetic contribution to the variance of several brain structures^{16, 17}. Development of WM tracts associated with higher cognitive function follow a more protracted trajectory, with larger FA increases per year than sensory and motor tracts¹⁵. We therefore investigated if WM tracts that exhibit high maturation rates would be more susceptible to detrimental effects of elevated BP and would therefore show more genetic overlap with the BP measurements.

Methods

Results

WB-FA values showed a significant negative linear trend with both PP and SBP (r²=0.20 and 0.11, respectively, p<10⁻⁴) with no significant relationship observed between WB-FA and SBP (Figure S1, please see http://hyper.ahajournals.org). Quantitative genetic analyses estimated that over 50% of the intersubject variance for the whole-brain average FA (WB-FA), PP and SBP and 17% of the variance for the DBP were attributed to additive genetic factors (Table 2). Age and Age² were significant covariates for PP and WB-FA. Age was a sole significant covariate for the SBP. DBP did not have significant covariates. There were no significant (p<.05) age by sex or age² by sex interactions detected.

Genetic correlation analyses between BP and the WB-FA values reported that PP and SBP shared 36% (obtained from squaring ρ_G) and 22% of genetic variance with the WB-FA values, respectively (Table 3). The negative sign of these correlation coefficients suggested that the same genetic factors associated with higher PP and SBP were linked to progressively lower FA values. Genetic correlation between WB-FA and DBP was not significant (Table 3). The estimates of the confounding effects of the antihypertensive were obtained by excluding 44 subjects who were taking antihypertensive medications and by adding 10 and 5 mg Hg to the SBP and DBP values for the treated subjects. There was no significant difference in either sign or magnitude of genetic correlations (Table 3). Therefore, all further, regional analyses were performed in the full pedigree, using the uncorrected BP measurements.

Genetic correlation analyses between BP and the tract-wise FA values (Table 4) indicated that six out of ten WM tracts (GCC, BCC, SCC, FO, SLF and SS) showed a statistically significant (p≤0.05) genetic correlation with PP. Genetic correlation failed to reach statistical significance (0.05≤p≤0.10) for three more tracts: cingulum, CR and EC (Table 4). Genetic correlation coefficients for FA of four tracts (GCC, BCC, FO and SS) showed a statistically significant (p≤0.05) genetic correlation with SBP and approached significance (p≤0.10) for SCC (Table 4). There were no significant or suggestive genetic correlation coefficients observed for the genetic correlation coefficients with DBP (Table 4)

The plot of the genetic correlation coefficients (ρ_G) for ten WM tracts versus tract-wise rates of cerebral maturation, taken from²⁰ (Figure 1) demonstrate that WM bundles with higher rates of maturation shared a progressively higher genetic overlap with both PP and SBP. Linear regression analysis reported that this relationship was statistically significant for both PP and SBP (Pearson’s r²=.55 and 0.57; p≤0.01, respectively) (Figure 1) but not for DBP (Pearson’s r²= .23; p=0.16).

Discussion

This study in a large, well-characterized sample of Mexican-American participants in the San Antonio Family Heart Study (SAFHS)⁹ demonstrated that the arterial pulse pressure (PP) and systolic blood pressure (SBP) shared 36% and 22% of genetic variance with the whole-brain (WB) fractional anisotropy (FA) of white matter (WM), respectively. In the past, the integrity of cerebral WM was commonly assessed using the T2-hyperintense white matter (HWM) lesion imaging techniques^5–8. HWM lesions are the regions of accumulation of extra-cellular water due to focal degradation of the myelin sheath¹⁴ and their volume is an important neuroimaging marker of cerebral integrity²⁶. DTI has an advantage over HWM-imaging techniques because it is capable of ascertaining subtle WM damage that precedes formation of HWM lesions¹⁰. Several recent studies confirmed that DTI is a sensitive neuroimaging marker of white matter integrity in hypertensive individuals^{10, 27, 28}. Specifically, FA values were inversely correlated with the arterial pulse pressure PP and systolic BP in both normotensive and hypertensive individuals²⁷

The PP shared more genetic variance with the WB-FA values than SBP. This is consistent with the pulse-wave encephalopathy (PWE) theory^{12, 13}, which suggests that elevation in the arterial pulsativity can lead to cerebral injury even in normotensive individuals²⁷. Similar pattern was observed in a prior whole-genome linkage study in this population where PP and the volume of HWM lesion shared significant genetic loci, while the association between SBP and HWM lesion volume only reached suggestive signficance⁵. The regional patterns of genetic overlap between BP and FA were in-agreement with the direct and indirect mechanisms of BP-related cerebral injury as described by the PWE theory. The direct mechanism is primarily responsible for the gliosis of the periventricular WM^{12, 13}. This explains the high negative (ρ_{G =}−.43 to −.67) genetic correlation between PP and periventricular WM tracts. The highest genetic correlation was observed for the sagittal stratum (SS), which is located in the area that is especially susceptible to the periventricular gliosis²⁷. The indirect mechanism of the PWE theory is suggested to be responsible for formation of subcortical WM lesions, most commonly observed in the frontal lobe^{11, 27}. This may explain high genetic correlations observed between the SBP and the FA of the genu and the body of CC, as these tracts contain commissural fibers connecting frontal lobes. The biological pathway of the indirect mechanisms remains unclear. Stenosis of small cerebral vessels, especially in the watershed areas²⁹ and high metabolic demands of oligodendrocytes of associative WM contribute to a high vulnerability of the frontal WM to the indirect PWE mechanism³⁰. Oligodendrocytes of the associative frontal WM are among the most metabolically active cells in the adult CNS and therefore, are highly susceptible to damages from hypoxia³⁰. In addition, the oligodendrocytes of the associated, frontal WM tracts have reduced rates (per axonal-segment) of myelin production and repair³¹ and this is thought to be responsible for the protracted age-related myelination and sharp age-related decline²⁰.

Previous findings suggested that the protracted development and high metabolic demands of associative WM make it more vulnerable to age-related neurodegeneration than sensory and motor WM^{19, 27, 32}. In agreement with these findings, our results showed that WM tracts that continue to meylinate into adulthood and therefore show higher age-related maturation rates shared progressively more genetic overlap with PP and SBP. In fact, over 50% of regional variability in shared genetic variance was explained by the regional rates of cerebral maturation. We interpret this finding as the evidence for genotype-by-age interaction with protracted development of associative WM contributing to its higher susceptibility to the neurodegeneration associated with elevated PP and SBP.

Unlike PP and SBP, significant genetic correlations were not observed between DBP and FA. However, given the lower observed heritability of DBP, the current study is relatively underpowered and therefore can fail to identify genetic correlations for this pair of phenotypes. The statistical power of the current study may also have been limited to detect significant genetic correlation for track-wise measurements of FA, where the magnitude of genetic correlation coefficients for several analyses approached statistical significance. Therefore, the lack of significant genetic correlations cannot be interpreted as evidence against that those pleiotropic relationships exist. Additionally, our study examined Mexican Americans, a population with significant Native American admixture. If relatively rare variants are involved in the determination of quantitative variance, we may expect considerable differences in the degree of shared genetic variance in other populations, such as European Americans³³.

Perspective

Our findings in a population characterized by an adverse cardiovascular risk profile demonstrated that genetic factors responsible for elevation in arterial PP and SBP were also responsible for declining integrity of cerebral WM. In agreement with PWE theory, the highest genetic association with WM integrity was observed for PP. Further, our data demonstrated that associative WM tracts that facilitate high-order cognitive functioning showed higher vulnerability to the elevated PP and SBP than motor and sensory WM tracts. The statistical power of this study was not sufficient to localize the individual genes responsible for the pleiotropy between BP and FA but previous studies in this population provide a likely candidate: selectin genes^{2, 5, 6}. The region harboring the constellation of selectin genes (SELP, SELL, and SELE) has been identified as a region of significant linkage (LOD=3.82) between PP and HWM volume. In particular, the adhesion molecule P-selectin³⁴ is a marker of potential endothelial dysfunction that has been implicated as a risk factor in essential hypertension^{35, 36} and stroke^{35, 37}. Increased blood levels of P-selectin have been implicated in formation of atherosclerotic plaque, loss of vascular reactivity and increase in arterial pulsativity and SBP^{35, 37}. In addition, platelet-derived gene expression levels of SELP were shown to be strongly and positively correlated with arterial BP ^{38, 39}

Limitations

Our data indicates that a genotype-by-age interaction is potentially responsible for genetic overlap between BP and FA. Hence, it may be useful to explicitly allow for the potential influences of genotype by age interactions. While advanced statistical genetic methods for family based data allow for the formal detection of such interactions within cross-sectional data, longitudinal family studies will have much greater power to localize and ultimately identify the specific genes involved in intersubject differences in the rates of WM atrophy. Therefore, further research is needed to confirm the cross-sectional trends observed here using a longitudinal design.

A potential limitation of this study is that the collection of BP measurements preceded the acquisition of brain images by 3.0±0.8 (maximum = 5.3) years. Individual subjects could have experienced a rise in BP during the period between BP and brain assessment suggesting caution when interpreting these findings.

Supplementary Material