Cerebral white matter hyperintensities indicate severity and progression of coronary artery calcification

Cerebral white matter hyperintensities (WMH) have been associated with subclinical atherosclerosis including coronary artery calcification (CAC). However, previous studies on this association are limited by only cross-sectional analysis. We aimed to explore the relationship between WMH and CAC in elderly individuals both cross-sectionally and longitudinally. The study population consisted of elderly stroke- and dementia-free participants from the community-based Austrian Stroke Prevention Family Study (ASPFS). WMH volume and CAC levels (via Agatston score) were analyzed at baseline and after a 6-year follow-up period. Of 324 study participants (median age: 68 years), 115 underwent follow-up. Baseline WMH volume (median: 4.1 cm3) positively correlated with baseline CAC levels in multivariable analysis correcting for common vascular risk factors (p = 0.010). While baseline CAC levels were not predictive for WMH progression (p = 0.447), baseline WMH volume was associated CAC progression (median Agatston score progression: 27) in multivariable analysis (ß = 66.3 ± 22.3 [per cm3], p = 0.004). Ten of 11 participants (91%) with severe WMH (Fazekas Scale: 3) at baseline showed significant CAC progression > 100 during follow-up. In this community-based cohort of elderly individuals, WMH were associated with CAC and predictive of its progression over a 6-year follow-up. Screening for coronary artery disease might be considered in people with more severe WMH.


Follow-up
115 participants underwent a median long-term follow-up of 5.8 years (range: 5.2-6.4years) including repeated brain MRI and CT of the coronary arteries.Vascular events occurred in 14 individuals during the follow-up period (12.1%; myocardial infarction: n = 7, stroke: n = 2, TIA: n = 5).

Discussion
In this cohort of stroke-and dementia-free elderly people, WMH volume on brain MRI was associated with calcification of the coronary arteries at baseline and predictive for its progression over long-term follow-up.Individuals with severe WMH, as indicated by a Fazekas score of 3, are at high risk for a substantial progression of CAC.
Our results are of interest as previous studies led to speculations about the mechanisms behind the association of cerebral WMH-a hallmark feature of cerebral small vessel disease-and (subclinical) large vessel atherosclerosis such as CAC [8][9][10][11] .A combination of shared classical vascular risk factors (predominantly arterial hypertension) and a genetic predisposition was suspected behind this phenomenon [8][9][10][11] .In this context, Johansen et al. identified differences in the strengths of the relationship between CAC and different WMH subtypes 9 .The more pronounced association observed between CAC and PVWMH, as opposed to DWMH, might be attributed to the typical vascular architecture of PVWMH-related short penetrating microvessels.These vessels may be more directly affected by arterial hypertension compared to the longer microvessels that supply the deep white www.nature.com/scientificreports/matter 14,15 .Atherosclerotic changes in the penetrating branches of the large intracranial arteries might lead to hypoperfusion and ischemia, representing a key mechanism underlying WMH development 16 .In addition, similar genetic variations were identified in patients with calcification of coronary arteries and a high burden of PVWMH 17 .
In our analysis, PVWMH also showed a stronger correlation with baseline CAC levels (in multivariable analysis), but the difference between both WMH subtypes was rather small in absolute numbers.Future studies might analyze ultrastructural white matter changes using diffusion tensor imaging (DTI) 18 and their association with macrovascular disease to improve the pathophysiological understanding of the presented association.
The unique longitudinal design of this study further allowed us to investigate potential associations between CAC as marker of (subclinical) atherosclerotic large vessel disease and WMH over a long-term follow-up period of 6 years.Apart from genetics and shared vascular risk factors, previous studies indicated a more direct link explaining the association between CAC and WMH.As atherosclerotic large vessel disease results in the stiffening of arteries, it is hypothesized that the subsequent increase in blood-flow pulsatility within the brain-supplying vessels directly damages the small cerebral vasculature, thereby promoting WMH 9,12,13 .For this reason, CAC was assumed to predict cerebral WMH progression, but previous studies only had cross-sectional data available [8][9][10][11] .
Our results do not support this hypothesis as we did not identify an association between CAC levels at baseline and WMH progression (neither total WMH nor WMH subtypes) during the follow-up period.Moreover, even participants with severe calcification of the coronary arteries at baseline (CAC score > 400) did not show a more pronounced WMH progression.
These results align with recently published data that failed to establish an association between intracranial pulsatility, measured in the middle cerebral artery, and both WMH volume at baseline and WMH progression during follow-up 19 .
Most notably, we found that baseline WMH volume predicted CAC progression during the follow-up period.Specifically, ten out of eleven participants with severe baseline WMH, as indicated by a Fazekas score of 3, showed substantial CAC progression of > 100.PVWMH and DWMH had a similar predictive value for CAC progression.Apart from a shared genetic predisposition, classical vascular risk factors (i.e., arterial hypertension) might be important factors behind this relation.In this context, all participants with severe baseline WMH (Fazekas 3)  and CAC progression during follow-up also had underlying arterial hypertension.Our results therefore point towards a subgroup of WMH patients, in which intense and continuous vascular risk factor control might be crucial to avoid further damage of the brain, but also to avoid macrovascular changes in other vascular beds such as the coronary arteries.Moreover, treating physicians should be aware of signs and symptoms of cardiac disease in patients with high WMH load and initiate cardiological exploration or even coronary artery disease screening in case of any clinical suspicion.

Limitations
Our study is limited by the fact that only a subgroup of study participants (27%) underwent follow-up brain MRI and coronary artery CT (n = 115).However, this should not have influenced our results to a relevant extent as there were no differences in demographics, vascular risk factors and CAC or WMH volume at baseline between patients with and those without follow-up (p > 0.1, data not shown).Based on the small-sized follow-up cohort, we cannot exclude that we have overlooked a (small) predictive value of baseline CAC levels on WMH progression.We also only observed few outcome events in this study not allowing to report on the predictive value of WMH, CAC and associated vascular events, which should be addressed in larger prospective studies.

Conclusions
This study reinforces the correlation between cerebral WMH and large artery atherosclerosis.Moreover, WMH serve as predictors for the progression of coronary artery disease during a long-term follow-up period.Clinicians should be aware of this observed association and may consider to screen individuals with severe WMH for coronary artery disease.Intense control of vascular risk factors is essential for all such patients.

Materials and methods
All methods were performed in accordance with the relevant guidelines and regulations.
The study was approved by the ethics committee of the Medical University of Graz (Approval number: 17-088 ex 05/06).Written informed consent was obtained by all included study participants.

Selection of participants and data collection
All included study participants derive from the Austrian Stroke Prevention Family Study (ASPFS), an extension of the Austrian Stroke Prevention Study (ASPS).ASPFS is a prospective population-based study that was designed to assess the effects of vascular risk factors on brain structure, function and vessel damage in different vascular beds 20 .Between 2006 and 2013, study participants of the original study-ASPS-and their first-degree relatives were invited to enter the ASPFS study.Inclusion criteria included the absence of a history of cerebrovascular disease (stroke or transient ischemic attack) or dementia, as well as a normal neurological examination.This is a single-center study.All participants were recruited at the University Hospital of Graz, Austria.
Baseline and follow-up assessments comprised blood pressure measurements, laboratory tests of vascular risk factors (such as blood glucose and serum lipid levels), a clinical evaluation of comorbidities as well as coronary artery computed tomography (CT) and brain magnetic resonance imaging (MRI).Vascular risk factors at baseline were defined according to latest guideline recommendations based on documented parameters (hypertension: systolic blood pressure > 140 mmHg, diastolic blood pressure > 90 mmHg; diabetes: fasting plasma glucose > 126 mg/dl; dyslipidemia: LDL cholesterol ≥ 116 mg/l) or if patients were already treated with respective medication (such as antihypertensives or statins) [21][22][23] .The guideline-based initiation of dedicated treatments was always verified by the study team.During the follow-up period, the same thresholds for increased blood pressure, hyperglycemia and dyslipidemia were used to identify patients with poor vascular risk factor control [21][22][23] .
All MRI scans were reviewed by blinded neuroradiological experts (C.E., R.S.).WMH, lacunes and (chronic) cortical infarcts were assessed on T2 and FLAIR images.WMH severity was rated according to the Fazekas rating scale in deep and periventricular locations 24 .For quantitative assessment, WMH areas were first segmented manually and consecutively added to a total lesion volume using FMRIB Software Library (FMRIB, Oxford, UK; freely available at https:// fsl.fmrib.ox.ac.uk) 25 .

Coronary artery calcification
Participants underwent cardiac CT with a 64-channel multidetector computed tomography (GE Imatron, San Francisco, USA) at baseline and follow-up.ECG triggering was used at 80% of the cardiac cycle to obtain images (slice thickness: 3 mm; image acquisition time: 100 ms).CAC was defined as a minimum of three contiguous pixels with a CT density ≥ 130 Hounsfield units.An experienced radiologist (GR) specialized in cardiac CT imaging and blinded to clinical data read all the CT images and calculated CAC scores according to the Agatston method 26 .

Patient and public involvement
Patients or the public were not involved in the design, conduct, reporting or dissemination plans of our research.

Statistics
Statistical analyses were performed using IBM SPSS Statistics, version 28.Aside from analyzing WMH volumes and CAC levels as continuous variables, WMH were dichotomized using the Fazekas visual grading scale scores 0-1 (i.e., no or mild WMH) versus 2-3 (i.e., moderate to severe WMH)) 27 .If WMH in either deep (DWMH) or periventricular (PVWMH) location were graded as Fazekas 2 or higher, they were classified as moderate to severe WMH.In a second step, participants were divided according to CAC score severity (quartile 4 versus quartiles 1-3).Pearson's chi-square or Fisher's exact test was used to compare dichotomous variables.All quantitative variables were first tested for Gaussian distribution with the Kolmogorov-Smirnov test and, if Gaussian distribution was identified, a two-sample independent t-test was utilized to compare the variables.The Mann-Whitney-U-Test was used for non-parametric data.
As WMH volumes and CAC levels at baseline and follow-up were not normally distributed, Spearman's rank correlation was performed for bivariable correlations including these parameters.A p-value less than 0.05 was considered statistically significant.
A multivariable linear regression model was fitted to identify factors that were independently associated with WMH volume and CAC levels at baseline.
Besides age and sex, the model included variables that were related to baseline WMH (target variable) and baseline CAC volume in univariable analysis (p < 0.05): arterial hypertension, diabetes and glomerular filtration rate (GFR).However, after testing for multicollinearity and interactions, GFR was removed from the multivariable analysis because of its strong correlation with age (variance inflation index > 10).
In a second step, the same model was again calculated with DWMH and PVWMH as the target variables (instead of total WMH volume) to test for the influence of different WMH subtypes.

Figure 1 .
Figure 1.Flow diagram of included study participants.

Table 1 .
Demographic and clinical data of ASPFS participants according to qualitative and quantitative WMH burden at baseline.MRI: magnetic resonance imaging; ASPFS: Austrian stroke prevention family study; WMH:

Table 2 .
Multivariable linear regression analysis with baseline WMH load (in cm 3 ) as the target variable.

Table 3 .
1emographic and clinical data of ASPFS participants according to CAC progression.MRI: magnetic resonance imaging; ASPFS: Austrian stroke prevention family study; WMH: white matter hyperintensities; SD: standard deviation; TIA: transient ischemic attack; MCI: myocardial infarction; LDL: low density lipoprotein; FU: follow-up.1Demonstratedp-value was determined by comparing participants with moderate/severe WMH to those with no/mild WMH. a Follow-up was available in 115 participants.

Table 4 .
Predictors of CAC progression in multivariable linear regression analysis.WMH: white matter hyperintensities; PVWMH: periventricular white matter hyperintensities, DWMH: deep white matter hyperintensities; SE: standard error; MCA: MRI: magnetic resonance imaging.a Multivariable linear regression model was recalculated with both WMH subtype volumes (PVWMH, DWMH) instead of total WMH volume.Correlations between Agatston score and WMH volume at baseline and their progression over a long-term follow-up period of 6 years.