Abstract
Background:
Few prospective cohort studies collect detailed information on stroke characteristics among individuals who experience ischemic stroke, including white matter hyperintensity volume, and thus cannot explore how prospectively collected biomarkers prior to the stroke influence white matter hyperintensity volume. We explored the association between a large panel of prospectively collected lipid and inflammatory biomarkers and white matter hyperintensity volume among participants in the Women’s Health Study with incident ischemic stroke.
Methods:
Among Women’s Health Study participants with first ischemic stroke who had baseline serum biomarkers and available magnetic resonance imaging, we measured white matter hyperintensity volume using a validated semi-automated method. Linear regression was used to explore the associations between biomarkers and log-transformed white matter hyperintensity volume.
Results:
After multivariate adjustment, a 1% increment in HbA1c% was associated with an increase in white matter hyperintensity volume (p-value=0.05). Evidence of a non-linear association between high density lipoprotein cholesterol levels and ApoA1 levels with white matter hyperintensity volume was noted (p-values for non-linearity=0.01 and 0001 respectively). No other biomarkers were significantly associated with white matter hyperintensity volume.
Conclusion:
Chronic hyperglycemia as evidenced by HbA1c levels measured years prior to stroke is associated with white matter hyperintensity volume at the time of stroke. Additional research is needed to explain why low levels of high density lipoprotein cholesterol levels and ApoA1 may be associated with similar white matter hyperintensity volume as high levels.
Keywords: white matter hyperintensity volume, biomarkers, stroke, epidemiology
Introduction
Larger white matter hyperintensity volume (WMHV) among individuals with ischemic stroke has been associated with increased risk of ischemic stroke recurrence(1) and poor outcomes after stroke.(2, 3) Understanding risk factors assessed prior to stroke which are associated with WMHV in ischemic stroke patients may identify strategies to reduce the overall burden of stroke-related morbidity.
One potential set of risk factors for WMHV may be cardiovascular disease biomarkers including cholesterol measures and inflammatory markers. Research from population-based stroke-free cohorts has observed associations between WMHV and several biomarkers including homocysteine(4), circulating interleukin-6(5), C-reactive protein (CRP) (5), fibrinogen(6), sICAM(7), cholesterol measures(6, 8–10), and Apo A1(10, 11).
However, few prospective cohort studies collect detailed information on stroke characteristics, including white matter hyperintensity volume (WMHV), among individuals who experience incident stroke events. Therefore they are unable to explore how prospectively collected biomarkers years prior to the stroke event influence WHMV. This is an important gap in knowledge because white matter hyperintensities (WMH) develop over time, so it may be that biomarker levels years before the stroke event are more relevant to WMH development and total WMHV than biomarker levels at the time of the stroke event. Thus far, only homocysteine and HbA1c measured at the time of the stroke event have been identified as risk factors for WMHV in ischemic stroke patients.(12–14) Some studies among individuals with stroke have also suggested that a history of hyperlipidemia may be associated with lower WMHV(15, 16) and that higher levels of fibrinogen or lipoprotein(a) at the time of the stroke event may be associated with WMH severity.(13)
We leveraged data from the Women’s Health Study (WHS), a prospective study of initially healthy women, to prospectively explore the association between biomarkers measured years prior to stroke and WMHV among individuals who experience an incident ischemic stroke.
Materials and Methods
Study Population
Participants in this study were enrolled in the WHS and experienced an incident ischemic stroke event during the course of the study. Briefly, the WHS was a randomized, double-blind, placebo-controlled trial among 39,876 female health care professional aged ≥45 years to test the effects of low-dose aspirin and vitamin E in the primary prevention of cardiovascular disease and cancer.(17, 18) Women were randomized from 1992 to 1994. The active treatment phase ended in 2004 and observational follow-up is currently ongoing. Data were available for this analysis through June 30, 2018. Prior to randomization, yearly during the trial, and biennially thereafter, WHS participants were mailed a questionnaire asking about demographic information, lifestyle characteristics, and health information.
Biomarker Ascertainment
Prior to randomization, women who were willing to provide a fasting venous blood sample were mailed a blood collection bit containing EDTA tubes. Of the 39,876 women enrolled in the WHS at baseline, 28,345 provided a blood sample prior to randomization. The baseline characteristics of women who provided a blood sample and those who did not were similar although those who provided a blood samples were more likely to be physically active and less likely to smoke.(19) Plasma samples were analyzed for lipids and a panel of inflammatory biomarkers as described previously in a core laboratory certified by the National Heart, Lung, and Blood Institute/Centers for Disease Control and Prevention Lipid Standardization Program.(20) The following biomarkers were studied in our analyses: total cholesterol, high density lipoprotein cholesterol (HDL-C), low density lipoprotein cholesterol (LDL-C), the ratio of total cholesterol to HDL-C, Apo A1, Apo B, triglycerides, Lp(a), CRP, fibrinogen, homocysteine, s-ICAM, HbA1c, and creatinine.
Ascertainment of Stroke
The mailed questionnaires asked the participants whether they had been diagnosed with a stroke by a physician since the last questionnaire. In the event that a woman reported a stroke on her questionnaire, an Endpoints Committee of physicians reviewed medical records to confirm or disconfirm the case according to the Trial of Org 10172 in Acute Stroke Treatment (TOAST) criteria. (21) The committee also classified confirmed events as ischemic, hemorrhagic or unknown stroke subtype.
WMHV Assessment from MRI
For each confirmed ischemic stroke since January 1, 2011, we requested permission to obtain electronic copies of any MRI performed as part of clinical care for the stroke event. We measured WMHV on the hemisphere contralateral to the cerebral infarct as observed on T2 FLAIR images using a semiautomated method which had been previously developed and validated in stroke patients.(22–26) Briefly, a single investigator (P.M.R.) reviewed the T2 FLAIR scans to exclude hyperintensities not caused by WMH such as those related to acute or chronic infarcts. Total WMHV was calculated by doubling WMHV from the hemisphere contralateral to the cerebral event. To correct WMHV for head size, we multiplied WMHV by the ratio of an individual’s sagittal midline cross-sectional intracranial volume (ICA) to mean ICA.(27, 28)
Statistical Analysis
Normalized white matter hyperintensity volume (nWMHV) exhibited a log-normal distribution and was log-transformed for all analyses. Since one participant had no WMHV visible, we added 1 to all WMHVs prior to performing the log transformation. We constructed histograms of each biomarker on their original scales and histograms of the natural log transformed (ln) biomarker values to evaluate normality. For biomarkers that were normally distributed, including total cholesterol, LDL-C, Apo B, s-ICAM, HbA1c, and creatinine, we constructed scatterplots of the biomarker on its original scale versus ln nWMHV. For biomarkers that demonstrated non-normality, including HDL-C, the ratio of total cholesterol to HDL-C, Apo A1, triglycerides, Lp(a), CRP, fibrinogen, and homocysteine, we constructed scatterplots of ln biomarker values versus ln nWMHV. We used the Jonckheere-Tepstra test to compare the median values of each biomarker across quartiles of nWMHV. We used linear regression to examine the association between these biomarkers (on the original or log transformed scales) and ln nWMHV adjusting for age at randomization. To explore potential non-linear associations between the biomarker and WMHV, we included a quadratic term for each biomarker in our age-adjusted models. In the event that the quadratic term was statistically significant, we further explored the association between the biomarker and WMHV using restricted cubic splines with four knots.
In addition to adjusting for age, we constructed multivariable models which adjusted for a number of variables measured at randomization, including: history of hypertension (yes/no), treatment for high blood pressure (yes/no), hormone use (yes/no), menopausal status (premenopausal, postmenopausal, or uncertain), smoking status (never, past, current), alcohol consumption (rarely/never, 1–3 drinks/month, 1–6 drinks/week, ≥1 drink/day), body mass index (<25 kg/m2, 25-<30kg/m2, ≥30 kg/m2), and physical activity (<200 kcal/week, 200-<600 kcal/week, ≥600 kcal/week).
Results
Between January 1, 2011 and June 30, 2018, 242 first ischemic stroke events were confirmed among women with a blood sample. Of these, 188 indicated in the medical record that an MRI was performed; MRIs were obtained from 151 of these events and WMHV measured in 146 participants. A comparison of the baseline characteristics of those with measurements of WMHV versus those without measurements of WMHV can be seen in Supplemental Table 1. Those without measurements of WMHV were more likely to be post-menopausal and were slightly older at randomization and at the time of their stroke event.
The median value for nWMHV among participants was 9.79 cm3 (25th percentile=4.75; 75th percentile=15.84). Table 1 shows the baseline characteristics of the participants overall and by quartile of nWMHV. The median age of the participants at baseline was 55.3 years (25th percentile=51.1; 75th percentile=62.1) and the median age of the participants at the time the MRI was performed was 75.3 years (25th percentile=71.1; 75th percentile=82.6). Those in the lowest quartiles of nWMHV were the youngest at randomization and at the time the MRI was performed.
Table 1.
Baseline characteristics of individual included in these analyses overall and by quartile of normalized WMHV (N=146).
| Overall | Quartile of normalized WMHV | |||||
|---|---|---|---|---|---|---|
| Characteristic at baseline | N (%) | Quartile 1 (N=37) | Quartile 2 (N=36) | Quartile 3 (N=37) | Quartile 4 (N=36) | p-value for trend |
| WMHV, median (IQR) | 9.8 (4.7, 15.8) | 2.9 (2.1, 4.0) | 7.2 (5.6, 7.9) | 13.6 (11.2, 14.6) | 26.7 (20.5, 49.5) | |
| Age at randomization, median (IQR) | 55.3 (51.1, 62.1) | 51.7 (47.8, 55.9) | 55.1 (52.3, 60.3) | 56.0 (52.6, 62.6) | 59.8 (51.5, 66.6) | <0.01 |
| Age at MRI scan, median (IQR) | 75.3 (71.1, 82.6) | 71.8 (68.7, 76.2) | 75.2 (72.7, 82.0) | 76.1 (73.1, 82.9) | 80.2 (72.0, 86.3) | <0.01 |
| Baseline treatment for hypertension | 23 (15.8) | 5 (13.5) | 5 (13.9) | 7 (18.9) | 6 (16.7) | 0.59 |
| Hx of hypertension | 41 (28.1) | 10 (27.0) | 6 (16.7) | 12 (32.4) | 13 (36.1) | 0.20 |
| No current use of postmenopausal hormones | 75 (51.4) | 23 (62.2) | 14 (38.9) | 19 (51.4) | 19 (52.8) | 0.66 |
| Menopausal status | 0.10 | |||||
| Premenopausal | 30 (20.6) | 12 (32.4) | 6 (16.7) | 8 (21.6) | 4 (11.1) | |
| Postmenopausal | 85 (58.2) | 20 (54.1) | 19 (52.8) | 23 (62.2) | 23 (63.9) | |
| Uncertain/unsure | 31 (21.2) | 5 (13.5) | 11 (30.6) | 6 (16.2) | 9 (25.0) | |
| Smoking status | 0.98 | |||||
| Never | 84 (57.5) | 22 (59.5) | 20 (55.6) | 21 (56.8) | 21 (58.3) | |
| Past | 51 (34.9) | 11 (29.7) | 15 (41.7) | 12 (32.4) | 13 (36.1) | |
| Current | 11 (7.5) | 4 (10.8) | 1 (2.8) | 4 (10.8) | 2 (5.6) | |
| Alcohol consumption | 0.15 | |||||
| Rarely/never | 71 (48.6) | 19 (51.4) | 11 (30.6) | 21 (56.8) | 20 (55.6) | |
| 1–3 drinks/month | 18 (12.3) | 5 (13.5) | 3 (8.3) | 7 (18.9) | 3 (8.3) | |
| 1–6 drinks/week | 38 (26.0) | 8 (21.6) | 11 (30.6) | 8 (21.6) | 11 (30.6) | |
| 1+ drinks/day | 19 (13.0) | 5 (13.5) | 11 (30.6) | 1 (2.7) | 2 (5.6) | |
| Body Mass Index | 0.60 | |||||
| Normal weight | 62 (42.5) | 17 (46.0) | 16 (44.4) | 14 (37.8) | 15 (41.7) | |
| Overweight | 62 (42.5) | 13 (35.1) | 17 (47.2) | 19 (48.7) | 14 (38.9) | |
| Obese | 22 (15.1) | 7 (18.9) | 3 (8.3) | 5 (13.5) | 7 (19.4) | |
| Physical activity | 0.65 | |||||
| <200 kcal/wk | 103 (0.6) | 26 (70.3) | 26 (72.2) | 26 (70.3) | 25 (69.4) | |
| 200–<600 kcal/wk | 20 (13.7) | 9 (24.3) | 3 (8.3) | 4 (10.8) | 4 (11.1) | |
| >=600 kcal/wk | 23 (15.8) | 2 (5.4) | 7 (19.4) | 7 (18.9) | 7 (19.4) | |
| Hx of diabetes | 1 (0.68) | 1 (2.7) | 0 | 0 | 0 | 0.18 |
| Hx of high chol | 52 (35.6) | 12 (32.4) | 14 (38.9) | 13 (35.1) | 13 (36.1) | 0.83 |
IQR=interquartile range (25th and 75th percentiles); SD=standard deviation; WMHV=white matter hyperintensity volume
Cutpoints for quartiles of normalized WMHV: 4.75, 9.79, 15.84
Table 2 shows the median and 25th and 75th percentiles of each biomarker by quartile of nWMHV. Levels of ApoA1, ApoB, fibrinogen, homocysteine, Lp(a), and s-ICAM could only be measured in 142 participants. HbA1c% were available for 144 participants; we further excluded 3 participants whose HbA1c% levels were above 6.5% and the one participant with diabetes at baseline.
Table 2.
Median and 25th and 75th percentiles of each biomarker by quartile of normalized WMHV (N=146)
| Quartile of normalized WMHV* | |||||
|---|---|---|---|---|---|
| Quartile 1 (N=37) | Quartile 2 (N=36) | Quartile 3 (N=37) | Quartile 4 (N=36) | p-value** | |
| Total Cholesterol mg/dL | 201.0 (185.0, 234.0) | 217.0 (187.0, 239.0) | 207.0 (184.0, 237.0) | 206.0 (183.0, 228.5) | 0.87 |
| HDL-C mg/dL | 47.0 (37.9, 63.8) | 50.8 (39.4, 63.5) | 46.6 (42.4, 58.1) | 52.7 (44.3, 59.9) | 0.37 |
| LDL-C mg/dL | 119.5 (104.8, 130.4) | 124.3 (98.3, 147.5) | 125.2 (102.4, 136.5) | 121.4 (97.5, 133.3) | 0.90 |
| Total cholesterol to HDL-C ratio | 3.8 (3.2, 5.6) | 4.0 (3.1, 5.3) | 4.4 (3.6, 5.1) | 3.7 (3.4, 4.7) | 0.52 |
| Apo A1 | 136.6 (124.5, 162.1) | 150.5 (129.8, 169.6) | 147.3 (130.5, 167.7) | 157.0 (139.4, 176.4) | 0.04 |
| Apo B | 108.4 (81.8, 122.3) | 106.4 (89.7, 123.0), | 106.2 (92.9, 128.8) | 98.6 (87.7, 121.5) | 0.60 |
| Triglycerides mg/dL | 133.0 (93.0, 181.0) | 119.5 (83.0, 232.0) | 163.0 (89.0, 194.0) | 139.5 (87.0, 189.5) | 0.93 |
| Lp(a) | 11.1 (5.1, 43.0) | 8.4 (4.9, 33.7) | 10.7 (5.4, 25.9) | 9.0 (3.8, 52.8) | 0.86 |
| CRP mg/L | 1.8 (0.7, 4.6) | 2.9 (1.1, 5.2) | 3.4 (1.7, 5.0) | 3.8 (2.0, 6.7) | 0.03 |
| Fibrinogen | 340.2 (296.8, 382.3) | 328.2 (309.2, 389.0) | 356.5 (310.7, 377.3) | 363.6 (326.2, 421.5) | 0.07 |
| Homocysteine | 9.8 (8.3, 13.2) | 10.3 (8.1, 13.1) | 10.6 (8.9, 13.7) | 10.4 (8.6, 12.5) | 0.44 |
| s-ICAM ng/ml | 344.9 (318.2, 416.2) | 339.4 (299.9, 363.7) | 345.3 (309.8, 404.5) | 354.8 (314.2, 410.7) | 0.78 |
| HbA1c % | 4.9 (4.8, 5.1) | 4.9 (4.8, 5.1) | 5.0 (4.9, 5.1) | 5.1 (4.9, 5.2) | 0.003 |
| Creatinine mg/dL | 0.7 (0.6, 0.8) | 0.7 (0.6, 0.8) | 0.7 (0.7, 0.9) | 0.7 (0.6, 0.8) | 0.96 |
Cutpoints for quartiles of normalized WMHV: 4.75, 9.79, 15.84
p-value is obtained from the Jonckheere-Tepstra test
We observed a significant trend in median values across quartiles of nWMHV for ApoA1, CRP, and HbA1c% and borderline significant trends for fibrinogen. For all of these biomarkers, the median value of the biomarker was the lowest in the first quartile and highest in the fourth quartile.
Based on the histograms, we modeled total cholesterol, LDL-C, ApoB, creatinine, s-ICAM, and HbA1c% on their original scales. We log-transformed the values of HDL-C, cholesterol ratio, ApoA1, CRP, fibrinogen, homocysteine, lipoprotein(a), and triglycerides for all models. Table 3 shows the associations between each biomarkers and log-transformed nWMHV. We observed no significant linear associations between our biomarkers and WMHV with the exception of HbA1c%. After adjusting for age, a 1% increment in HbA1c% was associated with a significant increase in ln nWMHV (p-value=0.04). This increase was similar in magnitude but of borderline significance after adjusting for other potential confounders (p-value=0.05).
Table 3.
Age-adjusted and multivariate*-adjusted associations between each biomarker and natural log transformed (ln) normalized white matter hyperintensity volume
| Age-adjusted model | Multivariate* adjusted model | ||||
|---|---|---|---|---|---|
| Biomarker | N | Beta (SE) | p-value | Beta (SE) | p-value |
| Total Cholesterol mg/dL | 146 | −0.0003 (0.0019) | 0.88 | −0.001 (0.002) | 0.70 |
| Ln HDL-C mg/dL | 146 | 0.19 (0.26) | 0.47 | 0.35 (0.31) | 0.27 |
| LDL-C mg/dL | 146 | 0.0004 (0.0024) | 0.88 | −0.0003 (0.0025) | 0.90 |
| Ln Total chol to HDL-C ratio | 146 | −0.16 (0.23) | 0.49 | −0.31 (0.27) | 0.24 |
| Ln Apo A1 | 142 | 0.44 (0.41) | 0.28 | 0.57 (0.46) | 0.22 |
| Apo B | 142 | −0.002 (0.002) | 0.34 | −0.003 (0.003) | 0.34 |
| Ln Triglycerides mg/dL | 146 | −0.14 (0.13) | 0.30 | −0.23 (0.15) | 0.14 |
| Ln Lp(a) | 142 | −0.03 (0.06) | 0.59 | −0.02 (0.06) | 0.67 |
| Ln CRP mg/L | 146 | 0.09 (0.07) | 0.18 | 0.07 (0.08) | 0.36 |
| Ln Fibrinogen | 142 | 0.30 (0.35) | 0.38 | 0.10 (0.39) | 0.80 |
| Ln Homocysteine | 142 | −0.02 (0.21) | 0.91 | 0.19 (0.22) | 0.41 |
| s-ICAM ng/ml | 142 | −0.0004 (0.0010) | 0.67 | −0.0008 (0.0012) | 0.52 |
| HbA1c % | 140 | 0.65 (0.32) | 0.04 | 0.66 (0.34) | 0.05 |
| Creatinine mg/dL | 146 | −0.20 (0.40) | 0.61 | −0.21 (0.42) | 0.61 |
Multivariate models are adjusted for age, history of hypertension, treatment for high blood pressure, hormone use, menopausal status, smoking status, alcohol consumption, body mass index, and physical activity.
We also used quadratic terms to explore potential non-linear associations between our biomarkers and WMHV. The quadratic terms for ln HDL-C, ln ApoA1, and HbA1c% were statistically significant (p-value=0.003, p-value=0.004, and p-value=0.04 respectively) so we further explored the shape of this association using splines. Splines showed an inverse U-shape between ln HDL-C levels and ln nWMHV (Figure 1), between ln ApoA1 levels and ln nWMHV (Figure 2), and between HbA1c% and ln nWMHV (Figure 3). After adjusting for age and potential confounders, tests for non-linearity indicated that the association between ln HDL-C levels and ln nWMHV (p-value=0.01) as well as between ln ApoA1 and ln nWMHV (p-value=0.001) were non-linear, even on the log scale. However, after adjusting for age and potential confounders, we did not find evidence of a non-linear association between HbA1c% and ln nWMHV (p-value=0.25). In post-hoc analyses, to further explore the shape of the association between HbA1c% and ln nWMHV, we categorized HbA1c% into the following groups: ≤4.75% (reference), 4.76–5.00%, 5.01–5.25%, and >5.25%. Compared to women with HbA1c% ≤4.75%, all other categories were significantly associated with increased ln nWMHV (Supplemental Table 2).
Figure 1.
Restricted cubic spline for the relationship between log transformed HDL-C and log normalized white matter hyperintensity volume.
Figure 2.
Restricted cubic spline for the relationship between log transformed ApoA1 and log normalized white matter hyperintensity volume.
Figure 3.
Restricted cubic spline for the relationship between HbA1c% and log normalized white matter hyperintensity volume.
Discussion
We observed a linear association between HbA1c% and WMHV suggesting that increased HbA1c% values were associated with higher WMHV a median of 20.4 years after measuring HbA1c%. The association between HDL-C and ApoA1 and WMHV appeared to have an inverse U-shape with the lowest WMHV being observed at the lower and upper ends of the HDL-C and ApoA1 distributions. No other cardiovascular disease biomarkers were associated with WMHV in this study.
A previous study among stroke patients which measured HbA1c% at hospital admission for stroke observed that increased HbA1c% was associated with increased WMHV.(12, 14) Our study expands upon this finding by demonstrating that HbA1c% levels measured a median of 20 years prior to the stroke event are associated with WMHV, suggesting that higher blood sugars – even in the absence of diabetes mellitus diagnosis - may have long term effects on WMH development. In addition, the participants in the current study had lower levels of HbA1c% than prior stroke cohorts and we excluded participants with HbA1c% values ≥6.5%. This demonstrates that relatively and chronically elevated blood glucose levels as reflected in HbA1c%, may be an important risk factor for WMH development. The exact mechanisms by which chronic exposure to relatively elevated blood glucose levels increases WMHV are unknown but it has been suggested that, as a surrogate marker of insulin resistance and chronic endothelial dysfunction, HbA1c% may affect WMHV progression through the dimethylarginine (ADMA) pathway.(29, 30) Increased ADMA is also associated with carotid artery atherosclerosis,(31) which may explain the association between HbA1c% and incident CVD.(32–34)
Prior studies in stroke cohorts have not examined measured levels of HDL-C or ApoA1 and WMHV. Instead, studies in stroke cohorts have observed associations between hyperlipidemia, defined as serum cholesterol >200 mg/dL, serum triglycerides >150 mg/dL, physician diagnosis, or use of medication to control hyperlipidemia and decreased WMHV.(15, 16, 35) In contrast, we observed no association between total cholesterol or triglycerides and WMHV and instead found a suggestion of a non-linear, inverse U-shaped association between HDL-C and ApoA1, a component of HDL-C, and WMHV. Other studies among healthy individuals (i.e. those without stroke) have examined the association between HDL-C or ApoA1 and WMHV. Evidence from healthy cohorts is inconsistent, with some studies observing no association between either HDL-C and ApoA1 and WMHV(36, 37), an inverse association for both HDL-C and ApoA1 among women only(10), or a positive association between HDL-C and WMH severity.(9) Higher levels of HDL-C are considered to be protective against the risk of ischemic stroke,(38) which may explain why we observed a decrease in WMHV for those with the highest levels of HDL-C. Our observation of a potential non-linear association between HDL-C or ApoA1 and WMHV is novel and will require further research to determine if this finding can be replicated in other cohorts and to explain the underlying biology. In addition, our biomarkers were measured in the early 1990s, prior to the widespread use of statins for cardiovascular disease prevention. It is possible that higher incidence of statin use over time among those with elevated HDL-C levels compared to those with normal HDL-C levels may have masked the true association between HDL-C or ApoA1 levels and WMHV. We performed sensitivity analyses in which we controlled for statin use after baseline but still observed evidence of an inverse U-shaped association between HDL-C or ApoA1 and WMHV.
Prior studies in stroke cohorts have also observed that elevated homocysteine levels as measured at hospital admission have been consistently associated with increased WMHV.(14, 25) We observed no significant associations between homocysteine measured years prior to the stroke event and WMHV in our cohort. However, our cohort is significantly smaller than previous stroke cohorts and our analyses may have been underpowered. In addition, our measurements of homocysteine were a median of 20 years prior to the stroke event.
Strengths of this study include the prospective design which allowed to measure biomarkers prior to the stroke event and avoid the potential for the stroke event itself to impact biomarker levels. In addition, the prospectively design of the study allowed us to control not only for several health conditions, but also for lifestyle factors such as physical activity and alcohol consumption which are often not captured in hospital-based cohorts and which may influence biomarker levels.
However, some important limitations of the study should be noted. Lipid levels were only measured at baseline and we were unable to explore whether changes in lipid levels are associated with WMHV. Our cohort only enrolled women so we were not able to explore whether there are sex differences in the association of cardiovascular disease biomarkers with WMHV. Our study was also smaller than many previous hospital-based studies of WMHV in ischemic stroke patients so we had limited power to detect small associations between the biomarkers and WMHV. We were unable to obtain MRIs on all ischemic stroke events which occurred in our cohort which may bias our results if biomarker levels and WMHV both impact the availability of MRIs. Additionally, we assumed that WMHV in the hemisphere contralateral to the infarction was a valid proxy for WMHV in the hemisphere with the infarct. Although this approach has been validated(22–26), it may result in some measurement error.
Summary and Conclusion
This is the first study to explore the association between biomarkers measured a median of 20 years prior to the incident stroke event and WMHV in a prospective cohort of previously healthy women with detailed lifestyle and cardiovascular health data. The ability to explore long-term effects of biomarkers on WMHV represents a novel contribution to the field and an important approach because WMHV may develop over the course of several years. Examining levels of biomarkers years prior to the event may offer novel insights into the pathological processes that lead to WMHV. However, such studies may be complicated to perform due to changes in treatment patterns over time as well as change in biomarkers levels and difficulties obtaining MRIs from all participants who experience a stroke. Future studies with detailed longitudinal assessments and high follow-up rates will be needed to determine if the suggestive associations seen between HbA1c, HDL-C, and ApoA1 and WMHV observed in this study can be replicated in other cohort with available biomarkers measurements years prior to the stroke event.
Supplementary Material
Acknowledgements:
This work was supported by the National Institutes of Health (CA047988, HL043851, HL080467, HL099355, CA182913 and HL128791).
Footnotes
All the work in this manuscript was performed at the Division of Preventive Medicine, Department of Medicine at Brigham and Women’s Hospital.
Conflict of Interest: The authors declare that they have no conflict of interest.
References
- 1.Henon H, Vroylandt P, Durieu I, Pasquier F, Leys D. Leukoaraiosis more than dementia is a predictor of stroke recurrence. Stroke. 2003;34(12):2935–40. doi: 10.1161/01.STR.0000103747.58719.59 [DOI] [PubMed] [Google Scholar]
- 2.Arsava EM, Rahman R, Rosand J, et al. Severity of leukoaraiosis correlates with clinical outcome after ischemic stroke. Neurology. 2009;72(16):1403–10. doi: 10.1212/WNL.0b013e3181a18823 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Kissela B, Lindsell CJ, Kleindorfer D, et al. Clinical prediction of functional outcome after ischemic stroke: the surprising importance of periventricular white matter disease and race. Stroke. 2009;40(2):530–6. doi: 10.1161/STROKEAHA.108.521906 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Wright CB, Paik MC, Brown TR, et al. Total homocysteine is associated with white matter hyperintensity volume: the Northern Manhattan Study. Stroke. 2005;36(6):1207–11. doi: 10.1161/01.STR.0000165923.02318.22 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Satizabal CL, Zhu YC, Mazoyer B, Dufouil C, Tzourio C. Circulating IL-6 and CRP are associated with MRI findings in the elderly: the 3C-Dijon Study. Neurology. 2012;78(10):720–7. doi: 10.1212/WNL.0b013e318248e50f [DOI] [PubMed] [Google Scholar]
- 6.Breteler MM, van Swieten JC, Bots ML, et al. Cerebral white matter lesions, vascular risk factors, and cognitive function in a population-based study: the Rotterdam Study. Neurology. 1994;44(7):1246–52. [DOI] [PubMed] [Google Scholar]
- 7.Han JH, Wong KS, Wang YY, Fu JH, Ding D, Hong Z. Plasma level of sICAM-1 is associated with the extent of white matter lesion among asymptomatic elderly subjects. Clinical neurology and neurosurgery. 2009;111(10):847–51. doi: 10.1016/j.clineuro.2009.08.018 [DOI] [PubMed] [Google Scholar]
- 8.Dickie DA, Ritchie SJ, Cox SR, et al. Vascular risk factors and progression of white matter hyperintensities in the Lothian Birth Cohort 1936. Neurobiology of aging. 2016;42:116–23. doi: 10.1016/j.neurobiolaging.2016.03.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Longstreth WT Jr., Arnold AM, Beauchamp NJ Jr., et al. Incidence, manifestations, and predictors of worsening white matter on serial cranial magnetic resonance imaging in the elderly: the Cardiovascular Health Study. Stroke. 2005;36(1):56–61. doi: 10.1161/01.STR.0000149625.99732.69 [DOI] [PubMed] [Google Scholar]
- 10.Yin ZG, Wang QS, Yu K, Wang WW, Lin H, Yang ZH. Sex differences in associations between blood lipids and cerebral small vessel disease. Nutrition, metabolism, and cardiovascular diseases: NMCD. 2018;28(1):28–34. doi: 10.1016/j.numecd.2017.10.001 [DOI] [PubMed] [Google Scholar]
- 11.Yin ZG, Li L, Cui M, Zhou SM, Yu MM, Zhou HD. Inverse relationship between apolipoprotein A-I and cerebral white matter lesions: a cross-sectional study in middle-aged and elderly subjects. PloS one. 2014;9(5):e97113. doi: 10.1371/journal.pone.0097113 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Rozanski M, Richter TB, Grittner U, Endres M, Fiebach JB, Jungehulsing GJ. Elevated levels of hemoglobin A1c are associated with cerebral white matter disease in patients with stroke. Stroke. 2014;45(4):1007–11. doi: 10.1161/STROKEAHA.114.004740 [DOI] [PubMed] [Google Scholar]
- 13.You CJ, Liu D, Liu LL, Liu QQ, Li GZ. Correlation between Fibrinogen and White Matter Hyperintensities among Nondiabetic Individuals with Noncardiogenic Ischemic Stroke. Journal of stroke and cerebrovascular diseases: the official journal of National Stroke Association. 2018;27(9):2360–6. doi: 10.1016/j.jstrokecerebrovasdis.2018.04.025 [DOI] [PubMed] [Google Scholar]
- 14.Cloonan L, Fitzpatrick KM, Kanakis AS, Furie KL, Rosand J, Rost NS. Metabolic determinants of white matter hyperintensity burden in patients with ischemic stroke. Atherosclerosis. 2015;240(1):149–53. doi: 10.1016/j.atherosclerosis.2015.02.052 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Jimenez-Conde J, Biffi A, Rahman R, et al. Hyperlipidemia and reduced white matter hyperintensity volume in patients with ischemic stroke. Stroke. 2010;41(3):437–42. doi: 10.1161/STROKEAHA.109.563502 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Zhang CR, Cloonan L, Fitzpatrick KM, et al. Determinants of white matter hyperintensity burden differ at the extremes of ages of ischemic stroke onset. Journal of stroke and cerebrovascular diseases: the official journal of National Stroke Association. 2015;24(3):649–54. doi: 10.1016/j.jstrokecerebrovasdis.2014.10.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Lee IM, Cook NR, Gaziano JM, et al. Vitamin E in the primary prevention of cardiovascular disease and cancer: the Women’s Health Study: a randomized controlled trial. JAMA. 2005;294(1):56–65. doi: 10.1001/jama.294.1.56 [DOI] [PubMed] [Google Scholar]
- 18.Ridker PM, Cook NR, Lee IM, et al. A randomized trial of low-dose aspirin in the primary prevention of cardiovascular disease in women. The New England journal of medicine. 2005;352(13):1293–304. doi: 10.1056/NEJMoa050613 [DOI] [PubMed] [Google Scholar]
- 19.Zhang SM, Buring JE, Lee IM, Cook NR, Ridker PM. C-reactive protein levels are not associated with increased risk for colorectal cancer in women. Annals of internal medicine. 2005;142(6):425–32. doi: 10.7326/0003-4819-142-6-200503150-00008 [DOI] [PubMed] [Google Scholar]
- 20.Kurth T, Ridker PM, Buring JE. Migraine and biomarkers of cardiovascular disease in women. Cephalalgia: an international journal of headache. 2008;28(1):49–56. doi: 10.1111/j.1468-2982.2007.01467.x [DOI] [PubMed] [Google Scholar]
- 21.Adams HP Jr., Bendixen BH, Kappelle LJ, et al. Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke. 1993;24(1):35–41. [DOI] [PubMed] [Google Scholar]
- 22.Irizarry MC, Gurol ME, Raju S, et al. Association of homocysteine with plasma amyloid beta protein in aging and neurodegenerative disease. Neurology. 2005;65(9):1402–8. doi: 10.1212/01.wnl.0000183063.99107.5c [DOI] [PubMed] [Google Scholar]
- 23.Gurol ME, Irizarry MC, Smith EE, et al. Plasma beta-amyloid and white matter lesions in AD, MCI, and cerebral amyloid angiopathy. Neurology. 2006;66(1):23–9. doi: 10.1212/01.wnl.0000191403.95453.6a [DOI] [PubMed] [Google Scholar]
- 24.Smith EE, Gurol ME, Eng JA, et al. White matter lesions, cognition, and recurrent hemorrhage in lobar intracerebral hemorrhage. Neurology. 2004;63(9):1606–12. doi: 10.1212/01.wnl.0000142966.22886.20 [DOI] [PubMed] [Google Scholar]
- 25.Rost NS, Rahman RM, Biffi A, et al. White matter hyperintensity volume is increased in small vessel stroke subtypes. Neurology. 2010;75(19):1670–7. doi: 10.1212/WNL.0b013e3181fc279a [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Rost NS, Sadaghiani S, Biffi A, et al. Setting a gold standard for quantification of leukoaraiosis burden in patients with ischemic stroke: the Atherosclerosis Risk in Communities Study. Journal of neuroscience methods. 2014;221:196–201. doi: 10.1016/j.jneumeth.2013.10.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Nandigam RN, Chen YW, Gurol ME, Rosand J, Greenberg SM, Smith EE. Validation of intracranial area as a surrogate measure of intracranial volume when using clinical MRI. Journal of neuroimaging: official journal of the American Society of Neuroimaging. 2007;17(1):74–7. doi: 10.1111/j.1552-6569.2006.00069.x [DOI] [PubMed] [Google Scholar]
- 28.Ferguson KJ, Wardlaw JM, Edmond CL, Deary IJ, Maclullich AM. Intracranial area: a validated method for estimating intracranial volume. Journal of neuroimaging: official journal of the American Society of Neuroimaging. 2005;15(1):76–8. doi: 10.1177/1051228404270243 [DOI] [PubMed] [Google Scholar]
- 29.Chen S, Li N, Deb-Chatterji M, et al. Asymmetric dimethyarginine as marker and mediator in ischemic stroke. International journal of molecular sciences. 2012;13(12):15983–6004. doi: 10.3390/ijms131215983 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Hoth KF, Tate DF, Poppas A, et al. Endothelial function and white matter hyperintensities in older adults with cardiovascular disease. Stroke. 2007;38(2):308–12. doi: 10.1161/01.STR.0000254517.04275.3f [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Bai Y, Sun L, Du L, et al. Association of circulating levels of asymmetric dimethylarginine (ADMA) with carotid intima-media thickness: evidence from 6168 participants. Ageing research reviews. 2013;12(2):699–707. doi: 10.1016/j.arr.2012.02.003 [DOI] [PubMed] [Google Scholar]
- 32.Pai JK, Cahill LE, Hu FB, Rexrode KM, Manson JE, Rimm EB. Hemoglobin a1c is associated with increased risk of incident coronary heart disease among apparently healthy, nondiabetic men and women. Journal of the American Heart Association. 2013;2(2):e000077. doi: 10.1161/JAHA.112.000077 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Selvin E, Steffes MW, Zhu H, et al. Glycated hemoglobin, diabetes, and cardiovascular risk in nondiabetic adults. The New England journal of medicine. 2010;362(9):800–11. doi: 10.1056/NEJMoa0908359 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Sarwar N, Aspelund T, Eiriksdottir G, et al. Markers of dysglycaemia and risk of coronary heart disease in people without diabetes: Reykjavik prospective study and systematic review. PLoS medicine. 2010;7(5):e1000278. doi: 10.1371/journal.pmed.1000278 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Ke D, Zhou F, Liang H, Xu Y, Lou H. Hypertriglyceridemia Is Associated with Reduced Leukoaraiosis Severity in Patients with a Small Vessel Stroke. Behavioural neurology. 2018;2018:1361780. doi: 10.1155/2018/1361780 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Schilling S, Tzourio C, Dufouil C, et al. Plasma lipids and cerebral small vessel disease. Neurology. 2014;83(20):1844–52. doi: 10.1212/WNL.0000000000000980 [DOI] [PubMed] [Google Scholar]
- 37.Guan J, Yan C, Gao Q, et al. Analysis of risk factors in patients with leukoaraiosis. Medicine. 2017;96(8):e6153. doi: 10.1097/MD.0000000000006153 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Sacco RL, Benson RT, Kargman DE, et al. High-density lipoprotein cholesterol and ischemic stroke in the elderly: the Northern Manhattan Stroke Study. JAMA. 2001;285(21):2729–35. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.