Hypertensive Disorders of Pregnancy Associated With Cerebral Small Vessel Disease Decades Later

Abstract

Background

Hypertensive disorders of pregnancy (HDP) are associated with elevated long‐term risks of cerebrovascular disease. Emerging data suggest a link between HDP and dementia, potentially mediated by cerebral small vessel disease. We aimed to evaluate whether women with a history of HDP have more cerebral small vessel disease compared with women without HDP.

Methods

This retrospective cohort study included 1602 women from the Utah Population Database who underwent brain magnetic resonance imaging after their index pregnancy. Magnetic resonance imaging scans were evaluated for white matter hyperintensities, remote brain infarctions, cerebral microhemorrhages, enlarged perivascular spaces, acute infarction, and cerebral volume loss. Multivariable regression models assessed the association between HDP and imaging findings, adjusting for confounding factors. Interrater reliability was evaluated using Cohen's kappa.

Results

Women with a history of HDP had higher white matter hyperintensities burden (adjusted mean difference, 0.57 [95% CI, 0.25–0.90]; P=0.001), more total remote brain infarctions (adjusted mean difference, 0.11 [95% CI, 0.00–0.21]; P=0.05) and enlarged perivascular spaces (adjusted mean difference, 0.14 [95% CI, 0.01–0.27]; P=0.03) compared with women without an HDP history. No significant differences were found in microhemorrhages or cerebral volume loss. Subgroup analysis revealed stronger associations between HDP and cerebral small vessel disease markers among women without history of midlife hypertension or stroke, and those in the earlier index pregnancy cohort; though, effect modification was not statistically significant (all Wald interaction tests P>0.10). Interrater agreement was high across imaging measures (weighted kappa range: 0.89–1.00).

Conclusions

Women with prior HDP show greater imaging evidence of cerebral microvascular injury decades after pregnancy, supporting HDP as a risk factor for later‐life cerebrovascular health.

Nonstandard Abbreviations and Acronyms

CBI

covert brain infarction

CSVD

cerebral small vessel disease

EPVS

enlarged perivascular spaces

HDP

hypertensive disorders of pregnancy

RBI

remote brain infarction

WMH

white matter hyperintensities

Clinical Perspective.

What Is New?

• In this retrospective cohort of 1602 women with long‐term follow‐up after pregnancy, women with a history of hypertensive disorders of pregnancy demonstrated greater white matter hyperintensity burden, more remote brain infarctions, and increased enlarged perivascular spaces on brain magnetic resonance imaging, consistent with cerebral small vessel disease.

What Are the Clinical Implications?

• These findings add to growing evidence that hypertensive disorders of pregnancy are not only acute obstetric complications but also clinically meaningful sex‐specific vascular risk factors associated with later‐life cerebrovascular injury, stroke risk, and potentially cognitive decline and dementia.

• The identification of cerebral small vessel disease markers on routine clinical imaging may provide an opportunity for earlier recognition, longitudinal surveillance, and targeted midlife vascular risk reduction strategies in women with a history of hypertensive disorders of pregnancy.

Hypertensive disorders of pregnancy (HDP), a spectrum of vascular disorders including pre‐eclampsia, gestational hypertension, eclampsia, and Hemolysis, Elevated Liver enzymes and Low Platelets syndrome, are a leading cause of maternal morbidity and mortality. In addition to the acute risks of HDP during the perinatal period, there is mounting evidence that women with a history of HDP have increased risk of later‐life cardiovascular and cerebrovascular diseases, such that HDP is now included as a risk factor for stroke by the American Heart Association. ¹ , ² , ³ , ⁴ , ⁵ With the prevalence of HDP doubling over the past 2 decades, understanding the long‐term consequences of HDP on cerebrovascular health has become increasingly important. ⁶ , ⁷

Beyond overt cerebrovascular events, there is emerging evidence that HDP is associated with later‐life cognitive impairment and dementia, particularly vascular dementia. ⁸ , ⁹ , ¹⁰ , ¹¹

White matter hyperintensities (WMH), a hallmark imaging manifestation of cerebral small vessel disease (CSVD), are strongly associated with cognitive decline, dementia, and stroke risk and are thought to reflect the cumulative effects of vascular injury over the life course. ¹² , ¹³ Established risk factors for WMH include increasing age, hypertension, prior stroke, and other vascular conditions, as well as sociodemographic factors such as race, ethnicity, educational attainment, and geographic context, which may reflect differences in vascular risk exposure and access to health care. ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷

CSVD has several imaging manifestations, the most common of which are WMH. with others including small deep gray matter or white matter lacunar infarctions, chronic microhemorrhages, and enlarged perivascular spaces. ¹⁸ HDP may represent an early vascular stressor that alters cerebrovascular structure or function, potentially accelerating the development of CSVD decades later. However, the extent to which HDP contributes independently to WMH burden, apart from traditional vascular risk factors and sociodemographic determinants, remains incompletely understood. Prior studies evaluating the association between HDP and neuroimaging markers of CSVD have reported mixed findings, with some recent studies demonstrating mixed or null associations with WMH. ¹⁹ , ²⁰ , ²¹ , ²²

In this retrospective cohort study, we evaluated women with versus without a history of HDP to determine if there were differences in common imaging manifestations of CSVD, acute infarction, and brain volume on routine brain imaging. Our hypothesis was that those with a history of HDP would be more likely to have imaging manifestations of vascular disease, specifically WMH.

METHODS

Patient Selection

The data that support the findings of this study are available from the corresponding author upon reasonable request. All participants were identified via the Utah Population Database. Complete description of the cohort has been reported previously. ¹¹ Briefly, individuals were included if they had both at least 1 singleton live birth in Utah (1939–2013) and lived in Utah until at least 45 years of age to ensure sufficient follow‐up time for observing fertility history. To capture CSVD outcome data adequately, we required that included women lived in Utah after 1996 when electronic medical records became available.

As has been done in prior studies on HDP and subsequent disease risk within the Utah Population Database, for our initial cohort, we chose a matched design to ensure an equal distribution among exposed and unexposed of the variables we believed to be confounding factors. ²³ , ²⁴ Women exposed to HDP were matched 1:2 to unexposed women by 5‐year age groups, year of childbirth, and parity at the time of the index pregnancy. Women were excluded if they were missing data on a key variable that would preclude matching, they died within 1 year of delivery, or they were diagnosed with any of the following comorbidities before the index child's birth date: myocardial infarction, ischemic heart disease, heart failure, stroke, diabetes, or dementia. The final study cohort included 59 668 women (19 989 women exposed to HDP and 39 679 women unexposed to HDP).

For this study assessing HDP and future risk of CSVD, we restricted to women who received clinical care at the University of Utah Health from 2000 onward (date at which magnetic resonance imaging [MRI] data became available in University of Utah Health electronic health records). All included women had brain MRIs as part of clinically indicated imaging using standard imaging techniques. Women were included if they underwent the following imaging sequences on either 1.5T or 3T MRI scanners: diffusion weighted imaging, axial T2, axial T2 fluid attenuated inversion recovery, and axial T2* imaging (either gradient recalled echo or susceptibility weighted imaging) of the brain after their index pregnancies. There were 1602 women who underwent 5102 brain MRIs. In cases of duplicate MRIs, we evaluated the most recent MRI for this analysis to have the longest interval between index pregnancy and brain MRI.

The University of Utah institutional review board approved this retrospective study (IRB #00116984; RGE #00003473) and issued a waiver of informed consent under 45 CFR 46.116(f) because the research posed no more than minimal risk and could not practicably be conducted without the waiver. Data were provided by Utah Population Database in deidentified form and handled in accordance with the Health Insurance Portability and Accountability Act, institutional policy, and the Declaration of Helsinki.

Exposure

Using birth certificate data, we assigned a diagnosis of HDP (yes/no) to each affected pregnancy. HDP diagnoses included gestational hypertension; preeclampsia; Hemolysis, Elevated Liver enzymes and Low Platelets syndrome; and eclampsia. Diagnoses were listed on the birth certificate for the affected pregnancy in the following manner: 1939 to 1977, text string; 1978 to 1988: International Classification of Diseases, Ninth Revision (ICD‐9) codes; 1989 to present, check boxes with additional text. When available, inpatient records were reviewed for exposed women (1996–2013). If there was a discrepancy between hypertensive diagnoses listed in the inpatient records compared with the birth certificate, the inpatient record diagnoses were used in place of the birth certificate diagnoses. Women with >1 pregnancy complicated by hypertensive disease were included only once, with the index pregnancy chosen using a 3‐step process: (1) most severe category; (2) same severity but delivered more preterm; or (3) same severity but not delivered preterm with the lowest birth order. ²⁵ Unexposed women had pregnancies during the study period, but none were complicated by HDP. ²⁵ , ²⁶

Outcomes

Images were reviewed by a blinded neuroradiologist using visual rating scales trained in CSVD measurements using standard definitions defined a priori. These were applied in a training session on 20 subjects initially. After this training session, the first batch of participants was reviewed for an image interpretation quality check to ensure adherence to clearly defined CSVD definitions. Thereafter, a consistency assessment was performed on each set of 100 brain MR exams. To assess the reproducibility of ratings, a second blinded neuroradiologist evaluated a subset of 100 MRI scans. Interrater reliability of the ratings was expressed using quadratic weighted Kappa coefficient for ordinal data and Cohen's kappa coefficient for binary data.

White Matter Hyperintensities

WMH were defined as hyperintense lesions on T2 fluid attenuated inversion recovery and T2 sequences and were evaluated in separate brain regions qualitatively using a scale adapted from the age‐related white matter changes scheme. ²⁷ Scoring was performed in each of the following regions with each hemisphere scored independently: frontal, parieto‐occipital, temporal, brainstem, and basal ganglia. Each region was scored on a scale from 0 to 3 (0=no lesions, 1=focal lesions, 2=beginning confluence of lesions, and 3=diffuse involvement of the entire region). The basal ganglia were scored on a similar scale (0=no lesions, 1=1 focal lesion (≥5mm), 2 = >1 focal lesion, and 3=confluent lesions).

Remote Brain Infarctions

Each MRI was evaluated for findings of remote brain infarctions (RBIs), which are commonly called covert brain infarctions (CBIs). CBIs are cerebral infarcts observed on brain imaging in the absence of any corresponding, clinically apparent cerebrovascular ischemic event. ²⁸ , ²⁹ In order to adhere to the strict definition of CBI, we called these findings of infarcts “RBIs” rather than CBIs because we could not definitively ascertain each patient had no corresponding symptom to the imaging findings. RBIs were categorized into 2 categories according to a previously used rating system: (1) lacunar cavitary infarcts and (2) cortical infarcts. ³⁰ , ³¹ RBIs were characterized by location as either in the basal ganglia, deep white matter, cortical, or pontomesencephalic by each hemisphere. Lacunes were characterized as 3 to 15 mm cerebrospinal fluid‐filled slit or wedge‐like cavities with irregular margins and a gliotic T2/fluid attenuated inversion recovery hyperintense rims in the basal ganglia or white matter. ³¹ The cortical subtype were >3 mm to >15 mm focal T2 hyperintense lesions in cortical gray matter with volume loss. ³⁰

Microhemorrhages

We evaluated for foci of chronic hemosiderin staining defined as small round or ovoid lesions of hypointensity with blooming on T2* weighted imaging. Given the very low number of microhemorrhages, we counted the total number of microhemorrhages per MRI.

Enlarged Perivascular Spaces

Enlarged perivascular spaces (EPVS) were defined as small, sharply delineated structures of cerebrospinal fluid intensity that follow the orientation of perforating vessels and run perpendicular to the brain surface using a previously published rating scale. ³² , ³³ The quantity of EPVS were rated in the (1) hippocampus and (2) basal ganglia and centrum semiovale as 0=no EPVS, 1=<10 EPVS, 2=10 to 20 EPVS, 3=21 to 40 EPVS and 4=>40 EPVS. For each patient, the most affected hemisphere was rated. ³² , ³³

Acute Infarction

Acute infarctions were rated as absent or present based on evaluation of diffusion weighted sequences. If present, the anatomical location of each infarct was noted.

Volume Loss

Cerebral parenchymal volume loss was defined as deep (enlargement of the ventricles, central atrophy) or peripheral (enlargement of the gyri) and rated on a subjective scale of 0 to 3 (0=absent, 1=mild, 2=moderate, 3=severe) against a reference MR brain template of normal subjects following a previously outlined method. ³³

Statistical Analysis

Sociodemographic and health history characteristics of the patient population, overall and by HDP status (prior history/no prior history), were assessed using descriptive statistics. We report frequencies (n [%]) for categorical CSVD outcomes and means±SD for continuous CSVD outcomes. We dichotomized the categorical variables based on the distribution of the data with frontal and parieto‐occipital WMH dichotomized to ≥2 versus ≤1; temporal, brainstem, and basal ganglia WMH dichotomized to ≥1 versus 0; basal ganglia, white matter, and cortical RBIs dichotomized to ≥1 versus 0, and any microhemorrhages, acute infarcts, volume loss dichotomized to ≥1 versus 0. We created a supratentorial WMH variable if frontal ≥2, parieto‐occipital ≥2, or temporal ≥1.

Although the original cohort was assembled using a matched design, the present analysis was restricted to a subset of women with available brain MRI data, which substantially disrupted the original matching structure; only 27 women (1.4% of those with MRI data) remained in intact matched or clustered sets. Consequently, analyses were conducted treating the data as an unmatched cohort.

We used generalized linear models to examine the associations between HDP and CSVD. Dichotomous outcomes were modeled assuming a binomial distribution with a log link to estimate unadjusted and adjusted risk ratios, while continuous outcomes were modeled assuming a normal distribution with an identity link to estimate adjusted mean differences. Ninety‐five percent CIs were calculated for all estimates. Covariates to adjust for were determined via prior literature and directed acyclic graphs (Figure S1) and included birth state (Utah versus not Utah), race (White versus races other than White), Hispanic (Hispanic versus non‐Hispanic), maximum education, birth year, and age at index pregnancy. To account for residual imbalance in follow‐up time, all models were additionally adjusted for time elapsed between the index pregnancy and the brain MRI. Models for continuous outcomes were estimated using generalized estimating equations with an independence working correlation structure, and robust (sandwich) SEs to provide valid inference for right‐skewed distributions and departures from normality. P values are reported for descriptive purposes only.

We tested for effect modification of the association between HDP and CSVD by history of hypertension, history of stroke, and birth cohort (before 1978, 1978–1988; and after 1988). Effect modification was assessed by including interaction terms between hypertensive pregnancy and each potential modifier in the generalized linear and linear regression models. Wald tests were used to evaluate the statistical significance of the interaction terms. Statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC).

RESULTS

At index pregnancy, the study population was on average 27.6 ± 6.8 years, predominately White (91%), non‐Hispanic (86%), born in Utah (68%), with some college or more (60%), urban residence (90%), and parous (mean 3.5 ± 1.9 births) (Table 1). Women with a history of HDP had on average 1.2 ± 0.4 HDP pregnancies with index pregnancy predominately preeclampsia (66%) followed by gestational hypertension (29%), eclampsia (4%), and Hemolysis, Elevated Liver enzymes and Low Platelets syndrome (1%). At MRI follow‐up, women were on average 60.9 ± 12.8 years, with an average follow‐up of 33.3 ± 14.6 years. In the intervening period between index birth and MRI follow‐up, 50% of women were diagnosed with hypertension (49.6% among those with prior HDP and 50.5% for those without prior HDP) and 5% diagnosed with a stroke (5.7% among those with prior HDP and 5.2% among those without prior HDP). Maternal age, birth year, and birth state were similar among women with and without a history of HDP (Table 1).

Table 1.

Demographic Characteristics at Time of Index Birth (n=1602 Women)

	Total (n=1602)	No HDP exposure (n=1056)	HDP exposure (n=546)
Baseline
Type of HDP at index pregnancy; n (%)
Preeclampsia	380 (66)	NA	380 (66)
Eclampsia	20 (4)	NA	20 (4)
Gestational hypertension	161 (29)	NA	161 (29)
Hemolysis, Elevated Liver enzymes and Low Platelets	5 (1)	NA	5 (1)
Total pregnancies with HDP (mean±SD)	1.2±0.4	NA	1.2±0.4
Parity at time of index pregnancy (mean±SD; median [25th%–75th%])	3.5±1.9; 3 (2–4)	3.5±2.0; 3 (2–4)	3.5±1.8; 3 (2–4)
Mother's age, y	27.6±6.8	27.9±6.7	27.1±6.8
Maternal birth, y (mean±SD)	1953±12	1952±12	1953±12
Child's birth, y (mean±SD)	1981±14	1981±14	1981±14
Race (White vs races other than White)
Yes	1459 (91)	971 (92)	488 (89)
No	143 (9)	85 (8)	58 (11)
Unknown	118	0	0
Ethnicity (Hispanic vs non‐Hispanic)
Yes	221 (14)	136 (13)	85 (16)
No	1354 (86)	902 (87)	452 (84)
Unknown	27	18	9
Born in Utah
Yes	1002 (68)	649 (67)	353 (69)
No	482 (32)	322 (33)	160 (31)
Unknown	118	85	33
Mother's education
Less than HS	155 (10)	88 (9)	67 (13)
HS graduate	464 (30)	303 (30)	161 (31)
Some college	487 (32)	315 (31)	172 (33)
College graduate	209 (14)	149 (15)	60 (12)
Post college	212 (14)	153 (15)	59 (11)
Unknown	75	48	27
Mother's residence at time of index birth
Frontier	23 (2)	12 (1)	11 (2)
Rural	130 (8)	88 (8)	42 (8)
Urban	1433 (90)	947 (90)	486 (90)
Unknown	16	9	7
Follow‐up
Maternal age at time of MRI, y	60.9±12.8	61.5±12.9	59.9±12.6
Average follow‐up between index pregnancy and MRI (mean ± SD; median [25th%–75th%])	32.7±14.6; 32.1 (22.5–42.5)	33.0±14.7; 32.1 (22.7–42.6)	32.2±14.3; 32.0 (22.0–42.4)
Hypertension diagnosis	804 (50)	533 (50)	271 (50)
Stroke diagnosis	86 (5)	55 (5)	31 (6)

HDP indicates hypertensive disorder of pregnancy; HS, high school; MRI, magnetic resonance imaging; and NA, not available.

Women with a history of HDP were more likely to have WMH on brain MRI compared with women without a history of HDP with a mean difference of 0.57 (95% CI, 0.25–0.90, P=0.001) after adjusting for birth state, race, ethnicity, maximum education, birth year, maternal age at index pregnancy, and time between index pregnancy and brain MRI (Table 2). When broken down by location, there was a pattern of higher WMH among women with a history of HDP compared with those without a history of HDP (all adjusted risk ratios >1.00).

Table 2.

Brain MRI Vascular Features Among Utah Women With and Without a History of Hypertensive Disorder of Pregnancy

	Total population N=1602	HDP history n=546 (34%)	No HDP history n=1056 (66%)	Unadjusted P value	Unadjusted RR (95% CI)	Adjusted P value	Adjusted RR (95% CI)
	No. (%)			HDP vs no HDP
Frontal WM*	172 (10.7)	71 (13.0)	101 (9.6)	0.04	1.04 (1.00 to 1.08)	0.12	1.02 (0.97 to 1.08)
Parieto‐occipital WM*	225 (14.0)	87 (15.9)	138 (13.1)	0.12	1.03 (0.99 to 1.08)	0.54	1.02 (0.96 to 1.07)
Temporal WM†	202 (12.6)	76 (13.9)	126 (11.9)	0.26	1.02 (0.98 to 1.07)	0.47	1.01 (0.97 to 1.05)
Supratentorial‡	349 (19.9)	131 (24.0)	218 (20.6)	0.13	1.04 (0.99 to 1.10)	0.37	1.02 (0.98 to 1.06)
Brainstem†	192 (12.6)	77 (15.0)	115 (11.4)	0.05	1.04 (1.00 to 1.09)	0.26	1.03 (0.98 to 1.07)
Basal ganglia†	539 (33.7)	187 (34.3)	350 (33.1)	0.64	1.02 (0.95 to 1.10)	0.38	1.02 (0.97 to 1.06)
RBIs basal ganglia†	121 (7.6)	38 (7.0)	83 (7.9)	0.53	0.99 (0.96 to 1.02)	1.00	1.00 (0.95 to 1.05)
RBIs WM†	37 (2.3)	17 (3.1)	20 (1.9)	0.12	1.01 (1.00 to 1.03)	0.72	1.01 (0.94 to 1.08)
RBIs cortical†	23 (1.5)	12 (2.2)	11 (1.1)	0.06	1.01 (1.00 to 1.03)	0.21	1.01 (0.97 to 1.06)
Any microhemorrhages	91 (6.5)	33 (6.9)	58 (6.3)	0.67	1.01 (0.98 to 1.04)	0.95	1.00 (0.96 to 1.05)
Any acute infarct	87 (5.5)	42 (7.7)	45 (4.3)	0.01	1.04 (1.01 to 1.06)	0.25	1.03 (0.98 to 1.07)
Volume loss†	553 (34.6)	196 (36.0)	357 (33.9)	0.41	1.03 (0.96 to 1.11)	0.37	1.02 (0.97 to 1.07)

	Total population N=1602	HDP history n=546 (34%)	No HDP history n=1056 (66%)	Unadjusted P value	Mean difference (95% CI)	Adjusted P value	Adjusted mean difference (95% CI)
	Means±SD			HDP vs no HDP
Total WM§	3.8±3.5	4.0±3.7	3.6±3.4	0.04	0.38 (0.01 to 0.75)	0.001	0.57 (0.25 to 0.90)
RBIs total|| mean ± SD	0.24±1.06	0.32±1.42	0.19±0.80	0.04	0.13 (0.00 to 0.26)	0.05	0.11 (0.00 to 0.21)
Total microhemorrhages¶	0.23±2.04	0.34±3.14	0.18±1.09	0.16	0.16 (−0.06 to 0.38)	0.12	0.20 (−0.04 to 0.44)
Total enlarged perivascular spaces#	3.5±1.2	3.7±1.3	3.5±1.2	0.43	0.05 (−0.08 to 0.18)	0.03	0.14 (0.01 to 0.27)

^*≥2 vs ≤1.

^† ≥1 vs 0.

^‡ If frontal, parieto‐occipital, or temporal are a “yes” as defined by 1 or 2.

^§ Continuous score from 0 to 30.

^|| Continuous score from 0 to 17.

^¶ Continuous score from 0 to 60.

^# Continuous score from 0 to 9; note, only n=6 with giant perivascular spaces, too few to assess (n=2 with HDP and 4 without HDP).

Adjusted for birth state (Utah vs not Utah), race (White vs races other than White), Hispanic (Hispanic vs non‐Hispanic), maximum education, birth year, age at index pregnancy, and time elapsed between the index pregnancy and the brain MRI. HDP indicates hypertensive disorder of pregnancy; MRI, magnetic resonance imaging; RBI, remote brain infarction; and WM, white matter.

Women with a history of HDP, compared with their counterparts, had higher total number of RBIs with mean adjusted difference of 0.11 (0.00–0.21), P=0.05 (Table 2) and higher total EPVS with a mean adjusted difference of 0.14 (95% CI, 0.01–0.27), P=0.03. Although crossing the null, women with a prior HDP pregnancy had 0.20 higher adjusted mean difference (95% CI, −0.04 to 0.44, P=0.12) in total microhemorrhages compared with women without a HDP history. Across stratified analyses by midlife stroke, midlife hypertension, and index pregnancy year, we found stronger associations between HDP and CSVD markers among women without a history of midlife hypertension, stroke, and those in the earlier index pregnancy cohort; however, these differences did not reach statistical significance for effect modification (all Wald interaction tests P>0.10). (Figures 1, 2, 3).

Figure 1. Adjusted associations between hypertensive disorders of pregnancy and markers of cerebral small vessel disease, stratified by index pregnancy year.

Adjusted for birth state (Utah vs not Utah), race (White vs races other than White), Hispanic (Hispanic vs non‐Hispanic), maximum education, birth year, age at index pregnancy, and time elapsed between the index pregnancy and the brain magnetic resonance imaging. HDP indicates hypertensive disorder of pregnancy; PVS, enlarged perivascular spaces; RBI, remote brain infarction; and WMH, white matter hyperintensities.

Figure 2. Adjusted associations between hypertensive disorders of pregnancy and markers of cerebral small vessel disease, stratified by midlife hypertension status.

Adjusted for birth state (Utah vs not Utah), race (White vs races other than White), Hispanic (Hispanic vs non‐Hispanic), maximum education, birth y, age at index pregnancy, and time elapsed between the index pregnancy and the brain magnetic resonance imaging. HDP indicates hypertensive disorder of pregnancy; PVS, enlarged perivascular spaces; RBI, remote brain infarction; and WMH, white matter hyperintensities.

Figure 3. Adjusted associations between hypertensive disorders of pregnancy and markers of cerebral small vessel disease, stratified by midlife stroke status.

Adjusted for birth state (Utah vs not Utah), race (White vs races other than White), Hispanic (Hispanic vs non‐Hispanic), maximum education, birth y, age at index pregnancy, and time elapsed between the index pregnancy and the brain magnetic resonance imaging. HDP indicates hypertensive disorder of pregnancy; PVS, enlarged perivascular spaces; RBI, remote brain infarction; and WMH, white matter hyperintensities.

We found high interrater reliability for our imaging interpretation with the lowest weighted Kappa coefficient of 0.89 (95% CI, 0.83–0.95) for cerebral volume to a perfect Cohen's Kappa coefficient of 1.00 for presence of acute infarction.

DISCUSSION

In a cohort of 1602 women, we found that those with a history of HDP were more likely to have WMH, RBIs, and EPVS compared with those without HDP on routine brain imaging. We found no differences in other markers of CSVD or cerebral volume loss in women with and without a history of HDP. Our findings are consistent with others in that women with a history of HDP are more likely to have findings of WMH and remote infarcts, further supporting the link between HDP and elevated later‐life risk of cerebrovascular disease.

HDPs are believed to induce lasting vascular dysfunction that may underlie the observed association with later‐life cerebrovascular disease. There are several shared pathways between HDP and later‐life cerebrovascular disease and vascular cognitive dysfunction including vascular and angiogenic dysregulation and inflammation. ³⁴ The exact mechanism by which HDP and later‐life cerebrovascular disease are related is still unclear. Given the established association between HDP and later‐life hypertension and stroke, it is plausible these vascular diseases may mediate the association between HDP and later‐life CSVD. Interestingly, we found a pattern that those without a midlife hypertension or stroke diagnosis had more pronounced differences between markers of CSVD compared with those with a midlife hypertension or stroke diagnosis. Though on the surface, this finding may suggest that midlife hypertension and stroke are not the primary drivers of the association between HDP and later‐life CSVD in our cohort, it is plausible this finding is due to the inclusion of women with undiagnosed and therefore untreated hypertension. Potentially, those without a diagnosis of midlife hypertension were still experiencing hypertension, without recognition or treatment, resulting in worse outcomes. ³⁵ Further investigation on the role of midlife hypertension as it relates to later‐life CSVD in this population is needed.

The stronger associations observed among earlier birth cohorts may reflect cohort effects related to historically poorer recognition and management of hypertensive disorders of pregnancy, resulting in greater severity or longer duration of untreated vascular injury and its downstream cerebrovascular consequences. In contrast, women in later birth cohorts likely benefited from earlier diagnosis, improved obstetric and cardiovascular care, and greater risk factor surveillance, potentially attenuating long‐term cerebrovascular sequelae of HDP. ³⁶

We found that women with a history of HDP had more acute infarctions on brain MRI but this difference was not present after adjustment for confounding factors, potentially due to relatively low numbers of infarcts. Many others have found strong associations with a history of HDP and acute infarctions. A large Taiwanese cohort found that those with HDP were more likely to have both ischemic and hemorrhagic strokes for years after childbirth. ³⁷ Others have found that those with a history of adverse pregnancy outcomes, including HDP, had earlier onset of first cerebrovascular event, including ischemic stroke. ³⁸ This confirms existing guidelines of recognizing HDP as a risk factor for stroke, as per the American Heart Association. ³⁹

Our findings confirmed some other reports of increased later‐life WMH in women with a history of HDP. ²⁰ , ²¹ , ²² WMH is the classic manifestation of CSVD and has the most robust evidence for later‐life cognitive impairment. It is also strongly linked with hypertension. Our study differed from others that evaluated WMH differences in women with HDP in that we evaluated WMH in the posterior circulation separately from the anterior circulation. We found the most robust differences in WMH in the brainstem between women with HDP and those without. This finding may highlight the role of cerebral dysregulation and blood–brain barrier permeability in women with a history of HDP, as the posterior circulation is often more likely to be affected in the setting of cerebral dysregulation. ⁴⁰

We found that women with a history of HDP were more likely to have more EPVS than those without. EPVS are increasingly recognized as imaging markers associated with cognitive impairment and dementia. ⁴¹ , ⁴² Though the exact pathophysiologic link between this newly recognized marker of CSVD and HDP is unclear, our findings warrant further investigation.

We found no differences in cerebral microhemorrhages in our cohort. There was a slight indication of more microhemorrhages in the HDP group, though the overall very low number of microhemorrhages precludes any conclusions from our study. Further work in older populations may be needed to see if there is a difference in these microhemorrhages as women age because of the known increased risk of hemorrhagic strokes in women with a history of HDP. ³⁷

Others have found lower brain volumes, specifically decreased gray matter volume in women with a history of HDP. ⁴³ , ⁴⁴ We did not find a significant difference in our cohort, though we were limited by the imaging technique used in our patients. Given the population‐based nature of our cohort, we did not have volumetric imaging available to quantitatively evaluate cerebral volume or cortical thickness, as was done in previous studies.

Though there are few to no data on the increased prevalence of CBIs in women with HDP, our findings of more RBIs in those with a history of HDP is consistent with the body of literature suggesting increased stroke risk with HDP. We divided RBIs into those found in the deep gray matter, deep white matter, and cortical RBIs. Individually, there were trends of increased risk ratios for RBIs in those with a history of HDP, but we found more total RBIs in those with HDP compared with those without. Many of these RBIs are more likely secondary to CSVD, specifically those in the deep white and deep gray matter, compared with those in the cortical gray matter, which are more likely to be thromboembolic in origin.

This retrospective study relied on clinically indicated brain MRIs obtained for diverse reasons, across different scanning platforms and acquisition parameters, which introduces heterogeneity that may add measurement variability and bias estimates toward or away from the null. Although our retrospective cohort design leveraging administrative health care records reduces differential bias by ensuring that women with and without prior HDP were all imaged in real‐world care, inherent selection bias remains. Patients referred for imaging may differ from nonimaged patients in access to care, symptom severity, and comorbidity burden, thus limiting generalizability to the broader HDP population. The population‐based use of medical records may also undercapture midlife hypertension, constraining inference about its role in later‐life CSVD. Our cohort was relatively homogeneous with a high prevalence of White women, which may limit applicability to other racial and ethnic groups. Some outcomes (eg, microhemorrhages and acute infarctions) were infrequent, limiting statistical precision. Finally, because this retrospective analysis uses administrative health care data, our ability to measure lifestyle, behavioral, and genetic factors is limited; therefore, despite multivariable adjustment, residual confounding cannot be excluded. Notwithstanding these constraints, leveraging real‐world clinical imaging allowed us to evaluate cerebrovascular markers in routine care settings that are difficult to capture prospectively, and associations persisted despite this heterogeneity, underscoring the clinical relevance of the observed differences between women with and without prior HDP. ³⁹ , ⁴⁰

Conclusions

In this retrospective cohort of women with long‐term follow‐up after HDP, we identified several differences in imaging markers of CSVD. Women with a history of HDP demonstrated higher WMH burden, greater prevalence of RBIs, and more EPVS. These findings collectively suggest chronic microvascular brain injury among women with an HDP history. These results reinforce growing evidence that HDP is not only an acute obstetric complication but also a sex‐specific risk factor for later‐life cerebrovascular disease. Importantly, the detection of such changes on routine clinical imaging highlights a potential opportunity for earlier identification and intervention in women at elevated lifetime stroke and dementia risk. Given accumulating evidence linking midlife HDP with later‐life cerebrovascular disease across diverse populations, HDP should be considered a clinically meaningful midlife vascular risk factor in women. Future research should further characterize these imaging biomarkers in diverse populations and investigate whether midlife risk factor modification can attenuate the long‐term cerebrovascular consequences of HDP.

Sources of Funding

We thank the Pedigree and Population Resource of Huntsman Cancer Institute, University of Utah (funded in part by the Huntsman Cancer Foundation) for its role in the ongoing collection, maintenance, and support of the Utah Population Database. We also acknowledge partial support for the Utah Population Database through grant P30 CA2014 from the National Cancer Institute, University of Utah and from the University of Utah's program in Personalized Health and Center for Clinical and Translational Science. Research was also supported by the National Center for Research Resources grant, “Sharing Statewide Health Data for Genetic Research” (R01 RR021746, G. Mineau, principal investigator) with additional support from the Utah Department of Health and the University of Utah. This work was supported by the National Institute on Aging grant “Hypertensive Disorders of Pregnancy and Subsequent Risk of Vascular Dementia, Alzheimer's Disease, or Related Dementia: A Retrospective Cohort Study Taking into Account Midlife Mediating Factors” (Project K01AG058781; PI: Karen Schliep). This work was also supported by the Doris Duke Charitable Foundation Utah Caregiver Assistance for Researchers in Early Stages Program.

Disclosures

The authors report no conflict of interests.

Supporting information

Figure S1

STROBE Checklist