Hypertensive Disorders of Pregnancy and Brain Health in Midlife: The CARDIA Study

Abstract

Background:

To understand the role of hypertensive disorders of pregnancy (HDP), including preeclampsia (PE) and gestational hypertension (GH), in brain health earlier in life, we investigated the association of HDP with midlife cognition and brain health.

Methods:

We studied a prospective cohort of women, baseline age 18–30 years, who were assessed at study Years 25 and 30 with a cognitive battery and a subset with brain MRI. A history of HDP was defined based on self-report. We conducted linear regression to assess the association of a history of PE, GH, or no HDP with cognition and brain MRI white matter hyperintensities (WMH).

Results:

Among 1,441 women (mean age 55.2 ± 3.6 years), 202 reported PE and 112 reported GH. GH was associated with worse cognitive performance: global cognition (mean score 23.2 vs. 24.0, p=0.018), processing speed (67.5 vs. 71.3, p=0.01), verbal fluency (29.5 vs. 31.1, p=0.033), and a trend for executive function (24.3 vs. 22.6, p=0.09), after multivariable adjustment. GH was associated with a greater 5-year decline on processing speed (mean change −4.9 vs. −2.7, p=0.049) and executive function (−1.7 vs 0.3, p=0.047); PE was associated with a greater 5-year decline on delayed verbal memory (−0.3 vs. 0.1, p=0.041). GH and PE were associated with greater WMH in the parietal and frontal lobes, respectively.

Conclusion:

GH and PE are associated with cognition and WMH during midlife, with differences in cognitive domains and brain lobes. Women with HDP may need to be closely monitored for adverse brain outcomes starting in midlife.

Graphical Abstract

INTRODUCTION

Hypertensive disorders of pregnancy (HDP), including preeclampsia (PE) and gestational hypertension (GH), can cause serious maternal complications, including eclampsia and cerebral edema, and have been associated with increased risks of cerebrovascular and cardiovascular disease (CVD).¹ In the United States, the prevalence of HDP has been increasing steadily from 13% in 2017 to 16% in 2019.² HDP may be related to worse brain health through various vascular mechanisms,³ but the association between HDP and more remote cognitive outcomes over the lifecourse remains understudied.

Several studies have shown that HDP is associated with mild cognitive impairment and dementia later in life.^4,5 Nevertheless, studies examining HDP and cognition among middle-aged women are sparse and have yielded mixed results.^6–8 It is crucial to determine whether the impact of HDP on cognitive aging might manifest as early as in midlife, a critical period when early divergence in cognitive trajectories might be prevented more effectively. Although brain vascular injury and response following HDP have been suggested as potential drivers to worse brain health,⁹ the association between HDP and brain white matter lesions (a marker of vascular injury) on MRI has rarely been studied.^10,11 HDP shares many key risk factors with adverse brain health, including obesity and diabetes, and is also associated with a higher risk of hypertension later in life.^12,13 However, it remains unclear whether HDP can provide additional information on brain health beyond these factors. Furthermore, previous studies have focused primarily on PE with limited consideration given to GH. Significant differences in placental and endothelial features between GH and PE suggest they are distinct entities with different etiologies.^14–16 Yet, GH is more strongly linked to future risks of hypertension and ischemic heart disease than PE, but its role in cognitive outcomes remains to be understood.

In a prospective cohort of women from the Coronary Artery Risk Development in Young Adults (CARDIA) Study, we sought to assess the association between a history of HDP, including PE or GH, and cognitive function and decline during midlife. In addition, we compared brain MRI burden of white matter hyperintensities (WMH) among middle-aged women who did and did not report HDP.

METHODS

Study Population

Requests to access the dataset from qualified researchers may be sent to the CARDIA Coordinating Center (coc@uab.edu). We studied participants in the CARDIA Study, a prospective cohort study of risk factors for CVDs among Black and White adults. In 1985–86, a total of 5115 Black and White men and women, aged 18–30 years, were recruited at baseline from population-based samples derived from 4 US cities (Birmingham, AL; Chicago, IL; Minneapolis, MN; and Oakland, CA), with a roughly balanced representation of participants at each site by sex, race (White and Black), and education (high school or less, more than high school) and age (18–24, 25–30 years). Participants had completed nine follow-up examinations every 2 to 5 years for 30 years with high retention rates.¹⁷ The study protocol was approved by each site’s Institutional Review Board. All participants provided written informed consent.

Among the 2,787 women in CARDIA, 2,157 reported a history of pregnancy without prior hypertension (Figure SI). We excluded 106 deaths and 610 women without cognitive follow-up at Year 30. Among the 1,441 parous women analyzed at Year 30, 314 reported a history of HDP and 1,127 reported no HDP history. Compared to the analyzed women, those who were alive but not examined were more likely to be current smokers and less educated (p<0.05) but did not differ on other demographic, baseline behavioral and cardiovascular risk factors (CVRFs), or APOE ε4.

Hypertensive Disorders of Pregnancy and Pregnancy History

Women were asked about their pregnancy history at each exam, including the number of pregnancies since the last exam, length of gestation, delivery dates, and age at first pregnancy. Participants were queried about whether they had “toxemia including all of the following: high blood pressure, albumin in the urine, and ankle swelling,” reflecting established definitions during the initiation of CARDIA.¹⁸ Women without hypertension in previous exams but reported developing hypertension without toxemia during one or more pregnancies were classified as having a history of GH. Women who reported developing toxemia during pregnancy were categorized as having a history of PE, which included those with hypertension (i.e., PE superimposed on chronic hypertension) and without hypertension (i.e., PE) in previous exams. Women reporting both PE and GH were classified as having PE (n=17). Women were classified as having no HDP if they had ≥1 pregnancy, no hypertension before any pregnancy, and reported no history of PE or GH.

Parity by Year 30 was defined as the number of pregnancies lasting ≥20 weeks of gestation. We also calculated the total number of HDP pregnancies and the time between Year 30 exam and the delivery date of the first HDP pregnancy (0 for those without HDP). Women who did not have diabetes before pregnancy but reported diabetes only during pregnancy were classified as having gestational diabetes.

Cognitive Assessment

At Year 30, five cognitive tests were administered by interviewers who underwent centralized training and certification. The Montreal Cognitive Assessment (range 0–30) assesses global cognition with tests on attention, executive function, memory, language, visuospatial skills, calculation, and orientation. The Rey Auditory Verbal Learning Test (RAVLT; range 0–15) assesses verbal learning and memory; participants were read a list of 15 words and asked to recall after 10 minutes in the delayed test. The Digit Symbol Substitution Test (DSST; range 0–133) assesses processing speed, executive function, and working memory; participants were given a key pairing symbols with numbers and 2 minutes to write the corresponding symbol under each number, with scores based on correct substitutions. The Stroop Test assesses executive function with three subtests: 1) reading color names in black ink, 2) naming the color of the sheets, and 3) naming the ink color for color names printed in incongruent ink colors. The Stroop Interference score was obtained by subtracting the subtest 2 score from the subtest 3 score, with higher scores indicative of worse performance. Verbal fluency was assessed using an average score of the Letter and Category Fluency tests; participants had 1 minute to name words starting with “A” and “S” in the Letter Fluency test and animals in the Category Fluency test. Higher scores indicate better performance for all tests, except for the Stroop Interference. We also estimated the 5-year cognitive change for DSST, RAVLT, and Stroop Interference administered at both Years 25 and 30, with a negative difference indicating a decline for each test.

Brain MRI White Matter Hyperintensities

The CARDIA Brain MRI sub-study enrolled participants during Year 25 and 30 exams, balancing race and sex from three of the sites: Birmingham, AL, Minneapolis, MN, and Oakland, CA.¹⁹ Participants with MRI contraindications or body size too large for the MRI tube were excluded. A subset of the parous women (N=367) had brain MRI-measured WMH at either or both Year 25 and 30; We used Year 30 measurements when both were available. Briefly, brain MRI was acquired on 3-T MR scanners at three sites (Oakland and Minneapolis: Siemens 3T Tim Trio/ VB 15 platform; Birmingham: Philips 3T Achieva/2.6.3.6 platform). The data were transferred to the MRI reading center at the University of Pennsylvania with image processing, quality control, and automated tissue volume computations as described previously.¹⁹ White matter was classified into regions of interest according to the Jakob atlas and further into normal and abnormal tissue.¹⁹ The WMH volume was estimated from the sagittal 3D FLAIR, T1, and T2 sequences that contain tissue damaged because of ischemia, demyelination or inflammation, and penumbra surrounding brain infarcts. Total intracranial volume was estimated from the sagittal 3D T1 sequence as a measure of head size. We examined the WMH absolute volumes overall and by lobe (frontal, temporal, parietal, and occipital lobes) such that the mean volumes overall and within specific lobes were averaged across the left and right hemispheres.

Covariates

Demographic characteristics (age, race, marital status), highest achieved educational attainment, and family history of dementia were obtained from self-report and interviewer-administered questionnaires. We assessed behavior and CVRFs at baseline. Physical activity was assessed by the CARDIA Physical Activity questionnaire and dichotomized as achieving the recommended activity level (≥300 exercise units). Alcohol drinks per week was obtained from self-report. Dietary quality was assessed using the CARDIA A Priori Diet Quality Scores. We defined obesity as a body mass index (BMI) ≥30 kg/m² based on height (meters) and weight (kilograms). Current cigarette smoking was defined as self-reported smoking of at least five cigarettes per week, almost every week. We defined dyslipidemia as triglycerides ≥150 mg/dL, high-density lipoprotein cholesterol <45 mg/dL, or low-density lipoprotein cholesterol ≥160 mg/dL. Depression was assessed at Year 5 using the Center for Epidemiologic Studies Depression scale ≥16. Elevated blood pressure was defined as systolic blood pressure (SBP) ≥130 mmHg, diastolic blood pressure (DBP) ≥85 mmHg, or taking antihypertensive medication. Diabetes was defined as fasting glucose ≥126 mg/dL, 2-hr post-load glucose ≥200 mg/dL from the 75-gram oral glucose tolerance tests, hemoglobin A1C ≥6.5%, or taking anti-diabetic medication. We also considered age at first pregnancy, hypertension, diabetes, and BMI at Year 30. Hypertension was defined as SBP ≥140 mmHg, DBP ≥90 mmHg, or taking antihypertensive medication. Uncontrolled hypertension or diabetes was defined as meeting the blood pressure or glucose/hemoglobin A1C criteria respectively, regardless of medication use. Traumatic brain injury was determined based on self-reports at Year 30. CVD events during follow-up, including coronary heart disease, stroke/TIA, congestive heart failure, and peripheral artery disease, were adjudicated based on medical records.

Statistical Analysis

Descriptive statistics of participant characteristics among different HDP groups were compared using χ² or Kruskal-Wallis tests. Post-hoc tests for specific group differences were conducted using Bonferroni multiple comparisons and unadjusted logistic regressions. Linearity between continuous covariates and cognitive outcomes was examined. Continuous covariates modeled linearly as generalized additive models with smoothing splines did not provide a better fit. We assessed the associations of HDP with cognition and WMH using linear regression. Due to skewness, WMH was modeled as a natural logarithm of WMH plus the smallest observed value. We considered demographics, education, baseline behavioral and CVRFs, and pregnancy-related factors that differed by HDP group in the models (p<0.05). In sensitivity analysis, we further adjusted for Year 30 hypertension, diabetes, and BMI, as well as CVD events. We also performed post hoc analyses among the no HDP, GH, and PE groups using the Tukey-Kramer test. To eliminate potential bias due to differential study attrition, we used inverse probability weighting (IPW) to upweight analyzed participants who were similar to those excluded. We also examined effect modification of HDP by race in fully adjusted models. We examined the number of HDP occurrences (0, 1, 2+) and the time since the first HDP (<15, 15–25, >25 years) by fitting categorical variables or single linear terms and tested for trends. Statistical analyses were conducted with SAS 9.4 and R 4.2.3.

RESULTS

Among the 1,441 parous women, the mean age at first pregnancy was 24 ± 7.1 years, and the mean parity was 1.9 ± 1.3; 202 (14%) women with PE, and 112 (8%) with GH. Women reporting GH or PE, compared to those without HDP, had higher parity and BMI at baseline and Year 30, and were more likely to have elevated blood pressure at baseline and develop hypertension during follow-up (Table 1). Post-hoc tests showed that those with GH were more likely to be Black women and obese, had lower diet quality and educational attainment, and consumed less alcohol than those without HDP. Women with PE did not differ from women without HDP in other baseline cardiovascular and behavioral risk factors, although they were younger at Year 30 and more likely to report gestational diabetes.

Table 1.

Characteristics of the 1,441 Parous Women

N (%) or mean (standard deviation)	No HDP N=1,127	Gestational hypertension N=112	Preeclampsia N=202	p-value
Baseline
Black	564 (50)	68 (60.7)†	115 (56.9)	0.029
Married/live with a partner	321 (28.5)	33 (29.5)	60 (29.9)	0.91
APOE ε4 carrier	301 (31)	25 (25.8)	58 (32.4)	0.5
Family history of dementia	279 (25.2)	21 (19.1)	51 (26)	0.34
Physical activity ≥300 units	552 (49)	44 (39.3)	101 (50)	0.13
Alcohol, drink/week	3.1 (5.4)	1.9 (3.1)†	2.4 (4.3)	0.006
Diet Quality	65 (13.4)	61.2 (12.2)†	63 (12.5)	0.003
Body mass index, kg/m2	23.9 (5)	26.4 (7.1)†	24.8 (5.7)†	<.001
Current smoking	308 (27.5)	29 (26.1)	47 (23.6)	0.52
Elevated blood pressure	16 (1.4)	8 (7.1)†	11 (5.4)†	<.001
Diabetes	3 (0.3)	0 (0)	1 (0.5)	0.72
Dyslipidemia	251 (22.5)	22 (19.8)	49 (24.4)	0.65
Obesity	121 (10.8)	23 (20.5)†	32 (15.9)	0.003
Depression	247 (23.8)	29 (29.6)	54 (28.9)	0.19
Age at first pregnancy, years	24.1 (7.4)	23 (6)	23.5 (6.2)	0.4
Year 30
Age at year 30, years	55.3 (3.6)	55.1 (3.6)	54.5 (3.8)†	0.027
Parity	1.8 (1.3)	2.3 (1.2)†	2.4 (1.3)†	<.001
Education, years	15.2 (2.5)	14.6 (2.5)†	15 (2.4)	0.047
Body mass index, kg/m2	30.5 (7.6)	33.9 (9.1)†	32.5 (8.5)†	<.001
Gestational diabetes	87 (7.7)	11 (9.8)	38 (18.8)†	<.001
Hypertension	392 (34.8)	64 (57.1)†	99 (49)†	<.001
Diabetes	154 (13.8)	18 (16.7)	36 (17.9)	0.26
Obesity	535 (47.6)	69 (62.7)†	111 (55)	0.003
Cardiovascular event	27 (2.4)	6 (5.4)	8 (4)	0.12
Traumatic brain Injury	149 (13.4)	14 (12.7)	32 (16.3)	0.53

HDP = Hypertensive disorders of pregnancy

^†p<0.05 in the post-hoc test using Bonferroni multiple comparisons (continuous variables) and unadjusted logistic regressions (categorical variables).

Midlife Cognition

At Year 30, the mean age was 55.2 ± 3.6 years and mean time between the first HDP pregnancy and cognitive assessment was 26.1 ± 6.9 years. Women reported GH had significantly worse scores compared to women without HDP on most cognitive tests (Table S1). After adjusting for demographics, education, alcohol, diet, BMI, elevated blood pressure, gestational diabetes, and parity, women with GH still showed worse performance on most cognitive domains than those without HDP (Figure 1, Table S2): global cognition (mean score 23.2 vs. 24.0, p=0.018), processing speed (67.5 vs. 71.3, p=0.01), verbal fluency (29.5 vs. 31.1, p=0.033), and a trend for executive function (24.3 vs. 22.6, p=0.09), but not for delayed verbal memory (9.5 vs. 9.0, p=0.12). Further adjustment for Year 30 hypertension, diabetes, and BMI, or alternatively, uncontrolled hypertension, uncontrolled diabetes, and BMI, did not attenuate the associations, nor did adjustments for CVDs. Post hoc analyses did not detect a difference between PE and GH. Using IPW to account for differential attrition did not substantively change the findings (Table S3). There were no differences in cognitive performance between those with PE and those without HDP. We did not observe differences in these associations by race (interaction p-values>0.05). Neither the time since the first HDP pregnancy nor the number of HDP pregnancies was associated with cognitive performance.

Figure 1. Adjusted Mean Cognitive Scores at Year 30 by Hypertensive Disorders of Pregnancy (HDP).

P-values were adjusted for age, race, education, alcohol, diet, body mass index (BMI), elevated blood pressure, gestational diabetes, and parity. †Differences (vs. no HDP) remain significant after additional adjustment for Year 30 hypertension, diabetes, BMI, and CVDs.

Five-year Cognitive Change

We assessed cognitive decline among the 1,270 women who had cognitive assessment both at Years 25 and 30. Compared to women without HDP, those who reported GH had a greater decline on processing speed (mean change −4.9 vs. −2.7, p=0.049) and executive function (−1.7 vs 0.3, p=0.047), after multivariate adjustment (Table S4). The association with executive function remained after additional adjustment for Year 30 hypertension, diabetes, and BMI or by using IPW to account for differential attrition but was attenuated after adjusting for CVDs (p=0.05). In addition, compared to women who did not report HDP, those who reported a history of PE had a greater decline on delayed verbal memory (−0.3 vs. 0.1, p=0.041) after multivariate adjustment, which remained significant after additional adjustment for Year 30 hypertension, diabetes, BMI, and CVDs or by using IPW weights.

White Matter Hyperintensities

Among the subset of 367 women, 52 (14%) women reported PE, and 18 (5%) reported GH. Women with GH and PE had greater WMH in the parietal and frontal lobes, respectively, after adjusting for age, race, and total intracranial volume. The differences in WMH between women with GH and those without HDP were more pronounced than those between women with PE and those without HDP (Figure 2). In addition, parietal and frontal lobe WMH were both associated with worse cognitive performance on global cognition, delayed verbal memory, and executive function in age and total intracranial volume-adjusted models (Table S5).

Figure 2. White Matter Hyperintensities by Hypertensive Disorders of Pregnancy (HDP).

*Significant difference (p<0.05) in comparison with women without HDP in adjusting for age, race, and total intracranial volume.

DISCUSSION

In a prospective cohort of women followed from young adulthood into middle age, we found that a self-reported history of GH was not only associated with worse midlife cognitive performance on multiple domains but also with greater midlife cognitive decline; a self-reported history of PE was associated with greater midlife cognitive decline on delayed verbal memory. These associations were independent of shared risk factors for GH, PE, and cognitive aging, including demographics, education, behavioral and CVRFs, and pregnancy-related factors. Moreover, a history of GH or PE was associated with greater WMH in some brain regions.

Several of our findings are novel. Importantly, we found an early divergence of midlife cognition among women who reported GH. Studies on HDP and cognition in population-based cohorts have been scarce, particularly regarding midlife cognitive decline. We found GH associated with a greater decline in processing speed and executive function, and PE with delayed verbal memory, suggesting a role for HDP in midlife cognitive aging. Among women with GH and PE, we observed greater location-specific WMH, which may explain some cognitive differences associated with HDP.

This study builds upon a prior analysis in CARDIA in key aspects.⁶ While the previous analysis primarily focused on PE, our study extends this to include GH. Furthermore, we investigated cognitive change, providing stronger evidence for the role of HDP in cognitive aging, and we used a more comprehensive cognitive assessment at Year 30, which allowed the detection of cognitive differences in additional domains. Additionally, MRI analysis of WMH advances our current findings, offering insights into potential mechanisms.

Previous studies on HDP and cognition in midlife are limited and have reported mixed findings.^6,8,10 Our results align with another large prospective cohort study that found an association between midlife cognitive performance and GH.⁷ Studies on PE have not found an independent association with midlife cognition and have not examined cognitive decline.^4,6,20,21 Our findings on the association with greater decline on delayed verbal memory suggest that PE may begin to contribute to cognitive decline during midlife. Nevertheless, prior studies and ours may be influenced by misclassification among HDP groups in the context of self-reports with low sensitivity. Differences in the tests and age of cognitive assessment, sample size, and model adjustment might also contribute to the different findings across studies.

The mechanism linking HDP and brain health remains unclear, primarily because the pathophysiology of HDP is poorly understood.¹⁴ Two possible explanations have been proposed, including 1) shared risk factors between HDP and brain health, such as sociodemographic factors, obesity, and diabetes,^12,13 and 2) vascular diseases developed after HDP. In our study, adjusting for sociodemographic, behavioral, and CVRFs did not alter the associations, suggesting that our findings may not be explained solely by these factors. However, pregnancy may be a stress test that reveals poor underlying cardiovascular health or subclinical conditions, more so in women with HDP, increasing their risks of vascular injury later in life.²² Unmeasured risks from other social, behavioral, and CVRFs may still have affected the association we observed. In addition, CVDs during follow-up explained cognitive decline associated with GH, supporting vascular disease as a possible mechanism.

We found a greater burden of WMH associated with GH and PE, suggesting an underlying role for small vessel disease (SVD). Although the exact mechanism remains unknown, increased white matter lesions among women with HDP may be attributable to the elevated risk of subsequent hypertension and CVD after HDP.^1,23 Some studies also suggested that infarction-related gliosis during delivery and vasogenic edema that persists postpartum may contribute to brain MRI WMH.^11,24 The associations we found in processing speed, executive function, verbal fluency, and global cognition but not verbal memory in GH resemble a pattern seen in vascular cognitive impairment. The association of decline in delayed verbal memory with PE may underscore changes seen both in vascular impairment and Alzheimer’s Disease.²⁵

Studies comparing the pathophysiology of GH and PE have been sparse, although emerging evidence supports that GH and PE are distinct entities with different etiologies.¹⁴ While PE typically involves placental and endothelial dysfunction, GH features normal endothelial function but an earlier increase in systemic stress.^15,26 Compared to PE, GH has a stronger genetic correlation with SBP and is associated with a greater risk of recurrent HDP, future hypertension, and ischemic heart disease, reflecting greater cardiovascular vulnerability.^14,27,28 In our study, GH is associated with higher 30-year average SBP and DBP compared to PE (Table S6). Additionally, GH showed a greater WMH compared to PE, indicating a higher burden of SVD. More research is needed to clarify whether potential differences in cardiovascular reserve and SVD in GH and PE could explain their varying associations with midlife brain health.

Although longer follow-up is needed to determine the clinical significance of the moderate cognitive decline observed in our study, this change in midlife long after pregnancy highlights a need to follow women with GH and PE more closely for long-term cognitive outcomes. A lifecourse approach in studying HDP and brain health may lead to a better understanding of underlying mechanisms and critical periods of risk. While more research has been dedicated to understanding the adverse effects of PE, given its known clinical significance, the outcomes of other HDP types, such as GH, should not be overlooked. In light of the research advances that may benefit PE prediction and prevention,^29,30 the landscape of HDP is likely changing. Thus, understanding the long-term outcomes of all HDP is becoming critical, especially given the rapid increase in incidence of all HDP types in the last decades.³¹ Recently, the US Preventive Services Task Force has updated recommendations to expand the scope of screening to all HDP, beyond a sole focus on PE.³² Given the significant implications for cardiovascular and brain health, further research is crucial to determine the optimal timing and duration for HDP screening, counseling, and treatment beyond pregnancy, which may involve not only obstetricians and gynecologists.

In one of the largest cohorts of middle-aged women, we assessed the association of HDP with cognitive function and decline. The strength of the study includes comprehensive cognitive assessment in multiple domains. We also examined brain MRI-measured WMH to explore potential mechanisms linking HDP to cognition, complementing our primary findings. In our analysis, we controlled for various confounders and evaluated the influence of subsequent hypertension, diabetes, BMI, and CVDs. We considered corrections for potential selection bias arising from differential attrition by HDP group. Using IPW in our models yielded consistent results, affirming that our results were not significantly impacted by loss to follow-ups.

This study has several limitations. CARDIA was not originally designed to study HDP, and defining PE and GH based on self-report introduces the potential for misclassification. Similar to other larger epidemiological studies, the self-report GH and PE were highly specific (93–98%) yet not sensitive (8–36%) based on a prior validation study.^33,34 Although the prevalence of HDP was comparable to other population-based cohorts, the misclassification between no HDP and HDP, as well as GH and PE, may contribute to a lower number of GH relative to PE when compared with other cohort studies.⁷ Obtaining more precise HDP data to confirm our findings is critical in future research. Nevertheless, since the pregnancy-specific information, including a history of HDP, was collected during each study exam long before cognitive assessment took place, the misclassification is likely nondifferential with respect to cognition and may potentially bias the observed associations towards the null. Given the young age at cognitive assessment, we may be underpowered to detect additional cognitive differences that may be more prominent later in life. Additional cognitive domains may be assessed in the future, especially for cognitive change. Despite the small sample size in MRI analyses, which may limit our power to find additional differences in WMH overall and by lobes, we were able to detect greater WMH in some brain regions associated with GH and PE. This suggests the need for future studies with larger MRI samples. Those who participated in the brain MRI substudy had lower BMI than the overall cohort, and the observed difference in WMH among women with and without HDP may have been underestimated. Other manifestations of SVD (e.g., microbleeds and silent infarctions) may also contribute to our findings, and we hope to investigate these associations in the future. Detailed information on specific types of PE (early or late onset), HDP severity, and labor induced was not available for a specific HDP pregnancy. Future investigations in a larger sample with more detailed pregnancy histories are needed to elucidate whether these factors may influence the association between HDP and brain health. Our findings may not be generalizable to other racial/ethnic groups as our study focused on Black and White women.

Perspectives

In this prospective cohort of women followed from young adulthood into middle age, women who reported a history of GH had worse cognition and greater cognitive decline in midlife. Women who reported a history of PE had greater midlife cognitive decline. GH and PE were associated with greater WMH in the parietal and frontal lobes, respectively. Women with a history of GH or PE may need to be closely monitored for long-term cognitive outcomes, and their modifiable risk factors should be rigorously optimized.

NOVELTY AND RELEVANCE.

What Is New?

Gestational hypertension (GH) and preeclampsia (PE) are associated with cognitive decline and brain white matter hyperintensities in midlife, with differences in cognitive domains and brain lobes.

What Is Relevant?

Our findings support the role of hypertensive disorders of pregnancy (HDP), including GH and PE, in contributing to cognitive aging among middle-aged women.

Clinical/Pathophysiological Implications?

GH and PE may be linked to cognitive aging via different pathophysiological pathways. Women with HDP may need to be closely monitored for long-term cognitive outcomes, and their modifiable risk factors should be rigorously optimized.

Non-standard Abbreviations and Acronyms:

HDP

hypertensive disorders of pregnancy

PE

preeclampsia

GH

gestational hypertension

WMH

white matter hyperintensities

CVD

cardiovascular disease

CARDIA

Coronary Artery Risk Development in Young Adults Study

APOE

apolipoprotein E

RAVLT

Rey Auditory Verbal Learning Test

DSST

Digit Symbol Substitution Test

BMI

body mass index

SBP

systolic blood pressure

DBP

diastolic blood pressure

CVRFs

cardiovascular risk factors

IPW

inverse probability weighting

LSMean

least squares mean

SVD

small vessel disease