Maternal Vascular Lesions in the Placenta Predict Vascular Impairments a Decade After Delivery

The data that support the findings of this study are available from the corresponding author upon reasonable request.

We enrolled women in 2016-2019 with deliveries in 2008-2009 identified from the Steve N. Caritis Magee Obstetric Maternal and Infant database, a clinical registry of women with detailed pregnancy data abstracted from medical records. About 45% of deliveries (n=4048) had a placental pathology evaluation at the time of delivery, and women with these births were the source population for the study.²⁹ Of the 3947 women who were eligible (alive, nonpregnant and without chronic hypertension or diabetes before the index pregnancy), 1070 declined to participate, 2379 were unable to be contacted, and 498 enrolled. Participation was monitored to reflect the prevalence of MVM in our source population. Enrolled women were, on average, slightly older, more likely to be of Black race and ethnicity and less likely to smoke compared with those who refused or were unable to be contacted. There were no differences in these groups according to rates of APOs (Table S1). Enrolled women attended a cardiovascular screening visit and comprised the study population. The study was approved by the University of Pittsburgh IRB and women provided written informed consent.

Delivery characteristics were abstracted from hospital records including gestational age (based mainly on prenatal ultrasounds), maternal age at delivery, race and ethnicity, prepregnancy body mass index, and smoking status. Preeclampsia was adjudicated via chart review as de novo BP elevations after 20 weeks gestation accompanied by evidence of end organ impairment such as proteinuria.³⁰ Preterm delivery was defined as delivery before 37 completed weeks gestation. SGA was defined as birthweight for gestational age lower than the tenth percentile based on race-specific nomograms.³¹

Presence of placental lesions was abstracted from clinical pathology reports and categorized as MVM using definitions in place at the time of our study. MVM was defined as presence of any of these lesions; we also considered severity by using a weighted score from most to least severe or absent (range, 9–0; vasculopathy > infarct > advanced villous maturation > perivillous or intervillous fibrin deposition) and summed. Diagnostic criteria^32–35 and severity scoring for these features are summarized in Table S2 and presented in Figure 1. Since the launch of our study the Amsterdam consensus statement has added low placental weight to this diagnosis, but our analysis revealed this feature alone is unrelated to maternal cardiometabolic health after delivery so we did not include it.^36,37 For births delivered 2008 to 2009 two perinatal pathologists prepared all evaluations following a standardized protocol using a uniform reporting approach and identical diagnostic criteria. Placental slides for enrolled women were reevaluated (n=459 were available to be retrieved) by study pathologist (W.T. Parks) who was blinded to all clinical information except gestational age. There was substantial agreement between the blinded review and the clinical report of MVM (κ=0.78) and our analysis used the clinical pathology report of MVM as this allowed all enrolled women to be included and our blinded review demonstrated these clinical pathology reports were reliable.

BP was measured during the study visit following a standardized research protocol. After 5 minutes of rest, trained research staff took 3 measurements separated by 1 minute using a validated automated device (Microlife A6 PC/BP 3GUI-8X) and appropriately sized cuff based on measured arm circumference. The average systolic and diastolic BP were calculated. Body mass index was calculated using measured height (via a stadiometer) and weight (Tanita scale TBF-300A) obtained at the study visit after removing shoes, socks and excessive clothing. Cardiometabolic markers were quantified in fasting serum at the UPMC Core laboratory following standardized protocols. Briefly, Beckman Coulter AU 5800 system procedures were used to measure high sensitivity CRP (C-reactive protein; run precision of <5% cardiovascular, and total precision <10% cardiovascular), glucose, HDL (high-density lipoprotein), LDL (low-density lipoprotein), and triglycerides (all cardiovascular <3%). Hemoglobin A1C was quantified with The Tosoh Automated Glycohemoglobin Analyzer HLC-723G8 (<2% cardiovascular) and insulin was measured using the Siemens immulite (IMMULITE 2000) method (total cardiovascular 4.1%–7.3%).

Sublingual microvascular functional density, perfusion, and endothelial glycocalyx barrier integrity (depth) were imaged (Figure 2) using a handheld sidestream dark field video microscope (MicroVision; Wallingford PA) and GlycoCheck software (Microvascular Health Solutions Inc, Salt Lake City, UT) to obtain and analyze video clips of red blood cell (RBC) flow.^38,39 Briefly, perfused vessels in the 5 to 25 µm diameter range were identified in 10 µm segments. Functional microvascular density was determined by the total length of vascular segments perfused with RBCs per area of tissue visualized, expressed as total length in micrometers per mm² of area of tissue (µm/mm²).³⁸ RBC filling was calculated as the percentage of time individual vascular segments were occupied by RBCs, reported as the average across all detected microvascular segments, and vessel diameter was estimated as the median RBC column width determined by the software. The glycocalyx on the luminal surface of endothelial cells is comprised of a negatively charged carbohydrate-rich layer, which contributes to both vascular homeostasis as well as vascular function and signaling.⁴⁰ The perfused boundary region was the portion of the glycocalyx permeable to RBC, with increases reflective of impaired glycocalyx barrier integrity.^41–43

Years of education completed, current smoking, and medications used for BP control were self-reported at the study visit 8 to 10 years after pregnancy. Women also reported sedentary time via a single item from the Global Physical Activity Questionnaire^44,45 and moderate/vigorous activity using the Paffenberger Physical Activity Questionnaire.^46,47 Sodium intake was assessed using the validated Sodium Screener developed by Block (©NutritionQuest 2011).⁴⁸

There were 170 women enrolled with prior MVM in the placenta, and 322 women with no MVM. Those with MVM tended to be heavier before pregnancy (body mass index 27.6 [SD 7.1] versus 26.6 [6.4], P=0.09) and, as expected, the MVM-affected pregnancy was more likely to have resulted in an APO compared with women without MVM (P<0.001; Table 1). For this study, women with MVM placentas were more likely to be evaluated a year sooner postdelivery than women without MVM (8.5 [0.7] years versus 9.5 [0.8], P<0.001), and they reported a lower average sodium intake (3077 [1259] mg/d versus 3310 [1186], P=0.044). There were no significant differences in other assessed lifestyle factors (daily sedentary time or physical activity below recommendations) or subsequent medical or pregnancy conditions (parity, adverse outcomes in other pregnancies, BP medication use, or type II diabetes).

Evaluated 8 to 10 years after delivery, measurements of sublingual microvascular features were notably different between the groups. Compared with women with no MVM, the 5 to 25 µm diameter microvessels of women with MVM had a smaller perfused boundary region (2.03 µm [interquartile range, 1.83–2.20] versus 2.07 [1.94–2.23], P=0.018) accompanied by a greater percentage of time that vessels were perfused with RBCs (RBC filling 73% [0.69–0.76] versus 71% [0.68–0.74], P=0.006). They also tended, on average, to have smaller diameter microvessels and lower vessel density (functional rarefaction) compared with women without MVM history although these unadjusted differences were not statistically significant (estimated mean vessel diameter 8.7 µm [8.2–9.4] versus 9.0 [8.3–9.6], P=0.087; vessel density 3610 [interquartile range, 2920–4850] μm/mm² versus 3930 [3110–4690], P=0.326). After adjustment for confounders (race, age, follow-up time, prepregnancy obesity, sodium, preeclampsia, SGA birth, preterm delivery, and antihypertensive medication use), women with MVM had significantly lower vessel density (−410 number/mm², [CI, −740 to −80] P=0.015), along with higher RBC filling percent (P=0.004) and smaller perfused boundary region (P=0.031).

There were also cardiometabolic differences 8 to 10 years after pregnancy in women with a history of MVM. Those with MVM had higher diastolic BP (77.7 [10.7] mm Hg versus 75.3 [10.9], P=0.018) compared with those without MVM. This diastolic BP difference according to MVM history persisted after accounting for covariates (Table 2). Both total and LDL-cholesterol were higher in women with MVM (184 versus 174 mg/dL for total cholesterol, P=0.025; 110 versus 101 mg/dL for LDL, P=0.010) and these differences persisted after adjustment for confounders (P=0.016 and 0.031, respectively). MVM was unrelated to other metabolic markers (body mass index, waist circumference, HDL-cholesterol, triglycerides, insulin, or glucose). Results were similar when analyzed as a MVM severity score, where the most severe lesions were weighted more heavily (mean severity score for women with MVM was 2.8 [SD 1.9]). MVM severity was correlated with more adverse microvascular features (r=−0.14 for estimated mean vessel diameter and r=0.14 for RBC filling percent), LDL-cholesterol (r=0.11), diastolic BP (r=0.17), and systolic BP (r=0.15; P<0.05 for all measures). Similarly, after accounting for confounders, systolic and diastolic BP, microvessel diameter and RBC filling were more adverse as MVM severity increased (Table 3).

We then examined whether these MVM associations varied by occurrence of each APO. The associations between MVM lesions, microvascular features and cholesterol were detected regardless of APOs (all P interaction >0.30). In contrast, the associations between MVM lesions and BP were more complex. While higher diastolic BP was detected in women with preeclampsia or SGA regardless of MVM (P interaction=0.507 and 0.528, respectively), diastolic BP was higher in women with preterm birth only when accompanied by placental evidence of MVM (preterm+MVM +4.7 mm Hg [95% CI, 1.02–8.34], preterm−MVM −1.9 mm Hg [−5.4 to 1.5], P interaction, 0.019; Figure 3). Results were similar but not as precise for SBP (P interaction=0.106). When evaluated according to strata of APO with and without MVM (Table S3), findings were consistent such that BP abnormalities were more strongly associated with APO occurrence while microvascular and atherogenic lipid findings were associated with MVM occurrence.

We present a novel approach to evaluating maternal cardiovascular risk a decade after pregnancy by focusing on occurrence of MVM lesions in the placenta as a marker of later life vascular susceptibility. These lesions comprise a group of related placental abnormalities that are thought to derive from early maldevelopment of the spiral arteries—the small arteries that supply the maternal blood to the placenta. While outside of pregnancy atherosis is a feature of large vessels (coronary arteries, aorta, etc), the spiral arteries (50–100 µm) are among the larger vessels of the uterine microvasculature. Early in gestation, deep migration of fetal extravillous trophoblasts into the maternal decidua and inner myometrium leads to the remodeling of these vessels, with dissolution of their smooth muscle walls and replacement of the walls by extravillous trophoblast-derived fibrinoid. Remodeling converts the otherwise narrow, tightly coiled vessels into looser funnel-shaped structures that have lost the capacity to respond to maternal vascular signals. When remodeling is incomplete or fails to occur, the spiral arteries retain their muscular walls and vascular tone. The resulting blood flow into the placenta retains its arterial character, with pulsatile, high velocity flow that mixes poorly and potentially damages the placental chorionic villi. The specific lesions of incomplete or failed remodeling are termed decidual vasculopathy and include incomplete remodeling of basal plate arteries, fibrinoid necrosis and atherosis of spiral artery walls, and mural hypertrophy of the spiral artery branches in the extraplacental membranes. Decidual vasculopathy may then lead to occlusion of spiral arteries with infarction of the overlying placenta or to hypoxic/ischemic damage to the placental parenchyma, including advanced (or accelerated) villous maturation, an alteration in villous growth in response to poor fetal perfusion, or increased fibrin deposition, a response to localized necrosis of the syncytiotrophoblast covering the chorionic villi.

Despite reasonable agreement about which individual lesions fall within the diagnostic category of MVM, there is little agreement on which features or how many individual lesions are necessary for a diagnosis of MVM. In fact, even the Amsterdam criteria, the most recent reassessment and recategorization of placental pathology features, only described the features of MVM and pointedly did not attempt to grade or stage this diagnosis. Given the lack of direction in this area, our study used a 2-pronged approach. Our primary approach was to follow the current clinical guidelines that any single MVM lesion was sufficient for a diagnosis of MVM. The risk with using this low-threshold, minimal criterion for MVM, however, is that contributing lesions likely do not represent equal probabilities for the development or severity of MVM, and less impactful criteria may dilute the significance of the more representative criteria. To mitigate this risk, we explored a severity index that assigned each placenta a score based on the types and numbers of individual lesions. Importantly, significant differences in 3 vascular domains examined (microvasculature, cholesterol, and BP) were detected using a minimal, dichotomous criteria for MVM as well as a severity score. These findings are consistent with our hypothesis that MVM may be manifestations of an underlying maternal vascular phenotype relevant to later vascular health. Supporting this interpretation is the lack of association between MVM and components of metabolic syndrome (waist circumference, HDL-cholesterol, glucose, triglycerides), pointing to a vascular rather than metabolic abnormality. We also note that the inclusion of low placental weight, in addition to the direct assessment of spiral artery-related abnormalities, did not improve the precision of associations with later life maternal vascular impairments. It is possible that low placental weight is the composite result of these impairments and that small placentas without these MVM findings are physiologically (versus pathologically) small. It is also possible that low placental weight may be more reflective of fetal rather than maternal sequelae or is a less precise marker of MVM. These possibilities warrant further study.

Our results present several notable findings. Aligned with our hypothesis that women with MVM would have an adverse vascular profile, most robust were the changes in the sublingual microvasculature as assessed by sidestream dark field imaging. Using only the minimal criteria for MVM, women with MVM placentas had lower vessel functional (perfused) density, higher RBC filling percent, and a lower perfused boundary region. Effectively, these women had functional rarefaction of their microvasculature with increased flow in this smaller number of perfused vessels. When these same variables were then evaluated in the context of MVM severity, those women with worse MVM (higher severity score) had even narrower vessel diameters. The other microvascular parameters also trended to severe values (further decreased vessel density and increased RBC filling percentage). Importantly, these variations in the microvasculature were independent of APOs, suggesting that the findings may represent an intrinsic propensity to an altered microvasculature in those women with MVM.

Microvascular impairments have been implicated in the occurrence of MVM in the placenta due potentially to underlying hemodynamic aberrations. For example, low plasma volume, a well-established feature of preeclampsia, may signal cardiovascular reserve capacity and reduced reserve may contribute to and be a consequence of microvascular damage.^12,49 Reduced functional capillary density 5 to 10 years after a hypertensive pregnancy has been reported using capillaroscopy.⁵⁰ To our knowledge, however, our study may be a first to assess microvascular features 8 to 10 years after pregnancy associated with specific placental pathological features. Interestingly, not all changes were in the hypothesized direction. Of note, loss of glycocalyx integrity quantified as deeper RBC penetration (larger perfused boundary region) has been associated with renal disease,⁵¹ lacunar stroke,⁵² and coronary heart disease,^38,53 and our prediction was that MVM cases would have larger perfused boundary regions. In contrast, our findings were opposite, such that women with MVM had a smaller perfused boundary region accompanied by a higher RBC filling percentage, lower vessel density and, on average, smaller diameter microvessels. We can only speculate about these unexpected findings. One possibility is that they reflect a maternal microvasculature that may be less able to respond to stressors, such as pregnancy, and later life adaptations required for healthy aging, but this warrants future study.⁵⁴ Aligned with this possibility, our group has reported that women with preeclampsia and MVM have lower cognition scores and blunted cerebrovascular reactivity in relevant brain regions after pregnancy compared with nonpreeclampsia cases or preeclampsia cases without MVM.⁵⁵

Consistent with our hypothesis that MVM will identify susceptibility to vascular impairments that may be distinct from the clinical outcomes of preeclampsia, preterm birth, and SGA, our findings indicate that diastolic BP, total and LDL-cholesterol, and small vessel sublingual derangements appear to be related to MVM occurrence that is not entirely explained by these APOs. Although the magnitude of these vascular differences assessed 8 to 10 years after pregnancy were modest, they were detected in young adult women at a mean age of 38 years and thus may be early markers of a high-risk profile warranting further study. It is noteworthy that large vessel (BP) and small vessel (sublingual) sequelae of placental impairments presented differently. While diastolic BP was higher in women with preeclampsia or SGA regardless of placental features, only the subset of women with preterm births accompanied by vascular placental lesions had higher BP. In contrast, women with MVM lesions had a distinct sublingual microvascular phenotype regardless of an accompanying clinical presentation of preeclampsia, SGA or preterm birth. Taken together, our findings raise the possibility that vascular impairments revealed in the placenta as well as the maternal and fetal responses which are manifest as clinical adverse outcomes may accrue in both the large and small vessels after delivery. For example, MVM was not associated with hypertension after delivery in our data, but whether progression to hypertension may converge with microvascular features detected in women following MVM and contribute to overt CVD warrants additional study. The evidence that traditional cardiometabolic risk factors, including hypertension, do not fully explain the link between APOs and CVD supports this possibility.⁵⁶

Several lines of evidence identify the placenta as a marker of preexisting susceptibility to vascular impairment as well as a contributor to vessel injury remote from the uterus and from the pregnancy event. First, our group and others have reported that Black race,^57,58 obesity,⁵⁹ and neighborhood deprivation⁶⁰ all are associated with higher risk of MVM and thus preconception susceptibility is likely. There is also accumulating evidence that a poorly perfused placenta leads to oxidative stress and placentally derived materials that are shed into the maternal circulation that may damage the endothelium systemically.⁸ How or if these injuries persist postpartum are unknown, but our data suggests they are detectable 8 to 10 years after delivery. While many prior studies have investigated individual APOs, we opted instead to study what we hypothesized would be a shared vascular pregnancy phenotype, a malperfused placenta.⁶¹ This approach makes comparisons challenging, but there is evidence that 7 months after delivery, women with preeclampsia with accompanying decidual vasculopathy (the most severe, chronic and likely underlying precursor lesion to other MVM features) had higher diastolic BP, lower left ventricular stroke volume, and higher total peripheral vascular resistance compared with women with preeclampsia but no decidual vasculopathy.¹² A large, prospective, community based US cohort has also reported that mural hyperplasia, a feature of decidual vasculopathy characterized by thickening of the smooth muscle wall of the spiral arteries, is associated with risk of hypertension a decade after delivery when accompanied by modest BP elevations in pregnancy.⁶² Consistent with our findings, we have reported in a separate cohort that women with normotensive preterm births accompanied by placenta vascular malperfusion have a more atherogenic profile a decade after delivery compared with those with preterm births without placental malperfusion.¹⁴ Other smaller studies that evaluated mothers soon after delivery have reported higher LDL-cholesterol and triglycerides in women with atherosis, a subtype of decidual vasculopathy.^63,64 Our data reporting modest BP elevations and a more atherogenic lipid profile following MVM that was independent of APOs are consistent with these findings and extend them to women assessed a decade after delivery.

Strengths of our study included a comprehensive approach to evaluating the maternal vasculature a decade after pregnancy that included novel (microvascular) and traditional measures. This is important because CVD has sex-specific features. Women present with more coronary microvascular disease and heart failure with preserved ejection fraction compared with men.^65,66 BP increases twice as fast from age 20 to 40 in women versus men, and these patterns predict CVD outcomes more commonly in women compared with men, so identifying the sex-specific contributors is essential.⁶⁷ APOs such as preeclampsia and delivery of preterm or SGA infants, which themselves are heterogenous outcomes, are associated with excess CVD risk but when added to risk prediction models do not appear to improve precision.⁶⁸ These findings suggest that subtypes and novel precursors to established risk factors, such as those with evidence of placental vascular impairments, may be informative.

Potential limitations of our study include the requirement that a woman have had a delivery from which the placenta was sent for examination. Since only 45% of placentas were evaluated during this time, this may have biased the findings towards the null as our non-MVM group was not completely normal. Future studies of nonselected evaluations are warranted to expand and replicate our findings. Another potential limitation is that no distinction was made among the different MVM lesions for inclusion into the study. Subsequent studies focused more tightly on specific lesions such as decidual vasculopathy or advanced villous maturation are still needed to more precisely delineate MVM risks. The use of placental pathology reports to identify study subjects is also a potential limitation as diagnosis of placental MVM lesions is known to have high interobserver variability. To mitigate this, the placental pathology slides from all available cases were rereviewed and we demonstrate good concordance between clinical and blinded research review. Although we considered occurrence of adverse outcomes in other pregnancies, placental evaluations for these were not available to be included. Future studies are needed to examine if recurrent MVM in subsequent pregnancies may signal a more severe postdelivery maternal vascular phenotype.

APOs are established harbingers of excess hypertension and CVD in women.¹ For each of these adverse outcomes (preeclampsia, preterm birth and a SGA infant) a subset of the placentas will show evidence of MVM on pathological examination. Studies such as ours are beginning to evaluate the relations between these APOs, an aberrant placental vascular phenotype and subsequent cardiovascular risk in women. Our study, which evaluated women 8 to 10 years after delivery, has demonstrated that MVM lesions are indeed associated with significantly worse vascular parameters, including rarefaction of the microvasculature, elevated total and LDL-cholesterol, and increased diastolic BP. These findings suggest that an underlying vascular phenotype may connect these different clinical entities, and they may help guide future studies attempting to define the molecular and genetic aspects of this vascular phenotype.

We gratefully acknowledge the study participants and study staff at the Magee-Womens Research Institute.