Biological Changes in the Pregnancy-Postpartum Period and Subsequent Cardiometabolic Risk--UPSIDE MOMS: A Research Protocol

Abstract

Multiple physiological changes occur in pregnancy as a woman’s body adapts to support the growing fetus. These pregnancy-induced changes are essential for fetal growth, but the extent to which they reverse after pregnancy remains in question. For some women, physiological changes persist after pregnancy and may increase long-term cardiometabolic disease risk. The National Institutes of Health-funded study described in this protocol addresses a scientific gap by characterizing weight and biological changes during pregnancy and an extended postpartum period in relation to cardiometabolic risk. We use a longitudinal repeated measures design to prospectively examine maternal health from early pregnancy until three years postpartum. The aims are: (1) identify maternal weight profiles in the pregnancy-postpartum period that predict adverse cardiometabolic risk profiles three years postpartum; (2) describe immune, endocrine, and metabolic biomarker profiles in the pregnancy-postpartum period, and determine their associations with cardiometabolic risk; and (3) determine how modifiable postpartum health behaviors (diet, physical activity, breastfeeding, sleep, stress) (a) predict weight and cardiometabolic risk in the postpartum period; and (b) moderate associations between postpartum weight retention and downstream cardiometabolic risk. The proposed sample is 250 women. This study of mothers is conducted in conjunction with the Understanding Pregnancy Signals and Infant Development study, which examines child health outcomes. Biological and behavioral data are collected in each trimester and at 6-, 12-, 24-, and 36-months postpartum. Findings will inform targeted health strategies that promote health and reduce cardiometabolic risk in childbearing women.

Summary

Results from this longitudinal cohort study will build on what is known and provide important information on how pregnancy related physical, biological, behavioral and psychosocial factors may influence the development of maternal cardiometabolic risk. Findings will offer novel insights into major sources of risk. Our focus on modifiable behavioral factors that mitigate adverse risk profiles may inform targeted preventive interventions.

Pregnancy adaptations that are essential for fetal growth and development include weight increases as well as immune, endocrine, vascular, and metabolic changes. For some women, these changes may persist well past the traditional 6-week postpartum period conferring increased risk for cardiometabolic diseases such as type 2 diabetes and cardiovascular disease (Christian & Porter, 2014; Gunderson, 2009; McClure et al., 2013). Higher body weight, altered lipid profiles, glucose dysregulation, elevated inflammation, and elevated blood pressure (BP), all of which may occur in pregnancy and can persist well beyond parturition, are predictive of cardiometabolic diseases (Cannon, 2007). Having at least one major cardiometabolic risk factor above the considered healthy range classifies women as at risk for future cardiometabolic diseases (Mosca et al., 2011).

Physiological alterations associated with pregnancy may become cardiometabolic risks when they extend beyond pregnancy and the traditional postpartum period. For example, it is estimated that nearly 50% of pregnant women gain weight in excess of established guidelines (the former Institute of Medicine now National Academy of Medicine) (Deputy et al., 2015), and nearly 75% of women remain above their pre-pregnancy weight at six months postpartum (Hollis et al., 2017). There is a 50–60% decrease in insulin sensitivity during pregnancy, which can persist after delivery (Catalano, 2014; Kitzmiller et al., 2007). Furthermore, approximately 30% of those who have hypertension in pregnancy remain hypertensive two years after pregnancy (Giorgione et al., 2020). It is unclear why some women, but not others, show persisting physiological alterations after pregnancy. The extent to which these factors confer subsequent cardiometabolic disease risk is also poorly understood. What is clear is that the pregnancy-postpartum period may be an especially sensitive time to study risk mechanisms that increase, or prevent, cardiometabolic risk.

We hypothesize that changes in body weight, body composition, and select biomarkers (i.e., immune, endocrine, and metabolic) across pregnancy and an extended postpartum period predict future cardiometabolic risk (Figure 1). Additionally, we also predict that certain health behaviors and risk factors (i.e., diet, physical activity, breastfeeding, sleep, and stress) during this period predict and/or modulate a woman’s future cardiometabolic risk (Gunderson, 2014; Renault et al., 2017). This National Institutes of Health/National Institute of Nursing Research (NINR)-funded study addresses a scientific gap by contributing longitudinal assessments of maternal weight, biology and behavior during pregnancy and an extended postpartum to examine the existence of, or resolution of, pregnancy-induced alterations and their association with cardiometabolic risk.

Figure 1:

Conceptual Model

Cardiometabolic Risk

In reproductive age women cardiometabolic health (or risk) is typically assessed by determining whether body weight, lipid profiles, glucose metabolism and BP are within a ‘healthy’ range. Refinement of these standard measures is potentially informative and may provide greater insight into cardiometabolic risk status. For example, markers highly predictive of future risk such as central fat mass, Apolipoprotein B100 (ApoB100), non-high density lipoprotein cholesterol (HDL-C), high sensitivity C-reactive protein (hs-CRP), and 1-hour oral glucose tolerance test (OGTT) can be used. Central fat mass is associated with chronic inflammation, insulin resistance, and undesirable metabolic effects (Gilmore et al., 2015). Atherogenic lipoproteins cause the atherosclerotic disease process in arteries and ultimately, cardiovascular disease events (Grundy et al., 2019). ApoB100, an apolipoprotein that exists on the surface of such atherogenic lipoproteins, is the gold standard for quantifying atherogenic lipoproteins, and is highly predictive of future cardiovascular disease events (Jacobson et al., 2015). High density lipoprotein cholesterol has anti-atherogenic effects. Total cholesterol is a measure of all cholesterol-containing molecules. Subtracting HDL-C from total cholesterol generates non-HDL-C, an accurate measurement of atherogenic lipoprotein particle number (Ridker et al., 2005). High sensitivity-CRP is an inflammatory marker that predicts cardiovascular disease outcomes; high levels indicate increased risk for a cardiovascular disease event (Grundy et al., 2019). An elevated 1-hour post-load glucose value (≥ 155 mg/dl) is associated with cardiometabolic risk (Bianchi et al., 2013; Fiorentino et al., 2016). Among subjects with a normal glucose tolerance, those with a 1-hour post-load glucose level ≥ 155 mg/dl have higher BP, lower HDL-C, impaired insulin sensitivity and β-cell function (Bianchi et al., 2013), and a higher risk of metabolic syndrome (Park et al., 2019).

Alongside the above specific biological markers are composite clinical risk measures, such as the 2018 American College of Cardiology/American Heart Association (ACC/AHA) lifetime cardiovascular disease event risk estimator. This clinical risk estimator takes into account race, age, sex, HDL-C, total cholesterol, a diagnosis of diabetes mellitus, treatment for hypertension, actual systolic BP, and smoking status, and is valid for individuals age 20 and older (Grundy et al., 2019). It is available at https://tools.acc.org/ascvd-risk-estimator-plus/#!/calculate/estimate/.

Maternal weight profiles and cardiometabolic risk

Pre-pregnancy obesity, excessive gestational weight gain (GWG) and postpartum weight retention are common in child-bearing women (Deputy et al., 2015; Hollis et al., 2017). Overweight and obesity prior to pregnancy and high GWG predict future health risks including long-term weight retention and obesity (Groth et al., 2013; Rasmussen & Yaktine, 2009), central adiposity (Fraser et al., 2011), type 2 diabetes, hypertension (Fraser et al., 2011; Groth et al., 2013), metabolic syndrome, and adverse metabolic profiles (Rooney et al., 2005). All of these outcomes are, in turn, associated with future cardiometabolic disease (Daviglus et al., 2004; Mosca et al., 2011; Pencina et al., 2009). A main limitation of prior studies of maternal weight is reliance on self-reported weight measures and lack of biological measures such as immune, metabolic and endocrine markers, which limits characterization of patterns of weight loss, retention, and/or gain in the early postpartum years (Mamun et al., 2010; McClure et al., 2013). Plus, there is only limited evidence suggesting that cardiometabolic profiles are related to weight patterns over the first year postpartum (Kew et al., 2014).

Immune, endocrine, and metabolic biomarkers and cardiometabolic risk

In non-pregnant adults, several biomarkers are associated with cardiometabolic risk. For example, levels of inflammatory cytokines, such as interlukin-6 (IL-6) are associated with obesity (Ellulu et al., 2017), visceral fat mass (Fontana et al., 2007), insulin resistance (de Luca & Olefsky, 2008), and elevated BP (Bautista et al., 2005). Likewise, cortisol, an endocrine biomarker, is associated with obesity (Björntorp & Rosmond, 2000), visceral fat mass (Drapeau et al., 2003), insulin resistance (Morais et al., 2019), and elevated BP (Whitworth et al., 2005). The metabolic markers leptin and adiponectin may directly and in combination (leptin/adiponectin ratio [Lekva et al., 2017]) predict an increase in cardiometabolic risk in non-pregnant individuals (Frühbeck et al., 2018). Immune (i.e., inflammatory cytokines), endocrine (i.e., cortisol), and metabolic (i.e., leptin, adiponectin, glucose-insulin homeostasis) biomarkers are altered in pregnancy, and these alterations may have a persistent pattern of disruption beyond pregnancy (Blackmore et al., 2011; Christian & Porter, 2014). Whether levels of immune, endocrine, and metabolic biomarkers during and after pregnancy predict future cardiometabolic risk is not clear.

Modifiable Health Behaviors and Cardiometabolic Risk

Modifiable health behaviors during postpartum may mitigate cardiometabolic risk development for some women, possibly by altering weight retention (Gunderson, 2014; Renault et al., 2017). The selected behaviors for this study, namely, diet, physical activity, breastfeeding, sleep, and stress, are modifiable factors that could be targeted to improve cardiometabolic risk in childbearing women. Diet has a role in weight regulation and cardiometabolic health by maintaining energy balance and macro- and micro-nutrient intake (Meng et al., 2018). Physical activity can improve cardiometabolic health by regulating weight, lowering inflammation and improving insulin sensitivity (Jelalian & Sato, 2012; Pedersen & Febbraio, 2008). Breastfeeding is thought to ‘reset’ maternal metabolism, reducing cardiometabolic risk (Peters et al., 2016; Stuebe & Rich-Edwards, 2009). Substantial sleep disruption is typical in the postpartum and is associated with energy intake (Crispim et al., 2011), physical activity (Kline, 2014), maternal physical and mental health, as well as adiposity, body weight, and inflammation (Taveras et al., 2011), Stress is a natural target for research on weight and biomarkers because the immune, endocrine, and metabolic systems are stress-responsive systems; stress may mediate how these biomarkers predict cardiometabolic risk (Slavich & Irwin, 2014). To date, research has considered these behaviors in relative isolation; we will consider the collective influence of these factors for modifying subsequent cardiometabolic risk.

Purpose and Aims

The purpose of this longitudinal study is to improve our understanding of the mechanisms by which biology and health behaviors during and after pregnancy may increase cardiometabolic risk in childbearing women. Therefore, we are examining relationships among maternal body weight/composition changes, immune, endocrine, and metabolic markers during pregnancy and an extended postpartum period, as well as maternal behaviors during the first two years postpartum (6-, 12-, 24-months postpartum) in relation to cardiometabolic risk measures at 36-months postpartum. The aims are to: 1) identify maternal weight profiles in the pregnancy-postpartum period [6m] that predict adverse downstream cardiometabolic risk (Hypothesis 1: Maternal weight profiles (pre-pregnancy body mass index [BMI], GWG, postpartum weight retention) will predict adverse cardiometabolic risk at 3 years postpartum); 2) characterize profiles of immune, endocrine, and metabolic biomarkers in the pregnancy-postpartum period, and determine their associations with downstream cardiometabolic risk (Hypothesis 2a. Maternal immune, endocrine, and metabolic biomarker profiles in pregnancy and postpartum will be associated with subsequent cardiometabolic risk. Hypothesis 2b. Maternal immune, endocrine, and metabolic biomarker profiles will mediate the association between weight profiles and cardiometabolic risk); and 3) determine how modifiable postpartum health behaviors (diet, physical activity, breastfeeding, sleep, stress) (a) predict weight and cardiometabolic risk in the postpartum period; and (b) moderate associations between postpartum weight retention and downstream cardiometabolic risk (Hypothesis 3a: Health promoting behaviors will be associated with weight and biomarker profiles in the postpartum period. Hypothesis 3b. Relationships between adverse weight and biomarker profiles in the postpartum period and downstream cardiometabolic risk will be attenuated among women who engage in health promoting behaviors).

Methods

Design

This ongoing study is a prospective longitudinal cohort design. UPSIDE MOMS has been approved by the Institutional Review Board at the participating institutions, and all participants provide written informed consent prior to participation.

Setting

UPSIDE MOMS participants are recruited from women in an ongoing longitudinal cohort study, Understanding Pregnancy Signals and Infant Development (UPSIDE), based at the University of Rochester Medical Center (O’Connor et al., 2021).

Target Population

The sample will include women enrolled in the UPSIDE study who delivered a full-term infant. UPSIDE began enrolling women in their first trimester of pregnancy in December 2015 and is following the resulting children who were born at term with regularly occurring study visits. Inclusion criteria for UPSIDE (and therefore for UPSIDE MOMS) are: <14 weeks gestation; age ≥18 years old, medically normal risk pregnancy, singleton pregnancy, plan to deliver at study hospitals, and ability to provide informed consent in English. Women are excluded if they have a history of major psychiatric issues, major endocrine disorders, or significant obstetric problems. Mothers participating in UPSIDE and who delivered a full term infant are invited to enroll in UPSIDE MOMS at a postnatal study child visit. UPSIDE MOMS began recruitment in August 2017 and is continuing to enroll participants.

Sample size

The targeted sample size for UPSIDE MOMS is 250 women. Our power analysis estimated the sample size needed to detect associations between postpartum weight change and cardiometabolic risk, as well as to detect associations among immune, endocrine, and metabolic biomarker profiles in pregnancy and postpartum and subsequent cardiometabolic risk. We also determined the sample size needed to examine associations among health promoting behaviors with weight and biomarker profiles in the postpartum period. All power calculations assume a 2-sided test with alpha=0.05, and are based on anticipated distribution across BMI categories as seen in prior work (Fernandez et al., 2015). With 250 women we will achieve 80% power to detect an association between postpartum weight and cardiometabolic risk if the correlation between these variables is 0.18. If the correlation was 0 for overweight/obese women and 0.24 for normal weight women, we would have 80% power to detect the association among normal weight women in an interaction model (Hypothesis 1). Similarly, we would achieve 80% power for Hypothesis 2a if the correlation between a biomarker and cardiometabolic risk was 0.18. For hypothesis 3a we would achieve 80% power to detect a significant interaction if the association between weight loss and risk was 0.24 in one tertile of modifiable health behavior, 0 in another tertile, and −0.24 in the third tertile. With 200 women, we would obtain 80% power for Hypothesis 1 and 2a/b with a correlation of .20, for Hypothesis 1 with a correlation of 0.27 among normal weight women, and for Hypothesis 3a/b with correlations of 0.26, 0, and −0.26 in the three tertiles.

Procedures

This study includes a total of three prenatal study visits, one in each trimester, and four postpartum study visits at 6-, 12-, 24- and 36-months (Table 1) that are conducted at the University of Rochester Medical Center in dedicated research space. Following consent, data collection at each study visit is completed by trained study team members.

Table 1.

UPSIDE MOMS Study Activities by Time Point

Activities	T1	T2	T3	6M	12M	24M	36M
Anthropometrics
-Weight, height, waist-hip circumferences	x	x	x	x	x	x	x
-Bioelectrical impedance analysis				x	x	x	x
Biological specimens: blood & urine collection	x	x	x	x	x	x	x
Modifiable behaviors: diet*, physical activity, breastfeeding**, psychosocial stress, sleep	x	x	x	x	x	x	x
Medical record: chart review/data abstraction	x	x	x	x
Questionnaires: demographics, health, lifestyle	x	x	x	x	x	x	x

T1, T2, and T3 are visits at the first, second and third trimesters during pregnancy. 6M, 12M, 24M and 36M are visits at 6, 12, 24, and 36 months postpartum.

^*Dietary recalls collected x2 at T2 and T3 & x2 at all postpartum time points;

^**Breastfeeding assessed postpartum.

Prenatal visits

Prenatal study visit data collection includes weight, waist and hip circumference (1^st trimester only), administration of questionnaires including demographics, general health, physical activity and behaviors, and blood and urine collection. Remote data collections after each prenatal visits includes saliva cortisol samples and two dietary recalls in the 2^nd trimester. At the end of a study visit, a review of cortisol collection is completed--mothers are instructed to provide saliva samples (via passive drool methods) at wake-up, 45 minutes after wake-up, 2.5 hours after wake-up, 8 hours after wake-up, 12 hours after wake-up and bedtime and return the samples to the study team (Table 1). There is also review of what to expect regarding dietary recall data collection via telephone.

Postpartum study visits

Postpartum study visit data collection includes weight, waist and hip circumference, body composition, BP, administration of questionnaires including changes in demographics, general health, physical activity and behaviors, and blood and urine collection. In addition, at the 24- and 36-month postpartum study visits, a one-hour oral glucose tolerance test is administered (unless a woman has been diagnosed with diabetes). Remote data collection after each postpartum study visit consists of two dietary recalls, 5-days of wearing an accelerometer, and diurnal saliva collections. These activities are described in greater detail in the variables and measures section. Briefly, at the end of a study visit, a study team member instructs participants on the use of the actigraph including wear time, use of the activity diary, number of wear days and hours/day, start time, and plans for returning the actigraph (i.e. via mail, team pick-up, or drop off). A review of cortisol collection and dietary recalls is also completed as is done during prenatal study visits. Participants receive remuneration at the end of a study visit and after completion of remote activities.

Variables and Measures

At the prenatal baseline visit, sociodemographic information is collected including age, race, education, current occupation and income, and marital status. Sociodemographic data that could change over time (e.g., current occupation) are updated at each study visit. Additional measures are collected as described below.

Prenatal and Birth Record Abstraction

Clinical data are abstracted from the electronic medical records. Relevant fields include reproductive history, prenatal visit records (e.g., BP, weight), medical, surgical, and gynecological history, ultrasound measurements, and clinical lab values for the index pregnancy. After delivery, data are abstracted regarding gestational age at admission, labor onset and duration, membrane rupture, highest intrapartum temperature, group B streptococcus status, maternal white blood cells, mode of delivery, any complications, medication use, and maternal morbidity.

Anthropometrics

Anthropometric measurements are collected using validated protocols procured from the PhenX toolkit (PhenXToolkit: Anthropometrics). Waist to hip ratio, a commonly used estimator of cardiometabolic risk (Perry et al., 2008; Qi et al., 2015), is calculated based on waist and hip circumference as measured in the first trimester and at each postpartum study visit. At all postpartum study visits mother’s weight is collected using a bioelectrical impedance analysis (BIA) machine (Tanita body composition analyzer, MC-780U) that provides output on BMI, total body and segmental fat percentage and weight, segmental muscle mass, fat free mass, basal metabolic rate, and visceral fat rating. Because BIA may not be valid in pregnant women (Widen & Gallagher, 2014), it is only used in the postpartum

Blood Pressure

Postpartum BP is measured using a manual sphygmomanometer after the participant has rested quietly in a seated position with feet flat on the floor for five minutes. Resting seated BP is measured three times on each arm, with 2 minutes of rest between measurements. The averages of BPs will be used for analysis and for entering in the ACC/AHA risk estimator.

Dietary Intake

Dietary recalls are collected remotely on two separate days following a study visit via telephone by a trained nutrition coordinator using a standardized approach, the United States Department of Agriculture’s (USDA’s) Automated Multiple Pass Method (Johnson et al., 1996; Moshfegh et al., 2008), which is then entered into Nutrition Data System for Research software (Feskanich et al., 1989). In addition to providing energy intake, and macro- and micro-nutrients, these data can be used to examine dietary patterns.

Physical Activity

Physical activity is assessed using the self-administered Pregnancy Physical Activity Questionnaire (PPAQ) (Chasan-Taber et al., 2004) in both pregnancy and postpartum at all study visits. The PPAQ assesses household/caregiving, occupational, sports/exercise, sedentary, light, moderate, and vigorous activities. The duration of time spent in each activity is multiplied by its intensity to arrive at a measure of average weekly energy expenditure (MET-h·week-1). The PPAQ is the only validated physical activity questionnaire specific to pregnancy. Although designed for pregnancy, we continue it into postpartum to maintain consistency of measurement. A lack of validation of the PPAQ for postpartum women is a limitation of continuing this measure after pregnancy.

In addition, accelerometers (Actigraph GT3X and GT3X+; Pensacola, FL, U.S.) are used to provide an objective assessment of physical activity following each postpartum study visit. Accelerometers are worn on an elastic belt around the waist during waking hours and removed for bathing or water activities. An elastic belt around the waist is the preferred method of collecting physical activity data with an accelerometer (Cleland et al., 2013). Devices have sensors sensitive and oriented to vertical acceleration and are worn at the hip to record physical activity because this allows detection of body acceleration and deceleration with movement (Quante et al., 2015). Women are asked to wear the accelerometer for a minimum of five consecutive days and to keep an activity diary for the days they wear the actigraph. Data are considered valid if at least 4 days of >=10 hours of wear time is collected, an amount considered sufficient to provide an estimate of habitual activity (Migueles et al., 2017). Activity based on accelerometer results will be categorized into sedentary, light, moderate, and vigorous activity (Kozey-Keadle et al., 2011; Sasaki et al., 2011).

Breastfeeding

Breastfeeding exclusivity, any combination of breast and formula fed, length of feedings, number of feedings/24 hours, child age when stopped breastfeeding, and reasons for stopping are collected at postpartum study visits. These data are collected until a mother is no longer breastfeeding.

Sleep

The Pittsburgh Sleep Quality Index (PSQI), an established validated 19-item questionnaire, assesses sleep disturbances and sleep quality over a 1-month time period (Buysse et al., 1989) and is administered at all study visits.

Psychological Symptoms and Psychosocial Stress Measures

Extensive data on maternal psychosocial stressors are collected during pregnancy and the postpartum. Trained study coordinators administer validated questionnaires and most assessments are repeated at each study visit. The constructs of interest include depression (the 10-item Edinburgh Postnatal Depression Scale that assesses depressed mood in the perinatal period and is not confounded with physical symptoms associated with pregnancy and the postpartum) (Cox et al., 1987), anxiety (the 16-item Penn State Worry Questionnaire) (Meyer et al., 1990), perceived stress (the 14-item Perceived Stress Scale) (Cohen et al., 1983), and social support (a 30-item modified version of the Interpersonal Support Evaluation List) (Cohen & Hoberman, 1983). Data on domestic violence, adverse childhood experiences (Felitti et al., 1998), neighborhood stressors (Ewart & Suchday, 2002), and stressful life events (Barnett et al., 1983) are collected at a single time point in the third trimester. Discrimination (Krieger et al., 2005) is assessed at the third trimester study visit and at each postpartum study visit. These tools have all been established as valid and reliable during and outside of pregnancy.

Biospecimen Collection and Analysis

Biospecimens are collected at all study visits (Table 2) and as described below. As part of the consent process study participants are asked to specify whether biospecimens collected for the study can be used for future research and to document accordingly on their signed consent form. After collection, all biospecimens are immediately stored at −80⁰ C pending analysis, unless otherwise indicated. In addition to the analyses described, additional aliquots are banked for future research. At all study visits participants undergo a blood draw of approximately 40 ml of blood and provide a urine sample. The specific gravity and temperature of the urine sample are measured using a handheld refractometer (Atago 4410 PAL-10S) and then frozen for future use. Participants are asked to fast for 4–8 hours prior to study visits if possible.

Table 2.

Key Biomarkers Assessed during the UPSIDE MOMS Study by Study Visit

Biomarker	T1	T2	T3	6M	12M	24M	36M
Inflammatory markers (IL-6, TNFα, IL-1β, IL-8)	x	x	x	x	x
Metabolic markers (leptin, adiponectin)	x		x	x	x
Cardiometabolic risk markers (non-HDL-C, ApoB100, hs-CRP, 1-hour OGTT)						x	x
Cardiometabolic risk marker (Central fat mass)				x	x	x	x
Cortisol (saliva)	x	x	x	x	x	x	x

Note. IL-6 = interleukin 6; TNFα = tumor necrosis factor alpha; IL-1β = interleukin 1beta; IL-8 = interleukin 8; non-HDL-C = non-high density lipoprotein cholesterol; ApoB100 = apolipoprotein B100; hs-CRP = high sensitivity C-reactive protein; 1-hour OGTT = 1-hour oral glucose tolerance test.

Biological Measures.

Inflammatory markers. A panel of cytokines including IL-6, TNFα, IL-1β, IL-8 is measured in maternal plasma in each trimester as well as at 6- and 12-months postpartum. We will use Milliplex MAP high sensitivity human cytokine magnetic beads Luminex platform to quantify circulating cytokines in a 96-well format according to manufacturer’s directions. Endocrine markers. Saliva samples are collected at all study visits for cortisol assessment. Because cortisol concentrations vary across the course of the day, diurnal saliva samples are collected using procedures developed by the MacArthur Research Network on Socioeconomic Status and Health (Stewart & Seeman, 2000). Samples are returned and stored in a −80º C freezer until assay. Saliva samples will be assayed using a high sensitivity cortisol enzyme-linked immunosorbent assay (ELISA) kit (Salimetrics, State College, PA). Metabolic markers. Leptin and adiponectin will be assayed in samples from the first and third trimester, 6- and 12-month study visits using ELISA kits and standard manufacturer protocols will be employed.

Cardiometabolic Risk Biomarkers.

Samples and data collected at the 24- and 36-month study visits will be used to assess cardiometabolic risk, including five salient biomarkers (central fat mass, ApoB100, non-HDL cholesterol, hs-CRP, and 1-hour OGTT).

Statistical Analyses

For Aim 1 we will determine the longitudinal profiles of maternal weight changes (pregnancy through postpartum) that are associated with an adverse cardiometabolic risk. In all aims we will use two measures of cardiometabolic risk. Our primary outcome measure will be the sum of Z-scores of the five cardiometabolic risk biomarkers described earlier. Summing Z-scores weights each of these outcomes equally and puts them on the same scale, while retaining the full information about the biomarker distributions. We will switch the sign of (multiply by −1) any variable for which smaller values are better, so that large values of each Z-score indicate adverse levels. Our secondary outcome measure will be the ACC/AHA lifetime CVD event risk estimator.

We will characterize maternal weight trajectories using four “weight profile” variables: pre-pregnancy BMI, GWG, postpartum weight retention at 6-months (difference between 6-month postpartum weight and baseline early pregnancy weight), and later postpartum weight change (difference between 6- and 12-month postpartum weight), each as a continuous variable. Using linear regression, we first will regress the sum of the cardiometabolic risk Z-scores on these four variables, adjusting for maternal and gestational age, smoking status, race, and income. Next, we will fit an additional regression model that will include the interactions between pre-pregnancy BMI and each of the other three weight variables. Here, BMI will be coded using indicator variables to distinguish between women with healthy weight, overweight, and obesity. These interactions allow us to assess whether the effect of weight gain and postpartum weight loss on cardiometabolic risk differs depending on pre-pregnancy BMI. All models will be repeated using the ACC/AHA risk estimator as the outcome.

In Aim 2 we will characterize longitudinal profiles of obesity-related predictors (inflammatory endocrine, and metabolic biomarkers) from pregnancy through postpartum, and examine their association with the two measures of cardiometabolic risk. For each obesity-related predictor separately, we will create four primary variables: the predictor at prenatal baseline and its changes over the three time periods: pregnancy (last trimester value minus baseline), early postpartum (6-month value minus baseline), and later postpartum (difference in 6- and 12-month values). We will regress the sum of the five cardiometabolic risk Z-scores on these four primary variables for each obesity-related predictor separately, adjusting for age, smoking status, race, and income, and repeat this using the ACC/AHA risk estimator in place of the Z-score sum.

In Aim 3 we will determine whether postpartum modifiable health behaviors impact postpartum weight retention and downstream cardiometabolic risk. Total energy intake and percentage of energy intake from carbohydrate and fat calculated from dietary recalls will be used as potential dietary risk factors. We anticipate that physical activity, sleep quality index and stress will be defined in tertiles. Sustained breastfeeding will be defined as whether or not the child was still breastfed at 6-months. We will regress the cardiometabolic risk Z-score on postpartum weight loss while also adjusting for tertiles of each of the four variables and for the interaction of these tertiles with postpartum weight change. We will repeat this analysis replacing postpartum weight change with the 6- to 12-month postpartum weight change. All models will be repeated using the ACC/AHA risk estimator.

Discussion

Changes in weight, body composition, and immune, endocrine and metabolic biomarkers across the pregnancy-postpartum period could be critical indicators of future cardiometabolic risk. Many women enter pregnancy overweight or obese and experience high GWG (Deputy et al., 2015). These weight-related factors lead to postpartum weight retention (Siega-Riz et al., 2009) that contributes to increasing body weight over successive pregnancies and accompanying elevated cardiometabolic risk. Our study focuses on characterizing weight and selected biomarker profiles over an extended pregnancy-postpartum timeframe to identify key predictors of sustained cardiometabolic risk after pregnancy. We also incorporate modifiable health behaviors and lifestyle factors that could modify cardiometabolic risk development during this sensitive time period. This study will provide valuable, clinically relevant, and actionable findings on how weight and physiological profiles during pregnancy and postpartum contribute to the cardiometabolic health of women. This prospective longitudinal study has several significant design innovations. First, it incorporates repeated measures across pregnancy and an extended postpartum period. Second, our biological model and analytic approach reflect the importance of considering multiple, related pathophysiologic pathways for disease risk. Third, our inclusion of anthropometric assessments of body composition beyond weight (e.g. fat type and location [Bertoli et al., 2016]) is especially pertinent given that in the first six weeks postpartum, fat mass does not decrease even with weight loss (Cho et al., 2011), and central fat accumulation can increase across successive pregnancies (Gunderson et al., 2008).

Progress to date

The protocol components are feasible and to date we have collected data on nearly 200 mothers into the early postpartum period. Enrollment has been highly successful with over 97% of women from UPSIDE enrolling in UPSIDE MOMS thus far. We have successfully collected blood specimens from approximately 90% of participants and saliva samples from 60–80%. Laboratory analysis of prenatal cortisol and cytokine levels was recently completed and analysis of 6- and 12-month postpartum biospecimens will begin in the near future. Thus far, 78% of participants who have reached the 24-month time point have completed a 1-hour glucose challenge test.

Lessons Learned

To date there are a number of lessons that will be informative for future research. From the beginning our approach has been to incorporate postpartum maternal data collection into visits scheduled for the child in the UPSIDE study. We believe this synergy contributed to our high success rate (97%) in consenting UPSIDE participants to participate in UPSIDE MOMS, and creates cost efficiencies. The approach increases convenience for scheduling for both the mother and the study team. At the same time, synchronization of data collection requires streamlining, which we have learned requires flexibility among study team members and extensive cross training to enhance our ability to interweave data collection within the visit to ensure essential maternal and child data are collected at each visit.

A continuing challenge that we were aware of prior to commencement of the study is that women may become pregnant again during the window of data collection--once re-pregnancy occurs they are no longer in a postpartum stage and therefore are withdrawn from the study. We anticipated a re-pregnancy rate of approximately 20%, and are closer to a 30% re-pregnancy rate. The majority of pregnancies that have occurred were between the first and second year after the index birth. Many of these women are re-enrolling (37 known re-pregnancies to date) and therefore there will be opportunity to follow these women postpartum. Consequently, we have realized it is important to build in outcome measures at more than one time point in longitudinal research. In this case, we have added an additional glucose tolerance assessment to the 24-month visit so that models could examine outcomes at 24-months as well as 36-months postpartum.

Taking advantage of the unexpected and building upon it can enhance research. By re-enrolling women who become pregnant again during the postpartum we have an opportunity to include inter-pregnancy data for a subset of women. These women will have two well-characterized pregnancies which may further inform development of maternal cardiometabolic risk. It has been purported that retained weight after pregnancy, followed by additional pregnancies, results in adverse metabolic consequences (Gilmore et al., 2015). Short inter-pregnancy intervals, typically defined as < 18 months from the birth of a child to the beginning of the next pregnancy, are associated with adverse maternal and infant outcomes (Appareddy et al., 2016). We will be able to assess not only the known adverse outcomes, but also the cumulative effects of successive pregnancies, biomarkers across consecutive pregnancies, and the development of cardiometabolic risk in these women.

The unexpected occurrence of a pandemic in the middle of the study has taught us how essential flexibility is in the context of a research study. Changing course to collect data virtually, via home visits or by mail, enables moving the project forward, albeit somewhat differently than anticipated. Building alternative approaches to data collection into original planning should be routinely considered in future work. The pandemic has affected progress in that visits were paused at times because of positivity rates in the community. Consequently, face to face data collection has been delayed and/or study visits relegated to remote visits. However, no participants have been lost to follow-up as a result of the pandemic.

Wearable data collection can be difficult and we have found it is important to prepare for trouble shooting and using creative approaches to obtain the data. . Having adequate actigraph devices available with the ebb and flow of participant visits and delayed return of devices can be challenging, particularly during an ongoing pandemic. Devices can malfunction, women might not wear the device for a sufficient amount of time, or women, although amenable to wearing the device find it cumbersome and/or annoying. Troubleshooting has included purchasing additional devices, using text messages to remind participants to wear and/or return devices, refined wear instructions, provision of pre-registered tracking mailers for return, and meeting participants at a location of their choice to retrieve the device. Despite these challenges we have valid data for approximately 60% of participants. Examination of the literature suggests that success rates of around 60% are common (Troiano et al., 2008) and as such our success rates thus far are as expected. We will examine how the women who provide actigraph data differ from those who do not.

Future directions are numerous and with the rich data collected and the foresight to collect the lessons learned thus far suggest several potential investigations to consider. For example, extensive placenta data are being collected at the time of delivery, providing an opportunity to include placenta mechanisms as predictors of maternal postpartum cardiometabolic health. The “placental syndrome” has been proposed as a key link between adverse cardiometabolic health in pregnancy (such as hypertensive disorders of pregnancy) and future heart disease and stroke (Ray et al., 2005). We aim to test that hypothesis because we have detailed placenta histological, molecular, and imaging data. Investigation of environmental exposures on maternal cardiometabolic risk development is another future investigation: increasingly, evidence shows that exposure to metabolism disrupting environmental chemicals such as phthalates and per- and polyfluoroalkyl substances (PFAS), can dysregulate the body’s ability to adapt to the demands of pregnancy (Borghese et al., 2020; James-Todd et al., 2018; Varshavsky et al., 2019). Furthermore, elevated concentrations of certain phthalate metabolites and PFAS are associated with adverse cardiometabolic outcomes including excessive gestational weight gain (Bellavia et al., 2017), gestational diabetes mellitus (Zhang et al., 2015), and impaired glucose tolerance (James-Todd et al., 2018). In postpartum readjustment to the non-pregnant state may be a second critical period of heightened vulnerability to endocrine disruption