Experiences of multiple psychosocial stressors and associations with oxidative stress biomarkers during pregnancy

Abstract

Background:

Oxidative stress is hypothesized to be one mechanism linking psychosocial stressor exposure to preterm birth and other adverse pregnancy outcomes. However, prior studies have focused solely on singular psychosocial stressors, which may not reflect real world exposures as pregnant women may experience multiple stressors simultaneously.

Methods:

Participants included a subset of the Chemicals in Our Bodies cohort, a prospective birth cohort in San Francisco, California (N=227). Self -reported psychosocial stressors were assessed via questionnaires administered during the second trimester that addressed financial strain, food insecurity, job strain, neighborhood quality, caregiving, stressful life events, unplanned pregnancy, and perceived community status. Oxidative stress biomarkers were measured during the second trimester of pregnancy and included 15-F_2t-IsoP, and its two major metabolites 2,3-dinor-5,6-dihydro-15-F_2t-IsoP, and 2,3-dinor-15-F_2t-IsoP, and PGF_2α. Linear regression models were used to examine associations between individual and pairwise combinations of psychosocial stressors in relation to each oxidative stress biomarker.

Results:

15-F_2t-IsoP, 2,3-dinor-15-F_2t-IsoP, and 2,3-dinor-5,6-dihydro-15-F_2t-IsoP were elevated among participants reporting experiences of low perceived community status, job strain, poor neighborhood quality, food insecurity, and stressful life events (e.g., β=0.36, 95% CI=0.00, 0.72 for food insecurity in association with 15-F_2t-IsoP). In models that included pairwise combinations of stressor exposures, nearly every combination was also associated with an increase in all oxidative stress biomarkers compared to those who experienced one or neither stressor. For example, stressful life events and poor neighborhood quality was associated with statistically significant increases in all biomarkers (e.g., β=0.94, 95% CI=0.17, 1.71 for 2,3-dinor-5,6-dihydro-15-F_2t-IsoP).

Conclusions:

Urinary oxidative stress biomarkers were elevated among pregnant women exposed to psychosocial stressors, and exposure to multiple stressors resulted in the strongest associations. These findings support oxidative stress as one potential biological pathway linking prenatal psychosocial stress to preterm birth and other adverse pregnancy outcomes.

Graphical Abstract

Introduction

Adverse pregnancy outcomes remain critical public health problems in the United States. For instance, preterm birth impacts approximately 10% of births in the United States each year,¹ with well-established disparities such that individuals who are racially marginalized and/or socioeconomically disadvantaged experience the greatest disease burden.² Causes of adverse pregnancy outcomes are complex and multifactorial, ranging from genetic pre-disposition,³ chronic medical conditions (e.g., diabetes),⁴ smoking and alcohol consumption during pregnancy,⁵ and exposure to environmental toxicants.⁶ Evidence also indicates that experiences of psychosocial stress likely increase the risk of adverse pregnancy outcomes, potentially through a variety of mechanisms. For example, pregnant women who experience discrimination are more likely to report elevated depressive and anxiety symptoms,^7,8 both of which have been shown to increase the risk of preterm birth.⁹ Individuals who report financial strain may be more likely to live in a deprived neighborhood, thereby limiting access to greenspace and healthy foods.^10,11 Studies also find that pregnant women with lower socioeconomic status report elevated levels of a multitude of different stressors relative to those who are more socioeconomically advantaged, as indicated by higher educational attainment.¹² While an abundance of research has linked experiences of singular stressors to an increased risk of adverse pregnancy outcomes,^9,13 research by members of our team have more recently shown that pairwise combinations of stressors (e.g., experiences of both food insecurity and unplanned pregnancy) are adversely associated with fetal growth, and that these associations are generally stronger than what is observed for a single stressor.¹⁴

The underlying biologic mechanisms linking psychosocial stress to adverse pregnancy outcomes are complex and comprise multiple different physiological processes. One possible mechanism includes oxidative stress, defined as the imbalance of reactive oxygen species (ROS) and antioxidants in the human body and commonly measured via urinary biomarkers of F₂-isoprotanes.¹⁵ Evidence from prospective birth cohorts finds that elevated levels of oxidative stress biomarkers (e.g., 8-OH-dG, 8-isoprostane-prostagnaldin-F_2α (15-F_2t-IsoP), and 2,3-dinor-5,6-dihydro-15-F_2t-IsoP) are associated with increased odds of preterm birth, preeclampsia, and fetal growth restriction.^16–21 Similarly, studies have observed that these same biomarkers are elevated among those who have less than a high school education, are unmarried, and are current smokers,²² suggesting a link between oxidative stress and indicators of socioeconomic status and behaviors that frequently co-occur with lower socioeconomic status (e.g., smoking).²³ However, results from prior research on associations between psychosocial stressor exposure and oxidative stress biomarkers have been mixed. For instance, adverse childhood experiences, depressive symptoms and poor psychosocial well-being have been associated with increased levels of F₂-isoprostanes among studies of pregnant women.^24–26 Similarly, perinatal cumulative risk scores (defined using maternal demographic information and maternal mental health indicators) were positively associated with maternal15-F_2t-IsoP levels measured repeated during the first year postpartum.²⁷ In contrast, other studies have observed no association between perceived stress, stressful life events, and oxidative stress biomarkers.^24,28 However, a limitation of these studies is that the majority have assessed each stressor individually, failing to account for combinations of potential co-exposures that could jointly impact perinatal health.

In the present study, we set out to address these knowledge gaps by examining combinations of stressor exposures that pregnant women experience either during or in the year prior to pregnancy in relation to oxidative stress biomarkers. We included eight individual psychosocial stressors and their 21 unique pairwise combinations. We hypothesized that oxidative stress biomarkers would be elevated among those who experienced multiple stressors, and that these associations would be greater in magnitude than what was observed for individual stressors alone.

Methods

Study Population

Our study utilized a subset of participants enrolled in the Chemical in Our Bodies (CiOB) study, a prospective birth cohort of pregnant persons recruited in the San Francisco Bay Area, California.²⁹ Participants included in this analysis were those for which urinary levels of oxidative stress biomarkers were available (N=227). Details on CiOB recruitment methods are provided elsewhere.²⁹ Briefly, participants were eligible for inclusion in the CiOB cohort if they were 18 years or older, spoke English or Spanish, and were pregnant with a singleton in the second trimester. Participants were recruited between 2014–2018 from three University of California, San Francisco affiliated hospitals (Moffitt Long, Mission Bay, and Chan-Zuckerberg San Francisco General Hospital). Upon enrollment, participants completed an interview questionnaire which was used to ascertain information on sociodemographic characteristics and psychosocial stressors before and during pregnancy. The Institutional Review Boards at the University of California, San Francisco (10–00861) and Berkeley (2010-05-04) approved this study and all participants provided written, informed consent prior to enrollment.

Psychosocial Stressors

Psychosocial stressors included in our analyses were quantified using validated, self-reported interview questionnaires. We asked participants about several stressors, including financial strain,³⁰ food insecurity,^31,32 job strain, poor neighborhood quality,^33,34 stressful life events,^35,36 unplanned pregnancy, self-perceived community status,³⁷ and whether or not they are a caregiver to a family member with more needs than usual.

Financial Strain.

Financial strain was assessed using a combination of annual household income and asking participants how hard it is for them to pay for basic necessities. Reponses were ranked on a 4-point Likert scale ranging from “not difficult at all” to “very difficult”. Participants were categorized as having experienced financial strain if their income was below the San Francisco poverty line, or they responded that it was “somewhat difficult” or “very difficult” to the question about paying for basic necessities.³⁰

Food Insecurity.

Food insecurity was assessed by asking participants if in the last 12 months they, or another adult in their household, had skipped meals because there wasn’t enough food, ate less than they should, or were hungry but did not eat due to not having enough food.^31,32 Responses were ranked on a 3-point scale ranging from “never true” to “often true”. Participants were categorized as having experienced food insecurity if they responded that it was “sometimes true” or “often true” to questions about not having enough food and balanced meals.

Job Strain.

Job strain was assessed by asking participants if their job allowed them to make a lot of decisions, develop their abilities, and not work excessive amounts, as well as by asking if their job left them too tired or stressed after work and if their pay was fair.⁷ Responses were ranked on a 5-point scale ranging from “strongly disagree” to “strongly agree”. Participants were categorized as having experienced job strain if they “somewhat disagreed” or “strongly disagreed” to questions about decision making, developing abilities, and fair pay. Participants were also categorized as experiencing job strain if their response was “somewhat agree” and “strongly agree” to the questions about being asked to do excessive amounts of work and were left tired and stressed after work.

Poor Neighborhood Quality.

Poor neighborhood quality was assessed by asking participants 15 questions which were used to rate their neighborhood for collective efficacy, safety, dissatisfaction, and physical disorder.^33,34 Positively stated questions were reverse coded so that higher scores always corresponded to lower perceived neighborhood quality. Responses were ranked on a 5-point scale ranging from “strongly disagree” to “strongly agree”. Participants were categorized as having poor neighborhood quality if their response was “somewhat agree” or “strongly agree” to questions about heavy traffic and wanting to move to another neighborhood. Participants were also categorized as having poor neighborhood quality if their response was “somewhat disagree” or “strongly disagree” to the questions about their neighborhood being a good place to live.

Caregiving.

Caregiving was assessed by asking participants to rate how often they were responsible for the care of a child or an older adult, either a parent or relative, with high medical or educational needs. Responses were ranked on a 5-point scale ranging from “never” to “very often”. Participants were categorized as having experienced caregiving stress if their responses were “often” or “very often” to the question about caregiving responsibilities.

Stressful Life Events.

Stressful life events were assessed by asking participants if in the last 12 months they had experienced any of the following events (yes or no): family member being sick, separation from a partner, moving, their partner losing their job, losing their job, arguing with a partner more than usual, their partner not wanting them to get pregnant, their partner having legal problems, someone close having a drinking problem, someone close dying, or a close family member having immigration problems. The number of events experienced was summed, and participants were categorized as having experienced stressful life events if they reported two or more events.

Unplanned Pregnancy.

Unplanned pregnancy was assessed by asking participants about how they felt about becoming pregnant with their current pregnancy. Participants were categorized as having an unplanned pregnancy if their responses were “I wanted to be pregnant later” or “I didn’t want to be pregnant then”.

Perceived Community Status.

Perceived community status was assessed by asking participants to rate their standing in their community on a ladder, ranging from 1 through 10 where higher scores indicate better self-perceived standing in their community.³⁷ Participants were categorized as having low perceived community status if their response was a 4 or below.

Measure of Oxidative Stress

Urine samples were collected during the second trimester of pregnancy, and were frozen at −80°C prior to analysis. The Eicosanoid Core Laboratory at Vanderbilt University Medical Center analyzed urinary levels of 15-F_2t-IsoP, its two major metabolites 2,3-dinor-5,6-dihydro-15-F_2t-IsoP and 2,3-dinor-8-iso-15-F_2t-IsoP, and prostaglandin-F_2α (PGF_2α) using liquid chromatography mass spectrometry (LC/MS), which has been previously described.^38,39 We focused on these specific F₂-Isoprostanes because they have been previously associated with adverse pregnancy outcomes.^20,21,40 Values below the limit of detection (LOD) were imputed using the LOD divided by the square root of 2. To account for urinary dilution, oxidative stress biomarkers were corrected for specific gravity (SpG) using the equation: ${\text{OX}}_{\text{c}}=\text{OX}(\frac{\left({\text{SpG}}_{\text{median}}-1\right)}{\left(\text{SpG}-1\right)})$. Using this equation, SpG_median is the median specific gravity among all participants in our study, SpG is the specific gravity level, OX is the uncorrected oxidative stress biomarker level and OX_c is the specific gravity-corrected oxidative stress biomarker level. All oxidative stress biomarkers were right skewed, therefore we natural log-transformed these biomarkers for downstream analyses to reduce skewness.

We additionally calculated a ratio of 15-F_2t-IsoP to PGF_2α in order to quantify the proportion of 15-F_2t-IsoP derived from prostaglandin-endoperoxide synthases and non-enzymatic lipid peroxidation pathways.⁴¹ The chemical fraction depicts non-enzymatic lipid peroxidation resulting primarily from oxidative stress, while the 15-F_2t-IsoP enzymatic fraction is created from prostaglandin-endoperoxide synthases and is thought to be more reflective of inflammation.⁴¹ While we utilize this ratio to quantitatively distinguish the proportion of 15-F_2t-IsoP derived from oxidative stress from inflammation pathways, we acknowledge that PGF_2α may be derived from pathways independent of cyclooxygenase (COX) activity.⁴² Considering this potential for outcome misclassification, we focus on the chemical and enzymatic fractions solely as secondary endpoints in sensitivity analyses.

Statistical Analysis

We examined the distribution of sociodemographic characteristics and psychosocial stressors (individually and as combinations) using means, standard deviation (SD), frequencies, and counts for continuous and categorical variables, respectively. We then examined the distribution of urinary oxidative stress biomarkers using geometric means, geometric SDs, and percentiles (5^th, 25^th, 50^th, 75^th, 95^th).

Next, we examined the associations between individual psychosocial stressors and urinary oxidative stress biomarkers using a series of minimally adjusted and adjusted linear regression models. Across all models, oxidative stress biomarkers were treated as separate outcomes in individual models. First, we examined these associations in models that were adjusted for covariates (maternal age, pre-pregnancy body mass index), which were identified via a literature review and determined to be the most strongly associated with exposures and outcomes in our prior work.^14,22 Second, we ran fully adjusted models which included the variables retained in the covariate adjusted models, as well as the remaining psychosocial stressors. Third, we added maternal education (a proxy for socioeconomic status) as a covariate adjusted in the covariate adjusted models. The rationale for not adjusted for maternal education in our primary covariate adjusted models is that we have previously observed that some of these stressors are correlated with maternal education (i.e., food insecurity and financial strain).¹⁴

We then assessed the pairwise joint effects of combinations of stressors in relation to oxidative stress biomarkers. In these models, the exposure was the pair of stressors, and the reference group was experiences of one or neither of the stressors. These models were similarly run as described for the individual stressors using unadjusted, covariate adjusted (maternal age, pre-pregnancy BMI), fully adjusted (covariates and remained stressors), and with the inclusion of maternal education in covariate adjusted models. Combinations of stressors experienced by less than 10 participants were not examined in these models.

Lastly, we conducted a sensitivity analysis where we additionally examined the chemical and enzymatic fractions of 15-F_2t-IsoP as secondary outcomes in linear regression models. All models were run on complete cases and analyses were conducted in R.

Results

The mean maternal age at delivery was 33 years (SD=5.4 years) and the mean pre-pregnancy BMI was 26 kg/m² (SD=5.5). Most participants were married (N=167, 74%) and college educated with either a college degree (N=55, 24%) or a graduate degree (N=92, 41%). Most participants self-identified as White (N=102, 45%) or Latina (N=68, 30%) (Table 1). The most common experiences of psychosocial stressors reported were stressful life events (N=124, 55%) and financial strain (N=69, 39%) while the least commonly reported psychosocial stressor was low perceived community status (N=18, 8%) (Table 1). Financial stress and stressful life events (N=47, 21%), and unplanned pregnancy and stressful life events (N=39, 17%) were the most common combinations of psychosocial stressors (Table S1). The least commonly reported combinations of psychosocial stressors included food insecurity and job strain, food insecurity and low perceived community status, job strain and caregiving, job strain and low perceived community status, job strain and poor neighborhood quality, caregiving and low perceived community status, unplanned pregnancy and low perceived community status (N<10 for each combination), therefore these combinations were not included in downstream models. Distributions of individual and combinations of psychosocial stressors were similar when restricting to complete cases of stressors (Table S1).

Table 1.

Distribution of sociodemographic characteristics and psychosocial stressors in a subset of the Chemicals in Our Bodies cohort, 2014–2018 (N=227).

	N (%) or Mean (SD)
Maternal age (years)	33 (5.4)
Pre-pregnancy body mass index (kg/m2)	26 (5.5)
Missing	9 (4.0%)
Maternal education
Less than high school education	24 (11 %)
High education or some college	54 (24 %)
College degree	55 (24 %)
Graduate degree	92 (41 %)
Missing	2 (0.9%)
Maternal race/ethnicity
White	102 (45 %)
Black	13 (6 %)
Asian/Pacific Islander	36 (16 %)
Latina	68 (30 %)
Multiracial	8 (4 %)
Parity
0	126 (56 %)
1+	96 (42 %)
Missing	5 (2.2%)
Marital status
Married	167 (74 %)
Living Together	38 (17 %)
Single	20 (9 %)
Missing	2 (0.9%)
Gestational age at delivery (weeks)	38 (2.7)
Financial strain
No	141 (62 %)
Yes	69 (30 %)
Missing	17 (7.5%)
Food insecurity
No	195 (86 %)
Yes	29 (13 %)
Missing	3 (1.3%)
Job strain
No	180 (79 %)
Yes	29 (13 %)
Missing	18 (7.9%)
Poor neighborhood quality
No	160 (70 %)
Yes	42 (19 %)
Missing	25 (11.0%)
Caregiving
No	187 (82 %)
Yes	37 (16 %)
Missing	3 (1.3%)
Stressful life events
No	99 (44 %)
Yes	124 (55 %)
Missing	4 (1.8%)
Unplanned pregnancy
No	166 (73 %)
Yes	56 (25 %)
Missing	5 (2.2%)
Low perceived community status
No	190 (84 %)
Yes	18 (8 %)
Missing	19 (8.4%)

Abbreviations: SD, standard deviation.

The distribution of urinary oxidative stress biomarkers is presented in Table S2. The geometric means were 0.71 for 15-F_2t-IsoP (geometric SD = 2.44), 1.12 for 2,3-dinor-5,6-dihydro-15-F_2t-IsoP (geometric SD = 6.55), 3.18 for 2,3-dinor-15-F_2t-IsoP (geometric SD = 3.18), and 1.91 for PGF_2α (geometric SD = 2.84).

In linear regression models that were adjusted for maternal age and pre-pregnancy BMI, we observed that participants who experienced low perceived community status, poor perceived neighborhood quality, job strain, food insecurity, and financial strain consistently had increased levels of all oxidative stress biomarkers compared to those who did not experience these stressors (Figure 1, Table S3). For example, low perceived community status and job strain were associated with elevated levels of 2,3-dinor-5,6-dihydro-15-F_2t-IsoP (β=0.80 95% CI=−0.08, 1.68; β=0.60, 95% CI=−0.14, 1.33, respectively), although confidence intervals indicated non-significance. Levels of 15-F_2t-IsoP and 2,3-dinor-5,6-dihydro-15-F_2t-IsoP were significantly increased in relation to poor neighborhood quality (β=0.34, 95% CI=0.02, 0.66; β=0.83, 95% CI=0.18, 1.49, respectively). Food insecurity was associated with significant increases in 15-F_2t-IsoP (β=0.36, 95% CI=0.00, 0.72), 2,3-dinor-15-F_2t-IsoP (β=0.58, 95% CI=0.15, 1.01) and PGF_2α (β=0.50, 95 % CI =0.06, 0.93). Unplanned pregnancy and caregiving were not strongly associated with oxidative stress biomarkers in covariate adjusted models. Associations were generally greater in magnitude in unadjusted models (Table S3). In contrast, effect estimates obtained from models mutually adjusted for covariates and other psychosocial stressors were more variable with wider confidence intervals compared to covariate adjusted models. For example, 2,3-dinor-15-F_2t-IsoP and PGF_2α were not associated with financial strain in fully adjusted models (e.g., β=0.33, 95% CI=−0.03, 0.70; β=0.06, 95% CI=−0.46, 0.58 for covariate adjusted and fully adjusted models which include PGF_2α as the outcome, respectively) (Table S3). When examining the chemical and enzymatic fractions of 15-F_2t-IsoP as additional outcomes, associations tended to mirror patterns observed in models which included the parent compound was included as the outcome (Table S3). In our sensitivity analysis additionally adjusting for maternal education in our models, most associations were attenuated to non-significance relative to what was observed in models were adjusted solely for maternal age and pre-pregnancy BMI (Table S4).

Figure 1.

Associations between individual psychosocial stressors and urinary levels of specific gravity corrected oxidative stress biomarkers (ng/mL) during the second trimester of pregnancy, estimated using linear regression models in the Chemicals in Our Bodies cohort, 2014–2018.

Note: Models are adjusted for maternal age (continuous) and pre-pregnancy body mass index (continuous).

When pairwise combinations of psychosocial stressors were examined, nearly every combination was associated with an increase in all measured oxidative stress biomarkers compared to those who experienced one or neither stressor, although not all combinations reached statistical significance (Figure 2, Table S5). Associations were greatest in magnitude for models which included 2,3-dinor-5,6-dihydro-15-F_2t-IsoP as the outcome. For example, in covariate adjusted models, stressful life events and poor perceived neighborhood quality were associated with significant increases in all oxidative stress biomarkers (e.g., β=0.94, 95% CI=0.17, 1.71 for 2,3-dinor-5,6-dihydro-15-F_2t-IsoP). Additionally, 2,3-dinor-15-F_2t-IsoP was increased in multiple combinations, such as stressful life events and unplanned pregnancy, poor neighborhood quality and stressful life events, and food insecurity and stressful life events (e.g., β=0.57, 95% CI=0.09, 1.04 for food insecurity and stressful life events). 15-F_2t-IsoP was also significantly elevated in response to poor neighborhood quality and stressful life events, job strain and unplanned pregnancy, poor neighborhood quality and caregiving, and financial strain and food insecurity (e.g., β=0.37, 95% CI=0.01, 0.72 for stressful life events and unplanned pregnancy (Figure 2, Table S5).

Figure 2.

Associations between combinations of psychosocial stressors and urinary levels of specific gravity corrected oxidative stress biomarkers (ng/mL) during the second trimester of pregnancy, estimated using linear regression models in the Chemicals in Our Bodies cohort, 2014–2018.

Note: Models are adjusted for maternal age (continuous) and pre-pregnancy body mass index (continuous). Statistically significant associations are denoted in black.

Discussion

Among a racially, ethnically, and socioeconomically diverse cohort of pregnant women in the San Francisco Bay Area of California, we observed that urinary levels of F₂-isoprostanes were elevated in relation to most psychosocial stressors, including low perceived community status, food insecurity, financial strain, poor perceived neighborhood quality and job strain. We also observed that a multitude of psychosocial stressors occurred in combination with one another, with the most prevalent combinations being financial stress and stressful life events, and unplanned pregnancy and stressful life events. Certain combinations of stressors were also associated with elevated levels of oxidative stress biomarkers, including poor neighborhood quality and stressful life events. In this study, associations were generally greatest in magnitude for 2,3-dinor-5,6-dihydro-15-F_2t-IsoP, suggesting that this oxidative stress biomarker may be the most sensitive to psychosocial stressors. Importantly, prior research suggests that of the oxidative stress biomarkers measured in this study, 2,3-dinor-5,6-dihydro-15-F_2t-IsoP is most strongly associated with preterm birth, suggesting that this specific F₂-isoprostane may be one of the better indicators of adverse pathways of relevance to adverse perinatal outcomes.^38,40,43

Our findings extend prior research examining psychosocial stress as a risk factor for elevated oxidative stress biomarkers.^{25–27,44,45} In particular, in this study we leverage multiple stressors that have not previously been studied within this context, including caregiving, poor neighborhood quality, food insecurity, and perceived community status. While we observed that poor neighborhood quality, low perceived community status and food insecurity were associated with elevated levels of oxidative stress biomarkers, and in particular 2,3-dinor-5,6-dihydro-15-F_2t-IsoP, we found no association between some of the other stressors, including caregiving. We also observed that stressful life events and unplanned pregnancy were only associated with increases in oxidative stress in models which included pairwise combinations, suggesting possible differential effects based on the type of stressor or joint effects. This notion is supported by prior research in prospective birth cohorts which has focused on objective stressors (e.g., stressful life events, adverse childhood experiences), as well as responses to psychosocial stress (e.g., depression, anxiety) and oxidative stress biomarkers.^22,24–26 Inverse associations between adverse childhood experiences and 15-F_2t-IsoP were observed among African Americans in the Stress and Health in Pregnancy cohort.²⁶ Two separate studies further observed no associations between experiences of stressful life events and F₂-isoprostane levels,^22,24 although one study found significant positive correlations.⁴⁶ While the stressful life events findings are in line with what we observed when examining stressful life events as an individual stressor, we did observe positive associations between stressful life events when examined in combination with other stressors (i.e., stressful life events and poor neighborhood quality). In particular, the effect of stressful life events and poor neighborhood quality on 2,3-dinor-5,6-dihydro-15-F_2t-IsoP and 15-F_2t-IsoP appeared to be greater than additive as compared to the individual effect of either stressor, suggesting a synergistic effect. However, effect sizes were less than additive and null for many of the other combinations examined, suggesting possible antagonistic interactions between certain stress exposures.

Findings from this study contribute to a greater understanding of the underlying mechanisms, specifically F₂-isoprostanes, that may be a key pathway linking psychosocial stress exposure to adverse maternal and child health outcomes. Prior studies have found that cord blood levels of 15-F_2t-IsoP and 8-OH-dG are higher among infants born preterm birth compared term.^16–19 However, in this study, 2,3-dinor-5,6-dihydro-15-F_2t-IsoP, one of the primary metabolites of 15-F_2t-IsoP, was the most sensitive to psychosocial stress exposures. This supports prior findings 2,3-dinor-5,6-dihydro-15-F_2t-IsoP suggesting that 15-F_2t-IsoP may be more sensitive biomarker of exogenous oxidative stress as compared to the parent compound.^40,47 For example, prior work pooling data from four birth cohorts, including CiOB, observed that levels of 2,3-dinor-5,6-dihydro-15-F_2t-IsoP and 15-F_2t-IsoP were associated with a 22% and 19% increase in odds of preterm birth, respectively.³⁸ Taken together, this suggests that 15-F_2t-IsoP metabolites may be the most relevant biomarkers for understanding key pathways leading to preterm birth.⁴⁷

An important strength of this study was the inclusion of a socioeconomically diverse study population and capturing a wide distribution of psychosocial stressors. Similarly, we included multiple different types of psychosocial stressors in our analysis. Relatedly, we also assessed combinations of stressors exposures, an approach that may better reflect real life exposure patterns, as evidenced by our findings that certain combinations of stressors are highly prevalent. This study also utilized multiple biomarkers of oxidative stress. The F₂-isoprostanes quantified in this study are stable throughout the day and unaffected by lipids in the diet,^48,49 thereby limiting residual confounding. Nonetheless, we acknowledge that 15-series F₂-isoprostanes and metabolites are not the only measures of F₂-isoprostane formation. Future work should consider additional isoprostanes, including the 5-series, as well as highly polar F₂-isoprostane glucuronide metabolites, which have recently been identified.⁵⁰ We were not able to measure the 5-series F₂-isoprostanes or other metabolites, in our study due to a limited budget. Our study is also limited by our relatively modest sample size. This limited our ability to assess certain stressor combinations. In addition, our analytic plan involved running a large number of regression models and we did not employ multiple comparisons corrections. While we acknowledge that this may increase the possibility of chance findings due to type I error, we also note that multiple comparison corrections are often not needed in epidemiologic studies as they may increase type II error.⁵¹ Lastly, our study population was comprised of pregnant participants from the San Francisco Bay Area, and findings may not be generalizable to pregnant people from other regions or non-pregnant populations.

Conclusions

Within the CiOB cohort, we observed that oxidative stress biomarkers, and in particular 15-F_2t-IsoP and 2,3-dinor-5,6-dihydro-15-F_2t-IsoP, are elevated in relation to psychosocial stressors, including low perceived community status, job strain, poor neighborhood quality, and food insecurity, as well as combinations of stressors, including poor neighborhood quality and stressful life events, and financial strain and poor neighborhood quality. These findings are of public health significance as they support the prioritization of 2,3-dinor-5,6-dihydro-15-F_2t-IsoP as one of the best indicators of adverse pathways linking psychosocial stressors to adverse perinatal outcomes.

Highlights.

• Oxidative stress biomarkers are elevated in response to psychosocial stress

• Joint exposure to multiple psychosocial stressors resulted in stronger associations

• Isoprostanes may reflect one pathway linking psychosocial stress to preterm birth