Maternal stress, low cervicovaginal β-defensin, and spontaneous preterm birth

Abstract

Background:

Spontaneous preterm birth (sPTB) is a major contributor to infant mortality and its etiology remains poorly understood. Host immunity and maternal stress may play a role in the pathogenesis of sPTB but mechanisms are poorly delineated. Antimicrobial proteins in the cervicovaginal space, such as beta defensins, modulate immune responses to bacteria and have been shown to modulate the risk of sPTB from non-optimal microbiota. While stress is known to induce immunological changes, no study has examined the interplay between maternal stress and the immune response in association with sPTB.

Objectives:

Our objectives were to determine whether psychosocial stress was associated with a mediator of the immune system in the cervicovaginal space, beta defensin-2, and to examine the combined impact of high stress and low cervicovaginal beta defensin-2 levels on the odds of sPTB.

Study Design:

From the Motherhood & Microbiome cohort study (n=2000), we performed a secondary, nested case-control study, frequency matched by race/ethnicity, of 519 pregnant women (110 sPTB and 409 term). Stress and cervicovaginal beta defensin-2 levels were measured at 16–20 weeks of gestation. Stress was dichotomized at a score of 30 on Cohen’s Perceived Stress Scale (PSS-14). We measured cervicovaginal beta defensin-2 levels with ELISA and dichotomized at the median. We modeled associations of high stress and low cervicovaginal beta defensin-2 levels using multivariable logistic regression. We also compared the proportion of women with high stress and low cervicovaginal beta defensin-2 levels among women with spontaneous preterm and term births using Chi-Square tests. We modeled adjusted associations of stress and cervicovaginal beta defensin-2 levels with odds of sPTB using logistic regression.

Results:

The majority of the study population was non-Hispanic black (72.8%), insured by Medicaid (51.1%), and had a PSS-14 score < 30 (80.2%). High stress was associated with reduced adjusted odds of low beta defensin-2 levels (aOR 0.63, 95% CI: 0.38 −0.99). In a model adjusted for race and smoking, both high stress (aOR 1.72, 95% CI: 1.03–2.90) and low beta defensin-2 (aOR 1.58, 95% CI: 1.004–2.49) were associated with increased odds of sPTB. We then built a model of the four possible combinations of low and high stress and low and high beta defensin-2 levels with the odds of sPTB. Women with either high stress (aOR 1.37, 95% CI: 0.68 – 2.78) or low beta defensin-2 (aOR 1.40, 95% CI: 0.83–2.34), had slightly elevated but not significantly increased odds of sPTB compared to women with neither exposure. However, women with both high stress and low beta defensin-2 had significantly elevated odds of sPTB compared to women with neither exposure (aOR 3.16, 95 % CI: 1.46 – 6.84).

Conclusion:

High perceived stress and low cervicovaginal beta defensin-2 levels are associated with higher odds of sPTB, and when present concurrently, they result in the highest odds of sPTB in a largely non-Hispanic black cohort. Our findings warrant further work to examine social determinants of health and the host cervicovaginal immune responses that may modulate the pathogenesis of sPTB.

Introduction

Preterm birth remains a major public health issue with approximately 380,000 infants born prior to 37 weeks of gestational age in the United States annually.¹ Preterm-related conditions are the second leading cause of infant mortality, and infants who survive have higher risk than term infants of chronic medical conditions such as lung disease and neurodevelopmental disabilities.² Preterm birth can occur spontaneously, following premature rupture of membranes and/or preterm labor, or it can be medically-indicated in the setting of maternal or fetal conditions such as preeclampsia or intrauterine growth restriction. There is some evidence that psychosocial stress is associated with adverse pregnancy outcomes,^3–9 but the mechanisms remain unknown. Stress can alter systemic immune function through hormonal pathways that may affect reproductive organs including the cervicovaginal space.^10–12

The cervicovaginal space is highly immunologically active. One family of mucosal antimicrobial peptides in the cervicovaginal space are beta defensins (β-defensins). β-defensins are cationic peptides produced by the innate immune system that inhibit fungal, viral, as well as bacterial function.¹³ One of the antibacterial mechanisms may be through increased permeability of the bacterial membrane to small molecules that impair DNA, RNA and protein synthesis impairing bacterial survival.¹⁴ Recently, high levels of β-defensin-2 was shown to reduce the risk of sPTB among black women with a non-optimal cervicovaginal microbiota characterized by a dominance of anaerobic bacteria and a paucity of Lactobaccilus species (community state type –IV).¹⁵ Psychosocial stress has been shown to increase the risk of bacterial vaginosis,^11,12 a sign of non-optimal vaginal microbiota and potential immunologic dysfunction.

Our hypothesis is that psychosocial stress may decrease the concentration of β-defensins in the cervicovaginal space and result in increased odds of sPTB. In this study, we analyzed the associations of perceived stress and cervicovaginal concentrations of β-defensins (CV-βD), as well as the combined impact of high stress and low CV-βD on the odds of sPTB.

Methods

Participants

The Motherhood and Microbiome (M&M) cohort was recruited during the course of prenatal care from December 2013 through December 2016 at Penn Medicine in Philadelphia.¹⁵ Eligible women were between 16 0/7 and 20 6/7 weeks of gestation with singleton gestations. Women were excluded if they had enrolled in the study with a prior pregnancy, had major fetal anomaly, had a history of organ transplantation, were on chronic glucocorticoids, or were HIV positive. A total of 2,982 pregnant women were approached for the study. Among those approached, 982 (49%) declined and 2,000 women were enrolled after written, informed consent. Participants included in this present, secondary analysis included the 519 women with perceived stress assessments, CV-βD measured, and either a sPTB or a term (≥ 38 completed weeks of gestation) delivery. Deliveries at 37 weeks were not chosen for CV-βD measurements to create distinct categories of preterm cases and term controls. sPTB was defined as a preterm birth that resulted from preterm labor or preterm premature rupture of membranes through a standardized adjudication process by a maternal-fetal medicine physician (ME). The study was approved by the Institutional Review Board at the University of Pennsylvania (IRB #818914).

Psychosocial Stress Assessment

Psychosocial stress was assessed using the Cohen’s Perceived Stress Scale (PSS-14).¹⁶ The PSS-14 includes questions regarding stress over the past month and includes questions such as “In the last month, how often have you felt nervous and ‘stressed’?” and “In the last month, how often have you felt confident about your ability to handle your personal problems?” Responses range from never to very often and each item is scored 0 to 4 with reverse coding for items that indicate lower levels of stress. We defined high stress as a score ≥ 30 consistent with a prior study of pregnant women,¹⁷ which was the 80^th percentile in our study.

Cervicovaginal β-defensin measurement

As previously described, we analyzed cervicovaginal fluid collected on dacron swabs between 16 0/7 weeks and 20 0/7 for human β-defensin-2 (CV-βD) via ELISA (Phoenix Pharmaceutical).¹⁵ Minimal detectable concentration was 15.6 pg/mL (inter-assay variation <15% and intra-assay variation <10%). We dichotomized women at the median (17,098 pg/mL) to maximize sample size in each group and thus maximize power since there are no published normal levels of CV-βD. Women below the median were considered as having low CV-βD and women above or equal to the median were considered as having high CV-βD.

Statistical analysis

We performed bivariate analyses of characteristics among sPTB cases and term controls. We also compared women with high and low CV-βD according to stress levels using chi square tests. We then used separate logistic regression models to calculate adjusted odds ratios of 1) low CV-βD among women with high (versus low) stress; 2) sPTB among women with high (versus low stress); 3) sPTB among women with high (vs low CV-βD); 4) sPTB among women with both high stress and low CV-βD (vs. women with neither). Even though we frequency matched by race, a priori we decided to control for race in our models due to the known associations of race with stress, CV-βD, and sPTB. We then considered multiple covariates in our models including age, BMI, smoking, parity, and insurance. Only variables that changed estimates by more than 10% were retained in the final model to achieve parsimony. We also used multiplicative interaction terms and stratified models to assess whether women with sPTB and term births differed with respect to associations of stress and CV-βD, as well as whether women with high and low stress differed with respect to the association of low CV-βD and sPTB.

Results

Of the 2,000 women who enrolled in the Motherhood and Microbiome study, we had delivery information for 1,921 women; 1,712 women delivered at term and 209 had PTB of which 124 were sPTB, 84 were sPTB, and one could not be categorized. Among the term and sPTB births, we analyzed CV-βD levels for 430 and 123 women, respectively, that were frequency matched by race/ethnicity. We had concurrent stress data for 409 women who delivered at term and 110 with sPTB who were included in the final analytic dataset.

Characteristics of women with sPTB and term births are presented in Table 1. The majority of women in the study were non-Hispanic black (72.8%), were insured by Medicaid (51.1%), and had PSS-14 scores <30 (80.2%). Counter to our hypothesis, high stress was associated with reduced odds of low CV-βD levels (aOR 0.63, 95% CI 0.38–0.99) (Table 2). The effect estimates between high stress and reduced odds of low CV-βD were similar among women with sPTB (aOR 0.66, 95% CI 0.28–1.58) and term birth (aOR 0.56, 95% CI: 0.56–0.90), interaction P= 0.41.

Table 1.

Characteristics of participants of the Motherhood and Microbiome Cohort with both perceived stress scale data and cervicovaginal beta defensin-2 levels at 16–20 weeks of gestation (n=110 spontaneous preterm cases and 409 term controls)

	sPTB (<37 weeks) (n=110)	Term (≥38 weeks) (n=409)
	n (col %)	n (col%)	P
Age (years)			0.20
< 25	26 (23.6)	130 (31.8)
25 – < 35	65 (59.1)	206 (50.4)
≥ 35	19 (17.3)	73 (17.8)
Body mass index (kg/m2)			0.95
< 25	38 (35.2)	150 (36.9)
25 – < 30	31 (28.7)	115 (28.3)
≥ 30	39 (36.1	142 (34.9)
Race/ethnicity^			0.75
Non-Hispanic black	78 (70.9)	300 (73.3)
Non-Hispanic white	28 (25.5)	91 (22.2)
Asian/Hispanic/other	4 (3.6)	18 (4.4)
Marital status			0.32
Married	38 (34.5)	121 (29.6)
Not married	72 (65.5)	288 (70.4)
Insurance			0.09
Medicaid/uninsured	64 (58.2)	201 (49.1)
Private	46 (41.8)	208 (50.9)
Parity			0.16
0	41 (37.3)	183 (44.7)
> 0	69 (62.7)	226 (55.3)
Smoked during pregnancy			0.03
Yes	15 (13.8)	30 (7.3)
No	94 (86.2)	379 (92.7)
Perceived Stress Scale			0.05
≥ 30	29 (26.4)	74 (18.1)
< 30	81 (73.6)	335 (81.9)
Cervicovaginal β-defensin 2 concentration			0.12
< Median (17,098 pg/mL)	63 (57.3)	200 (48.9)
≥ Median (17,098 pg/mL)	47 (42.7)	209 (51.1)

n=4 missing BMI, n=1 missing smoking; P value from Chi-Square test;

^{^}matching factor

Table 2.

Associations of cervicovaginal β-defensin-2 (< median [17,098pg/mL]) among women with high stress (PSS ≥ 30) versus low stress; women with high stress had reduced odds of low CV-βD.

				All births			Stratified by birth outcome
	Low CV-βD	High CV-βD					sPTB	Term
	< median	≥ median	Chi Sq. P	M1	M2	M3	M2s	M2t
Stress level	n (col%)	n (col %)	0.0008	OR (95 % CI)	OR (95 % CI)	OR (95 % CI)	OR (95 % CI)	OR (95 % CI)
High, PSS ≥ 30	37 (14.1)	66 (25.8)		0.47 (0.30–0.74)	0.60 (0.38–0.95)	0.63 (0.38 –0.99)	0.66 (0.28 – 1.58)	0.56 (0.34 –0.90)
Low, PSS < 30	226 (85.9)	190 (74.2)		ref		ref	ref	ref

M1: Unadjusted

M2: M1+ race/ethnicity + smoking

M3: M2 + age, BMI, parity, smoking status, insurance

M2s: M2 restricted to women with sPTB

M2t: M2 restricted to women with term births

In bivariate analyses, 26.4% of women with sPTB reported high stress compared to 18.1% of women with term births (p=0.05) (Table 1). We did not detect significant differences in the prevalence of low CV-βD between women with sPTB (57.3%) and term (48.9%) (p=0.12). Adjusted models similarly did not result in statistically significant associations of stress or low CV-βD with sPTB (Table 3). Additional adjustment for potential confounding variables did not substantially change effect estimates. However, when both high stress and low CV-βD were introduced into the same adjusted model, associations became significant. Specifically women with high (vs. low) stress had elevated odds of sPTB (aOR 1.72, 95% CI 1.03–2.90) and women with low (vs. high) CV-βD had elevated odds of sPTB (aOR 1.58, 95 % CI 1.004–2.49) (Table 3).

Table 3.

Associations of high stress (PSS ≥ 30) and low CV-βD (below the median of 17,098 pg/mL) with adjusted^* odds of spontaneous preterm birth

	Model 1	Model 2	Model 3
Exposure	aOR (95% CI)	aOR (95% CI)	aOR (95% CI)
PSS ≥ 30	1.62 (0.97–2.70)		1.72 (1.03–2.90)
PSS < 30	ref		ref
CV-βD < median		1.50 (0.96–2.35)	1.58 (1.004– 2.49)
CV-βD ≥ median		ref	ref

Model 1 includes PSS but not CV-βD

Model 2 includes CV-βD but not PSS

Model 3 includes both PSS and CV-βD

^*All models are adjusted for race and smoking

We explored whether associations of CV-βD and sPTB differed among women with high vs. low stress. While we found in stratified models, adjusted for race and smoking, that associations of low CV-βD and sPTB seemed stronger for women with high stress (aOR 2.24, 95% CI 0.90–5.58) than low stress (aOR 1.40, 95% CI: 0.83, 2.36), the interaction was not significant (P= 0.34).

When we examined the four possible combinations of high and low stress and high and low CV-βD, we found that women with sPTB (12.7%) were more likely than women with term births (5.6%) to have the combination of high stress and low CV-βD (Table 4). This association persisted after adjustment for race and smoking; women with both high stress and low CV-βD had higher odds of sPTB than women with neither exposure (aOR 3.16, 95% CI: 1.46–6.84) (Figure 1). Women with either high stress or low CV-βD appeared to have intermediate odds of sPTB compared to women with neither exposure, but these associations were not significant.

Table 4.

Combinations of high and low stress (PSS ≥ 30 and < 30) and high and low CV-βD (above and below the median) at 16–20 weeks of gestation among women who went on to deliver spontaneously preterm (sPTB) and term.

	sPTB (n=110)	Term (n=409)	Chi Square P
Combinations of stress and CV-βD	n (col %)	n (col %)	0.038
Low stress, high CV-βD	32 (29.1)	158 (38.6)
Low stress, low CV-βD	49 (44.5)	177 (43.3)
High stress, high CV-βD	15 (13.6)	51 (12.5)
High stress, low CV-βD	14 (12.7)	23 (5.6)

Figure 1.

Adjusted odds of spontaneous preterm birth among women in four categories: low stress (PSS < 30) and high CV-βD; low stress and low CV-βD, high stress and high CV-βD, and high stress and low CV-βD. Models adjusted for smoking and race ethnicity. Whiskers represent 95% confidence intervals.

Comment

Principal findings of the study

In this study, we demonstrated that maternal psychosocial stress was associated with reduced odds of low CV-βD, but that the combination of high stress and low CV-βD conferred significantly higher odds of sPTB. These findings suggest that CV-βD may serve as a biologic resilience factor to protect women from adverse exposures, including stress in pregnancy, and improve chances of a full term delivery.

Results - in the context of what is known

There are several studies demonstrating associations between psychosocial and mental health status with adverse birth outcomes.^8,9,18–26 However, specifically with respect to stress in pregnancy, recent, large studies have not demonstrated significant associations between stress and sPTB. A large prospective, multicenter study of over 10,000 nulliparous women found that those in the highest quartile (versus the other three) of PSS had non-significantly increased odds of sPTB (adjusted OR: 1.10, 95% CI: 0.88, 1.36).²⁷ In another study of over 5,000 Canadian women, higher PSS was also not significantly associated with sPTB (aOR 1.1, 95% CI 0.8–1.6).²⁸ Similarly, in our study, prior to adjustment for CV-βD, we did not detect a significant association of stress at 16–20 weeks of gestation and sPTB.

Regarding CV-βD, the literature exploring its role in sPTB is limited. The mechanism by which CV-βD may protect women from sPTB is not known but CV-βD is an epithelial-derived peptide with known antimicrobial activity, particularly against gram negative bacteria and candida species.²⁹ CV-βD has been shown to impair HIV replication and may affect viral transmission and pathogenesis.^30,31 Recently published primary findings from the Motherhood and Microbiome cohort (the cohort for present analysis) demonstrated that sPTB risk was higher with a cervicovaginal microbiota dominated by anaerobic bacteria with a paucity of Lactobacillus species, and that the risk of sPTB was mitigated, specifically among non-Hispanic black women, with high levels of CV-βD.¹⁵ To our knowledge, that was the first study to demonstrate the potential protective impact of CV-βD on sPTB risk. Herein, we demonstrated that high stress was associated with reduced odds of low CV-βD, but that women with low CV-βD levels in the setting of high stress were at particularly high risk of sPTB. While there are no data in the literature on stress and CV-βD, there are limited data on stress and beta defensin 2 in human saliva. In one study, 75 healthy volunteers underwent a psychological evaluation using a validated questionnaire (Questionnaire of Lipp‐ISS) and salivary beta defensin 2 and 3 were measured.³² There was no significant difference in levels among participants with stress (210.7 pg/ml) and without stress (242.3 pg/ml), but the sample size was small suggesting that the study may have been underpowered to detect differences. In another study of 15 health adults, salivary beta defensin 2 levels were higher after a yoga stretching intervention (165.4 pg/ML) than before (84.1 pg/mL) and that the elevation in levels was significantly different than changes after an intervention of resting only (p<0.05).³³ It is not clear why stress was associated with higher levels of CV-βD in our study, but further work to investigate exposures in relationship with CV-βD, which is potentially protective against sPTB, in the setting of high stress is warranted.

Clinical and Research Implications

A recent American College of Obstetricians and Gynecologists Committee Opinion publication emphasized that stress can contribute to depression and anxiety, and that pregnant women should be screened for these conditions during the course of prenatal care to improve outcomes.³⁴ It is possible that biomarkers such as CV-βD may eventually serve as additional screening and risk stratification tools for sPTB. While stress and low CV-βD may be independent risk factors additively resulting in higher sPTB risk, it is also possible that stress causes increases in CV-βD in some women who are thus protected from sPTB but not in others who remain at higher risk. In either case, if our findings are replicated in additional, larger cohorts, combining several risk factors including psychosocial and molecular biomarkers may improve prediction and eventually target interventions to reduce sPTB. Our findings warrant further studies to determine whether CV-βD or other antimicrobial peptides can serve as a resilience factors against environmental exposures that, if unmitigated, may lead to sPTB.

Strengths and Limitations

Our study has several strengths, including prospective enrollment of women in pregnancy, careful phenotyping of preterm births as sPTB, and racial/ethnic diversity. Non-Hispanic black women comprised the majority of the study population, which is important because this group of women has a fifty percent higher risk of preterm birth compared to white women in the United States.³⁵ Our study also combined psychosocial stress assessments with a biologic biomarker of immune function related to reproductive health as opposed to studies that often focus on one or the other. Limitations of our study include potential for residual confounding and reliance on perceived stress as opposed to a biomarker of stress. Additionally, our positive findings could be due to chance (Type I error) in the setting of multiple testing. Further we performed a secondary analysis of an already-completed cohort study and as such may not have been adequately powered to detect significant interactions as the cohort was not originally designed for the particular analysis.

Conclusion

We found that high perceived stress and low CV-βD levels were independently associated with elevated odds of sPTB, and when they co-occurred, they resulted in more than twice the odds of sPTB compared to women with neither exposure. Our findings warrant further work to examine the interplay of host environments, immune function, and their impact on the pathogenesis of sPTB. Combining psychosocial data with biomarker data may better predict sPTB and direct targeted therapeutic interventions that may bolster resilience and prevent sPTB.

AJOG at a Glance.

A. Why was the study conducted?

• We conducted this study to determine whether maternal psychosocial stress affected the risk of low cervicovaginal beta defensin-2 levels (a recently recognized risk factor for spontaneous preterm birth).

• We also analyzed the additive impact of concurrent high stress and low cervicovaginal beta defensin-2 levels on the odds of spontaneous preterm birth.

B. What are the key findings?

• High stress was associated with reduced odds of low cervicovaginal beta defensin levels.

• Women with the combination of high stress and low cervicovaginal beta defensin levels had increased odds of spontaneous preterm birth compared to women with neither exposure.

C. What does this study add to what is already known?

• Thus far, studies in the literature vary with respect to associations of perceived stress and spontaneous preterm birth.

• While stress alone was not significantly associated with spontaneous preterm birth in our study, when combined with low cervicovaginal beta defensin levels, women with high stress had substantially higher odds of spontaneous preterm birth.

• The combination stress and immune function on the pathogenesis of preterm birth warrants further study.