A PROPOSAL TO INVESTIGATE THE ROLES OF MATERNAL INFLAMMATION AND DIET IN ASSOCIATIONS OF MATERNAL PARABEN CONCENTRATIONS WITH GESTATIONAL LENGTH B y Diana C . Pacyga A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Epidemiology Master o f Science 2021 ABSTRACT A PROPOSAL TO INVESTIGATE THE ROLES OF MATERNAL INFLAMMATION AND DIET IN ASSOCIATIONS OF MATERNAL PARABEN CONCENTRATIONS WITH GESTATIONAL LENGTH B y Diana C . Pacyga The proposed research will fill a public health need by providing information about the role s inflammation and dietary interventions play in associations of maternal paraben concentrations with shorter gestation. Specifically, this study proposes to investigate the maternal inflammatory pathways linking parabens with shorter gestation, as well as the potential intervening effect of a maternal anti - inflammatory diet on shorter gestation in response to paraben s. The central hypothesis is that higher paraben concentrations are associated with shorter gestation due to elevated maternal inflammation and that an anti - inflammatory maternal diet mitigates these relationships. The proposed study will test this central hypothesis in 482 pregnant women enrolled in the Illinois Kids Development Study (I - KIDS), which is a n ongoing prospective pregnancy and birth cohort with the primary goal of evaluating the impacts of prenatal chemical exposures on infant neurodevelopment . The feasibility of testing these hypotheses has been determined by conducting several preliminary studies in a sub - sample of 294 I - KIDS women. Overall, f indings from this study will inform future research and clinical practice about the biological targets of parabens during pregnancy and guide prenatal healthcare professionals to make effective dietary recommendations to their pregnant pat ients. Copyright by DIANA C PACYGA 20 21 iv ACKNOWLEDGEMENTS I would like to thank my M.S. thesis committee ( Dr s . Rita Strakovsky , Joseph Gardiner , and Nicole Talge ) for all their help while I was preparing my defense and thesis . Specifically, Dr. Strakovsky provided critical guidance and iterative feedback during the writing process. Dr. Gardiner gave detailed guidance on statistical method selection and power analyses. Dr. Talge provided critical feedback during the oral defense and prior to thesis submission. I would like to acknowledge my collaborators from the Illinois Kids Development Stud y (I - KIDS) , especially Dr. Susan Schantz , the principal investigator who developed and designed the original parent study from which this thesis was developed. I would also like to acknowledge the I - KIDS staff for working long hours to collect all the data used for the preliminary studies in this proposal , as well as the pr egnant women who took the time to participate and contribute critical data. Finally, I would like to thank the Department of Epidemiology and Biostatistics at Michigan State University for providing a supportive environment and training that is invaluable for my professional development . v TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ...... vi i LIST OF FIGURES ................................ ................................ ................................ .. v ii i KEY TO ABBREVIATIONS ................................ ................................ ........................ ix A. OBJECTIVES ................................ ................................ ................................ ....... 1 B. SPECIFIC AIMS ................................ ................................ ................................ ... 3 B.1. Specific Aim 1 ................................ ................................ ............................... 3 B.2. Specific Aim 2 ................................ ................................ ............................... 4 C. BACKGROUND/SIGNIFICANCE ................................ ................................ ......... 5 D. PRELIMINARY STUDIES/PROGRESS REPORT ................................ ................ 8 D.1. Specific Aim 1. ................................ ................................ .............................. 8 D.1.1. Rationale. ................................ ................................ ............................. 8 D.1.2. Preliminary study 1: urinary paraben concentrations. ......................... 10 D.1.3. Preliminary study 2: parabens and gestational length. ....................... 11 D.1.4. Conclusions. ................................ ................................ ....................... 12 D.2. Specific Aim 2. ................................ ................................ ............................ 13 D.2.1. Rationale. ................................ ................................ ........................... 13 D.2.2. Preliminary study 1: maternal diet quality and gestational length. ...... 14 D.2.3. Preliminary study 2: parabens, gestational length, and diet quality. ... 15 D.2.4. Conclusions. ................................ ................................ ....................... 16 E. RESEARCH DESIGN AND METHODS ................................ ............................. 17 E.1. Introduction. ................................ ................................ ................................ 17 E.2. Research design and methods for Specific Aim 1. ................................ ...... 1 8 E.2.1. Outcome variable(s): maternal second trimester inflammation. .......... 19 E.2.2 Exposure variable(s): maternal urinary paraben concentrations. ......... 20 E.2.3. Covariates. ................................ ................................ .......................... 21 E.2.4. Expected re sults and alternative strategies for Aim 1. ........................ 21 E.3. Research design and methods for Specific Aim 2. ................................ ...... 22 E.3.1. Outcome variable(s): gestational length. ................................ ............. 22 E.3.2. Exposure variable(s): maternal urinary paraben concentrations. ........ 23 E.3.3. Effect modifying varia ble(s): maternal inflammatory diet index . .......... 23 E.3.4. Covariates. ................................ ................................ .......................... 24 E.3.5. Expected results and alternative strategies for Aim 2. ........................ 24 F. PROPOSED ANALYSIS AND PRESENTATION OF TABLES/FIGURES .......... 25 vi F.1. Statistical analysis. ................................ ................................ ...................... 25 F.1.1. Overview. ................................ ................................ ............................ 25 F.1.2. PCA to characterize inflammation (Aim 1a). ................................ ....... 25 F.1.3. Mediation by maternal inflammation (exploratory Aim 2b). ................. 26 F.1.4. Modification by maternal diet (Aim 2). ................................ ................. 26 F.1.5. Covariates. ................................ ................................ .......................... 27 F.1.6. Statistical power. ................................ ................................ ................. 28 F.2. Presentation of tables and figu res ................................ ............................... 29 F.2.1. Tables and figures related to both specific aims. ................................ 29 F.2.2. Tables and figures related to Specific Aim 1. ................................ ...... 32 F.2.3. Tables and figures related to Specific Aim 2. ................................ ...... 34 G. OVERALL CONCLUSIONS ................................ ................................ ................ 36 G.1. Strengths and future directions. ................................ ................................ .. 36 G.2. Scienti fic rigor. ................................ ................................ ............................ 36 REFERENCES ................................ ................................ ................................ .............. 3 8 vii LIST OF TABLES Table 1. Gestational associations of parabens with select maternal inflammatory markers. ................................ ................................ ................................ .......................... 9 Table 2. U rinary paraben concentrations in I - KIDS. ................................ ...................... 11 Table 3. Associations of AHEI - 2010 with gestational length. ................................ ........ 1 5 Table 4. Demographic and lifestyle characteristics of 482 I - KIDS pregnant women (2013 - 201 8 ) . ................................ ................................ ................................ .................. 18 Table 5. Power estimation for evaluating associations in 482 I - KIDS women. .............. 29 viii LIST OF FIG URES Figure 1. Summary of Specific Aims 1 and 2. ................................ ................................ . 3 Figure 2. Associations of parabens with gestational length. ................................ .......... 12 Figure 3. Associations of parabens with gestational length stratified by AHEI - 2010 ..... 1 6 Figure 4. List of cytokines proposed for analysis. ................................ .......................... 20 Figure 5. DAG with hypothesized associations of potential covariates with parabens, inflammation, and gestational length. ................................ ................................ ............ 2 8 ix KEY TO ABBREVIATIONS AHEI - 2010 Alternative Healthy Eating Index - 2010 BMI Body mass index BPA Bispheno l A CDC Centers for Disease Control and Prevention CI Confidence interval CRP C - reactive protein DAG Directed acyclic graph EDC Endocrine disrupting chemicals E - DII Empirical Dietary Inflammatory Index FFQ Food Frequency Questionnaire IFN Interferon I - KIDS Illinois Kids Development Study IL Interleukin NHANES National Health and Nutrition Examination Survey NHS PCA Principal Component Analysis SG Specific gravity TGF Transforming growth factor TNF Tumor necrosis factor U.S. United States 1 A. OBJECTIVES Shorter gestation is one of the leading risk factors associated with poor fetal development and adverse maternal and child outcomes. Even early - term infants (37 - 39 weeks), not just those born pre - term (<37 weeks), have higher risk of adverse birth and postnatal outcomes compared to full - term infants (39 - 41 weeks). Pregnant women are ubiquitously exposed to environmental pollutants, including parabens, which are anti - microbial ag ents found in personal care products and cosmetics. In pregnant populations with lower pre - term birth prevalence, we and others have observed that higher maternal paraben concentrations are associated with shorter gestation. While the precise mechanisms of parabens are poorly understood, several studies suggest that parabens may alter immune cell function , as well as mRNA expression or circulating levels of individual cytokines. This is concerning because trimester - specific inflammatory profiles play a crit ical role in regulating gestational length. Specifically, the second trimester is uniquely anti - inflammatory to support fetal development, but little is known about the impacts of parabens on maternal inflammatory patterns during the second trimester, and no study has evaluated whether associations of parabens with gestational length are due to inflammation. A healthy maternal diet is critical for proper fetal development, and the adverse consequences of some persistent chemical exposures (e.g. mercury, le ad, and air pollution ) have been shown to be mitigated by improving certain components of the maternal diet. However, previous studies only focused on individual nutrients rather than on overall maternal diet quality, making it difficult to prov ide relevant public health 2 messages to pregnant women. To address this, o ur preliminary findings suggest that better overall maternal diet, measured by an index that focuses on diet quality, is protective against shorter gestation in response to paraben s . This protective effect may be due to the anti - inflammatory nature of healthy diets, as numerous studies have shown that a healthy maternal diet is associated with reduced maternal inflammation and improved gestational length. However, it is unkno wn whether a maternal low - inflammatory diet, specifically, can mitigate the negative effects of paraben s on gestational length. Therefore, the overarching goal is to investigate the role of maternal inflammation in the relationship between paraben s and sh orter gestation, and evaluate how an anti - inflammatory maternal diet interacts with paraben concentrations . The central hypothesis is that paraben s are associated with shorter gestation due to elevated maternal inflammation and that an anti - inflammatory ma ternal diet mitigates these relationships. The central hypothesis will be tested in 482 women enrolled in the Illinois Kids Development Study (I - KIDS), a prospective pregnancy and birth cohort evaluating the impacts of prenatal chemical exposures on infant neurodevelopment in two specific aims that are summarized in Figure 1 . 3 B. SPECIFIC AIMS Figure 1. Summary of Specific Aims 1 and 2. Specific Aim 1a will evaluate associations of parabens with maternal inflammation, while Specific Aim 1b will assessed whether associations of parabens with gestational length can be explained by changes in maternal inflammation. Specific Aim 2 will determine whether associations of parabens with gestational length are different by maternal inflammatory diet quality. B.1. Specific Aim 1 . Specific Aim 1 will e valuate associations of maternal gestational paraben concentrations with maternal inflammation (Aim 1a). Maternal concentrations of four parabens (butyl, ethyl, methyl, propyl) have already been measured in a pool of 5 cross - pregnancy urine samples. A panel of 21 critical pro - and anti - inflammatory cytokines will be analy zed in maternal fasting plasma samples collected at median 17 (range: 13 - 22) weeks gestation, and p rincipal c omponent a nalysis (PCA) will be used to identify patterns of cytokine concentrations to characterize maternal second trimester inflammatory profi les . Our primary hypothesis is that higher paraben concentrations are associated with elevated maternal inflammation. If Aim 1a suggests that parabens are pro - inflammatory, we may pursue an exploratory Aim 1b to assess whether associations of parabens with shorter gestation can be partly explained by elevated maternal inflammation in response to 4 paraben s . The research accomplished in Specific Aim 1 will determine whether parabens can alter critical maternal inflammatory profiles implicated with shorter gest ation. B.2. Specific Aim 2. Specific Aim 2 will evaluate differences in associations of maternal paraben concentrations with shorter gestation by a maternal inflammatory diet index . To assess the anti - inflammatory potential of maternal diets in I - KIDS, th e validated Empirical Dietary Inflammatory Index (E - DII) will be calculated using average values from three - month food frequency questionnaires administered at 10 - 14 and 34 - 36 weeks gestation. Maternal paraben concentrations will be measured as described i n Specific Aim 1. Our primary hypothesis is that an anti - inflammatory maternal diet mitigates known associations of parabens with shorter gestation. Completion of Specific Aim 2 will ascertain the intervening effects of an anti - inflammatory maternal diet i n response to paraben s . 5 C. BACKGROUND/ SIGNIFICANCE T his project will investigate the role s of maternal inflammation and diet quality i n the associations of maternal paraben concentrations with shorter gestation in a population of pregnant women. Specifically, these studies will assess whether parabens are associated with alter ed maternal second trimester inflammation and whether a maternal anti - inflammatory diet protects against shorter gestation in response to paraben s . These studies will address a knowledge gap about the gestational molecular targets of parabens and will provide a clinically relevant dietary intervention for mitigating adverse outcomes in response to paraben s . The scientific premise is that parabens are pro - inflammatory compound s associated with shorter gestation and that mothers consuming an anti - inflammatory diet are protect ed against the adverse effects of paraben s . Shorter gestation is the leading risk factor for adverse fetal growth and development . Pre - term birth (<37 weeks gestation) is a leading cause of infant death worldwide and the U nited States (U. S. ) pre - term birth prevalence is ~ 10% (1) . Pre - term birth is associated with l ife - long adverse health outcomes in offspring, including poor birth outcomes, respiratory problems, cognitive problems, metabolic syndrome , and cardiovascular disease (2) . Although infant s born pre - term are at greatest risk of complications, even early - term infants born between 37 - 39 weeks have higher risk of adverse outcomes compared to full - term infants born between 39 - 41 weeks (3) . F or example, compared to infants born at 39 - 41 weeks gestation, those born before 39 weeks have higher risk of poor cognitive and motor development during the first year of life, and higher risk of respiratory problems and hypoglycemia right after birth (4, 5) . However, relatively f ew 6 studies have focused on evaluating risk factors for shorter gestation in predomina tely full - term populations. Inflammation play s a major role in shortening gestation. Pregnancy is characterized by distinct trimester - specific inflammatory profiles that regulate pregnancy progression and fetal development (6) . Elevated m aternal inflammation during t he first and third trimesters is critical for implantation and parturition, respectively (6) . Conversely, the second trimester is a n anti - inflammatory state that supports fetal development by ensuring maternal tolerance of fetal antigens (7) . Dysregulated maternal inflammation, especially during the second trimester, may be linked to adverse birth outcomes , including earlier bir th (8, 9) . Therefore, identifying factors associated with altered maternal inflammation is critical for develop ing intervention s against shorter gestation . Parabens may be pro - inflammatory compounds associated with shorter gestation . P arabens (butyl, ethyl, methyl, propyl) are anti - microbial agents used in personal care products , cosmetics, some food products , and medications (10) . Over 99 % of U.S. pregnant women have measurable levels of at least one paraben in their urine (11) . This is concerning because paraben s are endocrine disrupting chemicals (EDCs) that target reproductive tissue s and hormones (12 - 16) . Addition ally, h igher p araben concentrations are associated with shorter gestation (17 - 20) , which agrees with our preliminary findings . S ome observational studies in pregnan t women suggest that parabens alter inflammatory markers (21 - 23) , and experimental stud ies show that parabens impact immune cell fu nction and pro - inflammatory cytokine expression (24 - 26) . However, to our knowledge, 7 no study has evaluated associations of maternal paraben concentrations with maternal second trimester inflammation (Aim 1a) or assessed if inflammation part ly explains associations of paraben concentrations with shorter gestation (Aim 1b) . An anti - inflammatory maternal diet may protect against shorter gestation in response to paraben s . Appropriate m aternal nutrition is critical for fetal development , while p oor diet quality during key gestational periods is associated with adverse pregnancy outcomes , including shorter gestation (27) . O bservational studies and randomized controlled trials in non - pregnant populations found that h ealthier diets are associated with lower inflammation (28 - 31) , likely due to the anti - inflammatory nature of healthy diets. Importantly, studies in pregnant populations suggest that the adverse effects of some chemical exposures can be mitigated by improv ing maternal diet s (32 - 34) . However, these studies focused on indiv idual nutrients rather than overall diet quality, which does not reflect dietary patterns in pregnant women and is difficult to translate into actionable food - based intervention s . O ur preliminary studies suggest that better overall diet quality mitigate s a ssociations of parabens with shorter gestation , but it is unknown whether a n anti - inflammatory diet , specifically, protects against shorter gestation in response to paraben s (Aim 2) . This research will be among the first epidemiological studies to investigate the role s of maternal inflammation and diet i n associations of maternal paraben concentrations with shorter gestation. Th is proposal will provide insight into potential biological targets of parabens and inform an eff ective dietary intervention against parabens . 8 D. PRELIMINARY STUDIES/PROGRESS REPORT D.1. Specific Aim 1. D.1.1. Rationale. Shorter gestation is a leading risk factor for adverse fetal growth and development, and inflammation appears to play a major role (8, 9, 35) . The second trimester of preg nancy is an anti - inflammatory period responsible for fetal growth (6, 7) , and several studies suggest that elevated second trimester inflammation may predict shorter gestation (8, 9, 36) . Anti - microbial agents like parabens are thought to disrupt gestational inflammatory pathways ( Table 1 ) (21 - 23) . For example, a prospective study of Northern Puerto Rican women from the PROTECT birth cohort (n= 141 ) found negative associations of butyl and propyl paraben s with pro - inflammatory CRP (21) . A small study of Michigan women from the MMIP birth cohort (n=56) only found a negative association between butyl paraben and pro - inflammatory IL - 6 (22) . However, a nested case - control study of Boston women from the LIFECODES cohort (130 pre - and 352 full - term births) found negative associations of ethyl paraben with pro - inflammatory interleukin (IL) - necrosis factor (TNF) t positive associations of methyl and propyl parabens with pro - inflammatory IL - 6 and C - reactive protein (CRP), respectively (23) . Results from these studies are supported by mechanistic studies and an observational study in a non - pregnant population. One in vitro study found that exposing peripheral human lymphocytes to parabens (butyl, ethyl, methyl, propyl) inhibited the release of lysosomal enzymes compared to the control group at concentrations as low as 0.06 mmol/L, which could disrupt critical immune pathways (25) . Additionally, a study in 9 ( p ro - i nflammatory cytokine) after exposure to methyl and propyl paraben s , but lower mRNA levels of IL - 8 (pro - inflammatory cytokine) after exposure to propyl paraben at 1µM or 10µM compared to the control group (24) . In a cross - sectional analysis of data from U.S. adults, higher paraben concentrations were associated with reduced inflammatory bowel disease symptoms , which authors suggested could connect parabens to alteratio ns in gut inflammatory pathways (37) . The precise cellular mechanisms li nking parabens to inflammation are unknown, but may be related to the endocrine disrupting properties of parabens since immune cells express hormone receptors (38) . For cytokines, red and green denote pro - and anti - inflammatory, respectively. POSITIVE or NEGATIVE indicate the direction of significant associations between parabens and cytokines. NA, association not assessed. Ø, association was not significant at P <0.05. References for studies from LIFECODES (23) , PROTECT (21) , and MMIP (22) . 10 Studies in pregnant women provide preliminary evidence that parabens may alter maternal inflammation in populations with high pre - term birth rates (21 - 23) . However, additional r esearch needs to evaluate these relationships in lower risk populations, since early - term infants are also at risk for adverse outcomes (3) . Furthermore, two of the previous studies described above evaluated associations of parabens with inflammation across pregnancy (22, 23) . Given that t he second trimester is uniquely anti - inflammatory, evaluating associations of paraben s with maternal inflammation during this time is especially important. Lastly, two studies focused on evaluating the impact of parabens on a limited number of individual c ytokines (21, 23) . Because inflammation is characterized by a complex network of pro - and anti - inflammatory compounds that interact to create and maintain a favorable fetal environment, characterizing patterns of several cytokine concentrations may better represent maternal inflammation status than individual cytokine concentrations (39) . We will address these limitations in a low pre - term birth population and by assessing maternal second trimester inflammation using a composite of anti - and pro - inflammatory cytokines. The feasibility of completing Specific Aim 1 using I - KIDS data is supported by the following preliminary studies. D.1. 2. Preliminary study 1: urinary paraben concentrations. Paraben data needed to address the proposed aims are already available for all 482 women, and our preliminary findings suggest that 100% of I - KIDS women had measurable (non - zero) levels of urinary me thyl and propyl parabens, while 66% and 99% had measurable levels of butyl and ethyl parabens, respectively ( Table 2 ). Median urinary paraben concentrations in I - KIDS were highest for methyl and lowest for butyl paraben, 11 potentially because women have high er exposure to methyl paraben due to its widespread use in consumer products (40) . Median urinary paraben concentrations in I - KIDS women were somewhat lower than those in U.S. reproductive - aged women from the 2013 - 14 and 2 015 - 16 National Health and Nutrition Examination Surveys (NHANES) (41, 42) . Table 2. Urinary p araben concentrations in I - KIDS . I - KIDS women with measurable levels 1 I - KIDS 2013 - 201 8 (n=482) NHANES 2013 - 2016 (n=743) 2 Paraben % Median (25 th , 75 th percentile) in ng/mL Butyl 66 0.1 (0, 0.3) 0.1 (0.1, 0.5) Ethyl 99 1.3 (0.5, 6.8) 2.2 (0.7, 14.0) Methyl 100 50.2 (18.2, 133.8) 86.5 (19.0, 266.4) Propyl 100 8.4 (2.2, 27.7) 15.4 (2.9, 69.5) 1 Non - zero concentrations; 2 Urinary paraben concentrations of 18 - 40 - year - old U.S. females from 2013 - 14 and 2015 - 16 NHANES survey years . NHANES, National Health and Nutrition Examination Survey. D. 1.3 . Preliminary study 2: p arabens and gestational length. We preliminarily evaluated associations of four parabens with gestational length in the first 294 I - KIDS women who have paraben and gestational length data ( Figure 2 ). Overall, we found that parab ens tended to be associated with shorter gestation. Specifically, every 2 - fold increase in ethyl, methyl, and propyl paraben was non - significantly associated with 0.3 (95% CI: - 0.05, 0.6), 0.5 (95% CI: - 0.02, 1.0), and 0.4 (95% CI: - 0.1, 0.8) day decreases in gestational length, respectively. 12 Figure 2. Associations of parabens with gestational length. Parabens were assessed in a pool of five cross - pregnancy urines, specific gravity - adjusted, and ln - transformed. Multivariable linear regression models cont rolled for maternal age, pre - pregnancy body mass index ( BMI ) , race/ethnicity, education, parity, smoking since conception, second - hand smoke exposure, conception season, diet quality, and fetal sex. Data were back - transformed to represent change in gestati onal length (95% CI) in days for every 2 - fold increase in paraben concentration . CI, confidence interval. n=294. D.1.4. Conclusions . As previously discussed, several studies suggest that parabens may impact maternal inflammation (21 - 26, 37) . Given that higher second trimester maternal inflammation may be predictive of shorter gestation (8, 9, 36) , the primary objective of Aim 1a is to evaluate whether maternal paraben concentrations are associated with elevated maternal early second trimester inflammation. If Aim 1a suggests th at parabens are pro - inflammatory, we may pursue an exploratory Aim 1b to ask whether associations of parabens with shorter gestation are partly explained by elevated maternal inflammation in response to paraben s . 13 D.2. Specific Aim 2. D.2.1. Rationale. Maternal diet is an important determinant of fetal health, and in epidemiological studies, poor maternal diet quality during key gestational periods is associated with adverse birth outcomes, including shorter gestation (27) . Pregnant women are ubiquit ously exposed to parabens, which we and others have shown are associated with shorter gestation (17 - 20) . The adverse consequences of some chemical exposures (e.g. mercury, lead , and air pollution ) have been shown to be mitigated by improving ma ternal diet. For example, in pregnant Boston women, higher mercury levels were associated with poorer child cognitive performance at age 3, which was attenuated in women who consumed >2 weekly servings of fish a major source of polyunsaturated omega - 3 fa tty acids important for brain development (32) . Another U.S. study evaluated the importance of maternal calcium intake on the release of bone - stored lead into maternal circulation, and found that pregnant women with high calcium intake (>2000 mg/day) had lower blood lead levels than those with low int ake (<600 mg/day) (33) . Last ly, a prospective cohort study of Boston pregnant women with fertility problems found that associations of air pollution with decrease d probability of live birth were attenuated with higher supplemental folate intake ( ~ 1000 µg/day) (34) . Together, t hese studies suggest that maternal diet can intervene against the adverse consequences of chemical exposures, especially against exposures that we cannot avoid. However, these studies only focused on individual nutrients rather than on overall d iet quality (32 - 34) . This is a limitation since better overall maternal diet quality is protective against adverse gesta tional outcomes (27, 43) , potentially due to the combined anti - inflammatory properties of whole foods in healthy diets (44 - 46) . 14 Given previous studies (including our own preliminary findings) showing associations of parabens with shorter gestation (17 - 20) , and the known protective effects of healthy maternal diets (27) , the current study will investigate whether an anti - inflammatory maternal diet can mitigate associations of paraben s with shorter gestation. The feasibility of completing Spec ific Aim 2 using I - KIDS data is supported by the following preliminary studies. D.2.2. Preliminary study 1: maternal diet quality and gestational length. Maternal diet quality in I - KIDS is measured using the Alterative Healthy Eating Index 2010 (AHEI - 2010), which is a dietary index based on published research related to foods and nutrients predictive of chronic disease risk (47) . The AHEI - 2010 is scored o ut of 110 points using 11 dietary components, and a higher AHEI - 2010 score signifies healthier diet quality predictive of lower chronic disease risk. In I - KIDS, the score is calculated using data collected from a three - month food frequency questionnaire (d escribed later) administered at 10 - 14 and 34 - 36 weeks gestation. Consistent with other studies assessing associations of diet quality with pre - term birth risk (27) , our preliminary findings suggest that a better maternal AHEI - 2010 score is associated with longer gestatio n. Specifically, in fully adjusted models, each 10 point increase in AHEI - 2010 (at 10 - 14 weeks) was associated with 1.2 day (95% CI: 0.2, 2.2) longer gestation ( Table 3 ). These preliminary findings highlight the importance of high maternal diet quality for gestational length, and confirm that these associations exist in the I - KIDS population. 15 D. 2.3 . Preliminary study 2: parabens, gestational length, and diet quality . As presented in Figure 2 and described in D. 1.3 . , our preliminary data suggest that higher maternal paraben concentrations during pregnancy are associated with shorter gestational length. Our preliminary findings also suggest that the negative impacts of maternal paraben s on gestation length may be mitigated in women with better diet quality. Figure 3 demonstrates that previously observed overall associations only persisted in women who consumed a lower quality diet (AHEI - 2010 score < median; red diamond s ), such that 0.5 (95% CI: 0.1, 0.9) and 0.7 (95% CI: - 0.003, 1.4) day decrease s in gestation length were observed for every 2 - fold increase in urinary ethyl and methyl paraben c oncentrations, respectively. In women who consumed a better - quality diet (AHEI - 2010 score median), paraben s were no longer associated with gestational length (green diamond s ). Table 3. Associations of AHEI - 2010 with gestational length. Gestational age at diet assessment Model 1: Unadjusted (95%CI); P Model 2: Adjusted # (95%CI); P 10 - 14 weeks 0.9 (0.02, 1.8); 0.04 1.2 (0.2, 2.2), 0.02 34 - 36 weeks 0.3 ( - 0.6, 1.2); 0.48 0.4 ( - 0.6, 1.4); 0.42 Average 0.8 ( - 0.2, 1.7); 0.11 1.1 ( - 0.02, 2.2); 0.05 Gestational length change (days) for each 10 - point increase in AHEI - 2010 . # Controlling for maternal age, pre - pregnancy BMI, race/ethnicity, education, parity, smoking since conception, conception season, and fetal sex. AHEI - 2010, Alternative Healthy Eating Index 2010; CI, confidence interval. n=294. 16 Figure 3. Associations of parabens with gestational length stratified by A HEI - 2010. Parabens assessed in a pool of five cross - pregnancy urines were specific gravity - adjusted and ln - transformed. Multivariable linear regression models controlled for maternal age, pre - pregnancy BMI, race/ethnicity, education, conception season, smo king since conception, diet quality, parity, second - hand smoke exposure, fetal sex, and a multiplicative interaction between parabens and AHEI - 2010. Data were back - transformed to represent change in gestational length (95%CI) in days for every 2 - fold incre ase in paraben concentrations in women with AHEI - 2010 < median (red diamond) and AHEI - median (green diamond). AHEI - 2010, Alternative Healthy Eating Index 2010; CI, confidence interval. n=294. D.2.4. Conclusions. We have shown that maternal diet is an important predictor of gestational length, and that a healthier maternal diet may protect against shorter gestation in response to paraben s . These protective effects by better diet quality may be due to anti - inflammatory components of a healthy diet (30, 31) . Therefore, the objective of Aim 2 is to test the hypothesis that an anti - infla mmatory maternal diet mitigates known associations of parabens with shorter gestation. 17 E. RESEARCH DESIGN AND METHODS E.1. Introduction. The proposed stud ies will test the central hypothesis that elevated maternal inflammation partially explai ns associations of higher paraben concentrations with shorter gestation and that an anti - inflammatory maternal diet is protective against shorter gestation in response to parabens . To investigate this hypothesis, the proposed stud ies will expand on the aims of I - KIDS , a n ongoing prospective pregnancy cohort with the primary goal of evaluating the impact s of gestational EDC exposures on child health . I - KIDS participants were recruited at the first prenatal visit from two local obstetri c clinics in Champaign - Urbana, IL . Eligible women were 10 - 14 weeks gestation, 18 - 40 years old, fluent in English, not carrying multiples, not in a high risk pregnancy, residing within a 30 - minute drive of the University of Illinois Urbana - Champaign campus, willing to provide a fasting blood sample at 16 - 18 weeks gestation, not planning on moving out of the area before . T he proposed stud ies will utilize data from the first 482 women who enrolled in I - KIDS between 12/ 2013 and 09 / 2018 and remained in the study through the birth of their infant. Characteristics of these 482 women are as follows: 59 % are 3 0 years of age, 81% are college educate d, 80% are non - Hispanic white, 88% are married, 5 5 % have an annual 6 0,000, and 5 4 % have a BMI<25 kg/m 2 ( Table 4 ) . Additionally, these women have a median AHEI - 2010 of 56 (range: 28 - 83; out of 110), and median gestational age at birth of 39 weeks (range: 30 - 42). 18 Table 4. Demographic and lifestyle characteristics of 482 I - KIDS pregnant women (2013 - 201 8 ). Characteristic Category n=482 Maternal age < 30 years old 197 (40.9) 285 (59.1) Race/ethnicity Non - Hispanic White 385 (80.0) Others 96 (20.0) Education Some college or less 90 (18.7) College grad or higher 392 (81.3) Marital status Married 426 (88.4) Living as married/single 56 (11.6) Employment status Unemployed 67 (13.9) Employed 415 (86.1) Annual household income < $60,000 138 (28. 8 ) $60,000 - $99,999 182 (38.1) 158 (33.1) Conception season Winter 122 (25.4) Spring 133 (27.7) Summer 107 (22.2) Fall 119 (24.7) Parity No live births 246 (51.0) 236 (49.0) Pre - pregnancy BMI < 25 kg/m 2 256 (53.6) 2 222 (46.4) Smoking in 1 st trimester No 423 (87. 7 ) Yes 24 (5.0) Missing 35 (7.3) AHEI - 2010 Score < 56 240 (49.9) 241 (50.1) Missing: race/ethnicity, conception season, AHEI - 2010 (n=1); annual household income, pre - pregnancy BMI (n=4). E.2. Research design and methods for Specific Aim 1 . The overall objective of Specific Aim 1 is to evaluate associations of maternal paraben concentrations with maternal inflammation (Aim 1a) and determine if maternal inflammation partially explains associations of maternal paraben concentrations with 19 shorte r gestation (Aim 1b). E.2.1. Outcome variable(s) : m aternal second trimester inflammation . Maternal fasting blood samples were collected between 16 - 18 weeks gestation (early second trimester). Women fasted for 10 - 12 hours prior to the blood draw, and 30 - 35 mL of blood was collected by a certified phlebotomist at Carle Physician Group Christie Clinic in Champaign - Urbana, IL . To assure consistency across participants , all blood samples remained at room temperature for exactly 2 hours prior to processing. Sa mples were collected into glass heparin - containing v acutainer tubes, centrifuged at room temperature for 20 min, and the resulting plasma was aliquoted into cryovial tubes before immediate stor age at - 80° C . Plasma immune markers will be analyzed with a multiplex assay using Luminex xMap ® Technology at the University of Michigan Diabetes Research Center Clinical Core Chemistry Laboratory , which is a state - of - the art facility with extensive expertise in performing th ese types of analyses. The benefit of using a multiplex assay is that multiple analytes can be measured i n a single, small volume sample. Samples will be assessed for a comprehensive panel of early second trimester maternal pro - and anti - inflammatory cytok ines as shown in Figure 4 , which include s 1 2 analytes associated with shorter gestation and nine other critical inflammatory c ytokines (48, 49) . As will be described later in F.1.2 . , w e will utilize statistical methods to identify patterns of cytokine concentrations ( as previously reported in non - pregnant populations (50) ) that will approximate maternal second trimester inflammation status. 20 Figure 4 . List of c ytokines proposed for analysis. All 21 cytokines listed are involved in inflammatory pathways, while 12 have been implicated with shorter gestation (as specified in the figure) . C RP, C - reactive protein; IFN, interferon; IL, interleukin; TGF, trans forming growth factor; TNF, tumor necrosis factor. E.2.2. Exposure variable (s) : m aternal urinary paraben concentrations . I - KIDS collects five first morning urine samples across pregnancy at 10 - 14, 16 - 18, 22 - 24, 28 - 30, and 34 - 36 weeks gestation. Field blanks are obtained for every 10 samples . Urine sample s are collected using bisphenol A ( BPA ) - and phthalate - free materials approved by the Centers for Disease Control and Prevention (CDC). After collection, urine samples are vortexed and specific gravity is measured . Samples are then aliquoted and immediately stored at - 80°C . Additionally, a pool of all 5 urines is created for each participant by adding 900µL of urine from each gestational timepo int into one 5 mL polypropylene cryovial tube . Because parabens are non - persistent chemicals with short half - lives in the body (51) , urinary paraben concentrations for th is study were measured in pooled samples to provide a reliable biomarker of maternal paraben exposure across pregnancy (52, 53) . Urine has been shown to be most reliable for assessing non - persistent chemical concentrations (54) , and pooling urine samples is both cost effective and appropriate for evaluating non - persistent chemicals. U rinary concentrations of four Inflam matory cytokines associated with shorter gestation Other cytokines involved in inflammatory pathways IFN IL - 17A IFN - 2 IL - 27 IL - 1 TNF IL - IL - 31 IL - 1Ra TNF IL - 9 IL - 33 IL - 6 TGF IL - 13 IL - 10 IL - 18 IL - 22 IL - 12 CRP IL - 23 21 parabens (butyl, ethyl, methyl, and propyl) were analyzed at the CDC Divi sion of Laboratory Sciences using established protocols (55) . To account for urine dilution, all paraben concentrations will be specific gravity (SG) adjusted using the equation P c = P [( SG m SG i where P c is the SG - adjusted paraben concentration, P is the measured paraben concentration (ng/mL), SG m is the median SG of I - KIDS mothers , and SG i is the SG of each urine sample (56) . E.2.3. Covariates. Information regarding maternal characteristics in I - KIDS was collected using interviews and surveys . The current study will utilize maternal data collected at the baseline visit (10 - 14 weeks) , which included information about demographics (e.g. race, age) , health (e.g. body mass index) , lifestyle (e.g. smoking, alcohol use) , diet, depression, perceived stress, and interpersonal support . Health - related covariates, including diagnoses of medical conditions (e.g. pre - eclampsia) and gestational weight gain will be obtained from electronic medical record s. Aim - specific covariates used in statistical models are described in F.1.5 . E.2.4. Expected results and alternative strategies for Aim 1. Based on previous studies implicating parabens with gestational inflammatory pathways , we expect t hat parabens will be positively or negatively associated with specific individual pro - and anti - inflammatory cytokines, respectively. To provide additional insights into the paraben - inflammation relationship, we will utilize a statistical approach to ident ify patterns of cytokine concentrations to better predict overall maternal inflammation status. W e 22 expect to observe significant associations of parabens with distinct clusters of cytokines that represent apparent pro - and anti - inflammatory states. If we o bserve that parabens are associated with maternal inflammation , we will pursue an exploratory Aim 1b to assess whether associations of parabens with shorter gestation can be partly explained by elevated maternal inflammation . Given the known role of second trimester inflammation in predicting shorter gestation , we would expect that maternal inflammation will partly mediate associations of parabens with gestational length. Although we were underpowered to evaluate sex - specific associations of parabens with g estational length in our preliminary studies, we did observe a tendency for stronger associations in female infants (data not shown). Therefore, in Aims 1a and 1b, we may also consider sensitivity analyses to evaluate these associations stratified by fetal sex. E.3. Research design and methods for Specific Aim 2. The overall objective of Specific Aim 2 is to e valuate differences in associations of maternal paraben concentrations with shorter gestation by maternal inflammatory diet score. E.3.1. Outcome variable (s) : g est ation al length . In preliminary studies, gestational age at birth was calculated using the first day of a delivery date obt ained during a hospital study visit . For the proposed studies, birth date will be obtained from electronic medical records. 23 E.3.2. Exposure variable (s) : m aternal urinary paraben concentrations . Our approach for assessing maternal paraben concentrations is described in E.2.2. E.3.3. Effect modifying variable (s) : maternal inflammatory diet index . All w omen complete d a semi - quantitative food frequency questionnaire (FFQ) that was adapted from the full - length Block FFQ (NutritionQuest) to assess diet in pregnancy (57 - 59) . I - KIDS women completed the FFQ at 10 - 14 and 34 - 36 weeks gestation. Because the FFQ asks about food intake during the previou s three months, dietary information collected during these timepoints represents early and mid - to - late pregnancy maternal diet. The collected d ietary i n formation will be used to calculate the E - DII, which is an empirical score lth Study (NHS) that assesses overall inflammatory potential of whole diets (60) . The E - DII has been validated in different representative U.S. populations, including the NHS - II and the Health Professionals Follow - Up Study , and i s created using foods/food groups rather than individual nutrients, which makes this score a more relevant predictor of inflammatory dietary patterns than other published inflammatory diet scores (60 - 62) . Specifically, t he E - DII is calculated as the weighted sum of 9 anti - inflammatory (e.g. dark yellow vegetables, fruit juice) and 9 pro - inflammatory (e.g. processed meat, refined grains) foods/food groups shown to be predictive of 3 critical plasma - 6) (60, 62) . A ll dietary components necessary to calculate the E - DII are included in the Block FFQ completed by I - KIDS participants. For the proposed study, the E - DII will be calculated separately at 10 - 14 and 34 - 36 weeks gestation and averaged to represent maternal anti - inflammatory diet quality across pregnancy . 24 E.3.4. Covariates. Collecti on of maternal socio demographic and health - related characteristics is described in E.2.3 . E . 3.5 . Expected results and alternative strategies for Aim 2 . Because our preliminary findings showed that associations of maternal paraben concentrations with shorter gestation are attenuated in women with better diet quality , and because previous studies suggest that better diet quality is anti - inflammatory , we exp ect to observe differences in associations of parabe ns with shorter gestation by maternal E - DII . Although the E - DII has been validated in non - pregnant populations, no study (to our knowledge) has evaluated E - DII in pregnancy. Therefore, if we do not observe differences in associations of paraben s with gestational length by maternal E - DII, we may also consider other inflammatory dietary indices that have been validated in pregnancy (e.g. Dietary Inflammatory In dex (63) ) or use publishe d methods to create our own pregnancy - specific index (future directions) . While we plan to evaluate maternal E - DII across gestation ( average of 10 - 14 and 34 - 36 weeks ) , we will also consider maternal E - DII separately at the two timepoints because our prelim inary data suggests that early pregnancy diet may be more predictive of gestation length than late pregnancy diet . As with Aim 1, we may also perform sensitivity analyses to assess sex - specific associations, although we may be underpowered to detect such d ifferences in the current study. 25 F. PROPOSED ANALYSIS AND PRESENTATION OF TABLES/FIGURES . F.1. Statistical analysis . F.1.1. Overview. Our central hypothesis that paraben s ( ) are associated with shorter gestation ( Y ) due to elevated maternal inflammation ( ) and that an anti - inflammatory maternal diet ( W ) mitigates associations of with Y will be tested in an anticipated total of 482 women. The posited linear models are as follows: [ 1 ] and [ 2] for Aim 1 , as well as [ 3 ] for Aim 2 . and Y will be analy z ed as continuous measure s using multivariable linear regression models after considering transformations to mitigate skewness found in distributions . To ensure model assumptions are met, regression models will be checked for non - constant residual variance, influential points, and multicollinearity by standard methods. will be approximated from SG - adjusted urinary paraben levels and will be analyzed as continuous measures or in quantiles to evaluate non - linear relationships . A priori statistical considerations using directed acyclic grap h (DAG) will be used to identify potential confounders (see section F.1 .5 . for more details). All tests will be two - sided at type I error =0.05. F. 1. 2. P CA to characterize inflammat ion (Aim 1 a ). Maternal second trimester inflammation profiles will be created using PCA . PCA is a n unsupervised pattern identification method that linearly transforms a set of partially correlated variables into a joint set of uncorrelated factors (or components ) that are sorted in descending order based on the contribution to the total data variance (64, 65) . The 26 resultin g components representing concentrations of 21 measured cytokines will provide critical information about distinct inflammation patterns in I - KIDS women during the second trimester of pregnancy. C omponents generated from PCA will be used as continuous outc ome variables in generalized linear regression models to evaluate associations between maternal paraben s concentrations and cytokines patterns by testing the hypothesis [1] (64 - 66) . Additionally, c omponents may be used as continuous me diator variables in associations of maternal paraben concentrations with gestational length (exploratory Aim 1b) . F. 1. 3. M ediation by maternal inflammation (exploratory Aim 1b). We may examine whether associations of maternal paraben concentrations with shorter gestation can be partly explained by maternal inflammation using the VanderWeele method (67) . This regression - based approach assumes a flexible structural model for both Y and and relaxes the assumption that there is no interaction between and . This approach estimate s what proportion of th e total relationship between and Y is explained (mediated) by the natural indirect effect Y ( proportion mediated = ) . T his will be accomplished by testing the joint hypothesis by separately testing the hypotheses [1] and [2] . F. 1. 4. Modification by maternal diet (Aim 2 ) . We will evaluate the role of maternal anti - inflammatory diet (measured by the maternal E - DII) as an effect modifier of associations between maternal paraben concentrations and gestational length. E - DII will be dichotomized at the median as done in our preliminary 27 studies with AHEI - 2010 to maximize power and improve interpretation . We will test the hypothesis [3], which examine s the multipli cative interaction between maternal E - DII and paraben s . F. 1. 5. Covariates. Given that I - KIDS collects an extensive amount of data, there are numerous covariate s available to use for evaluating the associations proposed in Specific Aims 1 and 2. As mentioned in F.1.1., c ovariate selection for final statistical models will rely primarily on a priori considerations and previously published data , which will inform a DAG. Figure 5 present s a DAG that includes all covariates that are collected in I - KIDS that may be relevant for the proposed analyses. Statistical methods, including evaluating associations of potential covariates with parabens, inflammation, and gestation al length , as well as testing correlations between covariates that measure similar constructs, will be conducted to limit overadjustment of statistical models . 28 Figure 5. DAG with hypothesized associations of potential covariates with parabens, inflammation, and gestational length. The DAG includes a comprehensive list of covariates that are already available in I - KIDS that may be relevant confounders for evaluati ng associations of parabens with inflammation (Aim 1a), the mediation of associations between parabens and gestational length by inflammation (Aim 1b), and the modification of associations between parabens and gestational length by E - DII (Aim 2). F. 1.6 . Statistical power. We made our assessments of power on the basis that 482 women will be available to evaluate the relationships posited in linear models [1, 2, 3] having controlled for up to 4 covariates (68 - 70) . For these calculations we translated the model param eters to correlations that were derived from I - KIDS data or published in the literature (71) . Plausible correlation ranges for the proposed relationships were as follows: +0.13 to +0.15 for with and - 0.10 to - 0.12 for and with Y (23, 72) . Based on these values and sample size n =48 2 , we have 81 .2 % power for testing [1] to detect correlations of at least + 0.13 for the relationship (Aim 1a). Additionally, we have 29 over 97. 3 % power for testing [2] to detect correlations of at least - 0.10 for the relationship s Y and Y . As a result, we will have 7 9.1 % power for testing the joint hypothesis to detect the natural indirect effect Y and calculate the proportion mediated (Aim 1b) . We will also have over 80% power to for testing [3] for the relationship Y modified by W (Aim 2). Power calculations are summarized in Table 4 , where the shaded row represents the final estimated power for the proposed studies. Data are presented in the following order: columns 1 - 3, correlations between parabens, inflammation, and gestational length derived from the literature or calculated in preliminary analyses in I - KIDS; column 4, estimated percentage mediated based on correl ations in columns 1 - 3; columns 5 - 7, estimated power for testing hypotheses based on correlations in columns 1 - 3. X 1 , parabens; X 2 , inflammation; Y , gestational length. F. 2 . Presentation of tables and figures . F.2.1. Tables and figures related to both specific aims. Table 1 . Characteristics of I - KIDS pregnant women and those in analytic sample. This table will present characteristics of all I - KIDS pregnant women in the full 30 cohort compared to the 482 women who will be included in the final analytic sample. The participant characteristics presented in this table will include final covariates selected for statistical models evaluating associations of urinary paraben concentrations wi th maternal inflammation. We will use the chi - square goodness - of - fit test to determine if the final analytic sample is biased given the inclusion/exclusion criteria to select the final 482 women. Figure 1 . Urinary paraben concentrations in I - KIDS (n=482 ) . This figure will present urinary concentrations of butyl, ethyl, methyl, and propyl parabens . Additionally, this figure will include urinary paraben concentrations of same - age women from NHANES to compare whether urinary paraben concentrations in I - KID S are similar to those of a representative sample of U.S. women. Results from this figure will provide important information about the distribution of paraben concentrations in I - KIDS. Figure 2 . Correlations between urinary paraben concentrations (n=482) . This figure will present Pearson correlation coefficients between butyl, ethyl, methyl, and propyl parabens to determine how correlated parabens are with each other. Results from this figure will provide insights into which combinations of parabens women a re likely exposed to simultaneously. Table 2 . Urinary parabens and their associations with maternal sociodemographic and health - related characteristics (n=482) . 31 This table will present geometric mean (standard error) or median (95% CI) concentrations of butyl, ethyl, methyl, and propyl parabens by maternal sociodemographic and health - related characteristics. Parabens will be operationalized as continuous variables, while maternal sociodemographic and health - related characteristics will be operationalized as categorical variables. The non - parametric Kruskal Wallis test will be used to assess whether maternal characteristics are associated with paraben s . Results from this table will inform about potential confounders to include in statistical models evaluat ing associations of paraben concentrations with gestational length and maternal inflammation. Table 3 . Gestational length and its associations with maternal sociodemographic and health - related characteristics (n=482) . This table will present geometric me an (standard error) or median (95% CI) gestational age at birth by maternal sociodemographic and health - related characteristics. Gestational length will be operationalized as a continuous variable (in weeks or days) , while maternal sociodemographic and health - related characteristics will be operationalized as categorical variables. The non - parametric Kruskal Wallis test will be used to assess whether maternal characteristics are associated with gestational length . Re sults from this table will provide additional information about maternal characteristics that are associated with shorter or longer gestational length . These results will also inform about potential confounders to include in final statistical models evalua ting associations of paraben concentrations with gestational length . 32 Table 4 . Associations of urinary parabens with gestational length (n=482) . This table will present results from generalized linear regression models evaluating associations of four urin ary parabens with gestational length. The table will include unadjusted and adjusted - estimates (95 %CI ) that represent change (in days) in gestational length for every percent increase in paraben concentrations . Parabens will be operationalized as continu ous variables, as well as categorical variables (i.e. quantiles) to assess potential non - linear associations. These results will further provide evidence in the full analytic sample for overall associations of parabens with gestational length, which are th e basis of the proposed aims. F.2.2. Tables and figures related to Specific Aim 1. Figure 1 . Correlations between plasma cytokines (n=482) . This figure will present Pearson correlation coefficients between 21 cytokines that are components of critical in flammatory pathways. Results from this figure will provide insights about cytokines from common inflammatory pathways. These results will also provide preliminary evidence for PCA analyses. Figure 2 . Loading factors of plasma cytokines generated from PCA (n=482) . This figure will present several panels of bar graph s with PCA results. The figure will include loading factors for each plasma cytokine in each identified principal component to characterize patterns of cytokines . The PCA will be constrained to components that have eigenvalues >1.0 and together explain >80% of the variance. This figure will be important for interpreting results from statistical models 33 evaluating associations of paraben concentrations with maternal inflammation and the mediation of associations between parabens and gestational length by inflammation. Table 1 . Plasma cytokine principal components and their associations with maternal sociodemographic and health - related characteristics (n=482) . This table will present geometric mean (standard error) or median (95% CI) of each identified principal component by maternal sociodemographic and health - related characteristics. Identified cytokine principal components will be operationalized as continuous variables, while mate rnal sociodemographic and health - related characteristics will be operationalized as categorical variables. The non - parametric Kruskal Wallis test will be used to assess whether maternal characteristics are associated with cytokine components . Results from this table will provide important information about maternal characteristics that are associated with cytokines patterns. These results will also inform about potential confounders to include in final statistical models evaluating associations of paraben c oncentrations with maternal inflammation . Table 2 . Associations of urinary parabens with plasma cytokine principal components (n=482) . This table will present results from generalized linear regression models evaluating associations of four urinary para bens with several identified cytokine principal components. The table will include unadjusted and adjusted - estimates (95% CI ) 34 that represent change in principal component score for every percent increase in paraben concentration . Parabens will be operatio nalized as continuous variables, as well as categorical variables (i.e. quantiles) to assess potential non - linear associations. These results will determine whether parabens alter critical early second trimester maternal inflammatory profiles. Results from this table will also determine whether inflammation will be assessed as a mediator for associations between parabens and gestational length. Figure 3 . Associations of urinary parabens with gestational length mediated by plasma cytokine principal components (n=482) . Given findings from the previous table, t his figure will present associations of urinary parabens with gestational length mediated by inflammation. Specifically, this figure will include - estimates and 95%CI for the natural direct and indirect effects. The natural direct effect represents associations of paraben concentrations with gestational length . The natural indirect effect represents associations of paraben concentrations with gestational length explained by the associa tion between parabens and maternal inflammation. F.2. 3 . Tables and figures related to Specific Aim 2. Table 1 . E - DII and its associations with maternal sociodemographic and health - related characteristics (n=482) . This table will present geometric mean (standard error) or median (95% CI) of the E - DII by maternal sociodemographic and health - related characteristics. The E - DII 35 will be dichotomized at the median, while maternal sociodemographic and health - related characteristics will be operationalized as ca tegorical variables. A chi - square test will be used to assess whether maternal characteristics are associated with E - DII . Results from this table will identify the sub - samples of pregnant women that are more likely to consum e pro - vs. anti - inflammatory die ts. Figure 1 . Associations of E - DII with plasma cytokine principal components (n=482) . This figure will present associations of E - DII with plasma cytokine principal components. These results will determine whether the E - DII calculated in I - KIDS is associated with patterns of cytokines identified in PCA, and will validate whether the E - DII is correlated with gestational inflammation. Table 2 . Associations of urinary parabens with gestational length modified by the E - DII (n=482) . This table will present results from generalized linear regression models evaluating associations of four urinary paraben concentrations with gestational length modified by the E - DII . The table will include unadjusted and adjusted - estimates (95% CI ) that represent change in ges tational length (in days) for every percent increase in paraben concentration in women with E - DII scores median v s. < median . Parabens will be operationalized as continuous variables given power concerns. These results will determine whether a healthier, anti - inflammatory diet can intervene against the negative impacts of parabens on gestational length. 36 G. OVERALL CONCLUSIONS G.1. Strengths and f uture d irections . W e have foreseen some potential limitations /pitfalls and addressed several alternative strategies to our approach in sections E .2.4. and E .3.5 . T he proposed study has several strengths. One major strength is the ability to leverage biospecimens (maternal urine and blood ) and extensive covariate data already being collected for the ongoing I - KIDS parent study , thereby allowing for a highly efficient study design. While th is study is ambitious , all participants have been recruited and all prenatal data about participant characteristics are available to address our aims. Given that we currently work closely with the I - KIDS team, we do not expect major challenges for obtaining the remaining data needed to address these aims (e.g. electronic medical records) . While our results may not be generalizable to highly at - risk populations given the low pre - term birth prevalence in I - KIDS, these findings will provide novel information about risk factors for shorter gestation in a n understudied at - risk population. Once these aims are accomplished, findings will be published and utilized to develop new hypotheses about the adverse effects of paraben s in pregnancy and the intervening effect of an anti - inflammatory maternal diet on shorter gestation . No large - scale epidemiological studies have investigated the proposed relationships , which is a challenge , but also an opportunity to contribute new knowledge to the field. G.2. Scientific r igor. The scientific premise for this research is provided in earlier sections. Various measures will be taken to minimize b ias and to assure scientific rigor and reproducibility. First, our 37 participant selection relies on our ability to obtain the necessary measures for proposed analyses and not on any other factors, such as race / ethnicity. The selection of pregnant women year s 18 - 40 was done to minimize risk associated with pregnancies at younger or older ages and accomplish our scientific goals. Given that we will be evaluating overall associations and differences in associations by maternal diet quality , we will report overa ll and diet - specific relationships. While we may be underpowered to assess sex - specific associations in each aim, we will consider sex as a biological variable by performing sex - specific sensitivity analyses in each aim that will provide potentially import ant information for future studies. We have made substantial effort to develop protocols for reducing variability in paraben assessment (e.g. pooled sample, urine density adjustment) , as well as holistically assessing inflammation and diet quality (e.g. development of clusters or indices ). Importantly, we have adequate power to address our aims, thus enabling us to delve into potential biologic al pathways and dietary interventions of o ur hypothesized associations. Finally, we plan to report our results in a timely and transparent manner. 38 REFERENCES 39 REFERENCES 1. Centers for Disease Control and Prevention (CDC) . 10 Leading Causes of Death by Age Group, United States - 2017. WISQARS: National Center for Health Statistics, Centers of Disease Control and Prevention; 2017. 2. Markopoulou P, Papanikolaou E, Analytis A, Zoumakis E, Siahanidou T. Preterm Birth as a Risk Factor for Metabolic Syndrome and Cardiovascular Disease in Adult Life: A Systematic Review and Meta - Analysis. J Pediatr. 2019;210:69 - 80 e5. doi: 10.1016/j.jpeds.2019.02.041. PubMed PMID: 30992219. 3. Spong CY. Defining "term" pregnancy: recommendations from the Defining "Term " Pregnancy Workgroup. JAMA. 2013;309(23):2445 - 6. Epub 2013/05/07. doi: 10.1001/jama.2013.6235. PubMed PMID: 23645117. 4. Espel EV, Glynn LM, Sandman CA, Davis EP. Longer gestation among children born full term influences cognitive and motor development. P LoS One. 2014;9(11):e113758. Epub 2014/11/26. doi: 10.1371/journal.pone.0113758. PubMed PMID: 25423150; PMCID: PMC4244187. 5. Tita AT, Landon MB, Spong CY, Lai Y, Leveno KJ, Varner MW, Moawad AH, Caritis SN, Meis PJ, Wapner RJ, Sorokin Y, Miodovnik M, Carp enter M, Peaceman AM, O'Sullivan MJ, Sibai BM, Langer O, Thorp JM, Ramin SM, Mercer BM, Eunice Kennedy Shriver NM - FMUN. Timing of elective repeat cesarean delivery at term and neonatal outcomes. N Engl J Med. 2009;360(2):111 - 20. Epub 2009/01/09. doi: 10.10 56/NEJMoa0803267. PubMed PMID: 19129525; PMCID: PMC2811696. 6. Challis JR, Lockwood CJ, Myatt L, Norman JE, Strauss JF, 3rd, Petraglia F. Inflammation and pregnancy. Reprod Sci. 2009;16(2):206 - 15. Epub 2009/02/12. doi: 10.1177/1933719108329095. PubMed PMID : 19208789. 7. Mor G, Aldo P, Alvero AB. The unique immunological and microbial aspects of pregnancy. Nat Rev Immunol. 2017;17(8):469 - 82. doi: 10.1038/nri.2017.64. PubMed PMID: 28627518. 8. Vecchie A, Bonaventura A, Carbone F, Maggi D, Ferraiolo A, Carloni B, Andraghetti G, Affinito Bonabello L, Liberale L, Dallegri F, Montecucco F, Cordera R. C - Reactive Protein Levels at the Midpregnancy Can Predict Gestational Complications. Biomed Res Int. 2018;2018:1070151. doi: 10.1155/2018/1070151. PubMed PMID: 305334 23; PMCID: PMC6247639. 40 9. Ross KM, Baer RJ, Ryckman K, Feuer SK, Bandoli G, Chambers C, Flowers E, Liang L, Oltman S, Dunkel Schetter C, Jelliffe - Pawlowski L. Second trimester inflammatory and metabolic markers in women delivering preterm with and without preeclampsia. J Perinatol. 2019;39(2):314 - 20. doi: 10.1038/s41372 - 018 - 0275 - 8. PubMed PMID: 30518800; PMCID: PMC6760589. 10. National Biomonitoring Program . Parabens 2017 [cited 2019 09/04]. Available from: https://www.cdc.gov/biomonitoring/Parabens_BiomonitoringSummary.html . 11. Mortensen ME, Calafat AM, Ye X, Wong LY, Wright DJ, Pirkle JL, Merrill LS, Moye J. Urinary concentrations of environmental phenols in pregnant wome n in a pilot study of the National Children's Study. Environ Res. 2014;129:32 - 8. Epub 2014/02/18. doi: 10.1016/j.envres.2013.12.004. PubMed PMID: 24529000; PMCID: PMC4530794. 12. Ahn HJ, An BS, Jung EM, Yang H, Choi KC, Jeung EB. Parabens inhibit the early phase of folliculogenesis and steroidogenesis in the ovaries of neonatal rats. Mol Reprod Dev. 2012;79(9):626 - 36. Epub 2012/07/11. doi: 10.1002/mrd.22070. PubMed PMID: 22777679. 13. Aker AM, Watkins DJ, Johns LE, Ferguson KK, Soldin OP, Anzalota Del Toro LV, Alshawabkeh AN, Cordero JF, Meeker JD. Phenols and parabens in relation to reproductive and thyroid hormones in pregnant women. Environ Res. 2016;151:30 - 7. Epub 2016/10/21. doi: 10.1016/j.envres.2016.07.002. PubMed PMID: 27448730; PMCID: PMC5071140. 14. Kolatorova L, Vitku J, Hampl R, Adamcova K, Skodova T, Simkova M, Parizek A, Starka L, Duskova M. Exposure to bisphenols and parabens during pregnancy and relations to steroid changes. Environ Res. 2018;163:115 - 22. Epub 2018/02/13. doi: 10.1016/j.envre s.2018.01.031. PubMed PMID: 29433019. 15. Aker AM, Johns L, McElrath TF, Cantonwine DE, Mukherjee B, Meeker JD. Associations between maternal phenol and paraben urinary biomarkers and maternal hormones during pregnancy: A repeated measures study. Environ I nt. 2018. Epub 2018/01/26. doi: 10.1016/j.envint.2018.01.006. PubMed PMID: 29366524. 16. Lee JH, Lee M, Ahn C, Kang HY, Tran DN, Jeung EB. Parabens Accelerate Ovarian Dysfunction in a 4 - Vinylcyclohexene Diepoxide - Induced Ovarian Failure Model. Int J Enviro n Res Public Health. 2017;14(2). Epub 2017/02/18. doi: 10.3390/ijerph14020161. PubMed PMID: 28208728; PMCID: PMC5334715. 17. Aung MT, Ferguson KK, Cantonwine DE, McElrath TF, Meeker JD. Preterm birth in relation to the bisphenol A replacement, bisphenol S, and other phenols and parabens. Environ Res. 2019;169:131 - 8. Epub 2018/11/19. doi: 10.1016/j.envres.2018.10.037. PubMed PMID: 30448626; PMCID: PMC6347500. 41 18. Geer LA, Pycke BFG, Waxenbaum J, Sherer DM, Abulafia O, Halden RU. Association of birth outcomes with fetal exposure to parabens, triclosan and triclocarban in an immigrant population in Brooklyn, New York. J Hazard Mater. 2017;323(Pt A):177 - 83. Epub 2016/05/10. doi: 10.1016/j.jhazmat.2016.03.028. PubMed PMID: 27156397; PMCID: PMC5018415. 19. Messerl ian C, Mustieles V, Minguez - Alarcon L, Ford JB, Calafat AM, Souter I, Williams PL, Hauser R, Environment, Reproductive Health Study T. Preconception and prenatal urinary concentrations of phenols and birth size of singleton infants born to mothers and fath ers from the Environment and Reproductive Health (EARTH) study. Environ Int. 2018;114:60 - 8. Epub 2018/02/27. doi: 10.1016/j.envint.2018.02.017. PubMed PMID: 29477955; PMCID: PMC5899953. 20. Baker BH, Wu H, Laue HE, Boivin A, Gillet V, Langlois MF, Bellenge r JP, Baccarelli AA, Takser L. Methylparaben in meconium and risk of maternal thyroid dysfunction, adverse birth outcomes, and Attention - Deficit Hyperactivity Disorder (ADHD). Environ Int. 2020;139:105716. doi: 10.1016/j.envint.2020.105716. PubMed PMID: 32 283359; PMCID: PMC7275882. 21. Watkins DJ, Ferguson KK, Anzalota Del Toro LV, Alshawabkeh AN, Cordero JF, Meeker JD. Associations between urinary phenol and paraben concentrations and markers of oxidative stress and inflammation among pregnant women in Pue rto Rico. Int J Hyg Environ Health. 2015;218(2):212 - 9. Epub 2014/12/02. doi: 10.1016/j.ijheh.2014.11.001. PubMed PMID: 25435060; PMCID: PMC4323928. 22. Kelley AS, Banker M, Goodrich JM, Dolinoy DC, Burant C, Domino SE, Smith YR, Song PXK, Padmanabhan V. Ea rly pregnancy exposure to endocrine disrupting chemical mixtures are associated with inflammatory changes in maternal and neonatal circulation. Sci Rep. 2019;9(1):5422. doi: 10.1038/s41598 - 019 - 41134 - z. PubMed PMID: 30931951; PMCID: PMC6443771. 23. Aung MT, Ferguson KK, Cantonwine DE, Bakulski KM, Mukherjee B, Loch - Caruso R, McElrath TF, Meeker JD. Associations between maternal plasma measurements of inflammatory markers and urinary levels of phenols and parabens during pregnancy: A repeated measures study. Sci Total Environ. 2019;650(Pt 1):1131 - 40. Epub 2018/10/13. doi: 10.1016/j.scitotenv.2018.08.356. PubMed PMID: 30308801; PMCID: PMC6236678. 24. Bereketoglu C, Pradhan A. Comparative transcriptional analysis of methylparaben and propylparaben in zebrafish. Sci Total Environ. 2019;671:129 - 39. doi: 10.1016/j.scitotenv.2019.03.358. PubMed PMID: 30928742. 25. Bairati C, Goi G, Lombardo A, Tettamanti G. The esters of p - hydroxy - benzoate (parabens) inhibit the release of lysosomal enzymes by mitogen - stimulated peri pheral human lymphocytes in culture. Clin Chim Acta. 1994;224(2):147 - 57. doi: 10.1016/0009 - 8981(94)90181 - 3. PubMed PMID: 8004785. 42 26. Guzel Bayulken D, Ayaz Tuylu B. In vitro genotoxic and cytotoxic effects of some paraben esters on human peripheral lympho cytes. Drug Chem Toxicol. 2019;42(4):386 - 93. doi: 10.1080/01480545.2018.1457049. PubMed PMID: 29681198. 27. Chia AR, Chen LW, Lai JS, Wong CH, Neelakantan N, van Dam RM, Chong MF. Maternal Dietary Patterns and Birth Outcomes: A Systematic Review and Meta - A nalysis. Adv Nutr. 2019;10(4):685 - 95. Epub 2019/05/02. doi: 10.1093/advances/nmy123. PubMed PMID: 31041446; PMCID: PMC6628847. 28. Fung TT, McCullough ML, Newby PK, Manson JE, Meigs JB, Rifai N, Willett WC, Hu FB. Diet - quality scores and plasma concentrati ons of markers of inflammation and endothelial dysfunction. Am J Clin Nutr. 2005;82(1):163 - 73. doi: 10.1093/ajcn.82.1.163. PubMed PMID: 16002815. 29. Akbaraly TN, Shipley MJ, Ferrie JE, Virtanen M, Lowe G, Hamer M, Kivimaki M. Long - term adherence to health y dietary guidelines and chronic inflammation in the prospective Whitehall II study. Am J Med. 2015;128(2):152 - 60 e4. doi: 10.1016/j.amjmed.2014.10.002. PubMed PMID: 25305231; PMCID: PMC4315808. 30. Chrysohoou C, Panagiotakos DB, Pitsavos C, Das UN, Stefan adis C. Adherence to the Mediterranean diet attenuates inflammation and coagulation process in healthy adults: The ATTICA Study. J Am Coll Cardiol. 2004;44(1):152 - 8. doi: 10.1016/j.jacc.2004.03.039. PubMed PMID: 15234425. 31. Mena MP, Sacanella E, Vazquez - Agell M, Morales M, Fito M, Escoda R, Serrano - Martinez M, Salas - Salvado J, Benages N, Casas R, Lamuela - Raventos RM, Masanes F, Ros E, Estruch R. Inhibition of circulating immune cell activation: a molecular antiinflammatory effect of the Mediterranean diet . Am J Clin Nutr. 2009;89(1):248 - 56. doi: 10.3945/ajcn.2008.26094. PubMed PMID: 19056596. 32. Oken E, Radesky JS, Wright RO, Bellinger DC, Amarasiriwardena CJ, Kleinman KP, Hu H, Gillman MW. Maternal fish intake during pregnancy, blood mercury levels, and child cognition at age 3 years in a US cohort. Am J Epidemiol. 2008;167(10):1171 - 81. doi: 10.1093/aje/kwn034. PubMed PMID: 18353804; PMCID: PMC2590872. 33. Hertz - Picciotto I, Schramm M, Watt - Morse M, Chantala K, Anderson J, Osterloh J. Patterns and determi nants of blood lead during pregnancy. Am J Epidemiol. 2000;152(9):829 - 37. doi: 10.1093/aje/152.9.829. PubMed PMID: 11085394. 34. Gaskins AJ, Minguez - Alarcon L, Fong KC, Abu Awad Y, Di Q, Chavarro JE, Ford JB, Coull BA, Schwartz J, Kloog I, Attaman J, Hause r R, Laden F. Supplemental Folate and the Relationship Between Traffic - Related Air Pollution and Livebirth Among Women Undergoing Assisted Reproduction. Am J Epidemiol. 2019;188(9):1595 - 604. doi: 10.1093/aje/kwz151. PubMed PMID: 31241127; PMCID: PMC6736414 . 43 35. Romero R, Grivel JC, Tarca AL, Chaemsaithong P, Xu Z, Fitzgerald W, Hassan SS, Chaiworapongsa T, Margolis L. Evidence of perturbations of the cytokine network in preterm labor. Am J Obstet Gynecol. 2015;213(6):836 e1 - e18. doi: 10.1016/j.ajog.2015.07 .037. PubMed PMID: 26232508; PMCID: PMC4679519. 36. Jelliffe - Pawlowski LL, Rand L, Bedell B, Baer RJ, Oltman SP, Norton ME, Shaw GM, Stevenson DK, Murray JC, Ryckman KK. Prediction of preterm birth with and without preeclampsia using mid - pregnancy immune a nd growth - related molecular factors and maternal characteristics. J Perinatol. 2018;38(8):963 - 72. doi: 10.1038/s41372 - 018 - 0112 - 0. PubMed PMID: 29795450; PMCID: PMC6089890. 37. de Silva PS, Yang X, Korzenik JR, Goldman RH, Arheart KL, Caban - Martinez AJ. Ass ociation of urinary phenolic compounds, inflammatory bowel disease and chronic diarrheal symptoms: Evidence from the National Health and Nutrition Examination Survey. Environ Pollut. 2017;229:621 - 6. doi: 10.1016/j.envpol.2017.06.023. PubMed PMID: 28689150. 38. Nowak K, Jablonska E, Ratajczak - Wrona W. Immunomodulatory effects of synthetic endocrine disrupting chemicals on the development and functions of human immune cells. Environ Int. 2019;125:350 - 64. doi: 10.1016/j.envint.2019.01.078. PubMed PMID: 3074314 3. 39. Azizieh F, Dingle K, Raghupathy R, Johnson K, VanderPlas J, Ansari A. Multivariate analysis of cytokine profiles in pregnancy complications. Am J Reprod Immunol. 2018;79(3). doi: 10.1111/aji.12818. PubMed PMID: 29450942; PMCID: PMC5838769. 40. Matwi ejczuk N, Galicka A, Brzoska MM. Review of the safety of application of cosmetic products containing parabens. J Appl Toxicol. 2020;40(1):176 - 210. doi: 10.1002/jat.3917. PubMed PMID: 31903662. 41. Centers for Disease Control and Prevention (CDC). National Health and Nutrition Examination Survey Data. In: National Center for Health Statistics (NCHS), editor. Hyattsville, Maryland: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention; 2015 - 2016. 42. Centers for Disease Control and Prevention (CDC). National Health and Nutrition Examination Survey Data. In: National Center for Health Statistics (NCHS), editor. Hyattsville, Maryland: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention; 2013 - 2014. 43. Rodriguez - Bernal CL, Rebagliato M, Iniguez C, Vioque J, Navarrete - Munoz EM, Murcia M, Bolumar F, Marco A, Ballester F. Diet quality in early pregnancy and its effects on fetal growth outcomes: the Infancia y M edio Ambiente (Childhood and Environment) Mother and Child Cohort Study in Spain. Am J Clin Nutr. 2010;91(6):1659 - 66. doi: 10.3945/ajcn.2009.28866. PubMed PMID: 20410088. 44 44. Wirth MD, Hebert JR, Shivappa N, Hand GA, Hurley TG, Drenowatz C, McMahon D, Shoo k RP, Blair SN. Anti - inflammatory Dietary Inflammatory Index scores are associated with healthier scores on other dietary indices. Nutr Res. 2016;36(3):214 - 9. doi: 10.1016/j.nutres.2015.11.009. PubMed PMID: 26923507; PMCID: PMC4773655. 45. Kim Y, Chen J, W irth MD, Shivappa N, Hebert JR. Lower Dietary Inflammatory Index Scores Are Associated with Lower Glycemic Index Scores among College Students. Nutrients. 2018;10(2). doi: 10.3390/nu10020182. PubMed PMID: 29414858; PMCID: PMC5852758. 46. Sen S, Rifas - Shima n SL, Shivappa N, Wirth MD, Hebert JR, Gold DR, Gillman MW, Oken E. Dietary Inflammatory Potential during Pregnancy Is Associated with Lower Fetal Growth and Breastfeeding Failure: Results from Project Viva. J Nutr. 2016;146(4):728 - 36. doi: 10.3945/jn.115. 225581. PubMed PMID: 26936137; PMCID: PMC4807648. 47. Chiuve SE, Fung TT, Rimm EB, Hu FB, McCullough ML, Wang M, Stampfer MJ, Willett WC. Alternative dietary indices both strongly predict risk of chronic disease. J Nutr. 2012;142(6):1009 - 18. Epub 2012/04/2 0. doi: 10.3945/jn.111.157222. PubMed PMID: 22513989; PMCID: PMC3738221. 48. Gomes J, Au F, Basak A, Cakmak S, Vincent R, Kumarathasan P. Maternal blood biomarkers and adverse pregnancy outcomes: a systematic review and meta - analysis. Crit Rev Toxicol. 201 9;49(6):461 - 78. doi: 10.1080/10408444.2019.1629873. PubMed PMID: 31509042. 49. Murphy K. WC, Mowat A., Berg L., Chaplin D., Janeway C.A., Travers P., and Walport Gro up, LLC; 2017. 50. Tziakas DN, Chalikias GK, Kaski JC, Kekes A, Hatzinikolaou EI, Stakos DA, Tentes IK, Kortsaris AX, Hatseras DI. Inflammatory and anti - inflammatory variable clusters and risk prediction in acute coronary syndrome patients: a factor analys is approach. Atherosclerosis. 2007;193(1):196 - 203. doi: 10.1016/j.atherosclerosis.2006.06.016. PubMed PMID: 16857204. 51. Pollack AZ, Perkins NJ, Sjaarda L, Mumford SL, Kannan K, Philippat C, Wactawski - Wende J, Schisterman EF. Variability and exposure clas sification of urinary phenol and paraben metabolite concentrations in reproductive - aged women. Environ Res. 2016;151:513 - 20. Epub 2016/10/21. doi: 10.1016/j.envres.2016.08.016. PubMed PMID: 27567355; PMCID: PMC5071150. 52. Brookmeyer R. Analysis of multist age pooling studies of biological specimens for estimating disease incidence and prevalence. Biometrics. 1999;55(2):608 - 12. Epub 2001/04/25. PubMed PMID: 11318222. 45 53. Saha - Chaudhuri P, Weinberg CR. Specimen pooling for efficient use of biospecimens in stu dies of time to a common event. Am J Epidemiol. 2013;178(1):126 - 35. Epub 2013/07/04. doi: 10.1093/aje/kws442. PubMed PMID: 23821316; PMCID: PMC3698992. 54. Calafat AM, Koch HM, Swan SH, Hauser R, Goldman LR, Lanphear BP, Longnecker MP, Rudel RA, Teitelbaum SL, Whyatt RM, Wolff MS. Misuse of blood serum to assess exposure to bisphenol A and phthalates. Breast Cancer Res. 2013;15(5):403. Epub 2013/10/03. doi: 10.1186/bcr3494. PubMed PMID: 24083327; PMCID: PMC3978629. 55. Silva MJ, Samandar E, Preau JL, Jr., R eidy JA, Needham LL, Calafat AM. Quantification of 22 phthalate metabolites in human urine. J Chromatogr B Analyt Technol Biomed Life Sci. 2007;860(1):106 - 12. doi: 10.1016/j.jchromb.2007.10.023. PubMed PMID: 17997365. 56. Meeker JD, Hu H, Cantonwine DE, La madrid - Figueroa H, Calafat AM, Ettinger AS, Hernandez - Avila M, Loch - Caruso R, Tellez - Rojo MM. Urinary phthalate metabolites in relation to preterm birth in Mexico city. Environ Health Perspect. 2009;117(10):1587 - 92. Epub 2009/12/19. doi: 10.1289/ehp.080052 2. PubMed PMID: 20019910; PMCID: PMC2790514. 57. Boucher B, Cotterchio M, Kreiger N, Nadalin V, Block T, Block G. Validity and reliability of the Block98 food - frequency questionnaire in a sample of Canadian women. Public Health Nutr. 2006;9(1):84 - 93. doi: 10.1079/phn2005763. PubMed PMID: 16480538. 58. Subar AF, Thompson FE, Kipnis V, Midthune D, Hurwitz P, McNutt S, McIntosh A, Rosenfeld S. Comparative validation of the Block, Willett, and National Cancer Institute food frequency questionnaires : the Eating at America's Table Study. Am J Epidemiol. 2001;154(12):1089 - 99. doi: 10.1093/aje/154.12.1089. PubMed PMID: 11744511. 59. Donahue SM, Rifas - Shiman SL, Gold DR, Jouni ZE, Gillman MW, Oken E. Prenatal fatty acid status and child adiposity at age 3 y: results from a US pregnancy cohort. Am J Clin Nutr. 2011;93(4):780 - 8. doi: 10.3945/ajcn.110.005801. PubMed PMID: 21310834; PMCID: PMC3057547. 60. Tabung FK, Smith - Warner SA, Chavarro JE, Wu K, Fuchs CS, Hu FB, Chan AT, Willett WC, Giovannucci EL. Development and Validation of an Empirical Dietary Inflammatory Index. J Nutr. 2016;146(8):1560 - 70. doi: 10.3945/jn.115.228718. PubMed PMID: 27358416; PMCID: PMC4958288. 61. Tabung FK, Giovannucci EL, Giulianini F, Liang L, Chandler PD, Balasubramanian R, Manson JE, Cespe des Feliciano EM, Hayden KM, Van Horn L, Rexrode KM. An Empirical Dietary Inflammatory Pattern Score Is Associated with Circulating Inflammatory Biomarkers in a Multi - Ethnic Population of Postmenopausal Women in the United States. J Nutr. 2018;148(5):771 - 8 0. doi: 10.1093/jn/nxy031. PubMed PMID: 29897561; PMCID: PMC5972616. 46 62. Tabung FK, Smith - Warner SA, Chavarro JE, Fung TT, Hu FB, Willett WC, Giovannucci EL. An Empirical Dietary Inflammatory Pattern Score Enhances Prediction of Circulating Inflammatory Bi omarkers in Adults. J Nutr. 2017;147(8):1567 - 77. doi: 10.3945/jn.117.248377. PubMed PMID: 28659407; PMCID: PMC5525108. 63. Shivappa N, Steck SE, Hurley TG, Hussey JR, Hebert JR. Designing and developing a literature - derived, population - based dietary inflam matory index. Public Health Nutr. 2014;17(8):1689 - 96. Epub 2013/08/15. doi: 10.1017/S1368980013002115. PubMed PMID: 23941862; PMCID: PMC3925198. 64. Rao CR. The Use and Interpretation of Principal Component Analysis in Applied Research. Sankhya, Series A. 1964;26:329 - 58. 65. Kshirsagar AM. Multivariate Analysis. New York: Marcel Dekker; 1972. 66. Hotelling H. Analysis of a Complex of Statistical Variables into Principal Components. Journal of Educational Psychology. 1933;24:417 - 41, 98 - 520. 67. VanderWeele T J, Robins JM. Signed directed acyclic graphs for causal inference. J R Stat Soc Series B Stat Methodol. 2010;72(1):111 - 27. Epub 2010/01/01. doi: 10.1111/j.1467 - 9868.2009.00728.x. PubMed PMID: 25419168; PMCID: PMC4239133. 68. Baron RM, Kenny DA. THE MODERAT OR MEDIATOR VARIABLE DISTINCTION IN SOCIAL PSYCHOLOGICAL - RESEARCH - CONCEPTUAL, STRATEGIC, AND STATISTICAL CONSIDERATIONS. J Pers Soc Psychol. 1986;51(6):1173 - 82. doi: 10.1037/0022 - 3514.51.6.1173. PubMed PMID: WOS:A1986F285400010. 69. Pearle J. Causality, Models, Reasoning, and Inference, 2nd Edition. New York, NY: Cambridge University Press; 2009. 70. VanderWeele TJ. Explanation in Causal Inference: Methods for Mediation and Interaction. New York: Oxford University Press; 2015. 71. Maxwell SE. Sample size and multiple regression analysis. Psychological Methods. 2000;5(4):434 - 58. doi: 10.1037//1082 - 989x.5.4.434. PubMed PMID: WOS:000166233700004. 72. Kim H, Hwang JY, Ha EH, Park H, Ha M, Lee SJ, Hong YC, Chang N. Association of maternal folate nutrition and serum C - reactive protein concentrations with gestational age at delivery. Eur J Clin Nutr. 2011;65(3):350 - 6. doi: 10.1038/ejcn.2010.267. PubMed PMID: 21179048.