{"paper_id":"6f2c2715-2523-4606-9380-67049048793a","body_text":"Endometriosis is a common gynecologic disorder often associated with severe pelvic pain and infertility, affecting one in ten reproductive aged women throughout her life [ 1 ]. Endometriosis is thought to result from retrograde menstruation implanting endometriosis tissue on the peritoneal surfaces, but its etiology is not fully understood [ 2 ]. While studies have investigated risk factor profiles for endometriosis, there are very few known modifiable risk factors [ 3 , 4 ]. In addition, despite the robust evidence of symptom onset during adolescence and young adulthood for most adult women with endometriosis [ 5 , 6 ], few studies exist focusing on adolescents and young adults [ 7 , 8 ].\nAnimal and human studies demonstrate that  in utero  exposures affect reprogramming of the fetus which may result in higher risk of certain diseases later in life [ 9 ]. Early life exposures, such as mode of delivery, parental smoking, and exposure to breastfeeding, may modify the hormonal and/or immune status of the newborn, which could influence disease risk later in life [ 10 – 12 ]. Unlike all other impactful chronic diseases, there is minimal research on early-life exposures in relation to endometriosis risk with inconsistent findings [ 13 – 20 ]. Specifically, maternal exposure to diethylstilbestrol, smoking during pregnancy, and lower birthweight have been reported to be associated with greater risk of endometriosis diagnosed in adulthood in many but not all studies [ 4 ]. However, associations between early life factors and endometriosis have not been investigated within the adolescent and young adult population.\nHere, we investigated the association between  in utero  and early life exposures in relation to surgically diagnosed endometriosis in adolescents and young adults, defined as those aged <25 years old using the WHO definition [ 21 ]. We hypothesized that  in utero  and early life exposures contribute to the development of adolescent endometriosis.\n\nThe Women’s Health Study: From Adolescence to Adulthood (A2A), is a study of adolescents and women oversampled for those surgically diagnosed with endometriosis, enrolling participants from 2012 to 2018 [ 5 , 22 ]. Briefly, endometriosis cases were enrolled from Boston Children’s Hospital (BCH) and Brigham and Women’s Hospital (BWH) and were eligible if they were 1) female; 2) aged 7–55 years; and 3) had a surgical diagnosis of endometriosis. Controls at enrollment were selected through a combination of approaches that sampled the underlying population that gave rise to the cases [ 23 ]. Specifically, controls were individuals without a surgical diagnosis of endometriosis identified through clinics at BCH and BWH or recruited from the local Boston community served by these two hospitals through local advertisement, online postings, and/or word of mouth. Participants completed an extensive baseline questionnaire expanded from the World Endometriosis Research Foundation Endometriosis Phenome and Biobanking Harmonization Project (WERF EPHect) standard clinical questionnaire [ 24 ], providing information on demographic, lifestyle, and reproductive characteristics as well as clinical characteristics for endometriosis cases abstracted from medical records using the WERF EPHect standard surgical form [ 25 ]. Endometriosis staging was based on the revised American Society of Reproductive Medicine (rASRM) four-stage system [ 26 , 27 ]. This study was approved by the Institutional Review Board of the BCH and BWH (P00004267). All participants provided written assent or consent for study participation, with parental consent plus participant assent for girls < 18 years.\nAll study variables were based on self-reported data obtained from the baseline questionnaire. Participants were asked about their exposures  in utero  and during their early life.  In utero  exposures included whether the participants’ parents received fertility assistance at time of conception of the participant, mother’s age at delivery, whether their parents smoked during pregnancy, mode of delivery, birth weight, whether the participant was preterm (< 37 weeks gestational age), and whether the participant was one of a multiple gestation birth. Early life exposures included whether the participant was breastfed, whether their parents smoked while the participant was growing up (until age 13), and age at menarche.\nThe diagnosis of endometriosis in our adolescent and young adult population was based on surgical visualization, as supported by the American College of Obstetricians and Gynecologists [ 28 ]. Endometriosis in adolescents and young adults differs from adult diagnosed endometriosis. In adolescents, endometriosis lesions are more often clear or red, and biopsy of lesions suspicious of endometriosis do not always confirm endometriosis pathologically by the primary three histologic criteria established for blue/black “powder burn” lesion [ 29 ], and therefore visual diagnosis is suggested to be sufficient in this population.\nCovariate data collected in the baseline questionnaire included age at enrollment (years, continuous), body mass index (BMI) calculated from self-reported height and weight at enrollment (kg/m 2 ), race (dichotomous; white, non-white), age at menarche (years, continuous), and mother’s history of endometriosis (categorical; no, yes, unknown). For women aged ≥ 20, current BMI was categorized according to the World Health Organization criteria (underweight (BMI < 18.5 kg/m 2 ), normal weight (18.5–24.9 kg/m 2 ), overweight (BMI 25–29.9 kg/m 2 ), or obese (BMI ≥ 30 kg/m 2 )) [ 30 ]. For participants aged < 20, current BMI was categorized using age and female sex-specific BMI Z-scores (underweight (Z-score ≤ −2), normal weight (−2 < Z-score <1), overweight (Z-score 1–2), or obese (Z-score > 2) [ 31 ].\nFrom among the 1,017 participants enrolled in the A2A longitudinal study with data on early life exposures, per the aims of this analysis, we excluded those who were age 25 years or greater (n=373) at enrollment. Further to maximize the accuracy of our endometriosis case population, we additionally excluded those with unconfirmed endometriosis status (n=10), those who self-reported endometriosis but had no laparoscopic confirmation (n=15), and those who were defined as controls at enrollment but were subsequently diagnosed with endometriosis during follow-up (all within 2 years of enrollment) (n=4), and endometriosis cases missing age at diagnosis (n=11). In sum, our analysis included a total of 604 participants: 295 surgically confirmed endometriosis cases and 309 population-based controls.\nLogistic regression adjusted for age at enrollment (crude model) was used to calculate the odds ratios (OR) and 95% confidence intervals (CI) of endometriosis diagnosis for  in utero  and early life exposures. We conducted the main analyses restricting to cases with complete information on the exposures on interest (i.e. in utero and early life exposures) [ 32 ]. We also adjusted for  a priori  hypothesized potential confounders, selected based on prior literature or known associations with both the exposure and outcome [ 33 ], in the multivariable model including race/ethnicity, mother’s history of endometriosis diagnosis, and age at menarche. We additionally adjusted for birthweight when evaluating the association between preterm birth and endometriosis. Analyses for birthweight were restricted to full term births. We applied the missing indicator method [ 34 – 36 ] to account for and quantify the impact of missing covariate data, which included missing data on maternal history of endometriosis (n=57) and age at menarche (n=16).\nWe evaluated heterogeneity in the association between early life exposures and odds of endometriosis by mother’s history of endometriosis in some analyses (i.e. parent use of fertility treatment at time of participants’ conception and breastfeeding of participants) stratified by mother’s history of endometriosis and evaluated interactions by calculating likelihood ratio statistics comparing models with and without the cross-product terms. All statistical analyses were performed using SAS (version 9.4, SAS Institute, Cary, NC).\n\nA total of 604 adolescent and young adult participants aged < 25 (295 surgically confirmed endometriosis cases and 309 population-based controls) were included in the analytic population ( Table 1 ). Median age at enrollment was 17 years (range 12–24) in cases and 22 years (range 7–24) in controls, with 83% and 68% non-Hispanic whites, respectively. Median age at menarche was 12 years (range 8–15 and 9–15, respectively) for both groups. Among cases, 50% had mothers with endometriosis while only 9% of the controls did. Almost all cases (95%) were rASRM stage I or II at diagnostic surgery. The majority of the endometriosis cases (99%) had superficial peritoneal endometriosis only.\nUse of fertility treatment at time of conception or mother’s age at delivery was not associated with odds of endometriosis diagnosis in adolescents and young adults ( Table 2 ). Participants’ mothers who had been diagnosed with endometriosis were more likely to have conceived via fertility treatment compared to participants’ mothers without endometriosis (17% vs 6%), however, we did not observe a statistically significant interaction by maternal history of endometriosis. We were not able to examine the association by type of fertility treatment due to limited sample size.\nNeither exposure to smoking  in utero  due to maternal or paternal smoking was associated with odds of endometriosis, although  in utero  exposure due to maternal smoking association was based on a small number exposed (n=12). We did not observe any significant associations by different combinations of parental smoking status and odds of endometriosis when compared to those without  in utero  exposure to parental smoking (i.e. both mother and father of the participants were non-smokers when they were  in utero ), although power was limited by small sample size ( Supplemental Table 1 ). We did not observe significant associations by mode of delivery, preterm birth, and low birthweight ( Table 2 ), although only 16 participants among full term deliveries were low birthweight.\nParticipants who were breastfed had significantly lower odds of endometriosis compared to those not breastfed (OR=0.39, 95% CI=0.21–0.74)( Table 3 ). Since maternal smoking can be associated with the choice or ability to breastfeed or not [ 37 ], we additionally adjusted for smoking status of the participants’ mother during pregnancy and observed nearly identical results (OR=0.40, 95%CI=0.21–0.76). A higher proportion of participants were breastfed among those whose mother did not have a history of endometriosis (82% vs 72%), with the same protective pattern of breastfeeding within this group (OR=0.31, 95% CI=0.16–0.63). When restricted to women whose mothers had a history of endometriosis, the relation with breastfeeding attenuated and confidence intervals were much wider (OR=0.98, 95% CI=0.25–3.75).\nExposure to secondhand smoke due to maternal smoking during childhood was associated with more than two-fold increase in odds of endometriosis (OR=2.70, 95% CI=1.11–6.60), whereas exposure to secondhand smoke due to paternal smoking during childhood was not associated (OR=1.43, 95% CI=0.65–3.11) ( Table 3 ). When we further investigated by different combinations of parental smoking status, we did not observe any informative patterns ( Supplemental Table 1 ). We observed a four-fold, borderline significant, greater odds of endometriosis among those exposed to secondhand smoke due to both parents smoking during childhood compared to those not exposed to secondhand smoke during childhood (OR=3.48, 95% CI=0.96–12.62), although the confidence interval was wide ( Supplemental Table 1 ). As anticipated given the large, robust body of literature [ 4 , 38 ], age at menarche was inversely associated with odds of endometriosis diagnosed in adolescents and young adults ( Table 3 ).\n\nIn this investigation of  in utero  and early life factors in relation to endometriosis among adolescents and young adults aged less than 25, we observed that exposure to breastfeeding in early life was associated with lower odds of surgically diagnosed endometriosis and exposure to secondhand smoke due to maternal smoking during childhood was associated with greater odds of endometriosis.\nWe observed that early life exposure to breastfeeding was significantly associated with lower odds of endometriosis diagnosed in adolescents and young adults. Two prior studies, one prospective cohort study and one case-control study with small sample size using controls with other gynecologic diseases needing surgery, reported null associations between early life exposure to breastfeeding and endometriosis [ 16 , 17 ]. Two other studies reported regular formula feeding was associated with about 2-fold greater odds of endometriosis [ 18 , 19 ], which were both case-control studies with moderate sample size. Our results were in line with some of the prior studies despite the fact that all prior studies focused primarily on adult endometriosis cases whereas the current investigation focused on adolescent endometriosis. Moreover, our observation is biologically plausible. Breastmilk includes immunologic proteins such as IgA and immune cells [ 39 ] and exposure to breastmilk has been reported to enhance the immune system in the neonate [ 40 , 41 ]. Given the emerging evidence on immune dysfunction and endometriosis [ 42 , 43 ], it is possible that exposure to breastmilk early in life may positively influence the immune system, leading to reduced risk of endometriosis development later in life.\nExposure to secondhand smoke during childhood from maternal smoking was associated with greater odds of endometriosis, which was consistent with a prior study that reported childhood exposures to secondhand smoke being associated with 30% increased odds of endometriosis [ 44 ]. Although the association between exposure to smoking  in utero  and odds of endometriosis was not statistically significant in the current study, the point estimate suggested a positive association which was in line with some of the prior studies reporting the associations between  in utero  exposure to smoking and endometriosis risk [ 14 , 17 – 20 ]. Interestingly, we observed a high correlation between being exposed to maternal smoking  in utero  and being exposed to secondhand smoke during childhood due to maternal smoking (Spearman correlation coefficient=0.81). Given this high correlation, it is difficult to disentangle and conclude when the critical time window of smoking exposure is for endometriosis development. There were very few current (n=6) and former (n=6) smokers among the participants. Of the 12 participants who reported ever smoking, there was only one participant whose mother smoked during childhood and who themselves were a former smoker. Therefore, we were not able to assess correlation between parental smoking and participants’ smoking status in our data.\nWe did not observe a statistically significant association between prematurity or low birthweight and endometriosis in adolescents and young adults, although these factors have been observed to be associated with greater risk of adult endometriosis in previous studies [ 13 , 16 , 19 , 20 ]. While our full study sample size was amongst the larger, with over 600 participants, this population had low numbers with adverse early life exposures, which impact the power to yield statistically significant associations. Further investigation is warranted to understand the association between prematurity, birthweight, and risk of endometriosis in adolescents and young women.\nDespite that two-thirds of women diagnosed with endometriosis during adulthood report their symptoms emerged during their adolescence [ 38 ], research on endometriosis risk factors among adolescents and young adults have been very limited [ 7 , 8 ]. The strength of our study is restriction to girls and women younger than age 25 among whom endometriosis cases were strictly defined by surgical diagnosis and controls were enrolled from among the population catchment defined by these case participants with data collected from all using highly validated and internationally rigorously developed tools. Some prior studies comparing endometriosis cases to surgical controls may erroneously yield null results if the surgery indicating conditions among the control group, such as uterine fibroids or tubal ligation, have any of the same risk factors as endometriosis [ 45 ]. A strength of our design is that we included controls sampled from the underlying population that gave rise to the endometriosis cases [ 23 ].\nHowever, there are several limitations. Due to the relatively small number for some of the  in utero  and early life exposures, for those we had limited power to detect true statistically significant associations or interactions if they exist. The case-control study design may have influenced cases to have recalled exposures differently than controls leading to an over- or underestimate of the association. There is also the potential for endometriosis misclassification among controls, with possible asymptomatic or yet undiagnosed disease being included in the control group given the current diagnostic standard of surgical visualization. However, this is true for all clinical studies of endometriosis to date, leading to non-differential misclassification that would bias our results towards the null. Moreover, the likely prevalence of severe/symptomatic endometriosis in the community is less than 2% [ 46 ], and the characteristics of this small proportion of undiagnosed cases would be diluted among the true endometriosis-free comparison women. We also acknowledge that there is potential for exposure misclassification that is not different between cases and controls, thus driving associations toward the null. However, we used the same validated questions that were used in most of the previous studies, reporting high validity of self-reported early life exposures [ 16 , 18 ]. While unmeasured confounding may be possible, we anticipate this to be minimal since we have adjusted for potential confounders that have been considered in prior literature. More informatively, especially in our two statistically significant findings of exposure to breastfeeding and secondhand smoking, the age-adjusted and multivariable adjusted effect estimates were almost entirely unchanged, suggesting robust results. Our study results may be limited in generalizability since this study was based in a specific region of the United States and among those with high access to advanced medical care [ 47 ].\n\nIn summary, this novel examination conducted among those younger than age 25 suggest early life exposure to breastfeeding was associated with lower risk and exposure to secondhand smoke during childhood due to maternal smoking was associated with higher risk of endometriosis diagnosed in adolescence and early adulthood. These findings provide insight into additional potential etiologic pathways of endometriosis to be explored in this young population.","source_license":"CC0","license_restricted":false}