Measures of fetal growth and preterm birth and risk of endometriosis and adenomyosis in adult life: a systematic review and meta-analysis

All authors have contributed to the conception and design of the study. MB conducted the first draft. All authors have edited the draft, given their final approval of the version to be published and are accountable for all aspects of the work.

Figure  1 depicts the flow of the study selection process. In total, 843 studies were found in the initial search. After screening the studies by title and abstract, 14 were eligible for full‐text review; of these, 10 studies were ultimately included. In the search before submission one additional study was included. 43 Thus, a total of 11 studies were included in the systematic review; five cohort studies, 31 , 43 , 44 , 45 , 46 one matched cohort study 47 and five case–control studies. 18 , 32 , 33 , 34 , 40 All the studies investigated birthweight and eight of the studies also investigated preterm birth. Only two studies investigated children born small for gestational age. 31 , 43 None of the studies investigated other indices of fetal growth. The main characteristics and results of the included studies are presented in Tables  1 and 2 . The included studies received NOS scores ranging from three to seven. Preferred Reporting Items for Systematic Reviews and Meta‐analysis (PRISMA) flowchart depicting the study selection process. Characteristics of the 11 included studies investigating the association between gestational age at birth and birthweight and the risk of endometriosis or adenomyosis in adult life. Abbreviations: BMI, body mass index; BW, bodyweight; GA, gestational age; MRI, magnetic resonance imaging; n , total number. The study used a matched‐exposure cohort design. The exposure was considered to be a diagnostic laparoscopy or laparotomy regardless of indication. The exposure cohort (surgery cohort) was matched to a population by age and residence. The unexposed cohort had an MRI scan performed. Main results of the 11 included studies investigating the association between gestational age at birth and birthweight and the risk of endometriosis or adenomyosis in adult life. Abbreviations: Adeno., adenomyosis; aOR, adjusted odds ratio; BMI, body mass index; BW, birthweight; CI, confidence interval; cOR, crude odds ratio; DIE, deep infiltrating endometriosis; Endo, endometriosis; g, gram; GA, gestational age; HR, hazard ratio; kg, kilograms; LBW, low birthweight; OMA, ovarian endometriosis; OR, odds ratio; PTB, Index woman born preterm; ref, Reference; SES, socioeconomic status; SD, standard deviation; SGA, small for gestational age; SUP, superficial peritoneal endometriosis. All 11 included studies explored the association between birthweight and the risk of endometriosis in adulthood. The Swedish cohort study by Gao et al. 45 including 91 women with endometriosis (2.7% of the population) found a higher risk of subsequent endometriosis for every standard deviation decrease in standardized birthweight. In 2020, Gao et al. 31 published another cohort study including 8262 women with endometriosis (1.3% of the population) that further supported their previous findings. In 2004, Missmer et al. 46 conducted a cohort study including 1126 women with endometriosis (1.3% of the population) and found that low birthweight was associated with a higher risk of endometriosis. Likewise, a French case–control study by Borghese et al. 18 including 368 women with endometriosis, found that low birthweight was associated with a higher risk of endometriosis. Furthermore, they found that the risk of endometriosis decreased per 100 g increase in birthweight. When studying subtypes of endometriosis, they found a higher risk of deep infiltrating endometriosis and a tendency towards a higher risk of superficial peritoneal endometriosis and ovarian endometriosis among women born with low birthweight. In line with these findings, Vannuccini et al. 34 performed a case–control study including 161 women with endometriosis and found that low birthweight was associated with a higher risk of subsequent endometriosis. In contradiction to these findings, a large cohort study by Aarestrup et al. 44 including 1780 women with endometriosis (1.4% of the population), found no increased risk of endometriosis when comparing women with a birthweight between 2.0 and 3.25 kg with women with a birthweight between 3.26 and 3.75 kg. Similar to this finding, Rossi et al. 43 found no association between birthweight (adjusted for gestational age at birth) and endometriosis in adulthood in their cohort study including 348 women with endometriosis (9.1% of the population). Likewise, a Chinese case–control study by Liu et al. 40 including 440 women with endometriosis and an American case–control study by Upson et al. 33 including 310 women with endometriosis, found no association between low birthweight and endometriosis. An Italian case–control study by Somigliana et al. 32 including 91 women with endometriosis, showed no association between a birthweight below 3 kg and the risk of endometriosis. Finally, an American matched cohort study by Wolff et al. 47 including 204 women with endometriosis, found no association. In the meta‐analysis, six studies 18 , 31 , 33 , 34 , 40 , 46 investigating birthweight <2.5 kg were included. The meta‐analysis showed a higher risk of endometriosis in women born with a low birthweight with a pooled OR of 1.26 (95% CI 1.05–1.52). Heterogeneity was moderate ( I 2  = 44%) (Figure  2 ). The information gained from the funnel plot (Figure  S1 ) was sparse and publication bias could not be ruled out. Forest plot for the association between low birthweight and endometriosis. We included two cohort studies on birthweight and adenomyosis. The study by Gao et al. 45 including 24 women with adenomyosis (0.7% of the population), found no association between birthweight‐for‐gestational age (per standard deviation decrease) and adenomyosis. Similar to this finding, the large cohort study by Aarestrup et al. 44 including 1097 women with adenomyosis (0.8% of the population), found no association between birthweight per kg and the risk of subsequent adenomyosis. There were too few studies to perform a meta‐analysis. Eight of the included studies 31 , 32 , 33 , 34 , 40 , 43 , 46 , 47 investigated the association between preterm birth and endometriosis later in life, yet with different definitions of preterm birth. The case–control study by Vannuccini et al. 34 found a higher risk of endometriosis when being born preterm. Liu et al. 40 Somigliana et al. 32 and Upson et al. 33 found a tendency towards a higher risk of endometriosis in women born preterm. Rossi et al. 43 Gao et al. 31 and Missmer et al. 46 found no association between preterm birth and endometriosis. Likewise, Gao et al. 45 found no association between gestational age at birth (by week of gestation) and endometriosis. Wolff et al. 47 found no association between preterm birth and endometriosis when studying the surgery cohort overall, or when subdividing the cohort and only investigating women with endometriosis staged 3 to 4. However, when investigating endometriosis confirmed by histology in the surgery cohort, they found a lower risk of endometriosis in women born preterm. In the meta‐analysis, eight studies 31 , 32 , 33 , 34 , 40 , 43 , 46 , 47 investigating preterm birth were included. The meta‐analysis showed a higher risk of endometriosis among women born preterm with a pooled OR of 1.32 (95% CI 1.01–1.72). Heterogeneity was substantial ( I 2  = 72%) (Figure  3 ). Based on a funnel plot (Figure  S2 ), publication bias could not be ruled out. The study by Vannuccini et al. 34 included no definition of preterm birth; therefore to check the robustness of the findings from the meta‐analysis, we performed a sub‐analysis excluding the study. We found a slightly attenuated pooled OR of 1.14 (95% CI 0.94–1.38) (Figures  S3 and S4 ). Forest plot for the association between preterm birth and endometriosis. The cohort study by Gao et al. 45 was the only study investigating gestational age and the risk of adenomyosis. They found that duration of gestation was not associated with a higher risk of subsequent adenomyosis (for every additional week of gestation).

This systematic review involved studies with diverging results on the association between birthweight and gestational age and endometriosis and adenomyosis in adulthood. The meta‐analysis showed a 26% higher risk of endometriosis in women born with a low birthweight. Only two studies 44 , 45 assessed the association between birthweight and adenomyosis; no overall association was found. Further, the meta‐analysis on the association between preterm birth and endometriosis showed a 32% higher risk of endometriosis in women born preterm. The only study on preterm birth and adenomyosis found no association. These associations support the hypothesis of a developmental origin of endometriosis. However, based on the existing evidence, we cannot draw conclusions of causation between preterm birth or low birthweight and development of endometriosis based on this study. This systematic review and meta‐analysis has some important strengths. The study represents a comprehensive systematic review and meta‐analyses of all published studies on the association between fetal growth measures and preterm birth and endometriosis. To our knowledge, it is the first systematic review to focus on measures of fetal growth and preterm birth in relation to adenomyosis. A limitation of the meta‐analyses is the internal validity of the included studies. The quality of the studies was diverging with respect to NOS scores (two studies of low quality, eight of moderate and only one study of high quality). In the meta‐analyses, all studies were included regardless of scores. If more studies were available, a meta‐analysis restricted to high‐quality studies could be performed, potentially resulting in more valid estimates of the associations studied. Further, a major limitation is the lack of evidence regarding adenomyosis, making further conclusions impossible. The result of the meta‐analysis on the association between preterm birth and endometriosis was susceptible to the inclusion of the study by Vannuccini et al. , 34 which did not define preterm birth further. When including the study, we found a statistically increased risk of endometriosis, whereas excluding the study resulted in a slightly lower pooled OR with confidence interval overlapping the null. The overall result must therefore be interpreted cautiously. Furthermore, the investigated indices of fetal growth varied across the studies, which made the comparison difficult. These differences resulted in meta‐analyses where not all studies could be included. Finally, funnel plots were performed but, due to a sparse number of studies, the risk of publication bias could not be ruled out. Several methodological aspects of the included studies need to be empathized when assessing the results from this systematic review and meta‐analysis. Overall, the ascertainment of birthweight and preterm birth varied between the studies. Self‐reported information on birthweight and gestational age at birth is prone to misclassification. Differential misclassification may be present if women with and without endometriosis or adenomyosis are reported differently. However, we consider non‐differential misclassification to be more likely. Furthermore, the likelihood of being diagnosed with endometriosis or adenomyosis in adult life is most likely unrelated to birthweight or gestational age. Thus, the detection of the diseases will be independent of the exposure, and potential bias from classification error will be towards the null. Selection bias may occur if the association between birthweight, preterm birth and risk of endometriosis or adenomyosis differs between the population included and those eligible to participate. Selection bias in the register‐based cohort studies 31 , 43 , 44 , 45 , 46 was minimal because of no self‐selection and minimal loss to follow‐up. The higher‐than‐average incidence of endometriosis (34%) in the matched cohort study 47 may suggest selection bias. Likewise, the selection of the control group in the included case–control studies may have led to selection bias. In four of the case–control studies, 18 , 32 , 34 , 40 the controls were women undergoing gynecological surgery for diseases other than endometriosis. If the chosen controls differed from the background population with respect to the risk of being born preterm or with a low birthweight, selection bias might have occurred. Sub‐analyses excluding the mentioned case–control studies were performed; the association between low birthweight and endometriosis was lightly attenuated (OR 1.18, 95% CI 1.05–1.32). The sub‐analysis on the association between preterm birth and endometriosis revealed a similar result to the main result (OR 1.13, 95% CI 0.90–1.41). Information which might allow an evaluation of how the non‐participating women might differ from the included women in the studies, was unavailable. The matched cohort study by Wolff et al. 47 and the included case–control studies based the classification of endometriosis on surgery and histological confirmation when available. The cohort studies, except the study by Missmer et al. , 46 used health registers to identify women with endometriosis or adenomyosis based on International Classification of Diseases (ICD) diagnosis. The ICD codes do not specify whether surgery and confirmation by histology was done or whether the diagnoses were based on symptoms alone. If women with mild or no symptoms of the diseases did not present at a gynecological ward, the association between birthweight, preterm birth and endometriosis or adenomyosis might have been underestimated. This may be supported by the fact that all the cohort studies (except the study by Rossi et al. 43 ) found a relatively small percentage of women with endometriosis compared with other studies, suggesting that endometriosis affects approximately 10% of all women of reproductive age. 1 All of the included studies adjusted for potential confounding factors. However, the number of potential confounding factors included in the studies varied (Table  2 ). Only five of the included studies adjusted for some of the potential confounding factors we had identified as important a priori. Thus, confounding may still be present. Further Gao et al. 31 , 45 and Vannuccini et al. 34 adjusted for maternal endometriosis, although the results did not change after this adjustment. Endometriosis is considered a multifactorial disease and although the underlying causes are not well described, they are considered to be caused by both genetic and environmental factors. 26 , 27 Further, emerging evidence shows that pregnant women with endometriosis face a higher risk of having a child born preterm or with low birthweight. 48 If there is a strong heritability in endometriosis, the association found in this meta‐analysis, between preterm birth or low birthweight and risk of endometriosis in adulthood, may reflect that preterm birth or low birthweight are markers of an underlying genetic risk of endometriosis. Only three of the included studies 31 , 34 , 45 reported on maternal endometriosis and we were unable to pursue this issue further. We identified potential intermediate factors such as age at menarche and body mass index in childhood and adolescents. Adjusting for intermediate factors may result in an overadjustment and bias, the direction of which is unpredictable. Wolff et al. 47 adjusted for these intermediate factors but their results of no association remained with and without the adjustment. The external validity of the included case–control studies 18 , 32 , 33 , 34 , 40 and the matched cohort study by Wolff et al. 47 might be limited as they were based on selected populations, ie women undergoing surgery. Overall, we considered the results from four of the cohort studies to be generalizable to other similar populations, ie predominantly women from Northern European countries. 31 , 43 , 44 , 45 This systematic review and meta‐analysis support previous indications of an association between indices of fetal growth, preterm birth and development of endometriosis in adulthood. Future research may need to focus on valid information on indices of fetal growth restriction (eg birthweight‐for‐gestational age or fetal growth restriction defined by ultrasound during pregnancy), outcomes and potential confounders, facilitating systematic reviews and meta‐analyses of high‐quality studies with low risk of bias and residual confounding. Studies with large study populations and the possibility to separate subtypes of endometriosis with thorough measures of indices of fetal growth are recommended. Further, future research should aim to disentangle whether the associations found could be due to familial inherent factors or due to other fetal exposures.

This systematic review and meta‐analysis found that low birthweight and being born preterm were associated with a higher risk of endometriosis in adult life. The results must be interpreted cautiously due to many methodological weaknesses in the included studies. Importantly, it remains unknown whether the association reflects a causal relation between preterm birth or low birthweight and endometriosis or the association may be due to unrecognized confounding. Due to the small number of studies, no conclusion could be made related to adenomyosis and this remains to be investigated further. The current study supports the hypothesis of a potential early programming effect that may affect the risk of endometriosis in adulthood; however, this remains to be investigated further.

Endometriosis and adenomyosis are two common gynecological diseases affecting women in the reproductive age. 1 , 2 The diseases manifest with various symptoms and severities. Endometriosis can be divided into three subtypes: Superficial peritoneal endometriosis, ovarian endometriosis and deep infiltrating endometriosis. 3 A comprehensive classification of types of adenomyosis still ceases to reach consensus. 4 Generally, the symptoms of endometriosis and adenomyosis overlap. They include abdominal and pelvic pain, dysmenorrhea, dyspareunia and infertility. 5 , 6 The diseases are thought to be two different entities, though many women are diagnosed with both. 7 , 8 Endometriosis impairs health‐related quality of life and work productivity. 9 Furthermore, studies suggest that women diagnosed with endometriosis have an increased risk of developing cardiovascular, autoimmune and atopic diseases as well as gynecological cancers. 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 Despite the significant impact of endometriosis and adenomyosis and increasing research on the pathophysiology, the etiologies of the diseases remain enigmatic. This impairs development of preventive and therapeutic strategies and presents a major problem. 18 Various mechanisms for the development of endometriosis have been proposed. 19 Endometriosis is an estrogen‐dependent disease, 20 , 21 but also an immunological disease 22 , 23 , 24 , 25 with a genetic component. 26 , 27 Epigenetic changes in fetal life have been shown to influence both hormonal and immunological deviations. 19 The hypothesis related to the Developmental Origin of Health and Disease empathizes the importance of the intrauterine and early human environment on the risk of diseases in adult life. 28 Thus, epigenetic changes may be prompted by factors leading to fetal growth restriction or by preterm birth, and these changes may in turn also be associated with diseases in adult life such as endometriosis. The hypothesis of a developmental origin of endometriosis is novel and to date only a few studies have evaluated the association between fetal growth and preterm birth and the risk of endometriosis later in life. The current findings seem divergent. In a systematic review, Olšarová et al. 29 found that low birthweight was associated with a higher risk of endometriosis in adulthood, but evidence regarding preterm birth was inconsistent. In a meta‐analysis, Ottolina et al. 30 concluded that preterm birth and low birthweight were risk factors for developing endometriosis later in life. However, both studies only investigated the most severe types of endometriosis by including studies with surgically or histologically confirmed endometriosis, which may limit the generalizability. Also, none of the studies investigated adenomyosis. The underlying etiology of endometriosis is still unknown and more knowledge on the potential developmental causes is needed. With this systematic review and meta‐analysis, we aimed to review the literature of the association between measures of fetal growth and preterm birth and the risk of endometriosis and adenomyosis in adult life. We aimed to include studies with wider diagnostic criteria than the published systematic review and meta‐analysis and further to study adenomyosis and subtypes of endometriosis.

The authors have stated explicitly that there are no conflicts of interest in connection with this article.

We performed a systematic search on PubMed and EMBASE to identify original peer‐reviewed studies regarding fetal growth and preterm birth and endometriosis or adenomyosis. We used medical subject headings (MeSH terms) in PubMed and Emtree in EMBASE as well as free text terms. See Appendix  S1 for search strings. The reference lists of the included studies were furthermore searched for additional studies. An additional search was performed October 21, 2022 to make sure all relevant studies were included. We excluded studies without a reference group, eg case series, case reports as well as commentaries, letters and editorials. We included published case–control and cohort studies. The included studies had to investigate the association between measures of fetal growth or preterm birth and endometriosis or adenomyosis later in life. No language restrictions were applied. All relevant studies were transferred to Covidence ( https://www.covidence.org/ ), where duplicates were removed. Two authors (MB and KB) used Covidence to screen studies by title, abstract and full text. In case of disagreement, a third author (LH) was involved. To assess the quality of the cohort and case–control studies (the risk of selection and information bias as well as confounding) the Newcastle–Ottawa Scale (NOS) was used. The NOS for case–control studies was used to assess the quality of the matched cohort study included. A priori, we defined the criteria for scoring the studies on a scale from zero to eight (Appendix  S2 ). A study with a NOS score above six was assessed as a high‐quality study, four to six as moderate quality and zero to three as low quality. The scoring was done independently by two authors (MB and KB); if consensus was not reached by discussion, a third author (LH) was involved. The literature on potential causes of endometriosis and adenomyosis is limited, thus knowledge of possible confounding factors on the association between fetal growth and preterm birth and risk of endometriosis or adenomyosis is sparse. Based on the current knowledge, we defined potential confounders as maternal socioeconomic status, maternal ethnicity, maternal smoking during pregnancy and maternal age at birth. Intermediate factors were defined as age at menarche and body mass index in childhood and adolescents. 18 , 31 , 32 , 33 , 34 , 35 , 36 , 37 Information about bibliography, study design, study period, sample size, population, exposures, source of outcome, exclusion criteria, type of endometriosis studied, effect estimates and adjustment for confounding factors were extracted. Two authors (MB and KB) performed the data extraction from all studies. If disagreement occurred, a third author (LH) was involved. We used structured extraction sheets to identify number of women with endometriosis or adenomyosis in each group, crude and adjusted estimates with 95% confidence interval (CI). We used ReviewManager 38 to perform the meta‐analyses using a random‐effect inverse variance weighted model. 39 For the meta‐analyses, only studies with the same measure of exposure were included. First, we performed a meta‐analysis including studies investigating birthweight <2.5 kg. Secondly, we performed a meta‐analysis including studies on preterm birth, defined in the studies as birth prior to either 37 or 38 weeks of gestation. We allowed the study by Vannuccini et al. 34 to be included in the meta‐analysis, although they did not define preterm birth further. In a sub‐analysis, we repeated the meta‐analysis, excluding the study by Vannuccini et al. 34 Further sub‐analyses were made excluding the case–control studies, where the controls were women undergoing gynecological surgery for diseases other than endometriosis. 18 , 32 , 34 , 40 In the meta‐analyses, adjusted effect estimates were applied when available and, if not, crude estimates were used. With this method, effect estimates and the lower limit of the 95% CI were inserted in ReviewManager. The upper limit of the 95% CI was calculated by the program and was therefore not identical to the 95% CI stated in the original studies because of small differences in the rounding off of numbers. If both crude and adjusted estimates were unavailable, numbers of women born preterm and women born at term with and without endometriosis or adenomyosis were inserted in ReviewManager and crude odds ratios (OR) were manually calculated. The heterogeneity in study design, population and definition of exposure was evaluated and I 2 was used to assess the statistically heterogeneity between the included studies. An I 2 value of 0%–40% suggests a heterogeneity that might not be important, 30%–60% represents a moderate heterogeneity, 50%–90% a substantial heterogeneity and 75%–100% a considerable heterogeneity. 41 ReviewManager was also used to evaluate the risk of publication bias based on funnel plots. Reporting was done in accordance with the Preferred Reporting Items for Systematic Reviews and Meta‐analyses (PRISMA) checklist 42 and the protocol for this study was registered at PROSPERO with the ID: CRD42021249322. No institutional Review Board approval was needed.

Appendix S1. Click here for additional data file. Appendix S2. Click here for additional data file. Figure S1. Click here for additional data file. Figure S2. Click here for additional data file. Figure S3. Click here for additional data file. Figure S4. Click here for additional data file.