Perfluoroalkyl substances (PFAS) in drinking water and risk for polycystic ovarian syndrome, uterine leiomyoma, and endometriosis: A Swedish cohort study

A B S T R


Introduction
Perfluorinated substances (PFAS) belong to a large group of synthetic substances frequently used in industrial and household products because of their water-and oil-repelling abilities.However, PFAS are extremely resistant to degradation, resulting in global environmental pollution after decades of widespread use (Buck et al., 2011;Houde et al., 2006).Awareness of the possible contamination of drinking water sources from industrial waste and the use of PFAS-containing firefighting foam has increased (Domingo and Nadal, 2019;Guelfo Abbreviations: AFFF, aqueous film-forming foam; CI, confidence interval; EDCs, endocrine disrupting chemicals; HR, hazard ratio; PCOS, polycystic ovarian syndrome; PFAS, perfluoroalkyl substances; PFDoA, perfluorododecanoic acid; PFHxS, perfluorohexane sulfonic acid; PFOA, perfluorooctanoic acid; PFOS, perfluorooctanesulfonic acid.and Adamson, 2018;Hu et al., 2019).PFAS containing aqueous filmforming foam (AFFF), used for high-hazard flammable liquid fires has been used globally and uncontrolled, usually during trainings sessions for fire-fighters, since the mid-60th (Place and Field, 2012).At several firefighting training facilities there have now been a widespread leaching of PFAS from the soil to nearby water sources (Høisaeter et al., 2019;de Vries et al., 2017;Ahrens et al., 2015).
There is growing concern about the adverse effects of PFAS on human reproduction.As the female reproductive system is driven by a complex combination of endocrine mechanisms, the female reproductive organs can be sensitive to interference by endocrine disrupting chemicals (EDCs) (Piazza and Urbanetz, 2019).A recently published literature review of PFAS effects on ovarian function reported adverse effects of PFAS on ovarian folliculogenesis and steroidogenesis in experimental models, and provided some support for an association between PFAS exposure and ovarian dysfunction in a small number of epidemiological studies (Ding et al., 2020).
Most studies on PFAS and the female reproductive system are based on PFAS levels found in the general population from background exposure, but few studies have investigated the female reproductive system in relation to high PFAS exposure.In the C8 Health Project Study, which follows a US population with high perfluorooctanoic acid (PFOA) exposure, girls with higher levels of PFOA and perfluorooctanoic acid (PFOS) exposure were found to have a delayed onset of puberty (Lopez-Espinosa et al., 2011).Another study in the C8 Project detected a positive association between earlier age of menopause and higher serum levels of PFOA and PFOS, along with an inverse association between serum PFOS levels and estradiol levels (Knox et al., 2011).Moreover, an observational study of 146 young women from the Veneto region in Italy, an area heavily polluted with PFAS (primarily PFOA), showed a positive association between PFOA serum levels and self-reported delayed menarche and irregular menses (Di Nisio et al., 2020).
Polycystic ovarian syndrome (PCOS), uterine leiomyoma (also known as fibroids), and endometriosis are common gynecological diseases driven by gonadal steroid hormones that affect women of reproductive age (Carpinello et al., 2000;Escobar-Morreale, 2018;Stewart et al., 2016).The reported incidences (cases per 100 000 person years) among premenopausal women seeking healthcare varies largely between study designs and populations.Recent reports presents incidences between 65 and 100 for PCOS (Liu et al., 2021), 845-1348 for uterine fibroids (Stewart et al., 2017) and 20-600 for endometriosis (Rowlands et al., 2021).PCOS is characterized by polycystic ovaries, menstrual disturbances, and hirsutism caused by excess androgen hormone production and ovarian dysfunction (Escobar-Morreale, 2018).Uterine leiomyoma is a benign tumor that develops from the smooth muscular tissue in the uterus (Stewart et al., 2016).Endometriosis is defined as the presence of endometrial tissue outside of the uterus, causing inflammation and adhesions mainly in the pelvic cavity (Giudice, 2010).These benign diseases can, depending on the condition, cause much harm to the patient due to pelvic pain (Fauconnier and Chapron, 2005), heavy menstrual bleeding (Whitaker and Critchley, 2016), and reduced fertility (Balen et al., 2016;Haydardedeoglu and Zeyneloglu, 2015;Pritts et al., 2009), and may require surgical or medical treatment (Morgante et al., 2018;Stewart et al., 2016).
The etiologies of these diseases are considered to involve a complex combination of genetic and environmental factors (Asghari et al., 2018;Franks et al., 2006;Pavone et al., 2018).Their endocrine characteristics indicate that EDCs may play a role in the development of these diseases (Katz et al., 2016;Piazza and Urbanetz, 2019;Rutkowska and Diamanti-Kandarakis, 2016;Smarr et al., 2016).However, epidemiologic evidence considering PFAS exposure remain sparse and inconsistent, particularly for uterine leiomyoma, for which there has been only one study to date (Trabert et al., 2015).For PCOS (Heffernan et al., 2018;Vagi et al., 2014;Wang et al., 2019) and endometriosis (Campbell et al., 2016;Louis et al., 2012;Wang et al., 2017), there is some indication of an association at background exposure levels.
The aim of this cohort study was therefore to investigate the possible associations between high PFAS exposure and polycystic ovarian syndrome (PCOS), uterine leiomyoma, and endometriosis.

Study setting
In the fall of 2013, through a survey of groundwater quality in Blekinge County in southern Sweden, very high levels of PFAS were detected in one of the two waterworks in Ronneby, a Swedish municipality with 28,000 inhabitants (Li et al., 2018;Xu et al., 2021).The water from the contaminated waterworks, with sum of PFAS > 10,000 ng/L, supplied one-third of all households, thus exposing many Ronneby residents to PFAS for decades without their knowledge (Table S1).The other waterworks had far lower PFAS levels (sum of PFAS ~ 50 ng/L), lower than required by the present Swedish regulation covering PFAS (action limit = 90 ng/L) (Livsmedelsverket, 2020).However, the levels were still higher than those from the non-contaminated waterworks of the neighboring municipality of Karlshamn (sum of PFAS < 5 ng/L) (Table S1).Historical levels of PFAS in the drinking water before 2013 were not available as they had never been measured.
The highly contaminated waterworks was closed in December 2013, and water from the other waterworks was immediately provided.It was soon confirmed that the source of the PFAS contamination was groundwater contaminated by AFFF used in a fire drill site at the nearby military and civil airfield since the mid-1980 s (Xu et al., 2021).Extensive biomonitoring of the Ronneby population took place in 2014-2015, starting approximately 6 months after end of external exposure and revealed very high serum levels of perfluorohexane sulfonic acid (PFHxS), PFOS, and to a lesser extent PFOA compared to referents from Karlshamn and the general population in Sweden (Xu et al., 2021).

Study population
An open dynamic cohort, the Ronneby Registry Cohort, was defined as people who had ever been registered as living in Ronneby municipality, Sweden, during the time period 1985-2013.All females (n = 29,106) within the Ronneby Registry Cohort, and thus ever having resided in Ronneby between 1985 and 2013, were included in the present study.The annual registered residential address, vital status, and emigration status and, if available, the highest attained education level, for each participant were obtained from Statistics Sweden (www.scb.se).Educational level (information available from 1990) was defined as the highest degree accomplished at age 30 and was categorized into three groups: "low," defined as up to nine years of education; "medium," defined as 10-12 years; and "high," defined as > 12 years, and is a commonly used proxy for socioeconomic status.Through the use of the personal identity numbers and national patient registries (described in Section 2.4), a person's date of diagnosis was available regardless of their place of residence within Sweden (although not if they had emigrated from Sweden).

PFAS exposure assessment
The yearly exposure assessment was based on a person's yearly residence address linked to data from the municipality on water supply.Connecting all addresses to the waterworks supply data allowed us to determine whether a residential address had been provided with contaminated municipal water for each year during 1985-2013.Unfortunately, there is no clear information available as to when the drinking water initially became contaminated.It has been estimated by the Swedish Armed Forces that the use of AFFF started in the mid-1980s.In order not to misclassify any household waters as contaminated, we cautiously set 1985 as the start of contamination.We assume that PFAS levels in the drinking water were lower during the early years and S. Hammarstrand et al. increased over time.Cohort members were therefore assigned timevarying exposure, with a dichotomous exposure based on when they lived at an address with water from the highly contaminated water works ("never-high" or "ever-high", referring to the period 1985-2013) and a three-category alternative ("never-high"; "early-high", referring to the period 1985-2004; or "late-high", referring to the period 2005-2013).Note that many individuals in the late-high group had also been exposed earlier.The definitions for all exposure categories are listed in Table 1.Measurements in private wells in the municipality showed no indications of elevated PFAS levels of > 90 ng/L; residents with private wells were thus included in the "never-high" group (Xu et al., 2021).
The PFAS exposure group classification has been validated among 3511 subjects in a previous study (Xu et al., 2021), and was now repeated for women (Table S2).In 2014 women with "late-high" exposure had median; PFHxS 243 ng/mL, PFOS 279 ng/mL, PFOA 15.2 ng/mL, Table S2.On group level, there were marked differences between never vs ever high, and early vs late high, thus the crude classification was considered to be valid.
Individuals may have moved during the study period, and therefore the exposure status varied with time.However, as the internal exposure of PFAS will still be ongoing even after the external exposure stops, a change in category could occur only in one direction, i.e., as a woman may only move from the "never-high" to the "ever-high" category.We also calculated "duration of exposure at residency" i.e. the number of years spent living at an address with contaminated water, and the "time since first exposure" i.e. the number of years between the first year at an address with contaminated water and the diagnosis of disease.

Diagnostic outcomes
The Swedish National Board of Health and Welfare provided data from The Swedish National Patient Register, which includes complete information on inpatient care in Sweden from 1987 onward (Ludvigsson et al., 2011).Hospital outpatient visits from both public and private healthcare systems were additionally available from 2001.Thus, from 2002, the combination of in-patient-and out-patient data from hospitals likely captures most gynecological diagnoses, although not from primary care providers.For Blekinge County, in which Ronneby is located, inpatient data were already included from 1985, and therefore our observation period started on January 1, 1985.We obtained diagnosis data for incident cases of PCOS, uterine leiomyoma, and endometriosis according to the international classification ICD codes (Table S3) during the study period 1985-2013.All diagnoses from clinics of obstetrics & gynecology were assumed to be valid, whereas diagnoses from other clinics required additional information such as an associated relevant diagnosis, procedure, or surgery code to support the diagnosis in order for the case to be included in the analysis.

Statistical analysis
Through registers, the women were followed up to the first event of each disease.The associations between the time-varying exposure and diagnoses were analyzed using the Cox proportional hazard model, with calendar year on the time axis and the age groups (20-35, 36-50, 51-59, and 60 + years) were allowed to have separate baseline hazard functions (subcommand STRATA).Hazard ratios (HR) with 95% confidence interval (CI) were calculated to compare the different exposure (Table 1).At each year, the highest exposure category "up to now" was used in the model.The proportionality assumption was tested for all analyses by including the interaction term, exposure*calendar year, and testing if it was significant (p < 0.05).Proportionality was fulfilled for all main analyses.To assess unmeasured confounding we calculated the E-value for the main analyses (VanderWeele, 2017).The E-value is defined as the minimum strength of association, on the risk ratio scale, that an unmeasured confounder would need to have with both the exposure and the outcome to fully explain away a specific association, conditional on the measured covariates.
Our main analyses were restricted to women aged 20-50 years (assumed to be premenopausal) observed between 2002 and 2013.For uterine leiomyoma we also extended the age span with no upper age limit, due to the large number of cases with postmenopausal diagnosis.A person was censored at the event (PCOS, uterine leiomyoma, or endometriosis), emigration or death, or the end of the study period (December 31, 2013).Categories with fewer than seven cases were analyzed but not interpreted.All statistical analyses were performed using SAS 9.4 software (SAS Institute, Cary, NC, USA).The following sensitivity analyses were performed: (I) Adjusting for educational level as an indicator for socioeconomic status.(II) To assess how robust the findings are to potential unmeasured confounding, the E-value was calculated for the significant results.(III) Analyses were performed with age on the time axis where the three decades (1985-1994, 1995-2004, and 2005-2013) were allowed to have difference baseline hazards.(IV) Analyses were performed with exposure measured as the cumulative number of years that a person had lived at an address with contaminated water supply.(V) As an indication of severity of disease for endometriosis and uterine leiomyoma during 2002-2013, only cases with registered procedures or surgery codes (hospital inpatient and outpatient data) were analyzed.Examples of included surgery and procedures; laparoscopy, laparotomy, hysterectomy and transvaginal ultrasound.Separate analyses for inpatient data only covering the period 1985-2001 were performed, also in order to select more severe cases.(VI) Analyses of endometriosis in women 20-50 years, where cases of adenomyosis, (endometrial cells in the uterine musculature), were excluded.

Ethics approval
The study received ethical approval from the ethics Committee in Lund, Sweden (dnr 2014/267).

The study cohort
Of the 29,106 women in the Ronneby Registry Cohort, 7,823 (27%) were estimated to have been exposed to PFAS in drinking water at their a Through linkage between registered yearly residence addresses and the data on yearly water supply area obtained from the municipality water company, we determined, for each calendar year, whether the address received drinking water that was heavily or minimally contaminated.
home address sometime between 1985 and 2013 (Table 2).During this period, there were a total of 161 reported cases of PCOS, 1,122 cases of uterine leiomyoma, and 373 cases of endometriosis (Table 2).Women who had ever-high exposure at their residency were somewhat younger than the "never high" exposed women throughout the entire study period.Also, these women had lower educational level compared to women never exposed to PFAS in residential drinking water (Table 2).The large majority of cases were diagnosed at a clinic of obstetrics & gynecology (92% of PCOS cases, 99% of uterine leiomyoma cases, and 97% of endometriosis cases).Of the remaining cases of PCOS, most were diagnosed at either an endocrinology clinic or a surgical clinic.Two registered cases (one uterine leiomyoma and one endometriosis) were considered to be non-cases, as they were diagnosed at a psychiatric and an orthopedic clinic, respectively, with no further validation.

Main findings
Women diagnosed with PCOS were primarily found at ages 20-35, women with endometriosis between 20 and 50, whereas most women with uterine leiomyoma were found at ages 51-59 (Table 3).Information on in-and outpatient diagnoses are given in Table S4.In the main analysis we included 132 cases of PCOS, 393 cases (for extended analysis 719 cases) of uterine leiomyoma, and 156 cases of endometriosis (Table 3).For all three diseases, there were no reported cases younger than 12 years of age and very few below 20 years of age (Table S5).
Cox proportional hazards models showed an increased risk for PCOS in premenopausal women with "late-high" exposure compared to those with "never-high" exposure (HR = 2.18; 95% CI 1.43, 3.34; Table 4).We also found a nominally increased risk for uterine leiomyoma in premenopausal women with "late-high" exposure compared to the "never-high" exposed group (HR = 1.28; 95% CI 0.95, 1.74).No clear increased risk for endometriosis was found.

Findings; sensitivity analyses
In the analysis with further adjustment for the highest achieved education, the HRs remained similar for all three diseases (Table S6).
For the PCOS hazard ratio, HR = 2.18, the E-value was 3.78 and for the lower confidence limit, the E-value was 2.21, indicating that the lower confidence limit of 1.43 could be explained away by an unmeasured confounder that was associated with both the exposure and PCOS by a risk ratio of 2.21 each, but weaker confounding could not do so.
In the sensitivity analysis where calendar year was replaced by age on the time axis (Table S7), results remained similar as in the main analyses.Premenopausal women in the "late-high" exposed group showed a nominally increased risk of uterine leiomyoma with a registered procedure or surgery code (HR = 1.33; 95% CI 0.93, 1.89) compared to the "never-high" exposed group (Table S8).We detected no other overall consistent pattern of associations for uterine leiomyoma or endometriosis using inpatient data only (Table S8).Out of all cases of endometriosis, 52 cases (14%) were adenomyosis.The results remained similar when these were excluded (Table S8).
A majority of the women with PCOS and uterine leiomyoma with "late-high" exposure had been living at a residential address with contaminated water for a longer period of time than patients with PCOS in the "early-high" group (Table 5).Furthermore, a shorter time window separated first exposure from diagnosis in a majority of these "late-high" exposed women (Table 5).For women with endometriosis, the "earlyhigh" and "late-high" exposed groups had a similar duration of residency with contaminated water.The "late-high" groups was characterized by a shorter time from first exposure to diagnosis compared to the "earlyhigh" groups (Table 5).Using cumulative number of years at addresses with contaminated water gave risk estimates per 10 years: HR 1.18, (95% CI 0.87-1.58)for PCOS, HR 1.07, (95% CI 0.91-1.27)for uterine leiomyoma, and HR 0.76, (95% CI 0.53-1.09)for endometriosis.

Discussion
We found that high PFAS exposure, dominated by PFHxS and PFOS, was associated with an increased risk for PCOS and possibly also for uterine leiomyoma in women aged 20-50 years, an age range assumed to be premenopausal.We did not observe any increased risk for endometriosis.These findings are consistent with the hypothesis that PFAS may interfere with the female reproductive system.
The exposure situation in Ronneby is unique as serum PFOS and PFHxS levels in women aged 20-50 could reach more than 300-500 ng/ mL (95 percentile), with geometric mean levels around 60-100 ng/m (Xu et al., 2021), compared to most previous studies with PFAS levels representing background exposure.
Reproductive hormones play a significant role in the etiology of benign gynecological diseases, and PFAS exposure may affect hormone levels.However, the underlying molecular mechanism(s) have yet to be elucidated.Whereas both uterine leiomyoma and endometriosis are estrogen-dependent diseases (Borahay et al., 2017;Bulun et al., 2012), PCOS is characterized by an excess of androgens, in particular testosterone (Escobar-Morreale 2018).Thus, different exposure-response patterns can be expected.

PCOS
Our findings with PCOS are consistent with three observational casecontrol studies with PFAS levels representing background exposure, in which women with PCOS had significantly higher mean serum levels of PFOA (Vagi et al., 2014), PFOS (Heffernan et al., 2018;Vagi et al., 2014), and perfluorododecanoic acid (PFDoA) (Wang et al., 2019) than matched controls.As PCOS is driven by an excess of androgens, potential  biological mechanisms may include PFAS acting as an agonist of the androgen receptor or increasing the levels of circulating testosterone.Yet, the results from in vitro studies are inconsistent, as PFAS exposure has been associated with both an increase (Kraugerud et al., 2011) and a decrease (Du et al., 2013) in testosterone during steroidogenesis, and there have also been conflicting results regarding the effect of PFAS on androgenic activity (Du et al., 2013;Kraugerud et al., 2011).Human studies also suffer from the same overall inconsistencies regarding the effect of PFAS on reproductive hormones.A small observational study of 30 patients with PCOS and 29 matched controls from a fertility clinic determined that serum testosterone levels are positively associated with PFOA and PFHxS levels in controls but not in patients with PCOS (Heffernan et al., 2018).Yet, a systematic literature review of the effects of PFAS on reproductive hormones did not demonstrate a causal relationship (average PFOS levels ranging between 4 and 36 ng/mL and 2 and 6 ng/mL for PFOA) (Bach et al., 2016).

Uterine leiomyoma
This is the first study of uterine leiomyoma in a population with high PFAS exposure.The only previous study available on this topic, an observational study comprising 473 women of childbearing age undergoing laparoscopy or laparotomy that reported patients' background exposure levels (geometric mean PFOS: 6 ng/mL; PFOA: 2 ng/mL; PFHxS: 0.4 ng/mL), did not show any associations between uterine leiomyoma and PFAS levels (Trabert et al., 2015).However, the same study did find an association between polychlorinated biphenyls, another type of EDC, and uterine leiomyoma (Trabert et al., 2015).Two literature reviews on the effect of a broad range of EDCs concluded that several of these may be associated with increased risk for, and also the severity of, uterine leiomyoma (Bariani et al., 2020;Piazza and Urbanetz, 2019).

Endometriosis
No increased risk for endometriosis was observed which contrast to previous studies.A cross-sectional cohort study of 753 premenopausal women with self-reported endometriosis showed higher geometric mean levels of PFOA, PFOS, and perfluorononanoic acid (PFNA) (mean PFOA: 3 ng/mL, PFOS: 16 ng/mL, PFNA: 1 ng/mL; PFHxS; 1 ng/mL) compared to those not reporting endometriosis (Campbell et al., 2016).Another cohort study of 495 women undergoing laparoscopy or laparotomy, and a matched population-based cohort for age and residency (n = 131) also exposed to background PFAS levels, reported higher serum levels of PFOA and PFNA, but not PFHxS, among women with endometriosis (Louis et al., 2012).The same study detected a positive correlation between severity of endometriosis and PFOA (mean 3 ng/mL) and PFOS (mean 7 ng/mL) levels (Louis et al., 2012).Furthermore, a case-control study of 157 individuals with surgically confirmed endometriosis and 178 controls, with PFOA levels among the cases similar to those in our study (median 15 ng/ mL) but much lower levels of PFOS and PFHxS (median PFOS; 6 ng/mL; PFHxS; 0.30 ng/ml), found that only perfluorobutane sulfonic acid (PFBS) (median 1 ng/mL in women with endometriosis) was statistically significantly associated with endometriosis (Wang et al., 2017).However, the European Food Safety Authority (EFSA) concluded in 2018, based on two cross-sectional studies available at the time, that there was not yet enough evidence to establish a clear connection between PFAS exposure and endometriosis (Knutsen et al., 2018).

Strengths and limitations
The study is almost a "natural experiment" in a municipality with exposure ranging from background levels to very high levels of PFHxS and PFOS, allowing an internal comparison between inhabitants in the same geographical localization.This study has several strengths, including a long follow-up period (up to 28 years), case ascertainment through personal identification and national registers, and a large cohort size of 29,106 women.National health services are equally accessible to everyone in Sweden, such that access to healthcare should not differ substantially between exposure groups.Although we lack information on gynecological outcomes from the primary health care, assumed to be of slightly milder symptomatology, we do not expect differential loss of information in relation to exposure.We expect no reporting bias as the exposure was unknown to the population and, as all women were diagnosed by a physician, no major misclassification bias should have occurred.
The crude modelled exposure based on area characteristics can, on the one hand, be seen as a limitation.On the other hand, area-based modelling is less sensitive to individual confounding.We assume that municipal water was the predominant source of drinking water since the use of bottled water was not common in Sweden before 2013, but we lack data on individual water consumption at home and outside the home, as well as on the dietary habits of the cohort participants.In comparison to many other countries, Swedish tap water is usually of very good quality and consumption of bottled water in the Swedish household has historically been very low (Westrell et al., 2006).Although there has been an increase during the past 20 years, the consumption is still low in a global perspective thus we expect consumption of tap water to be dominating during the entire study period.True PFAS exposure levels is a function of internal accumulation (Pérez et al., 2013), and PFAS elimination rates vary widely between individuals (Li et al., 2018).To conclude, we consider the crude exposure assessment to reliably distinguish true exposure categories in the cohort, as these categories corresponded to measured serum PFAS levels determined in 2014 in a separate biomonitoring study.
Women in our cohort ever living in the highly exposed water district had a lower educational level compared to women living in the other water districts where exposure was classified as never high.Educational level has been well validated as a proxy for socioeconomic status, and correlates well with both body mass index (BMI) and smoking habits in fertile women in Sweden (Socialstyrelsen, 2016).The women in the highly exposed area also had a lower mean age compared to the never high area.After taking both these factors, education and age, into account our results still show an increased risk for PCOS in the late-high group.Individual information on smoking and BMI were not available.For PCOS and uterine leiomyoma, obesity is believed to be an important risk factor in the development of the disease (Escobar-Morreale, 2018;Pavone et al., 2018;Stewart et al., 2017), and smoking may aggravate PCOS (Xirofotos et al., 2016).By contrast, being underweight has been claimed to increase endometriosis risk (Peterson et al., 2013).However, both smoking and BMI are highly associated with socioeconomic status and educational level, which we have adjusted for.Furthermore, BMI does not seem to have much influence on the elimination rate of PFAS substances, (Li et al., 2018).Moreover, adjusting for overweight or obese status in PFAS studies have been shown to increase the risk for collider bias as there is possible bidirectional relationship between PFAS and obesity (Inoue et al., 2020).Instead, adjusting for sociodemographic factors such as educational level, less likely to be in the mediating pathway, has recently been recommended (Inoue et al., 2020).In our study, adjustment for educational level did not alter the risk estimates for developing specific benign gynecological diseases.For our main finding, the increased risk of PCOS among women with latehigh exposure (resided in the area with contaminated water during [2005][2006][2007][2008][2009][2010][2011][2012][2013], the E-value for the lower confidence limit was 2.21, which indicates that the effect of unmeasured or uncontrolled confounding would have to be quite strong to explain away the finding.

Conclusions
Our findings show an association between high PFAS exposure, dominated by PFOS and PFHxS from AFFF firefighting foam contamination of drinking water, and an increased risk for PCOS.They also indicate a potential increased risk for uterine leiomyoma.No increased risk for endometriosis was found.As we encounter widespread contamination by PFAS and other persistent chemicals of the environment, a better understanding of the effects of PFAS on the female reproductive system is of great importance.It must be empathized that internal PFAS levels may remain long after the source of external exposure has been removed, and that these substances may be transferred from the mother to the fetus during pregnancy.Research on the negative human health effects of PFAS exposure has led to international regulations for PFOA and PFOS (UNEP, 2009), comprising only a limited part of all PFAS compounds present in the environment.Future research on PFAS and the female reproductive system is encouraged, not only to better understand the etiologic mechanisms but also because, although benign, these gynecological diseases can cause much harm to those affected, along with healthcare and societal costs.

Table 1
Drinking water exposure groups and definitions.

Table 2
Characteristics of females (all ages) in the Ronneby Registry Cohort for different calendar periods and exposure groups.
a The cohort includes all individuals who ever lived in the Ronneby municipality during the study period.bResided in Ronneby, in the area with water supply from the highly contaminated waterworks at home address any time in1985-2013.cResided in Ronneby, but never resided in the area with water supply from the highly contaminated waterworks at the home address, or had own well in 1985-2013.

Table 4
Cox proportional hazard regression for the analysis of all diagnosed cases (both inpatient and outpatient care registers; 2002-2013), of polycystic ovarian syndrome (PCOS), uterine leiomyoma, and endometriosis, in the Ronneby Registry Cohort.All estimates from Cox regression models are with calendar year on the time axis, and separate baselines for age strata(20-35, 36-50, 51-59, and 60 +  years).The total number of women 20-50 years of age was 18,503, and 20 + years was 21,987.aResided in Ronneby, but never resided in the area with water supply from the highly contaminated waterworks at the home address, or had own well in1985-2013.bResidedin Ronneby, in the area with water supply from the highly HR: hazard ratio; CI: confidence interval; p: p-value for two sided test.Note:

Table 5
Duration of residency with contaminated water and time since first exposure until diagnosis in women 20-50 years of age with polycystic ovarian syndrome (PCOS), uterine leiomyoma or endometriosis in the Ronneby Registry Cohort (both inpatient and outpatient care registers; 2002-2013).
a Resided in Ronneby, in the area with water supply from the highly contaminated waterworks at home address, in 1985-2004 but not later.bResided in Ronneby, in the area with water supply from the highly contaminated waterworks at home address, in 2005-2013; may also have resided in same area in1985-2004.