Endometriosis and Organochlorinated Environmental Pollutants: A Case–Control Study on Italian Women of Reproductive Age

Of the 312 women who underwent laparoscopy between January 2002 and December 2005 at the Department of Gynaecology and Obstetrics, Policlinico Umberto I, University of Rome Sapienza, for endometriosis or other benign gynecologic conditions, 158 patients were enrolled in the study. The vast majority of women participating in the study underwent surgery between March 2004 and October 2005. The protocol of this study was approved by the Sapienza University, School of Medicine, Institutional Review Board. All patients met the inclusion criteria: 18–45 years of age, residence in Rome in the last 5 years, no breast-feeding history, absence of immunologic, hormonal disorders, or chronic diseases, and no occupational exposure to PCBs or pesticides. All enrolled women signed an informed consent form. A physician unaware of the indications to laparoscopy administered a questionnaire before surgery which documented age, education, job, medical, gynecologic and obstetric history, height and weight, and smoking and dietary habits. The questionnaire was designed to obtain information on potential confounders, including metabolic diseases, gravidity, parity, and weight changes in the last years. A detailed medical and gynecologic history was taken, and all patients underwent clinical and ultrasound examinations. For each woman, the body mass index (BMI) was calculated. Before laparoscopy, a blood specimen of approximately 30 mL was collected from the cubital vein in Vacutainer tubes and centrifuged. Serum specimens were stored at −20°C until subjected to analysis. A 10-mm laparoscopy was performed under general anesthesia. The presence of endometriosis was confirmed by histologic analysis of lesions, and the disease was staged in 80 women according to the revised American Society of Reproductive Medicine (ASRM) classification. The control group consisted of women without complaints of infertility or pelvic pain who were undergoing laparoscopy for benign gynecologic conditions and had no visual evidence and histologic features of endometriosis in random peritoneal biopsies. The questionnaire administered to document dietary habits included 16 questions on the frequency of consumption (times/month) of various milk, meat, and fish products. Serum samples were added with a mixture of 13 C-labeled internal standards (Cambridge Isotope Laboratories Inc., Andover, MA, USA) comprising PCB-28, PCB-52, PCB-101, PCB-118, PCB-138, PCB-153, PCB-156, and PCB-180, HCB, and p,p ′-DDE and kept overnight at 4°C. Before extraction, we thawed the spiked samples at room temperature, added a 4:1 (vol/vol) mixture of formic acid and iso -propanol (J.T. Baker, Phillipsburg, NJ, USA), and sonicated them. Extraction was performed with five 6-mL aliquots of n -hexane (Merck KGaA, Darmstadt, Germany) by manual shaking, followed by centrifugation at 3,500 revolutions per minute (rpm) for 5 min. We removed the n -hexane aliquots collected from centrifugation, pooled them in a centrifuge tube, and carefully concentrated them. Concentrated sulfuric acid (H 2 SO 4 , Carlo Erba, Milan, Italy) was added to the n -hexane extracts; the two phases were vigorously shaken and then separated by centrifugation (3,500 rpm for 20 min). We reduced the volume of the purified extracts and transferred them into 1-mL vials to undergo instrumental analysis. Instrumental analysis was carried out by using ion trap mass spectrometry (IT-MS) (Polaris Q; Thermofisher Scientific Inc., Waltham, MA, USA) coupled with high-resolution gas chromatography (GC). A RTX-5MS 60-m length, 0.25-mm i.d. capillary column (Restek Corporation, State College, PA, USA) coated with a 0.25-μm film was employed to separate the extracted compounds. The initial oven temperature of 70°C was increased to 190°C at a rate of 30°C/min, subsequently to 280°C at 5°C/min and to 330°C at 20°C/min, and maintained at 330°C for 4 min (total GC run time was 28 min). The injector was operated in the splitless mode with a 1.5-min splitless time. The initial injector temperature of 70°C was increased to 280°C at a rate of 14.5°C/sec. The transfer line and the ion source temperatures were set at 290 and 250°C, respectively. The IT-MS detector was operated in the electron ionization mode (70 eV) and MS/MS mode. Selected daughter ions used for quantification were (analyte, daughter ion): PCB-28, PCB-186; PCB-52, PCB-257; PCB-101, PCB-291; PCB-105, PCB-256; PCB-118, PCB-256; PCB-138 and PCB-153, PCB-325; PCB-156 and PCB-167, PCB-290; PCB-170 and PCB-180, PCB-361; HCB-249; p,p ′-DDE-248. Data were processed using the XCALIBUR software (Thermofisher Scientific Inc.). Based on the 13 C-labeled compounds employed, recovery ranges were within 75–110% for all compounds. Analytic reliability was warranted by the use of an in-house validated method ( Ingelido et al. 2008 ). The laboratory has a consolidated experience in the analysis of halogenated organic micro-contaminants and periodically participates in interlaboratory exercises concerning the analysis of PCDDs, PCDFs, PCBs, organo-chlorinated pesticides, and some brominated flame retardants in dietary, biological, and environ mental matrices. We conducted the analysis of compounds eliciting dioxin-like activity by the dioxin receptor (DR or AhR)-driven chemically activated luciferase expression bioassay (DR-CALUX; BioDetection Systems, Amsterdam, The Netherlands), a bioanalytic tool used to detect AhR active compounds, such as dioxins, present in different environmental and biological matrices ( Hoogenboom et al. 1999 ; Murk et al. 1997 ). DR-CALUX cells were purchased from BioDetection Systems. We cultured them routinely in a 5% CO 2 atmosphere at 37°C in Alpha MEM supplemented with 10% heat-inactivated fetal calf serum (GIBCO BRL, Gaithersburg, MD, USA). The luciferase assay system, Britelite kit, was purchased from PerkinElmer Life and Analytical Science (Boston, MA, USA). Analysis involved liquid-liquid extraction of blood serum with a 97:3 (vol/vol) mixture of n -hexane and diethyl ether and lipid removal by eluting the extract on a silica gel column concentrated H 2 SO 4 . The purified extract was quantitatively transferred to a vial, evaporated, and dissolved in dimethyl sulfoxide (DMSO) for DR-CALUX measurement. DR-CALUX cells were seeded in 96-multi-well plates (Packard ViewPlate, PerkinElmer) and incubated for 24 hr in a CO 2 incubator. We performed the cell treatments in triplicate, adding to each well 100 μL of sample extract diluted in culture medium just before use, using DMSO as vehicle (0.5%, vol/vol DMSO in culture medium). After 24 hr of incubation at 37°C, cell monolayers were checked under an inverted microscope to exclude any cytotoxic effects of the extracts. We removed the exposure medium, washed the cells with phosphate-buffered saline with calcium and magnesium (pH 7.4), and lysed them with 100 μL Britelite solution. Light production was immediately measured with a MicroBeta luminescence counter (MicroBeta Jet, 1450 LSC, PerkinElmer). We always included a standard calibration curve of 2,3,7,8-TCDD in the experiments and simultaneously analyzed it with the samples ( Sanderson et al. 1996 ; Windal et al. 2005 ). Eight 2,3,7,8-TCDD concentrations (0.3–300 pM/well), dissolved in DMSO and processed as described for sample extracts, were tested. Results expressed as relative light unit values were transformed into 2,3,7,8-TCDD toxicity equivalents (TEQs) using the BioDetection Excel file. Results were expressed in picograms of DR-CALUX-TEQs (C-TEQs) per gram of serum fat (pgC-TEQs/g fat). The limit of detection was calculated as the signal measured from the DMSO solvent control on each well plate plus three times its standard deviation. Concentrations of total cholesterol, phospholipids, and triglycerides were determined by enzymatic methods ( Ingelido et al. 2008 ) and the use of colorimetric kits (Futura System s.r.l., Rome, Italy). In setting the study design, sample size estimation was performed to determine the number of women per group sufficient to detect a true odds ratio (OR) between 2.5 and 3.0. With a power of 80%, type 1 error of 5%, and 0.20 probability of exposure in controls, we calculated that 64–94 subjects (ORs = 3.0 and 2.5, respectively) were required per group. A retrospective power analysis based on 80 cases and 78 controls enrolled in the study and a significance level of α set to 5% provided a power estimation of 72.6% to detect a risk of 2.5 and 88% to detect a risk of 3.0. Serum concentrations of all the analytes determined were subjected to statistical analysis. Differences in analyte levels between groups were investigated; geometric rather than arithmetic means were employed. The statistical significance of differences between cases and controls was assessed by Student’s t -test. The concentration distribution of each analyte was divided by tertiles, and women were classified at low, medium, and high exposure. Using unconditional logistic regression analysis to adjust for potential confounders, the adjusted ORs and 95% confidence intervals (CIs) were estimated for the second and third tertile of PCBs, C-TEQs, HCB, and p,p ′-DDE. Variables included in the final model were age (years), BMI (kilograms per square meter), smoking habits (nonsmokers, exsmokers, smokers), and evidence of relevant weight modifications in the last 5 years (> 10 kg). In addition, with the aim to investigate whether different histopathologic features of endometriosis were related to different etiologic risk factor patterns, patients were classified according to the presence of ovarian endometrioma, peritoneal lesion, and deep lesions, and the groups were compared. The relationship between the severity of disease, according to the revised American Society of Reproductive Medicine (ASRM) classification and the detected chemical levels was assessed by analysis of variance and chi-square test. Multiple linear regression analysis was used to test the association between log-transformed PCB concentrations in serum and dietary habits (milk, meat, fish) after checking for potential confounders.

A total of 158 women were enrolled in the study, 80 cases and 78 controls. Among cases, there were eight patients with stage I endometriosis, 5 with stage II, 44 with stage III, and 23 with stage IV. In particular, deep endometriotic nodules were present in 6 women (7.5%), ovarian endometriomas were observed in 72 patients (90.0%), and peritoneal implants in 45 cases (56.2%) ( Table 1 ). The control group consisted of 78 women with other benign gynecologic conditions, including eight uterine myomas and 70 adnexal masses, with no laparoscopic evidence of endometriosis and random peritoneal biopsies negative for the disease at histologic analysis. The main characteristics of the studied population, as derived from the administered questionnaires, are shown in Table 2 . Mean age at interview was slightly higher in cases (31.6 ± 6.0 years) compared with controls (29.5 ± 6.1 years). Moreover, cases had a lower BMI (21.1 ± 2.8) than controls (22.4 ± 4.6). Both these differences were taken into account using a multivariate analysis. Age at menarche, having been breastfed, smoking habits, and alcohol consumption did not significantly differ between the groups. Dietary habits, described by the analysis of the dietary intakes of the main categories of food-stuffs, were not significantly different in the case of meat and fish. Average monthly consumption of milk and dairy products was slightly different among cases and controls (41 vs. 33 times/ month), but the associated incremental risk of endometriosis could be considered negligible (OR = 1.02; 95% CI, 1.00–1.04). Geometric means and pertinent CIs of the analytes assessed in cases and controls are shown in Table 3 . The concentrations of all the analyzed PCBs were higher in cases than in controls. Significant differences ( p < 0.05) were found for PCB-101, PCB-156, and PCB-170, and highly significant differences ( p < 0.01) were found for PCB-52, PCB-118, PCB-138, PCB-153, and PCB-180. p,p ′-DDE and total PCB serum concentrations, as the sum of all the 11 congeners analyzed, were also significantly higher (relative increase: 44.9 and 48.4%, respectively) in cases than in controls. As for the AhR ligands that elicit a response of the DR-CALUX system [dioxins, DL-PCBs, and possibly other (pseudo) planar aromatic pollutants present in serum], no significant difference was detected between cases and controls, despite the observed increase of two DL-PCBs (PCB-118, p = 0.0002, and PCB-156, p = 0.01) in women with endometriosis. However, as noted above, C-TEQs account for all compounds with dioxin-like activity, with PCBs 118 and 156 likely providing a minor contribution. Table 4 shows the frequency distributions of all the analytes divided by tertiles, as well as the results of the multivariate analysis adjusted for age at interview, smoking habits, BMI, and weight loss. Concentrations of PCB congeners 118, 138, 153, and 170 are associated with a significant increased risk of endometriosis for the second and third tertile when compared with the lowest tertile. Risk of endometriosis appears to be significant (OR = 3.05; 95% CI, 1.25–7.42) also for the highest serum concentrations of PCB-180 (≥ 60.5 ng/g fat). As to the sum of all PCB congeners, patients in the mid and upper tertiles have a 4- to 5-fold increased risk of having the disease. Covariate-adjusted ORs showed that no increased risk of endometriosis was associated with serum levels of HCB, p,p ′-DDE, and total C-TEQs, the latter determined by the DR-CALUX bioassay. To explore a possible correlation between analyte serum concentrations and the disease type and localization, we stratified cases by peritoneal, deep, or ovarian endometriosis and compared levels of total PCBs, C-TEQs, HCB, and p,p ′-DDE among groups. No differences in mean analyte levels could be found in women with endometriosis with respect to the different kind of disease. Furthermore, the association between serum concentrations of the analyzed organo-chlorinated compounds and the stage of the disease (according to the revised ASRM classification) was investigated. No significant differences between groups were observed: Data analysis showed that pollutant concentrations did not correlate with disease severity. The relationship between different food intakes and values of PCBs, C-TEQs, HCB, and p,p ′-DDE was estimated by a multivariate regression model. The lack of significance ( p > 0.05) of slope coefficients suggests that food intake is not associated with the concentrations of the chemicals analyzed in the present study.

The results of the present study are in agreement with findings we had obtained in two previous investigations carried out on smaller samples of nulliparous women, in which cases resulted to have a significant increase in blood concentrations of PCB-118, PCB-138, PCB-153, PCB-180, the sum of the mentioned 11 PCBs ( Porpora et al. 2006 ; Quaranta et al. 2006 ), as well as of p,p ′-DDE ( Quaranta et al. 2006 ). In particular, a greater than 3-fold increase was found for these four PCB congeners and their sum. The observed absence of an increase in TEQ values in women with endometriosis with respect to controls confirms the results we obtained in a study carried out on pooled blood samples from two small groups of Italian and Belgian women of reproductive age ( De Felip et al. 2004 ), in which no correlation was observed between dioxin-like compounds and the disease, on a country basis. This agreement is observed although different analytical methodologies were used: High-resolution GC coupled with high resolution mass spectrometry was employed to determine dioxin-like compounds in our previous study, whereas DR-CALUX bioassay was used in this work. The present study and the three above have similar study designs. In fact, with the exception (in this study) of a small minority of non-nulliparous women who had never breastfed, only nulliparous women were enrolled, to avoid the confounding factor of breast-feeding, known to determine a significant decrease in organochlorine body burden. All women underwent surgical confirmation of the disease or its absence, and only women with no visual evidence or histologic signs of endometriosis in random peritoneal biopsies were included in the control group. In the present study, women with complaints of infertility were not enrolled as controls, because some organochlorinated pollutants have been hypothesized to be associated with infertility, in particular p,p ′-DDE ( Korrick et al 2001 ; Weiss et al. 2006 ). There are no data regarding a correlation between the other gynecologic conditions of the control group (uterine myomas, benign adnexal mass) and the aforesaid pollutants. A comparison of this study with other case–control studies designed to explore the association between persistent organo-halogenated pollutants and endometriosis is complicated by the variety of study designs, analytical methodologies used, and number and type of compounds or congeners assessed. Recent papers ( Anger and Foster 2008 ; Heilier et al. 2008 ) have presented a comprehensive overview of the studies carried out on this topic, the vast majority focused on assessing the association between dioxins and/or PCBs and peritoneal endometriosis, and have discussed their limited or null comparability. On the whole, no significant correlation was observed in case–control studies between NDL-PCBs and/or DL-PCBs and endometriosis ( Buck Louis et al. 2005 ; Fierens et al. 2003 ; Gerhard and Runnebaum 1992 ; Heilier et al. 2004 , 2005 ; Lebel et al. 1998 ; Pauwels et al. 2001 ; Tsukino et al. 2005 ), although a nonsignificant increase of the most abundant NDL-PCB-138, PCB-153, and PCB-180 was observed in one study ( Gerhard and Runnebaum 1992 ) and of the same three congeners plus the DL-PCB-118 in another study ( Pauwels et al. 2001 ). As to the sparse case–control studies available specifically focused on the AhR ligands (PCDDs, PCDFs, and DL-PCBs), a significant association with these chemicals and the disease was generally not observed ( Anger and Foster 2008 ; Pauwels et al. 2001 ), whereas a few studies reported increased but not significant ORs for the disease ( Buck Louis et al. 2005 ; Heilier et al. 2005 ; Mayani et al. 1997 ). In one study that considered peritoneal endometriosis and deep endometriotic (adenomyotic) nodules separately ( Heilier et al. 2005 ), a significantly increased risk was associated with dioxin and DL-PCB serum concentrations in women with deep endometriosis. As to the association between endometriosis and organohalogenated pollutants other than PCBs and/or dioxins, only a few studies are available, carried out on PBBs ( Hoffman et al. 2007 ) or organochlorinated pesticides, including HCB and p,p ′-DDE ( Tsukino et al. 2005 ). No evidence of an association was found between endometriosis and PBB or organochlorinated pesticide serum levels in these studies. In our study, no correlation was observed between the disease and HCB, a ubiquitous persistent pollutant identified in human tissues worldwide, although its use was discontinued decades ago. The effects of HCB on ovarian function and circulating ovarian steroids were demonstrated on exposed nonhuman primates ( Foster et al. 1995 ). With regard to p,p ′-DDE, the main metabolite of the pesticide p,p ′-DDT characterized by both immune and endocrine toxicity ( Halloway 2005 ; Wójtowicz et al. 2007 ), findings from the present investigation confirm an increase in serum concentrations in women with endometriosis observed in a previous investigation. In that study, on the basis of the reported observation of an immunologic dysregulation in women simultaneously exposed to p,p ′-DDE and PCBs ( Daniel et al. 2002 ), we evaluated the immunologic status of two small groups of women and determined the serum concentrations of the most abundant PCBs and p,p ′-DDE to evaluate their possible role in dysregulation of the immune function observed in patients with endometriosis. The results we obtained showed that peripheral blood NK cytotoxic activity and interleukin-1 beta and interleu-kin-12 production were significantly down-regulated in patients with endometriosis with respect to controls, and this matched with higher serum concentrations of PCBs and p,p ′-DDE in the same patients ( Quaranta et al. 2006 ). In addition to peritoneal endometriosis, deep endometriosis (adenomyotic) nodules of the rectovaginal septum, considered a distinct clinical entity by some authors ( Donnez et al. 1996 ), have also been studied as to their potential association with exposure to organo-chlorinated pollutants. For this type of endometriosis, Heilier and coworkers ( Heilier et al. 2004 , 2005 ) found a significantly increased risk associated with both NDL-PCBs and dioxin-like compounds in serum. When we analyzed the association between the serum concentrations of PCBs and pesticides and the different types of disease, no differences were found between cases with peritoneal implants, ovarian endometrioma, deep lesions, or combined lesions. Therefore, it can be hypothesized that PCBs and p,p ′-DDE are risk factors for developing any kind of endometriosis. Regarding the relationship between the severity of disease (according to the ASRM classification) and exposure to organochlorinated compounds, we did not find any significant correlation between the pollutant serum concentrations and the different stages of the disease. The reason for the observed increase in PCB and p,p ′-DDE serum concentrations in patients with endometriosis remains to be elucidated. Because no correlation with dietary habits was found, such an increase could be associated with a different capacity in bio-activation and/or detoxication due to both genetic makeup and/or induction/inhibition phenomena in the tested population. Work is in progress on genetic polymorphisms of glutathione S -transferase and cytochrome P-450, enzymes involved in organochlorine compound biotransformation, to explore the gene–environment interactions as a possible cause of the observed higher levels of the aforesaid compounds in patients with endometriosis and as a risk factor for the disease onset/progression, as suggested by some studies ( Ertunc et al. 2005 ; Hsieh et al. 2004 ; Tsuchiya et al. 2007 ). In summary, our data show that exposure to PCBs and p,p ′-DDE represents a risk factor for endometriosis. In particular, the observation that ORs increase with increasing PCB concentrations strongly supports the hypothesis of an association between exposure to these chemicals and the disease, although the specific mechanisms of actions remain to be characterized.