Women with endometriosis have a higher DNA repair capacity and diminished breast cancer risk.

Breast cancer (BC) and endometriosis are important reproductive diseases for women. Although carcinogenic transformation of endometriosis is rare ( Thomas and Campbell, 2000 ), certain aspects of endometriosis mimic a malignancy because it can migrate, invade, and damage other tissues ( Thomas and Campbell, 2000 ). Most notably, the implantation of extra uterine endometrial cells is characterized by inflammation, hormonal-dependent and dysfunctional immune response, which are precursors for many chronic illnesses, including cancers ( Bulun, 2009 ; Higashiura et al., 2012 ; Ness and Modugno, 2006 ). Indeed, endometriosis has been associated with an increased risk for ovarian cancer, non-Hodgkin’s lymphoma ( Gemmill et al., 2010 ; Olson et al., 2002 ; Vercellini et al., 2012 ), cervical and certain epithelial ovarian cancers, especially clear cell and endometrioid adenocarcinomas ( Modugno et al., 2005 ; Munksgaard and Blaakaer, 2011 ). Because endometriosis and BC (a) are fairly common disorders, (b) share some of the same risk factors, and (c) the incidence of both ( Munksgaard and Blaakaer, 2011 ) are increasing annually, it is important to ascertain if one affects the other. Preliminary evidence suggests that up regulation of inflammatory and hormonal mediators are common threads between endometriosis and BC ( Ness and Modugno, 2006 ; Wallace et al., 2010 ). However, data on the association between endometriosis and BC risk have been inconsistent and inconclusive, although the majority of the studies support an association between the two ( Bertelsen et al., 2007 ; Brinton et al., 1997 ; Melin et al., 2007 ; Moseson et al., 1993 ; Munksgaard and Blaakaer, 2011 ; Schairer et al., 1997 ). The potential relationship between endometriosis and BC has important clinical implications. Because both are common conditions, if one causes even a slight increase in the risk for the other, it could be significant ( Munksgaard and Blaakaer, 2011 ). Thus, discerning an association between the two diseases is worth studying. Worldwide, the incidence of BC is increasing by 3.1% annually (1980 to 2010 statistics) ( Ferlay et al., 2010 ). BC is the most common malignancy affecting women, accounting for 23% of all female malignancies, which translates to 1.38 million new cancer cases diagnosed in 2008 ( Bray et al., 2013 ; Forouzanfar et al., 2011 ). The prevalence of endometriosis is approximately 10% ( Higashiura et al., 2012 ; Louis et al., 2011 ; Olson et al., 2002 ), and an estimated 176 million women suffer from Gynecological disease as Endometriosis in different stages in worldwide ( Adamson et al., 2010 ). If endometriosis plays a role in predisposing or preventing BC, elucidating this relationship could improve prevention, screening, and early detection programs. A hallmark of cancer is genomic instability caused by faulty DNA repair. Numerous studies have shown that lower DNA repair capacity correlates with higher cancer risk overall ( Athas et al., 1991 ; Benhamou and Sarasin, 2000 ; Helleday et al., 2008 ; Helzlsouer et al., 1996 ; Hu et al., 2007 ; Jyothish et al., 1998 ; Matta et al., 2003 ; Patel et al., 1997 ; Ramos et al., 2004 ; Wei et al., 2003 ; Wei et al., 1993a , 1994 ; Wei et al., 1996 ). Our large scale population studies of breast cancer have shown that many women exhibit a decreased repair phenotype: on average, 36–60% lower than women without breast cancer ( Matta et al., 2012 ; Ramos et al., 2004 ). However, whether DRC is associated in any way with endometriosis has not previously been explored—and only a few studies have evaluated the role of DRC as a susceptibility or risk factor for BC ( Kennedy et al., 2005 ; Matta et al., 2012 ; Ramos et al., 2004 ). The present study represents the first effort aimed at evaluating the association of endometriosis and BC, while concurrently exploring whether DRC plays a role in such a relationship.

Results were based on participants who had a self-reported diagnosis of endometriosis based on the question “Has a doctor ever diagnosed you with endometriosis”? Table 1 includes the age distribution of BC and Low DRC between women with and without endometriosis. Women with BC, low DRC and endometriosis were older than those without those attributes. After adjusting by age, BMI, family history of BC, menopause, alcohol use, smoking, multivitamin use, marital status, and saturated fat consumption, BC cases had 50% lower odds of having history of endometriosis (OR = 0.5 95%CI: 0.3, 0.9, p = 0.038) compared to controls, which was statistically significant ( Table 2 and Figure 1 ). Other findings, while interesting, did not reach statistical significance including: (a) Women with a history of endometriosis had a slightly higher DRC than those without endometriosis (despite being the same age as the BC cohort). (b) Those with history of endometriosis had 10% lower odds of having a low DRC. (c) BC cases with history of endometriosis had 20% lower odds of having a diagnosis of endometriosis confirmed by or prior to age 38 than controls with history of endometriosis ( Table 3 and Figure 2 ). The rest of the associations with the variables used in the logistic model were similar to those reported in the literature. BC cases had 1.9 times greater odds than controls to have family history of BC; 30% less odds of taking multivitamin supplements ( p < 0.05); and were less likely to be married than controls (i.e., BC cases were 2.8 more likely to be single women rather than married as compared to controls). These associations were statistically significant. Interesting findings that did not reach statistical significance included the following: BC cases 1) were less likely to drink alcohol, 2) were 50% less likely to smoke (borderline statistical significance, p = 0.073); 3) had 30% greater odds of being menopausal; 4) had a slightly higher BMI ( p > 0.05); and 5) almost two-fold more odds (OR = 1.9) of eating saturated fat food as compared to controls. However, this last finding was also not statistically significant. Women with family history of BC were 1.6 times more likely to have low DRC. Menopausal women were 30% more likely to have low DRC, but this association was not statistically significant. Alcohol users were 10% less likely to have low DRC (not statistically significant), and smokers were 50% more likely to have low DRC than non-smokers (not statistically significant after adjustment). Other non-statistically significant associations with low DRC included use of multivitamin supplements, saturated fat consumption, and BMI. Single women showed the lowest odds for low DRC as compared to married, divorced and widowed women. However, marital status also was not statistically significant. Participants with endometriosis had 10% less odds to have low DRC. This last association was not statistically significant.

This is the first time that a statistically significant association has been reported between endometriosis and BC in the Puerto Rican population. Our findings indicate that, after adjusting for the effect of age and other confounders, endometriosis may have an important protective effect against BC. Although the diagnosis of endometriosis was self-reported, participants were asked specifically if the diagnosis was done by a physician. Although no data on endometriosis treatment was obtained in this study, the reduced odds we observed among women with endometriosis may reflect their exposure to drugs with anti-estrogenic effects (e.g., oral contraceptives) which, in turn, potentially protect against BC as suggested before ( Bertelsen et al., 2007 ; Gemmill et al., 2010 ). Gemill’s study of 4,331 members of the Endometriosis Association is notable because all participants had surgically diagnosed endometriosis, and nearly two-thirds of those women reported having one or more of the following: cancers, infections, or endocrine diseases. BC was less common in this group than in the general population ( Gemmill et al., 2010 ). In Bertelsen’s (2007) large, 21-year case-cohort study, 236 cases of endometriosis were diagnosed among 16,983 women who subsequently developed BC, while 7.3 times that many women were diagnosed with endometriosis among the remaining 97,344 cancer-free women selected randomly from the general population. Bertelsen concluded, “We found that women who had endometriosis at a young age had a deficit of BC, while those who had the condition at an older age had an excess of BC” ( Bertelsen et al., 2007 ). Although the article did not specify that the diagnosis was surgically confirmed, the data were linked to the Danish Hospital Registry, which alludes to surgical confirmation. In this study, the average age at the time of diagnosis was 40.6 years (range: 16 – 60 years). Interestingly, two earlier studies had found no association between postmenopausal BC and endometriosis ( Baron et al., 2001 ; Olson et al., 2002 ). Other studies’ results conflict with those results (and ours). Brinton’s large retrospective cohort study using the Swedish Multi-Generation Registry linked to the National Swedish Inpatient Registry (1996) found that women with endometriosis had a 30% greater risk of BC than women without endometriosis ( Brinton et al., 1997 ). However, Brinton conducted a similar study in 2005 on 12,193 U.S. women and found that those with endometriosis had a weak, though not statistically significant, protective effect for BC ( Brinton et al., 2005 ). Using Swedish data sources, Melin, Sparén and Bergqvist (2007) found a slightly higher risk (8%) of BC among those with endometriosis as compared to those without it ( Melin et al., 2007 ). Matalliotakis et al. (2008) found an increased risk of BC among women with endometriosis who have a first- or second-degree family history of BC ( Mataltiotakis et al., 2008 ). Similarly, our study results indicate that those with family history of BC (in their mother, sisters, maternal aunts, and paternal aunts) had higher odds of having history of endometriosis (results not shown). However, because of our small sample size, some of those familial associations did not reach enough power to be studied. Such results allude to factors in addition to endometriosis that might confer a protective effect, which is why we concurrently analyzed DNA repair capacity along with endometriosis and BC. The inverse association we saw between endometriosis and breast cancer in women is intriguing. While one could argue that oral contraceptive therapy for endometriosis—not endometriosis itself—could contribute to a degree of reduced BC risk, this does not necessarily explain this potential association. For example, both OC and the chemotherapeutic tamoxifen inhibit estrogen binding; however, tamoxifen’s effects are opposite in breast vs endometrial epithelial cells. Tamoxifen therapy in BC patients increases the risk for endometrial cancer and endometriosis ( Bardi et al., 1994 ; Cohen et al., 1994 ; Friedrich et al., 1999 ). There are also important differences between these two tissues regarding progesterone action: In normal mammary tissue and breast cancer, progesterone is pro-proliferative and carcinogenic while in in the endometrium it antagonizes growth promoted by estrogen-driven ( Kim et al., 2013 ). Thus, it is biologically plausible that endometriosis exerts an independent or shared effect on BC risk. By studying DRC concurrently with endometriosis and BC in this study, we are taking early steps to determine independent and shared contributors to BC risk or protection. DRC can be considered as a potential independent risk factor for BC because it is biologically plausible, shows a strong association with BC, and (after adjusting for all confounders simultaneously) exhibits a dose-response relationship ( Matta et al., 2012 ). Globally, the prevalence of endometriosis is difficult to estimate, partially because selection bias exists in the populations that are used to make the estimate ( Kim and Adamson, 2008 ). Prevalence estimates range from 10 to 20% ( Bertelsen et al., 2007 ; Buck Louis et al., 2011 ; Higashiura et al., 2012 ; Mataltiotakis et al., 2008 ; Melin et al., 2007 ; Munksgaard and Blaakaer, 2011 ). But, due to the proportion of women with mild or asymptomatic endometriosis, and because surgical confirmation is necessary for the diagnosis, the true prevalence of the disease is unknown ( Ballweg, 2004 ). In Puerto Rico, the prevalence of endometriosis has been estimated at 5% ( Flores et al., 2008 ). However, it is probably much higher because vague symptoms, cultural and socioeconomic factors may deter or delay diagnosis ( Flores et al., 2008 ). In the sample we studied, the prevalence of endometriosis was around 8%, which is very similar to Ballweg’s estimated prevalence of endometriosis at 10% ( Ballweg, 2004 ). Pathological assessment is necessary after confirmation diagnosis of endometriosis requires surgery, which may not be available to all patients, due to cost and limited access to healthcare resources. Munksgaard et al. (2011) noted the same in their multicultural review of epidemiological data on endometriosis ( Munksgaard and Blaakaer, 2011 ). Although endometriosis is most frequently diagnosed when women are in their mid- to late 20s, ( Olive and Schwartz, 1993 ), any combination of those factors may have contributed to the older age at which our study population was diagnosed. This raises a question that is an ongoing research interest at the Ponce School of Medicine Endometriosis Research Program: is endometriosis a health disparity? Whether ethnicity plays a role in modifying the susceptibility of endometriosis is still unknown. Most genetic studies on endometriosis lack consideration of population stratification; others were conducted on a single population/race ( Montgomery et al., 2008 ; Nyholt et al., 2012 ; Tempfer et al., 2009 ; Treloar et al., 2007 ; Zhao et al., 2006 ; Zhao et al., 2008a ; Zhao et al., 2008b ). Some epidemiological data indicate differences in rates of endometriosis in different races or ethnicities. A low overall prevalence of endometriosis could partially be due to access to care, partly accounting for a high number of undiagnosed cases ( Flores et al., 2008 ; Missmer and Cramer, 2003 ). Data from the Nurses’ Health Study II showed a 40% lower incidence rate of endometriosis in African-Americans and Hispanics living in the U.S. compared to Caucasians (Odds Ratios for endometriosis with no past or concurrent infertility = 1.0 for Caucasians, 0.7 for African-Americans, 0.5 for Hispanics)( Missmer et al., 2004 ). If proven, endometriosis may be a “reverse health disparity”—which would be valuable in helping determine if there is a biologic and/or access to care basis in endometriosis. Endometriosis cases potentially could have been underreported in our study; however, we don’t think this is the case. It is unlikely that women will underreport a diagnosis that is done via surgery. Also, women diagnosed with endometriosis may be more aware of female reproductive issues, seek medical attention more often, and hence have a greater chance of being diagnosed with any cancer—including BC. Additionally, DRC declines with age, which increases risk of BC ( Grossman, 1992 ; Wei et al., 1993b ). However, in our study we did not find women with endometriosis to have an increased risk of BC. Remarkably, the opposite was true. Endometriosis and BC share common risk factors including endogenous estrogen, reproductive ramifications, and high BMI ( Munksgaard and Blaakaer, 2011 ). Those, as well as surgical interventions, such as oophorectomy, could be confounders to studies like ours and the others discussed in this paper. Our data analyzed and controlled those and other covariates. In addition, our results with respect to age, BMI, marital status, family history, menopause, alcohol use, etc., mirror what has already been published in the literature ( Hunter et al., 1996 ; Key et al., 2001 ; Melin et al., 2007 ; Olson et al., 2002 ; Prentice, 2000 ; Vlahos et al., 2010 ; Wittenberg et al., 2010 ; Wynder et al., 1997 ). Our study is the first to concurrently compare DRC in endometriosis and BC. We show here that women with endometriosis had a 10% less chance of having low DRC as compared to those without endometriosis, and those whose diagnosis of endometriosis was confirmed by age 38 had 40% lower odds of having a low DRC. A low DRC is an important risk factor for BC reported in the literature and for the same group of women that participated in this study ( Matta et al. 2012 ). However, because of the small sample power, these findings were not statistically significant. This study of endometriosis/BC/DRC has several limitations. No information regarding treatment specifically for endometriosis was collected, making more difficult to determine the impact of hormonal therapy on the association found. The endometriosis participants self-reported being surgically diagnosed with that condition by a physician. Because no further verification was utilized to verify diagnosis (increasing the random error in ascertaining the presence or absence of endometriosis), this could lead to an underestimation of the association. A larger sample of confirmed cases of endometriosis is necessary to ratify whether DRC plays a role in the protective association between endometriosis and BC. Despite these limitations, the study has significant strengths. This is the first study in which direct measurement of DRC was used as an indicator of the protective effect for endometriosis and to demonstrate and establish an association with breast cancer risk. All BC cases were histopathologically confirmed, and the epidemiological data was collected using the same epidemiological questionnaire that whose variables were recently published as part of a large-scale population study of BC in Puerto Rico ( Matta et al. 2012 ). Finally, this is an ongoing study; thus we will continue collecting information from women with and without endometriosis until we reach an adequate sample power to re-analyze the data.

The mechanisms behind endometriosis’s protective association for BC and the positive association between endometriosis and DRC are largely unknown. Treatment for confirmed or presumptive cases of endometriosis often includes drugs with anti-estrogenic effects ( Bertelsen et al., 2008 ; Olson et al., 2002 ; Vlahos et al., 2010 ), and it is known that estrogen depletion is associated with decreased risk for BC ( Vlahos et al., 2010 ). It is also possible that some of these drugs may also modify DRC, and consequently may indirectly influence the risk of BC. As we learn more about the molecular similarities between endometriosis and BC, that knowledge should arm us with more effective strategies to prevent, screen, and treat both conditions. Additional studies with a larger sample of endometriosis cases are warranted to confirm our findings and elucidate the molecular mechanisms underlying this association.

Data from an incident-case case-control study of 991 Puerto Rican resident women representing approximately 83% of the island’s municipalities (counties) were collected. A total of 385 cases with BC and 606 controls, prospectively recruited over 5 years (2006 to 2012), were included in the study sample (See Matta et al. 2012 for a full description of this sample). Among them, 80 participants (age mean, range) self-reported being surgically diagnosed with endometriosis, 20 of whom had a diagnosis of primary BC. Women with BC secondary to another type of cancer were excluded from the study. Crude and multiple logistic regression adjusted odds ratio were used to assess the association between BC and endometriosis after adjusting for potential confounders. This study included only patients with incident primary BC that was recently confirmed histopathologically. Additionally, those patients were treatment-naïve; they had not received chemotherapy, blood transfusions, or radiotherapy yet. Women with metastatic BC or BC secondary to another type of cancer were excluded from the study. Controls were recruited consecutively from the same medical offices, clinics and hospitals as the cases. Puerto Rico offers universal health insurance coverage; so, if the controls eventually were to develop BC, they would be treated in the same places where the patients were recruited. This selection procedure minimized selection bias that could have otherwise been a confounding factor (site, screening/treatment modalities) if healthy women had been recruited from the general population by other means (e.g., through random-digit dialing, as noted by Rothman et al. [30], ( Szklo and Nieto, 2007 ). Refusal to participate rates was minimal (3%) and did not significantly differ between cases and controls. This study population represents women obtaining routine gynecological and primary care screening services at the same medical offices where patients with BC were being treated. They were recruited consecutively from Ponce School of Medicine and Health Sciences Outpatient Clinic, Auxilio Mutuo Hospital (San Juan), Damas Hospital (Ponce), and St. Luke’s Hospital (Ponce), as well as Yauco and other selected collaborating cities throughout Puerto Rico, representing 65 (83%) of the 78 municipalities (counties) on the island. Age differences between BC cases and controls with and without self-reported endometriosis were not statistically significant. However, BC case group was older than controls. This age difference was statistically significant (data not shown). Because of this difference, all the analyses performed were adjusted by age. Thus, we eliminated the possibility that any protective effect could be solely related to age. Data and peripheral blood samples were collected during a face-to-face self-report interview using a standardized questionnaire administered by an experienced research RN. Pathology reports were obtained from medical records to confirm diagnoses; blood samples were promptly processed and frozen by trained technicians in our laboratory for the DRC assays described below. Each study participant’s DRC level was measured by drawing a sample of venous blood and assaying its lymphocytes using the host-cell reactivation assay (HCR), as published by Matta et al. 2012 . This assay has also been described in previously published molecular epidemiological studies of cancer ( Cheng et al., 1998 ; Hall et al., 1994 ; Landi et al., 2002 ; Matta et al., 2003 ; Ramos et al., 2004 ; Spitz et al., 2001 ; Wei et al., 1993a ). The results of this assay reflect the host cells’ overall repair capacity, HCR primarily detects activity of the nucleotide excision repair (NER) pathway, which is known to be decreased in BC ( Qiao et al., 2002 ). Previously frozen peripheral blood lymphocytes from patients and women without BC were assayed in batches, as described by Ramos et al. [11]. Peripheral blood lymphocytes with > 95% viability were incubated for 72 h with phytohemagglutinin and then were transfected with undamaged or damaged plasmid DNA. Cells isolated from xeroderma pigmentosum patients corresponding to complementation groups C and D (XPC, XPD) were used as positive controls (cell lines GM 02246D and GM 02253F, respectively; Coriel Institute Medical Research; Camden, NJ). To confirm achievement of stable transfection, we utilized the Dual-GloR® Luciferase Assay System (Promega; Madison, WI), which uses the Firefly and Renilla luciferases proteins as co-reporters. The Dual-Glo assay enables us to use mammalian cells (e.g., lymphocytes) containing genes for Firefly and Renilla luciferases as positive controls. The issue of transfection efficiencies with the host-cell reactivation assay (HCR) has been in the literature for more than 15 years ( Cheng et al., 1998 ; Matta et al., 2012 ) and has created some confusion. Unlike the variability in transfection efficiencies that were characteristic of the “old” assay with the CAT plasmid that utilized radioactivity, this “new” HCR assay is based on a non-replicating plasmid expression vector (pCMVluc) of 4,863 base pairs containing LUC (the bacterial luciferase reporter gene)( Qiao et al., 2002 ). The LUC assay is faster, requires fewer steps, is not radioactive, requires half as many cells as the old assay, and is not affected by cell storage time, baseline expression of undamaged plasmids or expression of damaged plasmids. The PI was trained in, and has been using the LUC assay since 1999. The assay for the gene expression of luciferase activity was quantified using a luminometer (Turner Designs, model TD-20/20, Sunnyvale, CA). DRC was calculated as the percentage of luciferase activity present after repair of damaged plasmid DNA, compared to the DRC of undamaged plasmid DNA (100%). Resulting DRC values ranged from nearly 0 to 19% DRC. All samples were run in triplicate. If discrepancies are perceived in individual results, the assay was repeated. To determine whether our results would vary significantly between cryopreserved versus fresh blood samples, we took dual samples from 5 patients (total of 10 blood samples) and assayed 5 immediately after phlebotomy (fresh), then cryopreserved the other 5 samples at −80°C and analyzed those several weeks later. No statistical difference was seen in the results. All data were entered into the SPSS® Statistics 17 Multilanguage package, following a standardized procedure suited for data verification and correction of gaps and errors during data entry. DRC levels were dichotomized to “high” and “low,” using the median (3.185) as the cutoff between “high” and “low” for the population studied. All participants with high DRC were compared to those with low DRC regarding history of endometriosis and mean age when endometriosis was diagnosed. Crude and multiple logistic regression adjusted odds ratios were used to assess the associations between BC and endometriosis, as well as between low DRC and endometriosis. Because women with BC, endometriosis, or low DRC were older than those who were cancer-free or had high DRC or endometriosis ( Table 1 ), age was always adjusted in the logistic model. Data were analyzed also adjusting for body mass index (BMI), family history of BC, menopause, alcohol use, smoking, multivitamin use, marital status, and saturated fat consumption. A P value of 0.05 was set as the cutoff for statistical significance in this study.