Endometriosis-associated Malignancy

Abstract

Endometriosis is a common condition in women of reproductive age. According to several epidemiological studies endometriosis may be associated with increased risk of various malignancies. However, endometriosis-associated malignancy (EAM) is defined by certain histological criteria. About 80 % of EAM have been found in the ovary, whereas 20 % are localized in extragonadal sites like intestine, rectovaginal septum, abdominal wall, pleura and others. Some authors suggest that EAM arise from atypical endometriosis as an intermediate lesion between endometriosis and cancer. Moreover, a number of genetic alterations, like loss of heterozygosity (LOH), PTEN, ARID1 A and p53 mutations have been found in both endometriosis and EAM. Endometriosis-associated ovarian cancer (EAOC) is mostly a well or intermediately differentiated tumor of endometrioid or clear cell histological sub-type. Women affected by EAOC are on average five to ten years younger than non-EAOC patients; in most of the cases EAOC is a low stage disease with favorable clinical outcome. Since EAM is a rare condition systematic data on EAM are still missing. A systematic retrospective study on endometriosis-associated malignancies (EAM study) is currently being conducted by the Endometriosis Research Foundation together with the study groups on ovarian and uterine tumors of the working group for gynecological oncology (AGO) (gyn@mlk-berlin.de).

Abbreviations

CCC

Clear cell carcinoma

DFS

Disease-free survival

EAM

Endometriosis-associated malignancy

EAOC

Endometriosis-associated ovarian cancer

HR

Hazard ratio

LOH

Loss of heterozygosity

OC

Ovarian cancer

OR

Odds ratio

OS

Overall survival

PTEN

Phosphatase and tensin homolog

RR

Relative risk

SIR

Standardized incidence ratio

Introduction

Endometriosis is a common condition in women in which endometrium-like tissue is detected outside the uterine cavity 1. The incidence of the disease in females of reproductive age is estimated at 5 to 15 % 2; the main symptoms are dysmenorrhea, pelvic pain and infertility 3. Despite extensive research in this field, the exact pathogenesis of endometriosis remains not fully understood but one widely accepted hypothesis is the implantation of endometrial tissue in the peritoneal cavity due to retrograde menstruation 4. Although endometriosis is considered a benign condition, it has been demonstrated to share some characteristics with malignant tumors such as tissue invasion and damage, neoangiogenesis or spread to distant organs 5. Current knowledge about endometriosis-associated malignancies will be presented in this review.

Endometriosis and Cancer Risk – Epidemiological Data

According to several epidemiological studies endometriosis is associated with increased risk of various malignancies with the best evidence for ovarian cancer 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18. Furthermore, a number of cohort and case control studies demonstrated an epidemiological association of endometriosis with endometrial cancer, breast cancer, colorectal cancer, non-Hodgkin lymphoma and others 8, 19, 20. The clinical significance of the association between endometriosis and non-gynecologic malignancies is poorly understood 21.

The majority of the published studies on endometriosis-associated ovarian cancer have reported that ovarian cancer (OC) risk among endometriosis patients is moderately increased (RR, SIR or OR 1.32–1.92) (Table 1). The strongest association between OC and endometriosis was reported by Brinton et al. (SIR = 2.48) 8, Stewart et al. 22 (HR = 2.33) and Buis et al. 23 (HR = 12.4); interestingly, these three studies included only women suffering from infertility, which may be a potential confounding factor for their results. Moreover, a remarkably high increase of ovarian cancer risk (SIR = 8.95) has been demonstrated by Kobayashi et al. 24. In their retrospective cohort study ovarian cancer risk was analysed in women with sonographically diagnosed endometriomas based on time periods from first diagnosis of endometrioma to first diagnosis of ovarian cancer. Since endometriomas were histologically confirmed in only one-third of the cases, whereas the majority of diagnoses was made based on ultrasound image, the ovarian cancer risk has been possibly overestimated in this analysis: some of ovarian cysts identified as endometriomas might have been in fact ovarian cancers. In summary, there are a number of epidemiological studies, in which at least a modest association between endometriosis and ovarian cancer has been consistently demonstrated. Despite epidemiological evidence, a direct causal association between endometriosis and ovarian cancer has not been elucidated so far 25.

Table 1 Epidemiological studies on association between endometriosis and ovarian cancer.

Author	Type of study	Number of endometriosis cases	Number of cancer cases	Number of controls	OR, SIR, HR or RR (95 % CI)
CG: control group, RR: relative risk, OR: odds ratio, HR: hazard ratio, SIR: standardized incidence ratio
Brinton 1997 8	cohort	20 686	29		1.92 (1.3–2.8)
Ness 2000 10	case control	66 (85 CG)	767	1 367	1.7 (1.2–2.4)
Ness 2002 11	case control	51 (39 CG)	5 207	7 705	1.73 (1.10–2.71)
Olson 2002 12	cohort	1 392	3		0.78 (0.25–2.44)
Brinton 2004 13	cohort	12 193	45		2.48 (1.3–4.2)
Borgfeldt and Andolf 2004 14	nested case control	28 163	81	84 489	1.34 (1.03–1.75)
Modugno 2004 15	case control	177 (184 CG)	2 098	2 953	1.32 (1.06–1.65)
Melin 2006 16	cohort	64 992	122		1.43 (1.19–1.71)
Melin 2007 17	cohort	63 630	134		1.37 (1.14–1.62)
Kobayashi 2007 24	cohort	6 398	46		8.95 (4.12–15.3)
Pearce 2012 18	case control	738 (818 CG)	7 911	13 226	1.49 (1.24–1.65)
Stewart 2012 22	cohort	2 978	38		2.33 (1.02–5.35)
Buis 2013 23	cohort	3 657	34		12.4 (2.8–54.2)
Kok 2015 20	cohort	2 266	13		4.56 (1.72–12.11)

Histological Findings

The histological link between endometriosis and cancer has first been postulated as early as 1925 26. Sampson et al., based on microscopic observations, speculated that endometrial ovarian cancer may develop from endometriotic tissue and described the criteria for diagnosis of endometriosis-associated OC: (1) evidence of endometriosis close to the tumor, (2) exclusion of invasion from other sources, (3) presence of tissue resembling endometrial stoma surrounding characteristic epithelial glands 26. Scott et al., in their analysis from 1953, added a fourth criterion: (4) histological proof of transition from benign changes in endometriosis to malignant changes in cancer 27; all four criteria are still in use to define an endometriosis-associated malignancy (EAM).

About 80 % of EAMs are localized in the ovary, whereas extragonadal sites are affected in one-fourth to one-fifth of all cases 9, 28. In two large reviews of endometriosis-related neoplasms (222 and 205 cases, respectively), extragonadal tumors were reported in 21–24 % of women 29, 30. Since the malignant transformation can take place in every site affected by endometriosis, extragonadal EAMs can be found in the lower pelvis, gastrointestinal tract, abdominal wall, umbilicus, pleura and others 31. The most common localisations of extragonadal EAMs are the rectosigmoid, colon, rectovaginal septum and pelvic peritoneum; these sites are commonly involved by deep infiltrating endometriosis 31.

Based on histopathology and molecular features, ovarian cancers are divided into five main categories: high-grade serous, endometrioid, clear cell, mucinous, and low-grade serous carcinomas 32. The most common subtype is the high-grade serous cancer (70 %), followed by the endometrioid (10 %) and clear cell (10 %) type. Endometrioid and clear cell carcinomas usually present as low-stage disease; the former is typically well differentiated 33. Endometriosis-associated malignancies are most commonly cancers of these two histologic subtypes, rarely other ovarian malignancy types like borderline tumors, endometrial stromal sarcoma or adenosarcoma 18, 25, 28, 31. Recently published pooled analysis of case-controlled studies by Pearce et al. reported an increased risk for low-grade serous ovarian cancer among endometriosis patients as well (RR 2.11; 95 % CI 1.39–3.20, p < 0.0001) 18. No association between endometriosis and high-grade serous ovarian cancer or mucinous ovarian cancer has been reported to date 18, 28.

Theories of Pathogenesis

There are two main theories which attempt to explain the origin of endometriosis-associated malignancies. According to a number of histopathological studies endometriosis-associated ovarian cancer (EAOC) may arise from atypical endometriosis of the ovary 34, 35. This heterogeneous condition is histologically characterized by hyperplasia of endometrial glands with cytological atypia or presence of atypical hobnail cells within ovarian endometriosis 25, 36, 37. A direct association between atypical endometriosis and ovarian cancer was first demonstrated by La Grenade and Silverberg in 1988 34; in five cases of ovarian cancer (three clear cell carcinomas and two endometrioid carcinomas) accompanying atypical endometriosis was found in the ovary and in four out of five cases atypical changes were in contiguity with ovarian neoplasm 34. Furthermore, a significantly higher rate of cytological atypia within endometriotic lesion was demonstrated in cases of EAOC compared with endometriosis alone 38. Although the hypothesis that atypical endometriosis may be a premalignant condition is supported by the fact that it can be detected in up to 80 % of EAOC, there are still not enough data proving this model 35, 39, 40.

According to the other theory endometriosis is not a real precancerous lesion, but there is an indirect link involving common environmental, histological, immunological and genetic factors. Several studies have demonstrated that microenvironment of endometriosis and EAOC share similar mediators and cytokines 40, 41. Whether EAM arise by malignant transformation of endometriotic cells through intermediary lesions (atypical endometriosis) or similarity of microenviroment represents a link between these both entities has not been definitely understood 40, 42.

Genetic Alterations

In the last two decades, evidence has been accumulated that endometriosis-associated malignancies may arise within ovarian endometriosis. Possibly, periodic hemorrhage in ovarian endometriomas may lead to iron-triggered oxidative stress and thus induce genetic alterations. Yamaguchi et al. reported that the concentration of free iron in endometriotic cysts was over one hundred times higher than that in nonendometriotic cysts; the level of oxidative stress related markers, such as lactose dehydrogenase, and antioxidants was higher as well 43.

Accumulation of mutations in tumor suppressor genes and oncogenes is a crucial step during tumor development. Interestingly, numerous mutations in genes linked to carcinogenesis have been identified in endometriotic lesions 44, 45, 46, 47, 48, 49, 50, 51, 52. Several studies have reported genetic aberrations in tumor suppression genes, such as p53 and PTEN, in endometriosis specimens 53, 54, 55. Inactivation of DNA mismatch repair genes through hypermethylation has been observed in endometriotic tissue as well 53. Furthermore, genetic alterations associated with tumorigenesis are encountered more frequently in benign endometriosis samples from patients with synchronous ovarian cancer than in tumor-free patients who were diagnosed with endometriosis alone. Prowse et al. performed microsatellite analysis in ten patients with ovarian cancer and coexisting endometriosis; the tumor and endometriosis samples were analysed for common molecular genetic alterations 56. 63 events of loss of heterozygosity (LOH) were detected in cancer specimens; one-third of these was also identified in the corresponding endometriosis samples, suggesting that endometriosis might be a clonal precursor to a subtype of ovarian cancer. Hypothetically, endometriosis-associated and endometriosis-independent ovarian cancer may develop through different molecular pathways with distinct genetic alterations (Table 2).

Table 2 Genetic alterations associated with endometriosis and endometriosis-related malignancies.

Factor	Genetic alteration	Current data
Oxidative stress	8-OHdG	8-Oxo-2′-deoxyguanosine is a marker of oxidative DNA damage. Endometriosis-related ovarian cancer show significantly stronger staining of 8-OHdG than ovarian cancers not accompanied by endometriosis 43. Endometriotic cysts and atypical endometriosis also stain positive for this marker. 8-OHdG seems to play a role in pathogenesis of ovarian cancer and was linked to poor prognosis 44.
Tumor suppressor genes	PTEN	Phosphatase and tensin homolog is mutated in many cancer entities, particularly in endometrial and endometrioid ovarian cancer; its inactivation occurs early during tumorigenesis 53. PTEN somatic mutations are frequently found in endometriotic cysts 45.
	p53	As a negative cell-cycle regulator, p53 is involved in tumorigenesis of different malignancies. Several studies showed no expression in benign endometriosis but high expression in benign endometriotic lesions next to the endometrioid or clear cell carcinoma 54.
	ARID1A	ARID1A mutations are significantly more common in two ovarian cancer subtypes associated with endometriosis (clear-cell and endometrioid). In case of endometriosis synchronous with ovarian cancer, mutation was more frequent in clones derived from endometriosis samples directly adjacent of the tumor than in those from distant endometriotic lesions 46.
DNA repair	hMLH1	hMLH1 corrects errors in DNA replication; hypermethylation of its promoter occurs early in endometrial malignant transformation and can be identified in 10 % of typical and in 33 % of atypical endometrial hyperplasias 47. Abnormal methylation can be observed in endometriosis as well 53.
Oncogene	Bcl-2	Expression of this anti-apoptotic protein is significantly higher in endometriosis accompanying cancer (42–73 %) than in benign endometriosis (23 %) 54, suggesting its role in the early steps of tumorigenesis.
	KRAS	KRAS mutations are significantly more common in endometriosis-associated endometrioid adenocarcinomas (29 %) than in tumors not associated with endometriosis (3 %) 48.
Chromosomic aberrations	Aneuploidy	Aneuploidic frequency seems higher in endometriosis specimen from patients with advanced endometriosis when compared to the background frequency observed in normal specimens, particularly with regard to chromosome 17, on which tumor suppressor gene p53 is located 49, 50.
Loss of heterozygosity (LOH)		A trend of increasing LOH frequencies has been described between solitary endometriosis lesions, endometriosis-associated carcinoma and endometrioid ovarian cancer, respectively 51, 52. Common LOH events can be identified in endometriosis synchronous with ovarian cancer 45, 56.

Risk Factors

Several risk factors for EAM in patients with endometriosis have been reported to date. The size of endometrioma as well as postmenopausal status were demonstrated to be independent predictive factors for the development of ovarian cancer among endometriosis patients in the study by Kobayashi et al. 57. In this prospective cohort trial which assessed the risk of ovarian cancer in 6398 women with ovarian endometriomas, tumor size ≥ 9 cm in diameter was shown to be associated with increased OC risk. However, as mentioned before, one of the limitations of this study is that endometriomas were diagnosed mostly by sonography, and therefore some of the larger tumors might in fact have been ovarian cancers 57. Another risk factor for EAM is hyperestrogenism, both endo- and exogenous: in the study by Zanetta et al. obesity as well as therapy with unopposed estrogens after hysterectomy were shown to be a significant risk factor for the development of EAM 58. Similar effects of estrogen were demonstrated by others 40. On the other hand, hormonal contraception, childbearing, tubal ligation or hysterectomy were found to reduce the ovarian cancer risk among patients with endometriosis 15, 59.

Clinical Implications

EAOC is characterized by an early onset of the disease: Aris et al. reported in their retrospective cohort trial that the mean age of women with EAOC was 48.3 ± 10.8, on average 5.5 years lower than in non EAOC patients (p = 0.003) 60. In a retrospective study by Orezzoli et al. patients with endometriosis-associated clear cell cancer (CCC) of the ovary were even 10 years younger compared with those with non-endometriosis-associated CCC (95 % CI 0.6–18 years; p < 0.05) 61.

Interestingly, EAOC is commonly a low-stage and low-grade disease usually without ascites at initial presentation 62, 63. Furthermore, several authors reported that EAOC is associated with significantly better prognosis (DFS and OS) compared with non-EAOC, suggesting that EAOC represents a biologically distinct entity 28, 61, 64, 65. However, these findings have not been confirmed by others: after controlling for stage, age, grade and treatment no difference in overall survival between these two patient groups could be demonstrated 62, 66, 67; thus, better clinical outcome of EAOC might be explained by a high rate of well-differentiated early-stage tumors rather than by an association with endometriosis per se 62, 63, 67.

Based on existing data on incidence and risk factors of EAM, the possibility of malignant transformation should be included in diagnostic considerations for patients with endometriosis 21, especially in postmenopausal women who present a sudden recurrence of symptoms. Because of the malignant potential, endometriosis patients should, if indicated, receive a combined estrogen-progestin therapy (HRT, hormone replacement therapy) or tibolone even after hysterectomy; unopposed estrogens should generally be avoided in these patients 21, 68, 69. Clinicians should be aware of the increased risk of specific subtypes of ovarian carcinoma in endometriosis patients. A modest increase in cancer risk should be discussed with these patients 70. The risk of ovarian cancer is highest in women with endometriosis and primary infertility 13. However, there are currently no diagnostic options to predict the risk of malignant transformation in an individual patient. Future studies should aim at understanding the mechanisms underlying this phenomenon so as to identify patients who are most at risk for developing endometriosis-related malignancy.

Since EAM are mostly low-grade tumors, there is the question of the efficacy of chemotherapy in EAOC patients 29. However, in the study by Davis et al. no difference in response to chemotherapy in EAOC patients compared with controls with papillary serous ovarian cancer has been found 62. Due to the lack of appropriate data postoperative treatment of EAOC follows the standard chemotherapy guidelines for ovarian cancer 71. In case of extragonadal EAM, especially of rectum or rectovaginal septum, however, surgical resection followed by radiotherapy may be the treatment of choice 29, 31, 65.

Several epidemiological studies, case reports and case series on EAM have been published in the literature to date. However, a systematic analysis of a large population of patients with EAM is still missing. Thus, a systematic retrospective study on endometriosis-associated malignancies (EAM study) is currently being conducted by the Endometriosis Research Foundation together with the study groups on ovarian and uterine tumors of the working group for gynecological oncology (AGO). Histopathological second opinion by pathology reference laboratory is the core component of this study (for more information: gyn@mlk-berlin.de).