Atypical endometriosis has been observed in 12–35% of ovarian endometriosis, and it is estimated that around 60% to 80% of all EAOC occur in the presence of AE, often found in direct continuity with the tumor [55]. Evidence suggests that AE could be a transitioning entity from benign lesions to malignant variants. The pre-malignant, “atypical” lesions, are defined by several histologic criteria, including large nuclei with moderate to marked pleomorphism, increased nuclear-to-cytoplasmic ratio, cellular crowding, stratification, or tufting [55,56]. It has been reported that AE actually refers to two different histologic findings: cellular atypia, also known as cytologic atypia, and architectural atypia, commonly referred to as hyperplasia. The diagnosis of endometriosis with architectural atypia is important because patients with hyperplasia-type endometriosis may be at a higher risk of developing EAOC [49]. This variability in the incidence of the disease may be due to a difficult histological diagnosis, which is still far from being standardized among medical centers worldwide. Moreover, the new conceptualization of the histological pattern, and the differences that emerged in prognostic significance between cytological AE and hyperplastic AE should strongly encourage a revision of this classification in order to better understand which type of AE is actually to define a “high-risk” disease. Reflections can be made on the meaning of a “preneoplastic lesion”: determining whether the mere presence of endometrium at ectopic sites should be considered “per se” a premalignant condition seems crucial and constitutes the conceptual base of any strategy aimed at reducing EAOC mortality. Lesions are defined as “precancerous” based on definite epidemiologic, morphologic, molecular, and biologic criteria that imply the acquisition of genetic, karyotypic, structural, or functional changes in a cluster of cells that differentiate them from the surrounding normal tissue [57]. In other words, premalignant lesions such as AE should reflect an intermediate stage along the pathway leading to cancer. When enough genetic changes have occurred, modifications in appearance and function are observed but not yet associated with typical malignant behavior. In recent years, the literature has focused on the relationship between endometriosis and ovarian cancer. Indeed, sequencing and immunohistochemical studies have revealed that mutations found in endometriosis-associated cancers are found in adjacent endometriosis, supporting the theory of a clonal relationship between benign and malignant counterparts and confirming that the cancers have arisen from the endometriotic lesions probably through an “intermediate” premalignant step. In addition, gene encoding b-catenin (CTNNB1) mutations in 16–53.3%, PTEN mutations in 14–20%, and ARID1A mutations in 30–55% of cases were found in EnOC. PIK3CA mutations in 20–40% and ARID1A mutations in 46–57% of cases are found in ovarian CCC [58].

Unfortunately, in regards to the results of our systematic review, none of the above discussed biomarkers has such strong evidence that could justify their systematic clinical use in the management of endometriosis and AE. The most frequently detected mutations in AE are ARID1A, genes involved in PI3K pathway (i.e., PIK3CA), genes encoding for ER and PR, KRAS, and PTEN. Interestingly, somatic driven mutations in KRAS, PTEN, PIK3CA and ARID1A have been also observed in more than 26% of cases of deep infiltrating endometriosis lesions, which are associated with virtually no risk of malignant transformation [59]. Therefore, a specific role seems to be played by the ovarian microenvironment in increasing the risk of malignant derailment [60]. Karnezis concurs that the ovarian microenvironment seems to be essential for the malignant transformation of endometriosis because many endometriotic lesions are located outside the ovary, including the pelvic peritoneum, but carcinomas at such sites are rare [61]. Indeed, it was proposed for the endometrioma’s neoplastic transformation as a hypothetical model called the “two-hits” hypothesis [9]. Reactive oxygen species due to free heme and catalytic iron contained in the trapped blood in endometriomas may lead to increased oxidative stress and DNA damage in the epithelial layer of endometriomas. This may result in mutations and epigenetic changes, including mutations in the tumor suppressor gene ARID1A leading to AE. Possible second-hit mutations, as well as activation of the PI3K-AKT-mTOR pathway allow the mutated cell to escape apoptosis caused by increased oxidative stress. The accumulation of oncogenic mutations in AE may ultimately lead to the development of endometriosis-associated ovarian CCC and endometrioid carcinomas. ARID1A loss and activation of PI3K/AKT functionally cooperate in ovarian carcinogenesis and suggest that ARID1A-loss occurs early and that it may be “addicted” to PI3K/AKT oncogenic signaling [9]. In line with this hypothesis and with the reduction of the hormone dependency of CCC, oxidative stress has been shown to act as a physiological regulator of estrogen receptors. Contrary to CCC, EnOC is generally estrogen sensitive and associated with hormonal risk factors. ERa has been shown to represent an independent prognostic marker for EnOC, while nuclear ERa is barely detectable in CCC [62]. Inactivating ARID1A mutations are the most common molecular genetic alteration reported thus far in CCC and EnOC, but a higher frequency of ARID1A mutations has been detected in CCC (46–57%) compared with EnOC (30%). The fact that no differences in clinical behavior were observed comparing BAF250a-positive versus BAF250a-negative cancers may be the basis for supporting the idea of a marker for genomic instability without driving the phenotype: BAF250a appears to be an early event in most of these cases [46,63]. Based on results from the systematic review, there is a remarkable association between BAF250a loss in cases of AE and TE contiguous to BAF250a-deficient EAOC with a higher frequency of inactivating ARID1A mutations detected in CCC compared to EnOC [11,33,34,37,43,44,46,49]. Concordantly with these findings, also PI3K/Akt/mTOR molecular pathway seems to be altered in AE and in EAOC in all of the studies included, with a similar pattern between the two diseases [25,28,31,32,34,37,40,42,43,45,48,50]. Since ovarian cancer is considered a hormone-responsive cancer, its correlation to PR and ER immunoexpression has a major importance in clinic-pathological manifestations of ovarian carcinoma, including those associated with endometriosis [64,65]. Actions of estradiol are mediated by two isoforms of ERs (ERa and ERb) that differ in their tissue distributions, their ligand-binding specificity, and affinity: ERa has been shown to represent an independent prognostic marker for EnOC, while nuclear ERa was poorly detectable in CCC [62,66]. Since progesterone is modulated by the expression of both isoforms of the specific receptor (PR-A and PR-B), it is involved in the pathogenesis of endometriosis and EAOC [65]. Thus, from the results of the systematic review, it seems that alteration of steroid receptor immune expression is correlated to ovarian endometriosis and endometriosis-related carcinogenesis in a hormone-dependent manner with regards to EnOC and in a hormone-independent way concerning CCC.

Associated AE lesions seem to have the same biological expression as their adjacent Cancer histotypes [15,67,68,69,70,71,72,73,74,75]. Moreover, ER and PR expression seems to be higher in AE as compared to EAOC, and lower when compared to endometriosis. The gradual loss of ER and PR expression from endometriosis to EAOC carcinogenesis suggests that hormone receptor staining may be proposed as a marker for premalignant or malignant lesions in endometriosis. A reduced HNF 1-beta expression in AE as compared to EAOC, and in particular to CCC, and a higher HNF-1 beta expression in AE as compared to TE has also been reported. The main limit to the use of these molecular markers proposed for the early detection of endometriosis-preneoplastic lesions is that their clinical application may be difficult since they all need expensive molecular analyses. Moreover, strong evidence supporting their systematic use in clinical practice is still lacking. The presence of a serum non-invasive marker for the presence of AE could be more effective and easy to use. Inflammatory parameters, such as the neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR), have been found altered in the peripheral blood in patients with endometriosis, thus suggesting that systemic perturbations may contribute to the pathogenetic process of the disease [76]. NLR and PLR have also gained more and more space in the diagnostic and prognostic management of ovarian carcinoma [77]. Indeed, systemic inflammation contributes to cancer initiation and progression by promoting cell proliferation, angiogenesis, and gene repair [78]. To the best of our knowledge, there are no studies in the literature evaluating the possible role of NLR and PLR in predicting AE and its risk of malignant transformation. This is even more important if we think that, according to the literature, women with endometriosis had a tripled and doubled risk for CCC and EnOC subtypes, and a more frequent localized form of the disease when cancer arises in endometriosis [79]. Even if cancer arising in endometriosis seems to be characterized by a better prognosis, the early detection of preneoplastic lesions could really impact the quality of life of women with endometriosis. Our study has some limitations: (1) The non-homogeneous definition of AE in the studies included in the systematic review; (2) Heterogeneity between studies (i.e., study design) included in the systematic review; and (3) Low number of studies included in the systematic review. These limitations suggest that the results should be interpreted with caution until validated by future research projects providing more detailed, well-designed, and standardized data collection.