Origins of ovarian cancer: new insights

Keywords::

• borderline
• Fallopian tube
• high grade
• histology
• low grade
• molecular genetics
• ovarian cancer
• pathogenesis
• type I
• type II

In the USA, epithelial ovarian carcinoma (EOC) accounts for more deaths than all other gynecologic cancers combined. The high mortality to incidence ratio can be accounted for largely by the fact that, among common malignancies, EOC remains unusually enigmatic with respect to its early natural history, thus defying potential early detection strategies. This state of uncertainty applies to its histologic region of origin in addition to the existence of a premalignant or precursor lesion to EOC. What has become indisputable, however, is that EOC is an unusually heterogenous malignancy. The purpose of this editorial is to summarize the current theories related to EOC pathogenesis and to propose a model consistent with the published literature.

Historically, the notion that the epithelial component of the ovary gives rise to the four common histologic variants of EOC gained wide acceptance, with the major debate related to whether tumorigenesis occurs in the single-cell layer of the surface epithelium or architectural variations such as cystic inclusions [1]. However, as suggested by Dubeau, this theory is problematic from the perspective of embryologic development, in that epithelial ovarian tumors develop from serous, endometrioid, clear cell or mucinous histologies, like their gynecologic counterparts of Müllerian origin; the Fallopian tubes, uterus and upper vagina [2]. However, the ovary is not derived from the Müllerian duct; it develops separately on the urogenital ridge, with a surface epithelium that is actually a modified mesothelium [3], contiguous with, and morphologically resembling, the peritoneal mesothelial lining. Although metaplasia during malignant transformation is possible, the question persists as to whether EOC actually arises from the ovary, or perhaps from Müllerian remnants (e.g., rete ovarii) or from lesions such as endometriosis. This hypothesis is supported through the recognition that a proportion of endometrioid and clear-cell EOCs are associated with, and perhaps arise from, endometriosis [4,5].

Speculation regarding this potential shared pathophysiology is reinforced by a growing literature indicating that the four histologic types of EOC are distinct pathologic entities at the molecular genetic level. Recently, the ARID1A gene, encoding a key component of the SWI-SNF chromatin remodeling complex, was found to be a tumor-suppressor gene mutated in approximately half of all clear-cell EOCs [6,7] and in approximately one-third of endometrioid EOCs [6]. In two patients, the same ARID1A mutations were evident in tumors and contiguous atypical endometriosis [6]. No mutations of ARID1A were identified in high-grade serous EOCs. However, mutations of TP53 are present in essentially all high-grade serous EOCs [8], but are rare in other histologic types. Similarly, mutations in KRAS are confined largely to mucinous EOCs [9], and endometrioid EOCs are characterized by mutations in genes encoding components of the Wnt signaling pathway, including CTNNB1, PTEN and PIK3CA[10], as well as ARID1A[6]. From the perspective of gene-expression profiling, serous and endometrioid subtypes have strongly distinct expression profiles that correlate well with their uterine counterparts, whereas clear-cell tumors could not be distinguished from other EOCs or from clear-cell cancers of the kidney [11].

In addition to the well-recognized molecular and biological heterogeneity among ovarian cancer histologic variants, a more sophisticated model is emerging for the classification of EOCs based on histology and tumor grade. Kurman and colleagues suggest that the ‘type I’ tumor category consists of low-grade serous, low-grade endometrioid, clear cell, mucinous and transitional (Brenner) carcinomas [12]. They are suggested to arise from corresponding benign cystic epithelium, often through a borderline tumor (tumor of low malignant potential), supporting the classical paradigm of stepwise morphologic progression during tumorigenesis. Type I tumors are generally indolent, of early stage at diagnosis and genetically stable. Mutations of the TP53 gene are absent, but as discussed earlier, there exists a distinct molecular genetic profile between the various histologic types. Ironically, these tumors are also relatively chemoresistant, and thus prognosis is ultimately poor.

In contrast, ‘type II’ tumors are aggressive, generally of advanced stage and include high-grade serous (and other high-grade histologic type) carcinomas, undifferentiated carcinomas and carcinosarcomas [12]. As discussed previously, TP53 mutations are very common, if not ubiquitous, in type II tumors. These tumors are generally aneuploid and characterized by chromosomal instability. They tend to be diagnosed at stage III/IV, and although initially the majority are highly sensitive to standard platinum-based combination chemotherapy, the great majority recur and are ultimately fatal. There are now numerous drugs, however, that are effective in the salvage setting, to the extent that the median survival for advanced type II cancers is now 4–5 years. In contrast to type I tumors, a well-accepted premalignant lesion associated with this tumor has not been identified. This ‘type I/II’ classification is not dissimilar to that now widely accepted for endometrial carcinoma [13], and clearly suggests that different types of EOC develop along different molecular pathways.

Returning to the more obscure aspects of the early natural history of EOC, primarily relevant to type II tumors (as defined previously), an emerging theory posits that high-grade serous carcinomas actually originate in the fimbria of the Fallopian tube. This hypothesis was perhaps first developed by Piek et al. after observing a high incidence of dysplastic morphologic and preneoplastic molecular expression alterations in Fallopian tubes from women with BRCA-linked genetic predisposition to EOC [14], followed by their supporting literature review on this topic [15]. Subsequently, a number of research groups have reported on the striking prevalence of histopathologic and/or molecular genetic alterations in Fallopian tubes from women harboring BRCA mutations who have undergone prophylactic surgery or ovarian cancer surgery (e.g., [16–18]). The accumulated body of clinicopathologic and laboratory-based evidence begins to demand serious consideration of this model for the pathogenesis of at least some type-II EOCs.

Additionally, Crum, Drapkin and colleagues have developed a model suggesting that many, if not all, of high-grade serous carcinomas arise in the fimbria of the Fallopian tube (reviewed in [19,20]). In essence, this model involves the development of cells with a ‘p53 signature’ from normal tubal epithelium, leading to tubal intraepithelial carcinoma, which can develop into invasive serous carcinoma. The carcinomatous cells then exfoliate from the fimbria to the ovary or peritoneum [21,22]. Although compelling, these and other data related to the ‘tubal model’ of type II EOC pathogenesis remain subject to the proverbial chicken and egg conundrum.

This model is subject to several questions. The majority of data supporting this model of EOC were derived from tissues obtained from women with BRCA mutations; is the tubal model more relevant to this population? A very high proportion of the tubal abnormalities observed in this context are tubal intraepithelial carcinoma; is it possible that many or most of these lesions are ‘dead-end’ histopathologic entities and would never have progressed to invasive carcinoma, thus not having been discovered in the absence of prophylactic or ovarian cancer surgery? Even when co-existing with EOC harboring identical genetic mutations, is it not possible that the EOC seeded the tubal entity?

There is also a large body of data, both morphologic and molecular ([23] and references therein), supporting the origin of EOC within the ovary. We have demonstrated a morphologic and genetic continuum between normal epithelium, dysplasia and invasive high-grade carcinoma with cortical inclusion cysts of the ovary, in both BRCA mutation carriers and noncarriers. Furthermore, our data indicate that chromosomal instability in the form of aneuploidy is present in a large (9%) proportion of normal cystic inclusions in ovaries from women not at genetic risk of ovarian cancer [23]. Together, these data indicate that at least some type II EOCs arise within the ovary.

In conclusion, it is clear that EOC represents multiple diseases at the histopathologic and molecular genetic levels. Furthermore, a substantive proportion of ‘EOCs’ may arise outside the ovary. Further research is essential to fully understand EOC in all of its apparent complexity (especially the early natural histories of both type I and type II cancers) if morbidity and mortality are to be significantly reduced through early detection.

Financial & competing interests disclosure

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.