The notion that non-myoinvasive or superficially invasive ESC may potentially show extrauterine extension had long been recognized, even from some of the earliest descriptions of ESC (1). Serous EIC has been identified in 58–89 % of ESC [28, 35]. Although EIC is rarely identified without concomitant ESC, analy ses of small groups of patients have conclusively established that EIC may be associated with peritoneal carcinomatosis in the absence of any other serous neoplasm in the uterus [36, 37]. The frequency with which this occurs is not entirely clear. However, for the larger group of lesions defined as minimal or only very superficially invasive, up to two-thirds of cases may be associated with extrauterine extension [16]. Morphologically, the extrauterine deposits display the same features as the endometrial lesions [38] and are usually less than 2 cm each [39]. Similarly, both intrauterine and extrauterine lesions have similar immunophenotypes regarding estrogen receptor, progestin receptor, p53, WT1, and Ki-67 expression [38]. At the molecular level, both uterine and extrauterine lesions within a given patient have been shown to display identical TP53 mutations [39, 40]. However, the interrelationships between these lesions may be more complicated.

A very small subset of serous carcinomas in the endometrium may actually represent either drop metastases from the adnexa that manifest in the endometrium with an EIC-like intraepithelial growth pattern. Many of such lesions are associated with extrauterine serous carcinomas that are independent primaries as evidenced by the concomitant presence of corresponding precancerous lesions at the extrauterine site. It is known that some cases of EIC may be associated with serous tubal intraepithelial carcinoma (STIC ) in the fallopian tube. In one study of 22 consecutive cases of serous carcinomas involving the endometrium [41], the fallopian tubes were submitted according to the SEE-FIM protocol [42]. 50 % of cases showed adnexal involvement, including five wherein a STIC was identifiable. In all five cases, the endometrial lesions tended to be noninvasive or only minimally invasive of the myometrium , and identical TP53 mutations were shared by both tubal and endometrial lesions in a subset of cases [41]. In another study 8 % of ESC showed STIC when the fallopian tube was processed using the SEE-FIM protocol [35]. Extensive sectioning of the endometrium in patients with adnexal serous carcinomas has identified serous EIC in about 15 % of cases [43]. In our own recent analysis of samples from 21 patients with serous EIC, cellular lineage relationships between intrauterine and extrauterine lesions were assessed by TP53 mutation analysis. Based on the patterns of concordant or discordant mutation for this gene, we concluded that the extrauterine disease associated with serous EIC may be from the endometrium (47.6 % of cases), adnexa (23.8 %), or both (28.6 %) [39]. All of the aforementioned findings, in total, suggest either that some cases of ESC (EIC or otherwise) are of extrauterine origin, that the primary lesion accordingly contributes to the extrauterine disease burden, or that both the intrauterine and extrauterine lesions are independent, being simply reflective of an increased propensity of serous carcinogenesis that is generalized in the upper genital tract.

The conclusion by Ambros et al. in 1995 that serous EIC “is the likely precursor” for uterine tumors displaying serous differentiation was based on their intraepithelial growth pattern as well as their statistically significant, nearly exclusive association with carcinomas with a serous component [30]. Subsequently published studies have shown that serous EIC and ESC display broadly similar phenotypic patterns, including a comparable frequency of expression of proteins such as p53, p16, IMP3, HMGA2, Nrf2, hormone receptors, Cyclin E1, and HER2/neu [44–51]. This is not unexpected, since serous EIC is definitionally composed of cells that are identical to ESC.

At the molecular level, serous EIC and ESC have been shown to be very similar and to share a broad molecular profile. Kuhn et al. [52] performed a comparative mutational analysis in samples from nine patients with serous EIC and ESC pairs. All nine pairs showed concordant PIK3CA, PPP2R1A, and TP53 mutational profile, whereas eight of the nine pairs had concordant FBXW7 mutation status between these two components. The single discordant pair contained a FBXW7 p.Asp440Asn mutation that was absent in serous EIC but present in the associated ESC [52]. FBXW7 mutation may be associated with Cyclin E1 overexpression at the protein level, as can CCNE1 amplification, the underlying gene for the Cyclin E1 protein. It has subsequently been shown in another paired analysis that 45 % of ESC and 41 % of serous EIC showed CCNE1 amplification as assessed by fluorescent in situ hybri dization and that there was generally a high concordance in CCNE1 copy numbers between the pairs [53]. However, there is evidence that although serous EIC and ESC are similar, they are not identical, and this may lend credence to the postulation that serous EIC precedes ESC. The frequency of TP53 mutations is generally higher in ESC than in serous EIC (72–78 % in serous EIC vs. 90–96 % in ESC) [24, 54]. Loss of the wild-type p53 allele, as inferred from loss of heterozygosity at chromosome 17p, has been observed in 100 % of ESC and in 43 % of serous EIC [54]. Occasional cases of serous EIC and their associated ESC display discordant TP53 mutations (7 % of cases in our analysis [24]). Additionally, as previously indicated, FBXW7 mutations in the same patient are also occasionally discordant between these lesions. Also supporting the concept that serous EIC precedes ESC is the morphologic observation of their growth pattern and size. It is logical that a small, highly proliferative lesion with an intraepithelial growth pattern will eventually become a large lesion with an invasive growth pattern. Accordingly, serous EIC is placed before ESC in our model and is its likely precursor [16, 17]. However, this placement of serous EIC is merely a conceptual component of the framework by which endometrial serous carcinogenesis can be studied and understood. For practical and clinical purposes, this distinction is irrelevant, since the evidence indicates that serous EIC is already a cancer with significant potential for metastasis [36, 37], rather than a true precancer that is associated with an increased risk of cancer but which is in and of itself not malignant [55]. Serous EIC, when identified alone, is therefore best conceptualized as ESC at an early patholog ic phase and/or with an unusual intraepithelial growth pattern, and patients with serous EIC should be managed in the same manner as their counterparts with the conventionally invasive neoplasm [13, 32].

Studies published during the past decade have established endometrial glandular dysplasia (EmGD) as the most likely precancerous lesion for ESC [21–26, 59]. EmGD was originally described in 2004 [21], following works predicated on the hypothesis that there is a morphologically identifiable lesion that precedes serous EIC and which is precancerous. In the seminal reports that described the entity, the endometrium adjacent to a cohort of serous EIC, ESC, and endometrioid carcinoma cases was evaluated in detail. Atypical glandular lesions were identified 53 % of the ESC cases and 1.7 % of the endometrioid carcinomas [21]. These lesions were then studied and form the basis of the EmGD entity. The average age of the patients with an EmGD lesion was 65 years (range 57–79 years) [21].

A p53 signature is a morphologically unremarkable segment of epithelial cells displaying moderate to strong immunoreactivity for the p53 protein as assessed by immunohistochemistry (Fig. 8.22). These foci were identified by immunohistochemical analyses of the background endometrium in carcinomas and have been proposed as the first step in endometrial serous carcinogenesis in our model [16, 17]. They are multifocal in at least 89 % of cases [18]. Zhang et al. identified p53 signatures in the endometrium associated with 39.1 % of ESC, 37 % of serous EIC, and 3.3 % of endometrioid carcinomas and in 1.7 % of benign endometrial samples [18]. Jarboe et al. identified p53 signatures in 70 % associated with 70 % of serous EIC and in 4 % of 137 endometrial polyps [19]; other authors have reported less well-defined increases in p53 expression in endometrial polyps [68, 69, 73]. Koi et al. [70] identified p53 signatures in 11 % of the endometrial tissues of 82 women with a variety of benign diseases. Nguyen et al. identified p53 signatures in the background endometrium associated with 24 % of ESC, 0 % of CCC, and 20 % of carcinosarcomas [20]. In the latter study, of the eight non-endometrioid tumors with p53 signatures, 88 % were associated with serous EIC [20]. The p53 signatures are occasionally observed in direct continuity with serous EIC [19]. These signatures display a low proliferative index (4–5 %) [19, 20] and express estrogen receptor-alpha [20, 70]. In our analysis of microdissected samples of p53 signatures, 42 % of cases showed at least one TP53 gene mutation [18]. A high degree of concordance was identified between the EmGD, serous EIC, and ESC lesions within a given uterus, as at least one identical TP53 gene mutation was found in all three lesions in 50 % of cases [18]. Similarly, Jarboe et al. [19] found identical TP53 mutations in paired p53 signatures and serous carcinomas in two (67 %) of three cases analyzed.