SPC exhibits a complex papillary/glandular architecture with diffuse, prominent nuclear pleomorphism. TP53 is a transcriptional regulator to trigger apoptosis or cell cycle arrest under DNA damage, and TP53 mutation is observed in more than 90% of SPC and 75% of endometrial intraepithelial carcinoma (EIC, noninvasive SPC) while only in 30% of G3 and less than 10% of G1-2. As high-grade endometrial cancer arose through conditional uterine deletion of Tp53 using the Cre/loxP approach [13], TP53 mutation is reasonably considered as a primary event of SPC oncogenesis [14]. p16 ^INK4a mutation is another hallmark of SPC and EIC. p16 ^INK4a mutation is almost 100% observed in serous tubal intraepithelial carcinoma (STIC) and high-grade serous ovarian carcinoma (HGSOC) while not in the precursor of STIC, p53 signature, which is atypical tubular epithelium with TP53 mutation. These results indicate p16 ^INK4a mutation is also not a malignant phonotype driver but an early event of oncogenesis following TP53 mutation as p53 signature is occasionally identified in benign-appearing endometrium as a latent precancerous lesion of EIC [15, 16] even though oncogenic association of p16 ^INK4a mutation has not been clarified in endometrial p53 signature.

Mutations in PIK3CA (24%), FBXW7 (20%), and PPP2R1A (18%) in both SPC and EIC are also reported as early events [17]. PIK3CA regulates several malignant phenotypes, such as proliferation, survival, and mobility via PI3K/AKT/mTOR pathway. An F-box protein, FBXW7, is critical in the ubiquitination and targeting of tumor-promoting proteins cyclin E (CCNE1) and PPP2R1A. CCNE1 controls the G1 to S transition of the cell cycle, and CCNE1 amplification is common in primary resistant and refractory HGSOC [18], and more than half of USC harbor either a molecular genetic alteration in FBXW7 or CCNE1 amplification [17]. PPP2R1A is a regulatory unit of serine/threonine protein phosphatase 2A (PP2A), and its mutant promotes anchorage-independent growth and tumor formation in a dominant-negative manner [19]. EIC is not invasive but has a metastatic ability accompanied with anchorage-independent growth, and half of patients bearing EIC are found to have disease beyond the uterine corpus including omental involvement [20]. The alterations of PIK3CA, FBXW7, CCNE1, and PPP2R1A can regulate malignant phenotypes of SPC/EIC, but these alterations are observed only in a certain part of SPC/EIC, which indicates these alterations are not indispensable in the oncogenesis, but the accumulation of these alterations may define the malignant characters of each case and become therapeutic targets. For targeting PI3K/AKT/mTOR pathway, several mTOR and/or PIK3CA inhibitors are currently under evaluation in clinical trials. So far, the efficacy of mTOR inhibitor was demonstrated only for endometrioid carcinoma [21], but in this study, none of 11 SPC cases exhibited any clinical response. As dual blockade of PIK3CA and CCNE1 decreased tumor growth significantly in mouse model [22], multi-targeting might be also necessary in the clinical setting. Concerning the results that two out of two SPC patients showed clinical response to bevacizumab, the efficacy of combination of bevacizumab and temsirolimus, an mTOR inhibitor, was also investigated, but its toxicity was nonnegligibly high [23, 24].

HER2/neu is located upstream to the PIK3CA/AKT/mTOR pathway. HER2 (erbB-2, the epidermal growth factor Type II receptor) amplification or overexpression is observed most frequently in SPC [25] although the expression rate differs among studies depending on assessing techniques. HER2 expression status is at first determined by immunohistochemistry (IHC), but in cases with equivocal IHC results, fluorescence in situ hybridization (FISH) is employed. HER2 amplification is frequently observed in African-American patients with worse survival compared with Caucasian patients as those bearing HER2+ breast cancers exhibit poor survival due to metastasis and chemoresistance. HER2 receptor targeting therapy using humanized monoclonal antibody, trastuzumab, is effective in the treatment of HER2+ node-positive breast cancers [26], but it is not so effective for SPC; there were none of 11 HER2+ SPC cases to demonstrate any tumor response [27]. In SPC, frequent PIK3CA mutation/amplification would compensate trastuzumab effect, and truncated p95HER2 variant which lacks the trastuzumab-binding domain is frequently observed [28].

Endometrial CCC exhibits characteristic morphology similar to ovarian CCC, a combination of architectural patterns (papillary, glandular, solid, and cystic) and cytoplasmic features (clear and oxyphilic). Nevertheless, as endometrioid carcinoma and SPC display significant morphological overlap with CCC, interobserver discrepancy to differentiate CCC from its mimics occurs frequently [12, 29]. Then, immunostaining is usually employed for differential diagnosis although pathological diagnosis based on morphology is fundamental. In general, endometrial CCC as well as ovarian counterpart is positive for hepatocyte nuclear factor 1b (HNF1b:100%) and napsin A (93%) and negative for estrogen receptor (ER: 93%), and the combination of these markers is very useful for distinguishing solid/papillary pattern CCC from SPC or G3 [29, 30].

In contrast with the immune profiling, the molecular-genetic background of endometrial CCC is still obscure due to the rarity of pure CCC. Endometrial CCC is basically similar to ovarian CCC, but there exist differences between them: less alterations in ARID1A (13–24% vs. 50%) and PIK3CA (9–24% vs. 35%). Furthermore, TP53 mutation is relatively frequent compared with ovarian CCC (33–40% vs. 12%), and the aberrant p53 staining is designated as a poor prognostic factor [25, 29, 30]. Targeted genetic profiling of endometrial CCC identified mutations in genes involved in chromatin remodeling/transcriptional regulation (ARID1A, ZFHX3, and TSPYL2) [29], and loss of BAF250a expression occurs without ARID1A mutation in 26% of CCC [31]. Loss of BAF250a itself is not related to poor prognosis in endometrial CCC [32]. However, as ovarian CCC with loss of one or multiple SWI/SNF complex subunits including BAF250a exhibits aggressive behaviors and poor prognosis [33], in endometrial CCC, multiple loss of SWI/SNF complex subunits may also affect the prognostic outcome. In CCC, ubiquitin-mediated proteolysis (SPOP and FBXW7) and SPC-characteristic genes, TP53 and PP2AR1A, are also highly mutated [29], and HER2/neu amplification is identified [25]. These common genetic alterations would provide partial overlapping in tumor phenotypes between SPC and CCC.

CS is a biphasic tumor composed of Type II cancer with sarcomatous elements. Although this tumor is a wide-range admixture of nonspecific sarcomatous mesenchyme and high-grade epithelium, SPC, CCC, and G3, this tumor is considered to be of epithelial derivation as a masterpiece of epithelial-mesenchymal transition and to share genetic profiles with Type II cancers. CS harbors relatively high TP53 mutation (67%), but low PIK3CA mutation (22%), and the carcinoma component is considered responsible for the aggressive behavior of CS resulting in a poor survival, while the clinical impact of sarcoma component has been obscure [29]. Most recent integrated analysis revealed that CS shared proteomic features with SPC and EC, and sarcomas with epitherlial-mesenchymal transition features [34]. A multicenter retrospective study for 1192 CS cases demonstrated that carcinoma components tended to spread lymphatically, while sarcoma components tended to spread locoregionally, and high-grade carcinoma component was independently associated with decreased progression-free survival (PFS) [35]. In this study, postoperative chemotherapy was an independent predictor for improved PFS, and characterization of histologic pattern would make drug selection suitable in the treatment of each carcinoma/sarcoma combo: ifosfamide for low grade/homologous (HR 0.21, p = 0.005), platinum for high grade/homologous (HR 0.36, p < 0.001), and anthracycline for high grade/heterologous (HR 0.30, p = 0.001) [35].

Type I low-grade endometrioid carcinoma (EC) usually presents as an early-stage disease with or without shallow invasion resulting in an excellent outcome. Type I ECs generally exhibit myometrial invasion in a border-pushing expansile manner (usual pattern, Fig. 6.2b), but such invasion stays within inner half of the myometrium without lymphovascular space infiltration (LVSI) [36]. In such FIGO stage IA cases bearing usual pattern of Type I EC, extrauterine spread is so rare that hysterectomy, salpingo-oophorectomy, and pelvic node dissection (Fig. 6.2a) without adjuvant chemotherapy are generally accepted as the standard treatment. In contrast, some exhibit a different way of myometrial invasion in an infiltrative manner with microcystic, elongated, and fragmented glands surrounded by myxoid and inflamed stroma (MELF pattern, Fig. 6.2b), and this type of invasion is usually observed in the invasive front in the outer half of the myometrium. MELF frequency is around 13% (7–36%), and this morphologic pattern is highly associated with LVSI and node metastasis [37–39]. MELF has been considered as a kind of EMT feature due to the morphology of subtle sinus histiocyte-like invasion. Immunophenotyping studies revealed gain of S100A4 and L1CAM and loss or reduction of E-cadherin, CD147, MMP2, and Galectin-3, resulting in loss of cell-cell adhesion and polarity to endow migratory and invasive properties [38, 40–42]. These results indicate MELF is one phenotype of endometrioid carcinoma cells in the phase of EMT, but it is still not clarified whether low-grade ECs with MELF-type invasion are genetically distinct from usual Type I ECs as there is no comprehensive genomic analysis so far.

The clinical impact of MELF is controversial as well. MELF-pattern invasion is not associated with macrometastasis but micrometastasis or isolated tumor cells (ITC, Fig. 6.2b) [39]. Node metastasis is an infamous prognostic factor of endometrial cancer, but the clinical relevance of ITC is obscure since ITC does not increase the recurrence rate. The recurrence rate of cases with or without MELF is considered not different, but MELF cases trend toward decreased time to non-vaginal recurrence due to higher rate of node involvement [38]. MELF pattern invasion exists too sparsely in the tumor frontier to be detected at the preoperative diagnosis with endometrial biopsy or magnetic resonance imaging (MRI). Dynamic contrast-enhanced MRI is very useful for detecting the presence of myometrial invasion, but so far, deep myometrial invasion is detected in more than 10% cases preoperatively diagnosed as FIGO stage IA Type I EC [43]. Although the clinical significance of node dissection in the para-aortic lesion (PAN) is still controversial, underestimation would save PAN at primary surgery resulting in PAN recurrence/residue due to undertreatment. To avoid undertreatment following underestimation, establishment of preoperative diagnosis of MELF in the deep myometrium is warranted.

The International Cancer Genome Consortium (ICGC) was established to analyze the genome-wide abnormalities of 50 kinds of malignant tumors. The Cancer Genome Atlas (TCGA) is a study conducted as the ICGC project in the United States, which initially analyzed brain glioblastoma, non-small cell lung cancer, and high-grade serous ovarian carcinoma [44, 45]. As for endometrial cancer, whole-exome sequencing, SNP array assessing copy number alterations, mRNA expression microarray, DNA methylation microarray, and microRNA microarray were conducted for more than 200 cases [4]. This integrated analysis of somatic mutation rates, frequency of copy number alterations, and microsatellite instability (MSI) status along with the clinical information provided a new insight that endometrial cancers were classified into four distinct molecular subgroups. The subgroups are termed as POLE ultramutated, hypermutated (microsatellite unstable), copy number low (microsatellite stable), and copy number high (serous like) (Fig. 6.3a).