This study showed that PoRSEC may show morphological and immunophenotypical overlap with SEIC. However, a papillary architecture with confluent papillae is observed in SEIC but not in PoRSEC. Moreover, SEIC shows cytological demarcation with normal endometrium, while PoRSEC appears as a diffuse change which merges imperceptibly with normal glands.

The distinction between reactive/regenerative atypia and true neoplastic atypia can be challenging. In the endometrium, reactive changes may raise the concern of several types of premalignant and malignant lesions, such as atypical hyperplasia, complex papillary proliferations, and SEIC [7,8,9,10]. Atypical endometrial hyperplasia and complex papillary proliferations are characterized by glandular crowding and branching papillae, respectively, which are typically absent in reactive changes [11, 12]. By contrast, SEIC may replace the epithelial lining of endometrial atrophic glands without affecting their arrangement. The typical feature of SEIC is the presence of high-grade nuclear atypia with striking pleomorphism [13]. Endometrial reactive changes can show an “SEIC-like” atypia, including nuclear enlargement and pleomorphism, nuclear clarification or hyperchromasia, evident nucleoli, eosinophilic or clear cytoplasm, and hobnail changes [9, 10]. This kind of atypia was also observed in our series of PoRSECs.

Our study was the first to systematically assess the morphological and immunophenotypical features of PoRSEC and the differential diagnosis with SEIC.

We found that, although the degree of atypia was similar between PoRSECs and SEICs, the former showed no evident mitotic activity. This finding is coexistent with those of our previous study, which showed no mitotic activity in endometrial metaplastic/reactive changes coexistent with cancer [5]. However, in our experience, mitotic figures can occasionally be found in reactive changes; this can be concerning, especially when the amount of examined tissue is scarce.

Although SEIC lacks branching papillae and prominent glandular crowding (as discussed above), it does show architecture complexity in the form of simple papillae which are areas of confluence, typically restricted to the surface endometrium, which was not observed in PoRSEC. Moreover, SEIC appeared as a circumscribed lesion, with a sharp cytological demarcation between the SEIC area and the background atrophic endometrium. On the other hand, PoRSECs extensively involved the endometrium with no abrupt transition between areas with high-grade atypia and areas with low-grade or no atypia.

Immunohistochemically, SEIC is characterized by a mutation-type p53 pattern; this reflects the presence of underlying TP53 mutation and can also be observed in the earliest precursor of serous carcinoma, i.e., the so-called p53 signature [13]. Consistently, we observed a mutation-type p53 pattern in all included SEIC cases. As expected, all PoRSEC cases showed a p53-wt expression; however, more than one-third of PoRSEC cases had a “wt-high” pattern, that is, a p53 positivity in 50–80% of tumor cells. Before strict criteria to define a mutation-type pattern were defined, a wt-high pattern was often interpreted as aberrant [14]. However, even in the presence of strict criteria, the distinction between the two patterns can be difficult at times.

P16 and Ki67 are adjunctive markers in the diagnosis of SEIC. Indeed, SEIC characteristically shows a strong and diffuse (“block-type”) p16 expression and a high Ki67 expression (indicating a high proliferation index) [15]. A block-type p16 expression was observed in all SEIC cases and in almost half of PoRSEC cases. A similar finding was observed in metaplastic/reactive changes in our previous study and may constitute a further pitfall. Regarding Ki67, the mean value of SEIC was considerably higher than that of PoRSEC; however, the two groups showed a partial overlap, with the highest value among PoRSEC being 35% and the lowest value among SEIC being 5%.

Overall, PoRSEC showed a heterogeneous pattern of immunohistochemical markers, as opposed to SEIC. Therefore, SEIC shows both a morphological and immunohistochemical demarcation, which is generally absent in PoRSECs.

Finally, a complete absence of NapsinA and p504s staining could be useful to rule out a diagnosis of clear cell—EIN.

Remarkably, this study only included patients with LACC treated with CRT, in which PoRSEC can be expected. However, in our experience, PoRSEC can also be observed in benign uteri from patients who underwent CRT for other carcinomas. If the information regarding the previous CRT is missing, the pathologist might not consider the possibility of PoRSEC. It appears therefore necessary to recognize the crucial morphological and immunohistochemical features of PoRSEC, in order to avoid a serious misdiagnosis.