Ovarian EMA has been known for approximately 30 years to have a histological variant which is characterized by a resemblance to SCST [12, 13], but the precise frequency of EMA resembling SCST remains to be clarified. According to Young et al., 8 of 13 cases were initially misdiagnosed as SCST . A lack of accurate recognition about this variant may yield a misinterpretation which results in the misdiagnosis as SCST instead of EMA. Especially, at the intraoperative diagnosis, both the macroscopic and microscopic findings may mislead pathologists to raise the possibility of SCST. Although our experience is limited, the 4 present cases were encountered during the last 5 years since the opening of our institution, accounting for approximately 10% of the total ovarian EMAs. In fact, Case 1 (Figure 3a, b) was initially diagnosed by the referring pathologist as Sertoli-Leydig cell tumor with estrogen overproduction, mainly because of his misinterpreting the luteinized stromal cells as Leydig cells. One of the critical differential points between Sertoli-Leydig cell tumor and EMA is the occurrence age: 28 years for the former and over 55 years for the latter . But, interestingly, it should be noted that some Sertoli-Leydig cell tumors contain areas which resemble endometrioid tumor . As for Case 2 (Figure 3c, d), at the intraoperative diagnosis, the possible diagnosis of Sertoli cell tumor was made by the referring pathologist. In this case, the presence of granulosa cell tumor-like components also may have become a key to misdiagnosis.
It has not been pathologically defined as to what part of the components resembling SCST is supposed to be contained for the diagnosis of this variant. Reportedly, the proportions of components resembling SCST in the entire tumor vary considerably, ranging from 30 to 100% . In our review, too, the proportions ranged from 20 to 80%. Except for EMA which is almost entirely occupied by the components resembling SCST, as in Case 3 (Figure 3e, f, g), a combination with a variable amount of conventional EMA would favor the diagnosis of EMA resembling SCST. Though relatively rare, a coexistence of cribriform-like nests and solid nests may make it difficult to lead to the accurate diagnosis of a variant of EMA. In addition, the histologically reliable hallmarks of EMA are as follows: squamous differentiation of the tumor cells, complication of endometriosis within the tumor or in the background, and transition between adenofibromatous components of endometrioid type and carcinoma . In Case 2, endometriosis was speculated to be associated with tumor development. In Case 4 (Figure 3h, i, j), a transition with the benign component of adenofibromatous endometrioid tumor was considered to be strongly supportive of the diagnosis of a variant of EMA. But, it should be noted that mucinous cystadenoma may coexist not only in other epithelial tumors but also in SCSTs. Interestingly, in this case, the stromal cells were found to be more cellular in the vicinity of mucinous cystadenoma rather than within the endometrioid benign and malignant tumors.
When histologically challenging cases are encountered, certain immunohistochemical markers are thought to contribute greatly to making the diagnosis definite. The epithelial nature of EMA, in spite of the resemblance to SCST, could be determined by positive reactions for CK7 and epithelial membrane antigen [15–17]. Many EMAs are positive for ER  but inconsistently negative for inhibin-α [15, 17]. In the cases with pure SCST, inhibin-α is frequently expressed, while these epithelial markers are less frequently or weakly expressed if at all [15–17], and ER expression is usually negative .
The first descriptions about the ovarian functioning stroma go far back to the 1950s . This stromal change has been considered to be closely associated with clinical, biochemical or pathological evidence of endocrine function [4–7]. Several investigators demonstrated that many postmenopausal women with a variable epithelial ovarian tumor would be provided with a potential to produce more sex steroid hormone represented by estrogen than normal postmenopausal women [6–8, 20, 21]. The frequency of estrogen overproduction is reported to be over 70% by serological examinations [7, 8], and 47%  or 50%  by urine analyses. According to some reports, steroidogenic cells are observed more frequently in mucinous tumors [6, 8, 20], and predominantly in postmenopausal women (80%) . However, the difference in the incidence of estrogen overproduction among the variable types of ovarian tumors remains controversial . Concerning the difference among benign tumor, borderline malignancy, and malignant tumor, the previous reports do not seem to present confirmative data because of their statistically small scale of examined cases as well as their deviation in histological types [6–8, 20, 21].
In the report by Young et al., the association between EMA resembling SCST and estrogenic manifestation or presence of steroidogenic stroma was described as follows: 4 of 13 cases had the stroma consistent with that of luteinized stromal cells, but the evidence of possible estrogen overproduction was noted in only one case . There is another description that theca-like stroma and foci of vacuolated lutein-like cells were noted in one of 4 cases with EMA . To the best of our knowledge, with regard to postmenopausal EMA with estrogen overproduction which was confirmed by serological examination, there have been only 4 reported cases [11, 23–25]. These 4 cases included one case with EMA resembling SCST, and the remaining 3 cases showed conventional EMA. Based on our limited experience, it is speculated that EMA resembling SCST may have more estrogenic manifestations due to the presence of steroidogenic stroma. Focusing on the source of estrogen overproduction and its mechanism, in terms of the morpho-functional correlation, immunohistochemical and molecular-based analyses have been performed [9–11, 23, 26, 27]. Interestingly, Sasano et al. mentioned that SF-1 and steroidogenic enzymes are expressed in not only stromal cells but also epithelial tumor cells . However, the association between clinical hormonal abnormalities and presence or absence of luteinized stromal cells in ovarian tumor, including the relationship with the mechanism of extragonadal aromatization, has not been fully established yet .
In summary, EMA resembling SCST should be definitely differentiated from SCST from the clinical and pathological aspects. In order to clarify the significance of estrogen as a tumor marker for the ovarian tumor and its utility in the management of the patient, more precise clinical and pathological data must be gathered. Furthermore, the molecular-based mechanism with estrogen overproduction needs to be explored.