Diagnostic accuracy and prognostic factors of uterine serous carcinoma in Japanese women: a multi-center study.

Endometrial cancer (EC) is one of the most common gynecological malignancies in developed countries, accounting for approximately 65,000 newly diagnosed cases in the United States in 2020 [ 1 ]. An increasing incidence of EC, amounting to more than 20,000 new cases annually, has been observed in the last decade [ 2 ], which is likely attributable to the aging of the population and the increased prevalence of obesity in developed countries [ 3 ]. Endometrioid carcinoma of the endometrium is the most common histological subtype of EC, accounting for 85%–90% of cases [ 4 ], and is generally associated with a lower risk of progression and favorable prognosis, particularly for low-grade disease [ 4 ]. The most common non-endometrioid endometrial carcinoma is uterine serous carcinoma (USC), which accounts for approximately 10% of all ECs. Other types of non-endometrioid EC include carcinosarcoma (2%–5%), undifferentiated (5%), clear cell (2%–4%), and squamous cell (0.1%–0.5%) carcinoma [ 4 ]. Despite being the second most common type of EC, USC is considered a relatively rare tumor [ 5 ]. Patients with USC are frequently diagnosed at an earlier disease stage than those with the endometrioid subtype of EC. However, they have a higher risk of relapse and an overall worse prognosis when compared stage-for-stage to EC of the endometrioid subtype. Importantly, the rising incidence of USC, which accounts for 40% of the total number of EC-related deaths [ 6 ], has led to an increasing interest in understanding the biological features of USC [ 3 ]. Patients with USC are more likely to be diagnosed with early-stage disease with lymphovascular space invasion (LVSI), nodal involvement, and microscopic spread to the peritoneal surface, even in cases with limited myometrial invasion. Thus, the risk of early-stage USC being diagnosed with stage III or IV is 2.5 times higher than that of early-stage endometrioid subtype lesions (46% and 20% for USC and endometrioid subtype, respectively) [ 6 ]. The incidence of EC has increased markedly in Japan, with the number of new patients with EC increasing over 5 times in the last 25 years. In 2006, the incidence of USC in Japan was 3.8% of the 5,398 cases of EC [ 7 ]. Because of its rarity, a large number of studies have not been conducted in Japan; however, as the number of patients with EC has increased, the number of patients with USC has also increased, and in 2017, it accounted for 6.5% of all ECs [ 8 ]. Considering this trend, this study aimed to identify the prognostic factors of USC in Japanese patients.

Overall, 193 cases were examined. Of these, 188 cases, excluding 5 with no postoperative explanted specimens, were subjected to CPR. Of these, 144 (76.6%) and 43 (22.9%) cases were diagnosed with serous and non-serous carcinoma through CPR, respectively. Among the non-serous cases, 32, 5, 2, 2, and 2 were classified as endometrioid carcinoma (grade 1: 10 cases, grade 2: 11 cases, and grade 3: 11 cases), clear cell carcinoma, carcinosarcoma, poorly differentiated adenocarcinoma, and atypical hyperplasia, respectively. Of the 144 cases diagnosed with serous carcinoma, 120 (83.3%) were diagnosed with serous carcinoma using hematoxylin-eosin staining alone. Furthermore, 24 (16.7%) cases were finally diagnosed with serous carcinoma after additional immunostaining for p53. Serous carcinoma was diagnosed in 72 (50.0%) of the patients with preoperative excisional specimens. A final diagnosis of USC was made using CPR in 144 patients. The patient characteristics are presented in Table 1 . The median age was 65.7 (43–86) years, median body weight was 53.2 kg, and the mean body mass index was 22.8. Most of the patients were menopausal (91.8%), and 11% were nulliparous. Twenty-five (17.3%) patients had a history of cancer, and 14 (9.7%) had a history of breast cancer. The complications are presented in Table 1 . Only 9 (6.3%) patients had diabetes. Among these 72 patients, 38.2%, 15.3%, 45.8%, and 6.9% had FIGO stage I, II, III, and IV disease, respectively. BMI, body mass index; FIGO, International Federation of Gynecology and Obstetrics; LN, lymph node; LVSI, lymphovascular space invasion; mRH, modified-radical hysterectomy; NAC, neo-adjuvant chemotherapy; RH, radical hysterectomy; TAH, total abdominal hysterectomy. Total hysterectomy was performed in 142 patients, including modified radical hysterectomy in 28 and radical hysterectomy in nine patients. Pelvic and/or para-aortic lymphadenectomy was performed in 75 patients. The remaining 69 patients did not undergo lymphadenectomy because they aged >80 years, had apparent early disease, or had abundant peritoneal dissemination. Most patients underwent systemic chemotherapy with doxorubicin, cisplatin, paclitaxel, carboplatin, or other adjuvant regimens. None of the patients underwent adjuvant radiotherapy (RT). Seventy-six (52.8%) patients had stage 3 or 4 USC. Extrapelvic disease was also present in 38 (26.4%) patients, including distant metastases in 20 (13.9%). Pathologically, 56 (38.9%) patients had extrauterine lesions. Forty-one (28.5%) patients had lymph node metastases, including 16 with para-aortic node metastases. Seventy-eight (54.2%) patients had LVSI, and 56 had extrauterine lesions, including 38 with extrapelvic lesions. Among 84 patients with <50% myometrial invasion, LVSI, extrauterine disease, and lymph node metastasis were found in 29 (34.5%), 22 (26.2%), and 13 (15.5%) patients, respectively ( Table 2 ). Values are presented as number of patients (%). LVSI, lymph vascular space invasion. The outcomes were available for 144 patients, with a median follow-up period of 45.2 (2.8–86.8) months, and 60 (42.5%) patients died of the disease. Temporary remission was observed in 125 (86.8%) patients. Among them, 58 (46.4%) patients experienced recurrence. Fifty-nine (41.8%) patients survived, with no evidence of disease (NED). The 2-year survival rate was 56.0%. Among patients with stage I and II disease who underwent lymphadenectomy or biopsy (48 patients) and did not undergo lymphadenectomy (20 patients), the incidence of NED was 77.1% and 35.0%, respectively. Thirteen patients were diagnosed with stage I and II disease following complete staging surgery, including pelvic lymphadenectomy and omentectomy. Twelve (92.3%) patients had NED. The other 55 patients with stage I and II disease (incomplete staging) showed a trend toward a worse prognosis ( Fig. 1 ). CI, confidence interval; HR, hazard ratio; OS, overall survival; PFS, progression-free survival. The association between surgical technique and survival was analyzed: for PFS, pelvic lymphadenectomy was associated with prognosis in univariate analysis; for OS, pelvic lymphadenectomy was associated with prognosis in univariate analysis; and omentectomy was associated with prognosis in multivariate analysis ( Table 3 ). Bold values denote statistical significance at the p<0.05 level. CI, confidence interval; HR, hazard ratio; OS, overall survival; PFS, progression-free survival.

This is the first large multi-center retrospective study of USC data involving Japanese women. Endometrioid endometrial adenocarcinoma (EMCA) occurs most commonly in women aged >50 years. In this study, the mean patient age was 65.7 years. The patient characteristics were similar to those observed in other studies. Most patients with USC are postmenopausal, and a few women are nulliparous, are not obese, and have a history of cancer, especially breast cancer [ 10 ]. Women with breast cancer have a 2.6-fold risk of developing USC compared to endometrioid carcinoma [ 11 ]. Furthermore, 5% of patients with USC have mutations in tumor suppression genes, such as BRCA1 , CHEK2 , and TP53. Patients with USC and breast cancer should undergo genetic testing because 9% of them have BRCA1/2 mutations [ 12 ]. Thus, family history should be considered during the diagnosis of USC. USC has a worse prognosis than other types of EC, such as grade 3 EC or clear cell EC [ 13 ]. According to a Japanese Society of Obstetrics and Gynecology (JSOG) report, the 5-year OS rates of patients with EMCA G1, G2, and G3 in 2008 were 94.6%, 88.2%, and 80.2%, respectively. Comparatively, the 5-year OS rates for clear cell EC and USC were 64.1% and 49.5%, respectively [ 14 ]. In this study, the 2-year NED rate for USC was only 56.0%. Various reasons have been proposed for this poor prognosis, including advanced stage, deep myometrial invasion, lymph node metastasis, or LVSI. Patients with USC have more advanced-stage disease at diagnosis. In this study, patients with USC had a higher incidence of stage III and IV EC than all patients with EC in the JSOG report in 2017 [ 8 ]. Patients with stage III and IV disease represented 52.8% and 21.3% of the patients, respectively. LVSI and deep myometrial invasion were associated with lymph node metastasis. USC has a high incidence of lymph node metastasis regardless of myometrial invasion. Low-grade EC with <50% myometrial invasion and <2 cm tumor size has a 0% risk of lymph node involvement [ 15 ]. In contrast, the USC cases with <50% myometrial invasion had a high incidence of LVSI (34.5%), extrauterine disease (26.2%), and lymph node metastasis (14.3%). Another unsolved problem is the diagnostic difficulty associated with USC. We excluded 25% of the patients because they were not diagnosed with USC using CPR. In Western countries, the consensus among pathologists on this subtype of high-grade EC is only 62.5% [ 16 ]. Owing to the rarity of this disease, it is possible that in the early 2000s, there was an insufficient diagnostic consensus among Japanese pathologists. Furthermore, the diagnosis of USC using preoperative samples may be more difficult. In this study, only 50% of the patients were diagnosed with USC preoperatively. As shown in the ASTEC trial, lymphadenectomy can be avoided at the physician’s discretion in early EC [ 17 ]. Therefore, without information on the patient’s diagnosis of USC, physicians tend to treat them as early endometrioid carcinomas. Consequently, most patients undergo staging surgery. In fact, we had a case that appeared to be an apparently low-grade early-stage endometrioid carcinoma; however, the patient experienced recurrence within a short period. Finally, we diagnosed the patient with USC after recurrence [ 18 ]. Each institution determined the FIGO staging of the patients in this study. Among the 68 patients with stage I and II tumors in this study, only 12 underwent complete staging. Their prognosis was excellent because 92.3% of the patients had NED. However, in patients who did not undergo staging surgery, the NED rate was only 33.3%. Therefore, the underestimation of the stage may have caused a worse prognosis in patients with stage I and II tumors in this study. Growdon et al. [ 19 ] reported a significant difference in survival between patients with stage I disease, with and without surgical staging. On long-term follow-up, the OS of the patients with surgical staging was 16.4 years, whereas that of those without surgical staging was 2.8 years [ 19 ]. Accurate prognostic prediction can be achieved under conditions such as correct preoperative diagnosis of USC and complete staging surgery. Postoperative RT is the standard adjuvant treatment for EC. In contrast, systemic chemotherapy has become the standard adjuvant treatment for EC in Japan. A previous Japanese randomized study showed the partial benefit of adjuvant chemotherapy in a limited number of patients with EC [ 20 ]. In patients with stage I disease, systemic chemotherapy plays a major role in reducing the risk of extrapelvic recurrence. Fader et al. reported that patients with stage I tumors who were treated with platinum/taxane chemotherapy ± RT showed favorable prognosis. The 5-year PFS was 81.5 and 64.7 in the chemo± radiation and RT alone groups, respectively [ 21 ]. Vogel et al. reported that adjuvant treatment, including systemic chemotherapy, RT, or a combination of both, was associated with improved OS in stage IB–II but not in stage 1A EC in a large retrospective study [ 22 ]. In this study, patients with early-stage EC who received complete staging plus adjuvant platinum/taxane chemotherapy showed a favorable prognosis. Tate et al. [ 23 ] identified another low-risk characteristic in patients with stage I disease. In their study, 49 patients with negative pelvic cytology and absence of cervical stromal invasion had an extremely favorable prognosis without adjuvant therapy [ 23 ]. However, the prognostic advantage of adjuvant chemotherapy in patients with advanced-stage disease remains unclear. Lin et al. [ 24 ] showed that multimodal treatment that includes adjuvant RT in addition to chemotherapy has a survival benefit for patients with stage IIIc USC, based on data from the national cancer database. The median OS was 33.6 and 42.6 months for chemotherapy and chemotherapy + RT groups, respectively [ 24 ]. Frimer et al. [ 25 ] reported the results of a prospective phase II study on three cycles of adjuvant paclitaxel 175 mg/m 2 + carboplatin (area under the curve [AUC] 6–7.5) followed by pelvic RT and an additional three cycles of paclitaxel + carboplatin (AUC 5) for stages I–IV in 132 patients with USC. This sandwich method demonstrated better survival than the commonly used single modality with OS probability of 0.64 and 0.18 at 2 and 5 years, respectively; although the toxicity was severe, it was tolerable [ 25 ]. However, all patients with stage III/IV disease underwent complete resection, and those with distant metastases were excluded from the study. Recently, molecular targeting agents and immunotherapies have been employed for various cancers because of the high incidence of human epidermal growth factor receptor 2 overexpression. Trastuzumab plus carboplatin/paclitaxel combination treatment showed promising results in a randomized phase II trial. The median PFS was 9.3 and 17.3 months in the standard (carboplatin/paclitaxel) and experimental (plus trastuzumab) arms, respectively, among 41 patients with stage III and IV disease [ 26 ]. Regarding immunotherapy, although most USC were microsatellite-stable, combination therapy with pembrolizumab and the multikinase inhibitor lenvatinib showed a 36% overall response rate in microsatellite-stable tumors, including 33 patients with USC who had previously received primary treatment [ 27 ]. Furthermore, poly(ADP-ribose) polymerase inhibitors may be therapeutic candidates because homologous recombination deficiency has been observed in 24% of non-endometrioid carcinomas [ 28 ]. Further research is required to confirm this hypothesis. This study had some limitations. First, this was a retrospective study, and the patient records were slightly old. Nevertheless, the survival data are reliable because there have been no significant changes in the treatment strategy for USC over the past 10 years in Japan. However, the future introduction of immune checkpoint inhibitors [ 29 30 31 32 ] is mandatory and is expected to significantly improve survival data. Second, no study exists on molecular classification. Currently, the differentiation of EC subtypes is based on immunohistochemistry and molecular techniques. EC can be identified as one of the following four molecular subtypes: mutations in the gene encoding polymerase epsilon, deficiency of mismatch repair and high microsatellite instability, mutations in the TP53 gene and abnormal expression of p53 protein, and no specific molecular profile. It is considered that patients’ prognosis and the implementation of appropriate treatment depend on the cancer subtype [ 33 34 ]. In conclusion, USC has many characteristics that differ from those of conventional endometrioid carcinoma. Developing a unique treatment strategy and achieving an accurate preoperative diagnosis with a comprehensive complete staging procedure are required to predict the prognosis of these patients.

The medical records of patients with EC who were diagnosed with USC and treated between 2006 and 2008 in 24 participating institutions of the Gynecologic Cancer Study Group of the Japan Clinical Oncology Group (JCOG-GCSG) were examined. The Kurume University School of Medicine approved this study (Institutional Review Board approval registration number: 11149), and it was registered with the University Hospital Medical Information Network (protocol number: UMIN 000011052). Ethical committee approval was obtained from each participating center. Patients were included in the study if they underwent a biopsy or curettage to confirm the diagnosis of EC. The participants underwent initial treatment with total hysterectomy, bilateral salpingo-oophorectomy, and pelvic and para-aortic lymphadenectomy, along with the collection of formalin-fixed neoplastic and non-neoplastic tissues for investigations. Adjuvant therapy and treatment for metastatic or recurrent disease were not specified in this protocol. For each patient, a local pathologist reviewed the material from the biopsy or curettage and hysterectomy and provided slides for a central review by five experienced pathologists (HT, MY, MF, AS, and TK). USC was diagnosed using hematoxylin-eosin-fixed specimens and immunohistochemical staining for estrogen receptor (ER), progesterone receptor (PR), and p53. The clinical data of patients who were diagnosed with USC were compared with those of patients who were not diagnosed with USC using CPR. Expert pathologists were not blinded to the local diagnoses and could review the original reports (the identifiers were redacted). The patients’ medical information, including age, operative method, International Federation of Gynecology and Obstetrics (FIGO) 2008 stage, myometrial invasion, washing cytology, adjuvant therapy, progression-free survival (PFS), and overall survival (OS), was collected. The χ 2 test was used for the univariate analysis of prognostic factors. The log-rank test was used to compare the PFS and OS of patients with and without USC. PFS was defined as the time from the initial therapy to disease recurrence or progression, death, or last contact, whichever occurred first. PFS was censored for patients who were alive and had not experienced disease progression or recurrence at the last contact. OS was defined as the time from initial therapy to death owing to any cause or last contact. The relationships of PFS and OS with the baseline characteristics (whenever feasible) were also examined using the log-rank test. A Cox proportional hazards model was used to estimate the corresponding hazard ratios (HRs) [ 9 ]. Statistical significance was set at p<0.05, and variables with p<0.05 in univariate Cox regression analysis were entered into a multivariate Cox regression model. HRs and 95% confidence intervals (CIs) were calculated for each model to determine the associations and significant predictors of survival. Statistical analyses were performed using SAS software (version 9.4; SAS Institute, Cary, NC, USA).