Endometrial cancer (EC) is the most common gynecologic malignancy in the United States (1, 2) and is increasing at an alarming rate. Based on clinical and histological variables, EC have been divided into two types. Type I tumors, or endometrioid EC, represent approximately 85% of cases, and are usually low-grade with favorable outcomes. Type II tumors, or non-endometrioid EC, typically arise in postmenopausal patients, and are frequently of high-grade thus contributing to a relatively poor prognosis (3, 4). Type II EC is largely comprised of uterine serous carcinomas (USC), which represents less than 10% of all endometrial cancers, yet accounts for more than half of deaths attributed to EC (4–6). While type I are estrogen-dependent, the role of estrogens in type II EC is less clear, although studies have shown that the pathogenesis of type II EC may at least partially depend on the level of estrogens, differently from what was previously believed (7).

Pathological reporting of EC has limitations due to poor reproducibility of tumor typing and to accurately identify patients at risk for recurrence or metastatic disease (8). The identification of the underlying molecular background of EC has resulted in the development of new molecular based classifications of EC, namely The Cancer Genome Atlas (TCGA), which stratifies EC into four distinct clusters with prognostic significance: polymerase ε (POLE) ultramutated, copy-number low, and microsatellite instability (MSI) hypermutated, and copy-number high (9, 10).

The genomic characterization of USC is distinct from endometrioid EC, with USC exhibiting a high frequency of genetic aberrations involving TP53, FBXW7, PPP2R1A and ARID1A.USC are mostly classified as copy-number high according to TCGA (10, 11). In contrast, endometrioid EC typically demonstrates a higher frequency of microsatellite instability, frequent activation of WNT/CTNNB1 signaling, and mutations of POLE, KRAS, and CTNNB1. Endometrioid EC are mostly classified as copy-number low according to TCGA (9–11).

Although EC is typically diagnosed early and associated with a high 5-year survival of 80-90%, USCs have higher recurrence rates and carry a poor prognosis with significantly lower survival (4, 12). USCs have high risk of recurrence (up to 80%) and are associated with an increased incidence of extrauterine disease on presentation (5, 12–15). Hence, the ability to reduce mortality of EC largely depends on developing tailored therapy and management for recurrent and advanced USC (16).

Patterns of recurrence and metastasis may provide prognostic information for EC. For example, patients with single-site local or nodal recurrence of EC have been associated with improved survival compared to those with pelvic recurrence or distant metastasis (17, 18), while patients with peritoneal carcinomatosis or multiple sites of recurrence have significantly worse post-relapse survival rates (18).

Comprehensive genomic analysis of USC provides a clearer understanding of the molecular pathways involved in oncogenesis (19). Knowledge of the somatic mutations may help predict patterns of metastases and recurrence in various cancers, including urothelial cancer, where patients with TP53 mutations are at a higher risk of lymph nodes metastases (12, 20–22).

The primary objective of this study is to investigate the association of somatic mutations occurring in a diverse patient population diagnosed with USC to the patterns of metastases, recurrence, overall survival (OS) and recurrence free survival (RFS).

With advances in genomic testing and improved understanding of cancer biology, an increasing number of cancer patients undergo mutational testing to identify potential targeted treatment options. As we gain more insight into the impact of the mutational make-up of cancer on prognosis, understanding its associations to known clinicopathologic factors is crucial. Our study cohort was largely comprised of minority Hispanic and Black women that were poorly represented in prior TCGA data and in the PanCancer Atlas (26). Furthermore, 40% of our patients had evidence of recurrence during follow-up while 49% of patients had metastatic disease at diagnosis. This is significantly higher than the TCGA cohort, where 32% presented with recurrence and only 11% had metastatic disease at diagnosis (10).

We showed that TP53, PIK3CA and ERBB2 are often mutated in USC, consistent with TGCA data. In our cohort, 79% of cases were classified as copy-number high, consistent with TCGA data, where 77% of cases with serous histology were classified as such (10). In a recent study by Watanabe et al. on 100 EC cases, TP53 mutations were associated with non-endometrioid histology (12). Several studies have demonstrated that human epidermal growth factor receptor 2 (HER2) a tyrosine kinase increasing cell proliferation encoded by ERBB2, is frequently overexpressed in USC (27–29). HER2 overexpression has also been associated with advanced-stage disease, and poorer survival outcomes in USC (25). Few ARID1A gene mutations in were identified in our cohort, consistent with the findings of the PanCancer Atlas, where ARID1A gene mutations were identified in 13% of cases (26, 30). Mutations in ARID1A in EC have been associated with promoting tumor invasion and metastasis, which may shed light on the significant mortality among women diagnosed with USCs (30).

In our cohort, ER and PTEN are expressed in the majority of USC on IHC. A study on 1054 women with EC showed that ER, though more common in type I EC, was expressed in 72% of type II EC (31). A study on 56 high grade EC showed that majority of USC were positive for PTEN on IHC (32). A study on 221 women with EC, showed that loss of PTEN expression was associated with endometrioid histology and favorable survival (33).

In our study, FBXW7 was mutated in 15% of USC. No association between FBXW7 mutations, ethnicity and race was identified in our study. FBXW7 was mutated in 29 out of 109 (27%) USC included in the PanCancer Atlas. In a study on 66 USC, FBXW7 mutations were identified in 18.2% of cases (11). On the study by Watanabe et al., FBWX7 mutations were associated with late-stage, vascular invasion, and lymph node metastasis (12).

In our cohort, USC most commonly metastasized to lymph nodes, peritoneum and pelvic organs, while recurrences were mostly nodal, peritoneal and at the vaginal cuff. These findings are similar to two prior studies: a study on 841 EC, which showed that the most common sites of recurrence of USC was the abdomen, including ascites, followed by the vaginal cuff and a study on 50 USC, which showed that the most common sites of metastases of USC were the lymph nodes, pelvic organs and peritoneum/omentum (13, 14).

In our study, PR expression was associated with non-Hispanic ethnicity and nodal metastases. A study on 99 endometrioid EC showed that nodal metastases correlated with negative PR status on IHC (34). This discrepancy may be related to the different histologies. The association between non-Hispanic ethnicity and PR expression is unclear.

We found that ERBB2 alterations were more common in patients with vaginal cuff recurrence and PIK3CA mutation was more common in patients with liver metastases. ERBB2 amplifications are associated with higher stage, chemoresistance, and lower survival, especially in Black patients (35, 36). Trastuzumab, a HER2/neu receptor inhibitor monoclonal antibody, demonstrated increased progression free survival when added to carboplatin paclitaxel, in patients with USC and ERBB2 overexpression (37). This highlights the significance of somatic tumor testing either at diagnosis or at recurrence to aid in prioritizing treatment options.

Women with presence of recurrence or metastases to the liver and ARID1A mutations had lower OS. A study on 86 recurrent EC, showed that recurrence to the liver was associated with lower OS, with an HR of 10 [3.72–26.81 95% CI] (14). A metanalysis on the prognostic significance of ARID1A in endometrium-related gynecological cancers showed that negative ARID1A expression predicted shorter progression free survival (38). ARID1A mutations affects multiple pathways, and may mediate resistance to platinum chemotherapy, possibly explaining the lower OS observed in our study (39). Therapies targeting the pathways affected by ARID1A mutations, such as poly(ADP-ribose) polymerase inhibitors, immune checkpoint inhibitors and mTOR inhibitors, have shown activity in preclinical models and in patient, and could be implemented in patients with ARID1A-mutated USC (40, 41).

Finally, on multivariable Cox proportional hazards regression analysis, the presence of recurrence or metastases to the liver and/or the peritoneum were independently associated with shorter OS regardless of their mutational status, race, ethnicity or other sites of recurrence or metastases. Various studies attempted to build predictive survival models for USC, including a study by Chen et al. on 110 women with USC which showed that a combination of mutated genes, a 4-gene signature, was predictive of OS (42). Differently from our study, their model did not include sites of metastases or survival as potential predictors. Knowledge of the associations between survival data and sites of recurrence or metastases of USC is valuable: it allows for a more accurate risk stratification and helps the oncologists and radiologists to potentially formulate more appropriate follow-up strategies (43).

Some limitations of this study should be noted. This is a retrospective study performed at a single institution with inherent selection bias. The relatively small patient cohort and the widely variable follow-up period may have limited the power in detection of some of the associations between mutations and patterns of metastases and recurrence, as well as the prognostic values of mutations. Furthermore, our sample included only patients with USC selected from a tertiary cancer center and may not be representative of patients treated at a community hospital. We grouped different genetic aberrations involving the same gene under that gene to facilitate analysis until additional data are available. This methodology has also been previously utilized and reported (11, 21).

One of the significant strengths of this study is the cohort of Hispanic and Black women diagnosed with USC with genomic testing. Our study showed that mutational status of USC had implications on pattern of metastases and survival, and that sites of recurrence and metastases influence survival. These findings should be assessed in larger studies, to confirm our findings and may be valuable for future trial design.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by University of Miami. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

MH: Supervision, ideation, and draft editing. SW: statistical analysis and draft editing. FA: ideation, statistical analysis, draft writing, and editing. CL: data collection and draft editing. NG: data collection MM: data collection and draft editing. All authors contributed to the article and approved the submitted version.