Recently, owing to next-generation sequencing technology, the molecular characteristics and mutational profiles of EC have been identified. Specifically, TCGA has classified EC into four categories: POLE (ultramutated), MSI (hypermutated), copy-number low (endometrioid), and copy-number high (serous-like) and found these classifications to be associated with prognosis [13].
During initial management of EC, surgery is often the standard of care, and identifying the risk of relapse based on the pathological findings of excised specimens has a strong influence on decision-making during the course of postoperative treatment [5, 18]. The pathological findings useful for determining risk include histology, the degree of muscle invasion, lymph vascular space invasion, and the presence or absence of extrauterine lesions [18]. However, these factors do not account for the importance of the molecular and mutational signatures of EC in prognosis. Molecular characteristics of endometrioid carcinoma were first mentioned in the revised 5th edition of the WHO classification of female genital tumors [20].
Although many studies have investigated molecular genetics using next-generation DNA sequencing, this technique is expensive and impractical to apply in routine clinical settings. Intra-tumor heterogeneity is thought to be present in the earliest days of carcinogenesis [11, 12], and metastatic lesions may have heterogeneous characteristics even at the start of treatment. Therefore, analysis of only primary lesions might lead to a misunderstanding of metastatic tumors.
TP53 mutation occurs early in tumorigenesis and is one of the most important molecular factors associated with unfavorable prognosis [2, 12, 21]. According to TCGA, analysis of dMMR is also crucial for prognosis and MMR deficiency arises in the early stages of tumorigenesis. Moreover, the inherited pathogenic germline variants in MMR genes cause Lynch syndrome [5, 7, 19, 22, 23]. Most previous studies, including those using TCGA data, were based on data from primary lesions only; no study has yet investigated the molecular features of each metastatic lesion. In contrast to next-generation sequencing, IHC is easy to perform and apply in routine clinical practice. Moreover, IHC can reveal the status of TP53 mutation and MMR deficiency [21, 23]. To the best of our knowledge, this is the first study to analyze the association of clinical outcomes with p53 and MMR protein status of both, primary lesions and all metastatic LNs using IHC and clinicopathological data.
TP53 mutation is the single most influential factor affecting prognosis [21] and aberrant p53 expression was observed in 39.4% of the patients in our cohort. Previously reported frequencies of p53 overexpression vary widely for a range of reasons, including the study cohort. For example, Haraga et al. reported a p53 overexpression rate of 14.7% in a cohort of patients with all stages of EC, and Saijo et al. reported aberrant p53 status in 63% of patients with endometrial carcinosarcoma [19, 24].
William et al. reported a variation in the number of mutations in primary lesions and their paired metastases; even among common driver mutations, the concordance rate of primary and metastatic lesions was only 83% [2]. The concordance rate of truncal mutations, such as those in TP53 was also less than 100% [2]. For example, in the current study, the concordance rate of p53 status between primary and metastatic lesions was 93.9%; and even within the same classification (both primary and metastatic lesions as wild or aberrant), six cases (19.4%) showed heterogeneous expression. Three out of six (50%) cases exhibited low-grade endometrioid histology, and two out of the three cases exhibited dMMR. Köbel et al. reported the possibility of later acquisition of TP53 mutation in endometrioid carcinoma, especially in mutator phenotype like dMMR [21]. The heterogeneous expression revealed in our study could be attributable to the subsequent occurrence of TP53 mutations, especially in low-grade endometrioid and dMMR cases.
Many studies have evaluated MMR status in EC. Approximately 20–40% of patients with EC demonstrate dMMR, largely resulting from promoter hypermethylation and silencing of MLH1 [3, 23, 25, 26]. Ida et al. reported a dMMR rate of 60% in mixed endometrial carcinomas, while Saijo et al. reported a dMMR rate of only 10.5% in only endometrial carcinosarcoma [3, 19, 24–26]. Our cohort displayed a relatively high dMMR rate of 48.5%, which is likely due to the inclusion of different cohorts. The variation in tumor volume among the cohorts is another possible cause of this variation. For example, Casey et al. reported that epigenetic silencing of MLH1 is significantly associated with large tumor volume [27]. Because our study evaluated patients with EC at an advanced stage, we observed considerably large tumor volumes with a median tumor size of 42.0 cm3. Tumor size is a significant prognostic factor, and the cutoff value between large and small tumors is a diameter of 2 cm [14]. Our study observed diameters much longer than the cutoff value, and in all cases, the tumor size was ≥2 cm.
We speculate that the large tumor volume is also related to the heterogeneous staining pattern of MMR proteins. Of the 16 cases of dMMR in this study, eight presented homogenous loss of MLH1/PMS2 and heterogeneous expression of MSH2/MSH6. This same expression pattern is often observed in colorectal cancer, and the possibility of secondary MSH2/MSH6 inactivation has been proposed [28]. Because stage IIIC EC is advanced by definition, the primary lesion volume tends to be large owing to repeated cell division. Additionally, because patients with EC with dMMR have a faulty DNA mismatch repair system, more cell division occurs, resulting in increased DNA replication errors and mutations, ultimately resulting in heterogeneous MMR protein expression.
Johannes et al. reported higher levels of heterogeneity in LN metastases than in distant organ metastases [29]. Our findings of inter-LN heterogeneity corresponded to previous findings [29], and in the metastatic LNs, we could validate the polyphyletic evolution of different mutational signatures from primary lesions occurring even in fundamental molecules, such as p53 and MMR proteins.
Regarding the prognostic value of p53, Kaplan–Meier and Cox univariate analyses showed that p53 expression in metastatic lesions was significantly associated with PFS in our study. Because of its advanced nature, stage IIIC EC has relatively poor prognosis, and a significant difference according to p53 expression could be difficult to observe. However, even under these circumstances, aberrant p53 status in metastatic lesions could be a superior prognostic predictor to that in primary lesions.
The association between MMR deficiency and survival outcomes remains controversial [3]. Because EC with dMMR is characteristically hypermutated according to a TCGA study [7], it displays large amounts of neoantigens and perhaps high immunogenicity. These signatures could result in prognostic differences between patients with MMR proficient and deficient EC. Among the dMMR cases, although we failed to demonstrate significant differences between the observed classifications ((1) MLH1-PMS2 or MSH2-MSH6, (2) MLH1-PMS2 and MSH2-MSH6) (p=0.197), the recurrence rate in cases from classification two (44.4%) was higher than in cases from classification one (5.9%). We believe that the observed classifications could explain the prognostic differences.
For clinical data, LNR has been proposed as a meaningful prognostic factor in patients with stage IIIC EC [15–17]. Ali et al. found an LNR >0.15 to be an independent prognostic factor for PFS and OS [15]. Moreover, Fleming et al. reported that patients with an LNR >0.5 had significantly worse PFS than those with an LNR ≤0.5 [16]. Additionally, Polterauer et al. reported that only LNR was significantly associated with PFS and OS in multivariate analysis [17]. Under our cutoff value by ROC curve analysis, an LNR ≥0.11 showed a marginal association with PFS, though the correlation was not significant in Kaplan–Meier or Cox univariate analyses (p=0.052 and p=0.062, respectively). Thus, based on our findings, molecular characteristics have a superior prognostic impact compared with clinical factors.
In conclusion, our study suggests that aberrant p53 expression in metastatic lesions could provide superior prognostic information than that in primary lesions in patients with stage IIIC EC. Polyclonal development from the primary lesion to individual LNs is possible, even in fundamental molecular genetics, such as p53 and MMR proteins. Further study is required to verify the association between MMR protein expression and prognosis. Although our study cohort was relatively small and confined to a single institution, this study provides a thorough evaluation and interpretation of p53 and MMR proteins in primary and all involved LNs in patients with stage IIIC EC. The technical use of this study result was for the IHC only; therefore, we believe these findings can be applied to daily medical practice easily.