Background
Endometrial carcinoma is the sixth most common cancer in females worldwide and the third most common cancer in the female genital tract in Southeast Asia, with an annual incidence of 32,000 new cases [
2]. The major factors influencing postsurgical treatment decision are surgical staging and pathological findings [
3]. Histological type has traditionally been an important parameter for risk stratification [
3], but it has been shown to be poorly reproducible, with a significant disagreement rate as high as 35%, particularly in high-grade carcinoma [
4,
5].
The Cancer Genome Atlas Research Network has identified four molecularly distinct subgroups of endometrial carcinoma with different clinical outcomes:
POLE ultramutated (
polymerase-E exonuclease domain-mutant: POLE-mut), microsatellite instability hypermutated (mismatch repair-deficient: MMR-d), copy number high (p53-mutant/-abnormal: p53-abn), and copy number low (no specific molecular profile: NSMP) [
6]. This new approach provided better diagnostic reproducibility than the traditional histologic typing and grading, thus improving the precision of prognostication and therapeutic decision-making [
4,
5].
The European Society for Medical Oncology (ESMO), European Society for Radiotherapy & Oncology (ESTRO), and European Society of Gynaecological Oncology (ESGO) consensus guidelines for risk-based management of endometrial carcinoma have duly incorporated this molecular-based classification and introduced significant differences in management between each molecular subgroups [
9]. This paradigm shift has also been reflected in the updated 2023 International Federation of Gynecology and Obstetrics (FIGO) staging system [
10]. Besides, the extremely high mutation burdens seen in POLE-mut and MMR-d tumors, which greatly increase antigenicity, make immunotherapy an especially appealing treatment option [
11]. While some may be concerned about the economic implications of these molecular studies, particularly in resource-constrained settings, studies have shown that TCGA-based molecular categorization can be seamlessly integrated into routine clinical practice by using immunohistochemistry (IHC) for MMR proteins and p53, and
POLE mutation testing, rather than comprehensive molecular testing [
7,
8].
The present study aimed to assess the prevalence and clinical outcomes among the different molecular subgroups of endometrial carcinoma in Northern Thailand. We also compared the financial costs of the molecular information-based strategy to those of the conventional approach.
Materials and methods
Study population
The study cohort included women diagnosed with endometrial carcinoma at Chiang Mai University Hospital during January 2015 to December 2017. Inclusion criteria were patients with informed consent, available hysterectomy specimens for evaluation, and available clinical information and pathological results. We excluded individual without histologic materials and formalin-fixed paraffin-embedded (FFPE) tissue blocks and those with suspected cervical cancer. In cases where there was suspicion of cervical cancer extending to the endometrium, IHC panel was performed, including p16, estrogen receptor (ER), progesterone receptor (PR), and/or vimentin.
Pathology reports of eligible cases were reviewed (WD and TP). Pathologic findings included tumor size, histological type, extent of uterine wall invasion, presence of lymphovascular space invasion, endocervical stromal invasion, and lymph node metastasis. To confirm the tumor origin and the histological classification, a gynecologic pathologist (TP) reviewed the histological slides in accordance with the current World Health Organization Classification (WHO) guidelines [
12]. In cases where there were diagnostic discrepancies, a consensus was achieved through consultation with another gynecologic pathologist (SK). In each case, 1–2 representative histologic slides with the highest-quality tumor preservation were selected, and the corresponding FFPE blocks were retrieved for further testing.
Clinical information, comprising patient age, FIGO stage, and follow-up data up to January 2023, was obtained from electronic medical records and Chiang Mai Cancer Registry. Disease progression was defined as tumor recurrence or progression confirmed by imaging studies or histology. Progression-free survival (PFS) denoted the duration from the date of surgery to the date of disease progression, while overall survival (OS) was defined as the interval between the date of surgery and the date of last follow-up or death from any cause.
Immunohistochemistry for mismatch repair proteins and p53
Immunohistochemical staining was performed using BenchMark ULTRA IHC/ISH platform (Ventana Medical Systems, Roche Diagnostics, Tucson, AZ, USA), following the manufacturers’ instructions. To assess MMR protein status, a streamlined approach using two key IHC markers (PMS2 and MSH6) were employed instead of the full panel of four markers (MLH1, PMS2, MSH2, and MSH6). This approach was chosen based on established evidence demonstrating its cost-effectiveness and reliability [
13,
14]. Primary monoclonal antibodies against PMS2 (A16-4 Ventana clone ready to use; Optiview kit revelation with amplification) and MSH6 (SP93 Ventana clone ready to use; Optiview kit revelation) were used. Evaluation of MMR protein status was performed in cases exhibiting nuclear positivity in internal non-neoplastic tissues, serving as an internal positive control. The loss of MMR protein expression was defined as the complete absence of nuclear staining in tumor cells, while internal non-neoplastic tissues retained their staining. Cases showing the loss of at least one MMR protein expression were classified as MMR-d.
For p53 expression, we used a primary monoclonal antibody against the p53 protein (DO-7 clone, DAKO, dilution 1:100). The results were categorized into two groups: wild-type expression and abnormal expression. Specifically, wild-type p53 expression was characterized by a mixture of tumor cells displaying variable nuclear staining. Abnormal p53 expression included three main patterns: the diffuse pattern (strong positivity in at least 80% of tumor nuclei), the null pattern (absence of tumor nuclear staining), and the cytoplasmic pattern (unequivocal cytoplasmic staining of tumor cells accompanied by variable nuclear staining) [
15]. Cases with any type of abnormal p53 expression were classified as p53-abn.
DNA extraction and POLE mutation analysis
Genomic DNA was extracted from three ribbons of five µm-thick FFPE sections selected from regions with ≥ 60% tumor nuclei and < 10% necrosis, the extraction was performed according to the manufacturer’s recommendations using the QIAamp DNA FFPE Tissue kit (Qiagen, Toronto, ON, Canada). The concentration and purity of the DNA were quantified using a UV spectrophotometer (Nanodrop 2000, Thermo Fisher Scientific, Waltham, MA, USA) with A260/280 and A260/230 ratios that were expected to be within the range of 1.80–2.30. GAPDH was used as a reference gene for quantitative DNA analysis, and cases with no GAPDH amplification were not further processed for POLE mutation sequencing.
The
POLE gene exons 9–14 were then amplified using previously published primers [
16]. Following confirmation of the presence of target amplicons and absence of non-specific amplification products, bi-directional Sanger sequencing was performed according to standard protocols using BigDye™ Terminator v 3.1 Cycle Sequencing Kit (Thermo Fisher Scientific Baltics UAB, Vilnius) and Seqstudio genetic analyzer (Lifetechnologies holdings Pte Ltd, Singapore). Using the NM_006231.4 reference sequence, all sequences were evaluated for the presence of pathogenic mutations, including five hotspot mutations (P286R, S297F, V411L, A456P, and S459F) and other rare mutation types [
12]. To confirm the results, new PCR products from all mutation-positive samples were re-sequenced, and samples containing confirmed pathogenic mutations were classified as POLE-mut.
Molecular-based classification of endometrial carcinoma
The tumors in our cohort then were categorized using the Proactive Molecular Risk Classifier for Endometrial Cancer (ProMisE) algorithm, which has been proposed to be appropriate for clinical implementation [
17]. This model follows a stepwise approach, using MMR IHC,
POLE sequencing, and p53 IHC. Cases with suboptimal DNA quality were excluded.
Cost estimation for the application of molecular-based classification
We acquired cost information for IHC and adjuvant therapy from our institution’s cost announcements. The cost of
POLE mutation analysis (Sanger sequencing for five hotspot mutations) was calculated using the costs of prior in-house testing at our institution. First, for each patient, we evaluated the direct medical costs for adjuvant therapy according to ESMO/ESTRO/ESGO recommendations [
9]. These individual costs were summed to provide a reference cost. We then estimated the costs for the molecular-based strategy by combining the cost of IHC and/or molecular testing with the cost of adjuvant treatment. Finally, we compared the reference cost to the estimated cost derived after incorporating molecular information and management modification. Only patients with stage I-II were included in this comparison because molecular findings caused substantial management changes in this group, such as the omission of adjuvant therapy in the POLE-mut subgroup or the addition of chemo-radiation therapy in the p53-abn subgroup [
9]. We also assessed the costs of different strategies for EC molecular-based approach, including WHO algorithm [
12], ProMisE model [
17], and
POLE mutation testing alone.
Statistical analysis
All statistical analyses were performed using STATA version 16 (STATA Corp., Texas, USA). Continuous data were compared using one-way analysis of variance (ANOVA), while categorical data were assessed using the Chi-square test to analyze univariable associations of molecular subtypes of endometrial carcinoma. Kaplan-Meier methods and log rank test were used to explore the association between baseline prognostic variables and survival endpoints. A p value < 0.05 was considered statistically significant.
We conducted a flexible parametric regression analysis to elucidate the association of molecular subgroups and patient’s OS and PFS [
18]. For patients who had not experienced an event, they were censored at their last follow-up. To account for confounding factors, we generated a confounder summary score (CFS). This score was constructed by incorporating baseline prognostic variables (i.e., age, FIGO stage, histological type, degree of uterine wall invasion, and lymphovascular space invasion) into the flexible parametric model, excluding the molecular results [
19]. Subsequently, we estimated the model’s linear predictors to generate the CFS, which was then included in the main flexible parametric model to compute adjusted restricted mean survival time (RMST) for each molecular subgroup at five-year follow-up interval [
20].
Discussion
We found that the distribution of molecular subgroups in endometrial carcinomas was consistent with previous reports (Table
4) [
21‐
27], with the majority falling into the NSMP subgroup (39–64%), followed by the MMR-d (17–39%), the p53-abn (8–21%), and the POLE-mut (4-13.6%). Our findings also support the prognostic significance of molecular-based classification when applied to the Northern Thai population, indicating its broad applicability. Similar to previous reports [
26,
28], patients in the POLE-mut subgroup exhibited most favorable survival outcomes, while the p53-abn subgroup had the worst prognosis. These findings support the use of the recent ESMO/ESTRO/ESGO management guidelines in Northern Thailand.
Table 4
Comparison of the previously reported distribution of molecular subgroups of endometrial carcinoma
Talhouk et al. (2015) [ 21] | North America (Canada), n = 143 | 28.7 | 8.4 | 17.5 | 44.1 |
Stelloo et al. (2016) [ 22] | Europe, n = 834 | 26.3 | 5.9 | 8.9 | 59 |
Cosgrove et al. (2018) [ 23] | North America (USA), n = 982 | 38.6 | 4.0 | 8.5 | 48.9 |
Kommoss et al. (2018) [ 24] | Europe (Germany), n = 452 | 28.1 | 9.3 | 12.2 | 50.4 |
| Oceania (New Zealand), n = 88 | 17.1 | 9.1 | 10.2 | 63.6 |
| Asia (Korea), n = 240 | 19.6 | 10.8 | 17.1 | 52.5 |
| Asia (Japan), n = 265 | 26.4 | 13.6 | 21.1 | 38.9 |
Present study | Asia (Thailand), n = 138 | 28.2 | 5.1 | 13.8 | 52.9 |
All seven POLE-mut tumors in our cohort had endometrioid histotype, and 42.9% of them were FIGO grade 3, which aligns with the findings in a recent systematic review (43.4%) [
28]. Without
POLE mutation testing, patients with grade 3 tumors would typically be classified as having at least an intermediate risk, which warrants adjuvant therapy [
9]. Despite their high histologic grade, patients with POLE-mut tumors had a much lower recurrence rate when compared to the other subgroups [
26]. The favorable prognosis of POLE-mut tumors provides a compelling rationale for avoiding adjuvant therapy in early-stage patients, as such treatment is unlikely to improve their outcomes [
11,
26].
The p53-abn subgroup, on the other hand, is the most aggressive, necessitating adjuvant therapy. Serous histology is seen in a significant proportion of tumors in this subgroup. It should be noted that serous adenocarcinoma has a broad histological spectrum and is sometimes confused with low-grade endometrioid carcinoma [
28]. In cases where molecular testing is not performed, p53 IHC may be useful for screening for p53-abn endometrial carcinoma in tumors with high-grade histology or endometrioid carcinoma with high nuclear grade or overlapping features with serous adenocarcinoma [
28]. However, p53-abn tumors may account for 2 to 5% of low-grade endometrioid carcinomas [
12], and patients with these tumors may benefit from more aggressive treatment [
12]. The 2023 FIGO staging system designated stage I-II low-grade endometrioid endometrial carcinoma with p53-abn as stage IICm
p53abn [
10], and, as a result, p53 IHC screening is likely to be beneficial in all patients with early-stage disease.
MMR-d and p53-abn may be found in 9.8% and 15.7% of POLE-mut tumors, respectively, since
POLE mutations cause multiple subsequent mutations, including MMR and
p53 genes [
29]. Without
POLE mutation testing, some POLE-mut tumors in our study could have been misclassified as MMR-d. This is significant because
POLE mutations have been shown to be a prognostic driver even when MMR deficiency or
p53 abnormalities are present [
30].
One of the challenges of implementing molecular-based classification is the cost of molecular testing [
11], which is especially significant in resource-limited settings. While we anticipated additional costs, our findings suggest that incorporating
POLE mutation testing into early-stage patient management may not result in a significant increase in overall costs when compared to the conventional approach. This was primarily due to the fact that omitting adjuvant therapy in POLE-mut patients saved money. Interestingly, by carefully selecting patients with at least intermediate risk, doing
POLE mutation testing alone could result in a 1.6% cost reduction, suggesting that testing for
POLE mutations could be useful in early-stage patients, even in resource-constrained settings. In such cases, a single molecular testing center could serve as a reference laboratory. As technology advances and the volume of testing increases, the cost tends to decrease.
POLE mutation status can be determined using DNA sequencing techniques, such as next-generation sequencing, Sanger sequencing, or hotspot mutation analysis [
12]. In our case, the cost of next-generation sequencing testing was estimated to be 3–4 times that of five hotspot Sanger sequencing. Recently, quantitative PCR and droplet digital PCR to detect
POLE mutations have been developed [
27]. Table
5 compares the cost of testing, detected mutations, advantages, and limitations of each method, including Sanger sequencing in exon 9–14 used in this study.
Table 5
Comparison of detection techniques for pathogenic POLE mutation in terms of costs, detected mutations, advantages, and limitations
Next-generation sequencing | Highest | Full rangea | - High sensitivity of detection - Simultaneous testing for multiple mutation types (high throughput) | - Requirement of high technology equipment and expert bioinformatics in the interpretation - Time-consuming |
| Intermediate | 11 mutations | - High sensitivity of detection - Rapid processing | - Limited data - Further validation needed |
Droplet digital PCR assay [ 27] | Intermediate | Five hot spot mutations | - Rapid processing | - Limited data - Further validation needed |
Sanger sequencing in exon 9,10,11,12,13, and 14 b [ 34] | Intermediate | Full rangea | - High accuracy - Rather simple technique | - Limit of detection; not suitable for samples with a low proportion of mutation sequence - Not suitable for testing a large volume of samples |
Sanger sequencing for five hotspot mutations | Lowest | Five hot spot mutations, accounting for 95% of POLE-mut cases [ 31] | - High accuracy - Rather simple technique | - Limit of detection; not suitable for samples with a low proportion of mutation sequence - Not suitable for testing a large volume of samples |
In this new era of endometrial carcinoma treatment, molecular-based classification is becoming increasingly important in determining the best therapeutic approach. The treatment is rapidly evolving, driven by the development of novel strategies [
35‐
38]. For example, the discovery of new mutational pathways, such as the PI3K-AKT or FBXW7-FGFR2 pathways, could become a potential option for targeted therapy in patients with poor prognosis, particularly in the p53-abn or NSMP subgroups and non-endometrioid histotype [
37]. In addition, integrating preoperative radiomic analysis using MRI findings with molecular testing has the potential to improve risk stratification and enable more personalized treatments [
38].
This is the first study to look into the molecular data of endometrial carcinoma in Southeast Asia, and it provides information on the prevalence and outcomes of each molecular subgroup in this population. Because molecular-based classification has become a current topic of interest, this study focuses on its application in routine clinical practice in resource-limited settings. Limitations of our study include the study’s retrospective nature, small sample size, study design, and proportion of cases with suboptimal DNA preservation (25.8% of recruited cases), which may all affect the distribution of the subgroups. The DNA preservation factor could be attributed, at least in part, to our tropical environment and/or the duration of FFPE block storage. Given that FFPE blocks are typically transferred for testing immediately after diagnosis in clinical practice, DNA degradation is expected to be minimized. Because our classification algorithm begins with MMR IHC, cases with MMR-d and POLE-mut coexistence would be classified as MMR-d subgroup. These tumors are uncommon, accounting for only 5% of endometrial carcinomas, and data are limited [
10]. Another potential limitation is the use of a limited MMR IHC panel (PMS2 and MSH6). Yet, when compared to the full panel, this method has a less than 0.5% chance of missing MMR-d [
39] and this two-marker panel has now been proposed for clinical use in the 2023 FIGO staging system [
10].
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.