
==== Front
ESMO Open
ESMO Open
ESMO Open
2059-7029
Elsevier

S2059-7029(24)01480-7
10.1016/j.esmoop.2024.103710
103710
Original Research
Identification of molecular subtypes for endometrial carcinoma using a 46-gene next-generation sequencing panel: a retrospective study on a consecutive cohort
Guo Q. 12†
Tang S. 23†
Ju X. 12
Feng Z. 12
Zhang Z. 4
Peng D. 4
Liu F. 4
Du H. 4
Wang J. 4
Zhang Y. 4
Wang G. 4
Zhang Z. 4
Cai S. 4
Diao Y. 12
Zhong Y. 12
Wu X. wu.xh@fudan.edu.cn
12∗
Zhou X. xyzhou100@163.com
23∗
Wen H. wenhao_fdc@163.com
12∗
1 Department of Gynecologic Oncology, Fudan University Shanghai Cancer Center, Shanghai
2 Department of Oncology, Shanghai Medical College, Fudan University, Shanghai
3 Department of Pathology, Fudan University Shanghai Cancer Center, Shanghai
4 Burning Rock Biotech, Guangdong, China
∗ Correspondence to: Dr Hao Wen, Department of Gynecologic Oncology, Fudan University Shanghai Cancer Center, Shanghai 200032, China; Department of Oncology, Shanghai Medical College, Fudan University, 270 Dong-an Road, Shanghai, China. Tel: +86-21-64175590 wenhao_fdc@163.com
∗ Prof. Xiaoyan Zhou, Department of Pathology, Fudan University Shanghai Cancer Center, Shanghai 200032, China; Department of Oncology, Shanghai Medical College, Fudan University, 270 Dong-an Road, Shanghai, China. Tel: +86-21-64175590 xyzhou100@163.com
∗ Prof. Xiaohua Wu, Department of Gynecologic Oncology, Fudan University Shanghai Cancer Center, Shanghai, 200032, China; Department of Oncology, Shanghai Medical College, Fudan University, 270 Dong-an Road, Shanghai, China. Tel: +86-21-64175590 wu.xh@fudan.edu.cn
† These authors contributed equally.

16 9 2024
10 2024
16 9 2024
9 10 103710© 2024 The Author(s)
2024
https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Background

Traditional classification tools for endometrial carcinoma (EC), such as DNA sequencing, immunohistochemistry (IHC), or PCR, are cumbersome and time-consuming. Large next-generation sequencing (NGS) panels have simplified testing but are expensive. In this study, we propose a concise NGS panel as an effectively viable approach for classifying EC.

Materials and methods

We retrospectively enrolled a consecutive EC cohort of hysterectomy with bilateral salpingo-oophorectomy from Fudan University Shanghai Cancer Center between 2020 and 2022. A 46-gene NGS panel was utilized to identify POLE exonuclease domain mutations, microsatellite instability-high (MSI-H), TP53 mutations, and other clinically relevant targets.

Results

Tumor tissue samples from 331 EC patients were evaluated, with 284 (85.8%) cases classified as endometrioid endometrial carcinoma. The median follow-up time was 32.6 months (n = 303), during which 23 patients experienced recurrence or disease progression. Using the concise NGS panel, patients were stratified into four molecular subgroups according to the World Health Organization classification criteria: POLE mut (n = 47; 14.2%), mismatch repair deficiency (dMMR) (n = 79; 23.9%), non-specific molecular profile (n = 148; 44.7%), and abnormal p53 expression (p53 abn) (n = 57; 17.2%). POLE mut displayed the most favorable prognosis, while p53 abn had the worst prognosis (P < 0.001). The concordance between NGS and IHC was 91.8% (269/293) for detecting MMR status and 65.3% (201/308) for detecting p53 status. Patients detected solely by NGS had significantly worse prognosis than those detected solely by IHC, indicating higher accuracy of the NGS panel. With the molecular subtyping information, adjuvant treatment plans for 19.6% of patients could potentially be altered, mainly concentrated in the POLE mut and p53 abn subtypes. This panel also aids targeted therapy and poly (ADP-ribose) polymerase (PARP) inhibitor-related gene mutation detection, as well as auxiliary genetic screening.

Conclusion

Our study demonstrates that the concise NGS panel is an effective ‘one-stop’ strategy for precisely classifying EC with high clinical availability.

Highlights

• The NGS panel accurately categorizes patients into four molecular subtypes, showing 91.8% agreement with IHC for MMR status.

• Exclusive NGS identification of p53 abnormalities correlated with worse patient outcomes compared to IHC detection alone.

• Molecular subtyping can change 19.6% of treatment plans, notably in POLE mutation and p53 abnormality cases.

• The NGS panel is crucial for targeted therapy, gene detection for PARP inhibitor response, and genetic screening.

Key words

endometrial carcinoma
NGS panel
molecular subgroups
adjuvant treatment
==== Body
pmcIntroduction

Endometrial cancer (EC) is a malignancy that affects the inner epithelial lining of the uterus. With an increasing global incidence,1, 2, 3 it is the sixth most common cancer in women, accounting for >420 000 new cases worldwide in 2022.4 Approximately 18% of these new cases are reported in China.5 Based on cytomorphological features, the 2020 World Health Organization (WHO) classification of female genital tumors divides EC into seven main histotypes, guiding prognosis and treatment decisions.6 High-grade endometrioid, serous, and clear cell histotypes are aggressive with worse prognoses than low-grade endometrioid EC.7

Despite advances in morphologic and immunohistochemical characterization of EC, the histologic subtyping of EC remains subject to considerable interobserver variation.8, 9, 10 In recent years, there has been a concerted effort to develop clinically applicable and reproducible methods for molecular classification of EC. Since the establishment of The Cancer Genome Atlas (TCGA) classification based on genomic abnormalities in 2013, which identifies four molecular subtypes with distinct prognostic outcomes, the TransPORTEC classification and Proactive Molecular Risk Classifier for EC (ProMisE) have also been developed.11, 12, 13 According to the National Cancer Comprehensive Network (NCCN) guidelines for EC, molecular analysis for POLE mutations, mismatch repair (MMR)/microsatellite instability (MSI), and abnormal p53 expression (p53 abn) should be conducted to complement morphologic assessment of histologic tumor type.14 Furthermore, the updated 2023 International Federation of Gynecology and Obstetrics (FIGO) staging encourages the utilization of molecular classification [POLE mut, mismatch repair deficiency (dMMR), non-specific molecular profile (NSMP), p53 abn] for all EC cases for prognostic stratification and guiding treatment decisions.15

Although the four-tier molecular classification systems ProMisE/TransPORTEC have been well established to inform the prognosis of EC patients.12,13 there are still some issues that need to be addressed. Adjuvant therapy is the standard of care for surgically staged EC patients,14 yet it remains challenging to identify those who will benefit from adjuvant therapy. Whether the four-tier molecular classification systems can effectively identify surgically staged EC patients who may benefit from adjuvant therapy and guide adjuvant treatment decisions has not been well established. Additionally, the four-tier molecular classification systems often rely on multiple testing platforms, such as comprehensive genomic profiling, immunohistochemistry (IHC) testing, and PCR, which may introduce high cost, labor intensiveness, and long turnaround time. There is an unmet need to develop a cost-effective classification system for EC in routine clinical practice.

To address these issues, we aimed to develop a pragmatic molecular classification system for EC through a comprehensive one-stop NGS strategy and investigate the system’s value in predicting response to different adjuvant therapy regimens by retrospectively including a large cohort of Chinese EC patients who underwent 46-gene NGS panel testing. Additionally, the most appropriate adjuvant treatment in patients with resectable EC using this classification system was also explored.

Materials and methods

Patients and samples

In our study, we conducted a retrospective analysis of tissue samples from a consecutive EC cohort with 331 patients who underwent hysterectomy and bilateral salpingo-oophorectomy at Fudan University Shanghai Cancer Center between 2020 and 2022. Patients who underwent preoperative chemotherapy (CT) or other treatments before surgery were excluded. Tumor tissues, fixed in formalin and embedded in paraffin, were obtained from each patient after surgery, with a minimum tumor content of 50% after microdissection, and necrotic areas not exceeding 10%, meeting standardized criteria. Disease-free survival (DFS) was defined as the interval from the date of surgery to the occurrence of any defined events, such as local recurrence, distant metastasis, or death. If a patient did not experience any of these events during the study, DFS calculations were carried out until the latest follow-up, which was on 30 December 2023.

The study was approved by the institutional ethics and review board of Fudan University Shanghai Cancer Center (approval No. 2022251-11), and informed consent was obtained from the patients or their representatives for the use of their clinical data and tumor tissue specimens.

DNA sequencing and molecular classification of patients

DNA was extracted from formalin-fixed, paraffin-embedded (FFPE) tumor tissues using the QIAamp DNA FFPE Tissue Kit (Qiagen, Hilden, Germany) following the manufacturer’s instructions. A 46-gene NGS panel (ColonCore, Burning Rock Biotech, Guangzhou, China) was employed for sequencing, as previously described.16 The ColonCore panel was designed to simultaneously detect MSI status and mutations in 46 genes associated with Lynch syndrome (LS) (e.g., MLH1, MSH2, MSH6, PMS2) and key genes for the molecular classification of endometrial cancer (e.g., POLE, TP53). It also incorporates additional genes frequently altered in endometrial cancer (PTEN, CTNNB1, ARID1A, PIK3CA), as well as genes that are integral for targeted therapy and prognostic evaluation (e.g. BRCA1, BRCA2, ATM, CHEK2) (detailed gene list see Supplementary Table S1, available at https://doi.org/10.1016/j.esmoop.2024.103710). Sequencing was carried out on the NextSeq platform (Illumina Inc., San Diego, CA) at a depth of 1000× following the manufacturer’s instructions. The mutation detection process and definition refer to previously published article,17 variants were filtered using the VarScan filter pipeline, and loci with depths <100 were filtered out. Base calling in tissue samples required at least eight supporting reads for single-nucleotide variant (SNV) and two and five supporting reads for insertion–deletion variants (Indel), respectively. Variants with a population frequency >0.1% in the ExAC, 1000 Genomes, dbSNP, or ESP6500SI-V2 databases were grouped as single-nucleotide polymorphisms and excluded from further analysis.

MSI-ColonCore utilized a read-based approach for MSI phenotype detection, relying on the coverage ratio of specific repeat length sets as the primary feature for each microsatellite locus. If the coverage ratio at a locus fell below a given threshold, it was classified as unstable. The MSI status of a sample was determined based on the percentage of unstable loci in that specific sample. A tumor sample is considered to be MSI-H if >30% of the marker loci are length-unstable, and microsatellite stable (MSS) if the percentage of length-unstable loci is <30%.

Patients were molecularly classified based on genomic features determined by DNA sequencing, in the following order: POLE exonuclease domain mutations (POLE mut group), MSI-H (dMMR group), and TP53 mutations (p53 abn group). Patients not exhibiting any of these three features were categorized as non-specific molecular profile (NSMP) group.

IHC and PCR

For IHC, 4-μm thick FFPE tumor tissue slices were utilized. Monoclonal antibodies targeting MLH1, PMS2, MSH2, MSH6, and p53 were applied. Mismatch repair proficiency (pMMR) was determined when all four MMR markers (MLH1, PMS2, MSH2, and MSH6) exhibited nuclear expression. Mismatch repair deficiency (dMMR) was identified when there was a complete loss of nuclear expression in any of the four MMR markers within tumor cells. pMMR is considered consistent with the MSS determined by ColonCore panel testing; similarly. dMMR by IHC is considered consistent with MSI-H determined by ColonCore panel testing. The p53 IHC results were categorized as abnormal or normal expression. Abnormal expression was further classified into three patterns: (i) all tumor cells negative, (ii) >80% tumor cells with strong nuclear positivity, and (iii) cytoplasmic staining in tumor cells. The interpretation of IHC results was independently conducted by two professional pathologists.

MSI analysis was conducted through PCR using the Microsatellite Instability Detection Kit (Microread Gene Technology Co., Ltd, Beijing, China) based on six mononucleotide repeat markers (BAT-25, BAT-26, NR-21, NR-24, NR-27, and MONO-27). Tumors showing instability in two or more markers were classified as MSI-H, otherwise considered MSS.

Statistical analysis

All statistical analyses were carried out using R software (version 4.0.3, https://www.r-project.org/). Group differences for categorical variables were compared using chi-square tests or Fisher’s exact tests, while comparison of continuous variables was assessed with the Mann–Whitney test and Kruskal–Wallis test between two or more than two groups, respectively. Multivariable Cox regression analysis was conducted to adjust for confounding factors. Kaplan–Meier analysis was used to estimate survival outcomes, and log-rank tests were used to determine differences in DFS between groups. Median follow-up time was calculated using the inverse Kaplan–Meier method. A P value <0.05 was considered as statistically significant.

Results

Patient characteristics

A total of 331 patients diagnosed with EC were included in the present study. The baseline characteristics are displayed in Supplementary Table S2, available at https://doi.org/10.1016/j.esmoop.2024.103710. The age at diagnosis ranged from 26 to 85 years, with a median of 55 years. The majority of cases (85.8%) were classified as an endometrioid histotype, while 4.8% were carcinosarcoma, 3.0% were mixed carcinoma, and the remaining exhibited serous adenocarcinoma and clear cell histology. Among these patients, 204 (61.6%) were diagnosed with FIGO stage IA, 52 (15.7%) at stage IB, 20 (6.0%) at stage II, 39 (11.8%) at stage III, and 16 (4.8%) at stage IV (Supplementary Table S2, available at https://doi.org/10.1016/j.esmoop.2024.103710). Among 303 patients with available follow-up data, the median follow-up time was 32.6 months. Out of these 303 patients, 23 (7.6%) patients developed recurrent or progressive disease.

Molecular classification of EC via a one-stop NGS strategy

Utilizing the WHO classification criteria, we categorized patients into four molecular subgroups employing the streamlined NGS panel: POLE mut (14.2%), dMMR (23.9%), NSMP (44.7%), and p53 abn (17.2%) (Figure 1A). Consistent with previous studies,13 POLE mut demonstrated the most favorable prognosis, while p53 abn exhibited the worst (P < 0.001, Figure 1B).Figure 1 Molecular classification of EC by NGS and comparison of clinical features. (A) Flowchart of molecular classification of patients by NGS. (B) Kaplan–Meier survival curves for DFS among patients with different molecular subtypes. (C-F) Comparison of clinical features, including adjuvant therapy (C), FIGO stages (D), pathological subtypes (E), and lymphovascular space invasion status (F) among the four subtypes. ADT, adjuvant therapy; CCC, clear cell carcinoma; CRT, chemo-radiotherapy; CS, carcinosarcoma; CT, chemotherapy; DFS, disease-free survival; EC, endometrioid carcinoma; MEC, mixed endometrioid carcinoma; dMMR, mismatch repair deficiency; MSI-H, microsatellite instability-high; MSS, microsatellite stable; NGS, next-generation sequencing; NSMP, non-specific molecular profile; RT, radiotherapy; SC, serous carcinoma; UC, undifferentiated carcinoma.

We then explored the association between molecular classification and clinical characteristics. P53 abn had higher adjuvant therapy rates (61.4%) versus POLE mut (31.9%), dMMR (43.0%), and the NSMP (29.1%) (P < 0.001, Figure 1C). Besides, the p53 abn subtype displayed more patients with FIGO stage III-IV (35.1%, P = 0.002, Figure 1D), higher incidence of subtypes like carcinosarcoma (19.3%) and serous adenocarcinoma (15.8%, P < 0.001, Figure 1E), and increased lymphovascular space invasion (43.6%, P = 0.001) (Figure 1F). These results partially elucidate the poorest prognosis in p53 abn subtype patients.

Association between the molecular subtypes and genomic alteration

In general, the most common genetic alterations occurred in PTEN, PIK3CA, ARID1A, TP53, CTNNB1, and others (Figure 2A). The POLE mut subtype had the highest mutation count than the others (P < 0.001, Figure 2B). We further excluded the POLE mut subtype and compared the genomic landscape with the remaining three subtypes. Notably, the dMMR subtype had higher mutation frequencies for most genes compared to other subtypes such as PTEN, ARID1A, PIK3CA, POLE, and MSH6 (Figure 2C). However, CTNNB1 mutation frequency was higher in the NSMP subtype than that in the dMMR subtype (P = 0.006, Figure 2C). DFS showed that patients with CTNNB1 mutations tended to have a better prognosis than those with wild-type CTNNB1 in the NSMP subtype (P = 0.067, Figure 2D).Figure 2 Genomicprofiles of EC patients with different molecular subtypes. (A) Genomic alterations of EC patients. (B) Comparison of the number of alterations among the four subtypes. (C) Comparison of genomic alteration detection rates of high-frequency mutated genes among dMMR, p53 abn, and NSMP subtypes. (D) Kaplan–Meier survival curves for DFS compared with CTTNB1 mutations with the wild-type in EC patients with NSMP subtype. (E) Genomic alteration profile of patients within p53 abn subtype. Different colors on the oncoprint represent the mutation types. Only the genes with a mutation rate of ≥5% were included in the oncoprint. (F) Correlation of PTEN and ARID1A within the p53 abn subtype. (G) Mutual exclusivity of mutations in PTEN and KRAS within the p53 abn subtype. Comparison of clinical features, including myometrial invasion (H), lymphovascular space invasion (I), and pathological subtypes (J). CCC, clear cell carcinoma; CS, carcinosarcoma; DFS, disease-free survival; EC, endometrioid carcinoma; MEC, mixed endometrioid carcinoma; dMMR, mismatch repair deficiency; NSMP, non-specific molecular profile; SC, serous carcinoma; UC, undifferentiated carcinoma; WT, wild-type; LGR, large genomic rearrangement.

Specifically, in the p53 abn subtype, notable mutations were observed in PTEN, PIK3CA, and ARID1A (Figure 2E). PTEN co-mutated with ARID1A (P = 0.03, Figure 2F) and was mutually exclusive with KRAS (P = 0.005, Figure 2G). PTEN mutations in this subtype were associated with lower myometrial invasion (P = 0.005, Figure 2H), reduced lymphovascular space invasion (P = 0.012, Figure 2I), and a higher proportion of endometrioid carcinoma (P < 0.001, Figure 2J), while in the dMMR subtype, PTEN, ARID1A, and PIK3CA had the highest mutation frequencies (Supplementary Figure S1A, available at https://doi.org/10.1016/j.esmoop.2024.103710). RNF43 co-mutated with PTEN (P = 0.04) and APC (P = 0.02), while TP53 co-mutated with AKT1 (P < 0.001, Supplementary Figure S1B, available at https://doi.org/10.1016/j.esmoop.2024.103710). dMMR with TP53 mutation showed a greater proportion of myometrial invasion compared to those without TP53 mutation (P = 0.024, Supplementary Figure S1C, available at https://doi.org/10.1016/j.esmoop.2024.103710). In the NSMP subtype, PTEN, ARID1A, PIK3CA, CTNNB1, and KRAS were the most frequently mutated genes (Supplementary Figure S1D, available at https://doi.org/10.1016/j.esmoop.2024.103710). CTNNB1 was mutually excluded with KRAS (P = 0.009) and PTEN (P = 0.03), and was associated with AKT1 (P = 0.007, Supplementary Figure S1E, available at https://doi.org/10.1016/j.esmoop.2024.103710). NSMP with MET amplification, mutually exclusive with PIK3CA, had a higher lymph node metastasis proportion (30.8% versus 5.5%, P = 0.014, Supplementary Figure S1F, available at https://doi.org/10.1016/j.esmoop.2024.103710). In the POLE mut subtype, 61 POLE mutations were observed with five hot spot mutations (Supplementary Table S1, available at https://doi.org/10.1016/j.esmoop.2024.103710). Patients with multiple POLE mutations had a significantly higher mutation count (P < 0.001, Supplementary Figure S1G, available at https://doi.org/10.1016/j.esmoop.2024.103710). POLE mut with multiple mutations tended to display higher FIGO stage (P = 0.057, Supplementary Figure S1H, available at https://doi.org/10.1016/j.esmoop.2024.103710) and a greater proportion of lymphovascular space invasion (P = 0.08, Supplementary Figure S1I, available at https://doi.org/10.1016/j.esmoop.2024.103710).

Concordance between NGS-based detection, IHC, and PCR results

For the molecular classification of EC, MSI status assessment commonly involves IHC and PCR-based assays. Our study utilized a streamlined NGS panel for MSI analysis, comparing results with MMR IHC in 293 cases, and for the discrepant results, PCR detection was involved.

The majority of cases (91.8%, 269/293) exhibited concordant results between IHC and NGS for MSI results. However, there were 15 cases that showed MSI-H results with NGS but pMMR status with IHC (Figure 3A). Additionally, nine cases demonstrated MSS results with NGS but had dMMR status with IHC (Figure 3A). Further validation by PCR was carried out on the 24 discordant samples. Of these, 18 (75%) displayed consistency between PCR and NGS results (Figure 3B), whereas only 6 (25%) cases were concordant between PCR and MMR IHC (Figure 3C). Notably, after excluding the POLE mut subtype, the concordance between NGS and IHC in detecting MSI status increased to 92.5% (233/252) (Figure 3D), and for the POLE mut subtype, the consistency between IHC and NGS regarding MSI in the POLE subtype has reached 87.8% (36/41).Figure 3 Concordance between NGS-based detection and IHC/PCR on post-operative specimens. (A) Concordance of MSI status between NGS and IHC methods. (B) Concordance of MSI status between NGS and PCR methods in samples with discordant results from NGS and IHC methods. (C) Concordance of MSI status between PCR and IHC methods in samples with discordant results from NGS and IHC methods. (D) Concordance of MSI status between NGS and IHC methods after exclusion of the POLE mut subtype. (E) Concordance of p53 expression status with IHC and TP53 alteration status with NGS. (F) Concordance of p53 expression status with IHC and TP53 alteration status with NGS after exclusion of the POLE mut and dMMR subtype. (G) Kaplan–Meier survival curves for DFS between patients with and without p53 abn identified by NGS or/and IHC. DFS, disease-free survival; dMMR, mismatch repair deficiency; IHC, immunohistochemistry; MSI-H, microsatellite instability-high; MSS, microsatellite stable; NGS, next-generation sequencing; pMMR, dMMR, mismatch repair proficiency.

Subsequently, we explored the concordance between p53 expression status with IHC and TP53 alteration status with NGS. The concordance rate between p53 IHC and TP53 mutational analysis was 65.3% (201/308) (Figure 3E). After excluding the POLE mut and dMMR subtypes, the concordance rate slightly increased to 67.2% (127/189) (Figure 3F).

Delving into the accuracy of the NGS panel in TP53 alterations, we grouped patients using IHC and NGS by p53 status and compared the prognostic differences. Interestingly, patients identified as p53 abn by both methods showed the poorest prognosis, followed by those identified as p53 abn by NGS alone. Both of the two groups displayed significant differences between the p53 normal group by both methods (P < 0.001 and P < 0.001, Figure 3G). While the prognosis of samples defined as p53 abn solely by IHC was better than the p53 abn group by both methods (P = 0.001, Figure 3G), it displayed a similar prognosis compared to the p53 normal group by both methods (P = 0.771, Figure 3G). Altogether, these findings suggested that NGS-detected p53 abn might be a more accurate prognostic biomarker than IHC.

NGS molecular subtypes to guide adjuvant therapy

We further examined whether NGS-classified molecular subtypes could guide adjuvant therapy in EC. Patients without adjuvant therapy had a significantly better prognosis than those with it, with no significant differences among different adjuvant therapy regimens (P < 0.001, Figure 4A). Among patients without adjuvant therapy, those in the p53 abn subtype exhibited a worse prognosis than the others, suggesting a potential necessity of adjuvant therapy (P = 0.001, Figure 4B). For patients with adjuvant therapy, patients in the p53 abn subtype tended to have a better DFS with chemo-radiotherapy (CRT) (Figure 4C). The POLE mut subtype consistently had an excellent prognosis, implying adjuvant therapy might be unnecessary (Figure 4D). Within the NSMP subtype, patients who received radiotherapy (RT) showed better outcomes (Figure 4E), while those in the dMMR subtype benefited from CT (Figure 4F).Figure 4 Application of molecular classification in the adjuvant therapy for EC. (A) Kaplan–Meier survival curves for disease-free survival (DFS) among patients with varying adjuvant therapy regimens or without adjuvant therapy. (B) Kaplan–Meier survival curves for disease-free survival (DFS) among patients with no adjuvant therapy across four molecular subtypes (n = 188). (C-F) Kaplan–Meier survival curves for DFS among patients with varying adjuvant therapy regimens within p53 abn (C), POLE mut (D), NSMP (E), and dMMR subtype (F). (G) Modification of adjuvant therapy regimens based on molecular classification. (H-K) Modification of adjuvant therapy regimens within POLE mut subtype (H), p53 abn subtype (I), dMMR subtype (J), and NSMP subtype (K). ADT, adjuvant therapy; CRT, chemo-radiotherapy; CT, chemotherapy; DFS, disease-free survival; dMMR, mismatch repair deficiency; NSMP, non-specific molecular profile; RT, radiotherapy.

Based on the above results, we provided a revised adjuvant treatment strategy (Figure 4G-K), potentially affecting 19.6% (63/321) of patients (Figure 4G). Specifically, 27.7% (13/47) of POLE mut subtype patients may avoid adjuvant therapy (Figure 4H), while 77.8% (42/54) of p53 abn subtype patients were reclassified for CRT (Figure 4I). Changes in dMMR and NSMP subtypes were relatively fewer (Figure 4J and K). Integrating subtyping information can guide aggressive treatments for p53 abn subtype patients and avoid overtreatment for POLE mut subtype patients.

Potential therapeutic targets and assisted genetic screening in EC

We further evaluated whether the 46-gene NGS panel could provide more information to guide targeted treatment. Among 14 patients with ERBB2 (HER2) variants, five had ERBB2 (HER2) amplification, a well-established target for anti-HER2 therapies in EC. Additional cases revealed ERBB2 SNVs (R678Q, V842I, V773M, and T862A), potentially benefiting from anti-HER2 tyrosine kinase inhibitors (TKIs), antibody–drug conjugate (ADC), or monoclonal antibodies (Supplementary Table S3, available at https://doi.org/10.1016/j.esmoop.2024.103710). Variants with clinical significance in PTEN, PIK3CA, ARID1A, and KRAS were frequently observed, aligning with the Standards and Guidelines for the Interpretation and Reporting of Sequence Variants in Cancer 2017 (Supplementary Table S4, available at https://doi.org/10.1016/j.esmoop.2024.103710).

The concise NGS panel covered four genes linked to homologous recombination repair (HRR): BRCA1, BRCA2, ATM, and CHEK2, detecting pathogenic or likely pathogenic variants at rates of 3.6%, 9.7%, 12.7%, and 2.7%, respectively (Figure 5A), Individuals with these variants may benefit from poly (ADP-ribose) polymerase (PARP) inhibitors. Analysis revealed 117 pathogenic or likely pathogenic variants within HRR genes, with 12 in BRCA1 and 34 in BRCA2 (Figure 5B).Figure 5 Potential therapeutic targets and assisted genetic screening in EC. (A) Percentage of pathogenic or likely pathogenic variants within HRR genes. (B) Percentage of two hit events within HRR genes. (C) Tumor-only germline variants number. EC, endometrioid carcinoma; HRR, homologous recombination repair.

Among 331 patients, 88 showed MSI-H. Due to tumor-only testing, germline mutation status remains uncertain. However, suspicious pathogenic/likely pathogenic germline mutations in four MMR genes (MLH1, MSH2, MSH6, and PMS2) were found in nine patients, sharing the same mutation MSH6 p.F1088fs (Figure 5C). No EPCAM mutation and no suspicious pathogenic/likely pathogenic germline mutations of PTEN and DNA polymerase genes (POLE and POLD1) were detected among 331 patients.

Discussion

Molecular subtyping analysis on POLE mut, MMR/MSI, and p53 abn has been widely utilized to inform prognosis of primary EC. However, current approaches have several disadvantages. Firstly, IHC exhibits high interobserver variability. Studies have shown that consensus diagnosis among seven pathologists in p53 abn EC patients is <40%,18 and agreement among three pathologists in high-grade EC is ∼60%.9 Secondly, different cut-offs are used to define p53 abn. Thirdly, PCR, with low throughput, is unsuitable for identifying unknown genes. Fourthly, it involves multiple testing platforms, introducing a delay in obtaining results. Therefore, precisely identifying the subset of EC patients who will benefit from different adjuvant therapies through a convenient approach is an unmet need. Furthermore, Chinese EC patients exhibit unique molecular features unlike Western patients, with AKT1 alterations being more frequent and PMS2 less in the Chinese.19 This distinctiveness highlights the essentiality of developing feasible molecular subtyping classification for Chinese patients. Thus, in this study, we conducted targeted sequencing in a large cohort of consecutive Chinese EC patients using a 46-gene NGS panel. The roles of molecular classification in guiding adjuvant treatment decisions were explored.

Recently, several DNA-based targeted NGS classifiers are developed, including NGS-based ProMisE, 1021-gene NGS panel, and targeted AmoyDx EC Panel classification.20, 21, 22 In this study, targeted sequencing by a 46-gene panel was utilized, yielding 14.2%, 23.9%, 17.2%, and 44.7% of patients categorized into POLE mut, dMMR, p53 abn, and NSMP subgroups, respectively. The distribution pattern of EC identified by the panel was similar to that identified by ProMisE.13 Additionally, POLE mut had the best prognosis, while p53 abn had the worst prognosis. These findings were consistent with previous reports.11,23 Among the patients who did not receive adjuvant therapy, the p53 abn subtype displayed the most unfavorable outcome. Notably, the DFS among POLE mut, dMMR, and NSMP subtypes was comparable. These findings emphasize the necessity of adjuvant therapy for p53 abn subtype. In this study, 32 EC patients with p53 abn received adjuvant therapy. Although there was no significant difference in DFS among the different adjuvant therapy regimens received by p53 abn patients, those receiving CRT had the most favorable outcome, followed by those with CT and radiotherapy alone. A similar result has been reported in the PORTEC-3 trial.24 These observations suggest that CRT might be an effective adjuvant therapy option for p53 abn subtype. Of note, POLE mut subtype had an excellent outcome without adjuvant therapy, comparable with those treated with adjuvant therapy, implying the necessity of de-escalation of adjuvant treatment for POLE subtype. Previous research shows that dMMR EC patients with adjuvant radiotherapy had a significantly longer overall survival and DFS than those without it.25 However, in this study, dMMR patients with adjuvant radiotherapy had a shorter DFS than those with adjuvant CT or CRT. Due to the limited number of dMMR patients who received adjuvant radiotherapy (n = 2), we cannot conclude that dMMR patients cannot benefit from adjuvant radiotherapy. These findings emphasize the need for further investigation into the role of adjuvant therapy in EC patients with different molecular subtypes. dMMR has emerged as a key biomarker in predicting response to immunotherapy in solid tumors.26,27 Numerous clinical trials have established the effective antitumor activity of programmed cell death protein 1/programmed death-ligand 1 inhibitors in advanced or recurrent EC patients with dMMR/MSI-H who have failed first-line CT.28, 29, 30, 31, 32 The objective response rate ranges from 40% to 64%, highlighting the potential benefits of immunotherapy in this setting. However, whether immunotherapy is the feasible paradigm in the adjuvant therapy setting for medically operable EC patients with dMMR needs to be addressed in the further work. For NSMP patients, those with radiotherapy had the best DFS compared to those with other regimens in the adjuvant treatment setting. This suggests that radiotherapy might be an effective adjuvant therapy option for NSMP subtype. However, further confirmation is needed to establish the efficacy of adjuvant radiotherapy for NSMP subtype in a prospective, large cohort study. Collectively, our findings support the adoption of adjuvant CRT for p53 abn subtype and the potential de-escalation of adjuvant therapy for POLE mut subtype. Further research is necessary to identify the optimal adjuvant therapy regimens for dMMR and NSMP subtypes.

IHC and PCR are two common approaches to detect the MSI/MMR status. The overall consistency rate between IHC and PCR for MSI or MMR detection ranges from 88.9% to 93.4% in endometrial cancer.33, 34, 35 Discrepancies in the remaining 10% of cases are mainly due to the inherent differences in their detection mechanisms. Relatively, compared to PCR, IHC is more convenient, cost-effective, and easier to obtain results. Besides, according to previous research, this panel demonstrated a 97.9% consistency in MSI detection with PCR in colorectal cancer.16 Taken together, we focused the comparison between IHC and the NGS panel in detecting MSI/MMR status.

The feasibility of the 46-gene NGS panel in distinguishing MSI-H/dMMR from MSS/pMMR tumors has been demonstrated.14,36 Furthermore, its potential for guiding targeted and immunotherapy treatments, as well as understanding familial genetic risks, has also been explored.37 Consequently, the NGS panel is a mature product and recommended for MSI status detection and the identification of potential therapeutic targets. Consistent with previous reports,14,36 the concordance of NGS and IHC in identifying MSI-H/MMR status achieved 91.8%. Among the 24 patients with discordant NGS and IHC results, 18 (75.0%) showed concordant NGS and PCR results. These data indicate the feasibility and reliability of the 46-gene NGS panel for distinguishing MSI-H from MSS tumors in EC. Regarding p53 detection, the concordance between NGS and IHC was 65.3%, which was consistent with that reported ranging from 60% to 92.3%.21,22,38, 39, 40 The various concordance across studies might be attributed to the different definitions of p53 abn and TP53 alterations.

Distinguishing LS-related EC cases from sporadic cases is crucial because relatives of LS patients might benefit from intensive management and surveillance. The gold standard for detecting germline variants is through testing white blood cells (WBCs).36 However, in clinical practice, obtaining matched WBC or non-cancerous samples from some patients are challenging, making it an unmet need to accurately distinguish germline from somatic alterations by tumor-only NGS. A recent study has shown the feasibility of using the ColonCore panel with tumor-only NGS to predict LS-related germline variants.36 In this study, among the 79 patients identified as MSI-H, 9 patients harbored pathogenic/likely pathogenic germline MSH6 p.F1088fs. These findings highlight the need for conducting germline MMR gene variant testing for the relatives of these patients. In this cohort, the most frequently altered genes were PTEN (83.4%), PIK3CA (59.2%), ARID1A (58.6%), and TP53 (29.3%), consistent with that reported in Chinese and Asian EC patients.19,41 Fourteen patients harbored actionable ERBB2 variants, who might benefit from anti-HER2 TKIs, ADC, or monoclonal antibodies. Additionally, 12 patients harbored BRCA1 and 32 carried BRCA2 variants, indicating that these patients potentially respond to PARP inhibitors.

This study included a large sample size of consecutive Chinese EC patients without selection bias, which highlights the representativeness and reliability of the findings. Although the size of the 46-gene panel is relatively small, it can be used for molecular subtyping, identification of therapeutic targets (ERBB2, KRAS, PIK3CA), prediction of response to immunotherapy, and screening of hereditary EC (LS, Cowden syndrome, polymerase proofreading-associated polyposis). Notably, it is cost-effective and convenient, which might overcome the disadvantages of four-tier molecular classification systems. Collectively, this simplified panel could meet the current needs for the diagnosis, treatment, and molecular subtyping of EC in clinical practice.

There are some limitations in this study. Firstly, a large, multicenter, prospective cohort of EC patients is warranted to fully evaluate the efficacy of adjuvant CRT for p53 abn subtype and to determine the feasibility of the omission of adjuvant therapy for POLE mut subtype. In addition, further studies are required to determine the optimal adjuvant treatment regimen for dMMR or NSMP patients. Secondly, due to the limited panel size and gene coverage, the 46-gene NGS panel can only identify certain molecular biomarkers for targeted therapies. NTRK fusion and tumor mutational burden-high are undetectable by this panel.

In conclusion, our study emphasizes the necessity of adjuvant CRT for p53 abn subtype and the potential benefits of withholding adjuvant treatment for POLE mut subtype. Our study also suggests the 46-gene NGS panel is a feasible tool for molecular classification of medically operable EC patients. This tool aids in predicting the prognosis of patients, identifying certain actionable variants that may be targeted by precision treatment, and screening for LS-related EC patients who require clinical attention.

Supplementary data

Supplementary data

Funding

None declared.

Disclosure

ZZhe, DP, FL, HD, JW, YZ, GW, ZZho, and SC report a relationship with Burning Rock Biotech Limited that includes employment. All other authors have declared no conflicts of interest.

Data sharing

The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.
==== Refs
References

1 Crosbie E.J. Kitson S.J. McAlpine J.N. Mukhopadhyay A. Powell M.E. Singh N. Endometrial cancer Lancet 399 10333 2022 1412 1428 35397864
2 Siegel R.L. Miller K.D. Wagle N.S. Jemal A. Cancer statistics, 2023 CA Cancer J Clin 73 1 2023 17 48 36633525
3 Amant F. Moerman P. Neven P. Timmerman D. Van Limbergen E. Vergote I. Endometrial cancer Lancet 366 9484 2005 491 505 16084259
4 World Health Organization Data Visualization Tools for Exploring the Global Cancer Burden in 2022 2024 World Health Organization Geneva, Switzerland
5 Han B. Zheng R. Zeng H. Cancer incidence and mortality in China, 2022 J Natl Cancer Center 46 3 2024 221 231
6 McCluggage W.G. Singh N. Gilks C.B. Key changes to the World Health Organization (WHO) classification of female genital tumours introduced in the 5th edition (2020) Histopathology 80 5 2022 762 778 34996131
7 Urick M.E. Bell D.W. Clinical actionability of molecular targets in endometrial cancer Nat Rev Cancer 19 9 2019 510 521 31388127
8 Plotkin A. Kuzeljevic B. De Villa V. Interlaboratory concordance of ProMisE molecular classification of endometrial carcinoma based on endometrial biopsy specimens Int J Gynecol Pathol 39 6 2020 537 545 32281774
9 Gilks C.B. Oliva E. Soslow R.A. Poor interobserver reproducibility in the diagnosis of high-grade endometrial carcinoma Am J Surg Pathol 37 6 2013 874 881 23629444
10 Han G. Sidhu D. Duggan M.A. Reproducibility of histological cell type in high-grade endometrial carcinoma Mod Pathol 26 12 2013 1594 1604 23807777
11 Kandoth C. Schultz N. Cherniack A.D. Integrated genomic characterization of endometrial carcinoma Nature 497 7447 2013 67 73 23636398
12 Stelloo E. Bosse T. Nout R.A. Refining prognosis and identifying targetable pathways for high-risk endometrial cancer; a TransPORTEC initiative Mod Pathol 28 6 2015 836 844 25720322
13 Kommoss S. McConechy M.K. Kommoss F. Final validation of the ProMisE molecular classifier for endometrial carcinoma in a large population-based case series Ann Oncol 29 5 2018 1180 1188 29432521
14 NCCN NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines®) Uterine Neoplasms Version 1.2024. Plymouth Meeting: NCCN Available at https://www.nccn.org/patientres
15 Berek J.S. Matias-Guiu X. Creutzberg C. FIGO staging of endometrial cancer: 2023 J Gynecol Oncol 34 5 2023 e85 37593813
16 Zhu L. Huang Y. Fang X. A novel and reliable method to detect microsatellite instability in colorectal cancer by next-generation sequencing J Mol Diagn 20 2 2018 225 231 29277635
17 Wang M. Chen X. Dai Y. Concordance study of a 520-gene next-generation sequencing-based genomic profiling assay of tissue and plasma samples Mol Diagn Ther 26 3 2022 309 322 35305253
18 Hoang L.N. Kinloch M.A. Leo J.M. Interobserver agreement in endometrial carcinoma histotype diagnosis varies depending on the cancer genome atlas (TCGA)-based molecular subgroup Am J Surg Pathol 41 2 2017 245 252 28079598
19 Li J. Li X. Quan C. Li X. Wan C. Wu X. Genomic profile of Chinese patients with endometrial carcinoma BMC Cancer 23 1 2023 888 37730563
20 Jamieson A. McConechy M.K. Lum A. Harmonized molecular classification: assessment of a single-test ProMisE NGS tool Gynecol Oncol 175 2023 45 52 37321155
21 Li Y. Chen R. Yuan M. One-stop molecular classification of endometrial carcinoma using comprehensive next-generation sequencing Int J Cancer 151 11 2022 1969 1977 36036365
22 Rao Q. Liao J. Li Y. Application of NGS molecular classification in the diagnosis of endometrial carcinoma: a supplement to traditional pathological diagnosis Cancer Med 12 5 2023 5409 5419 36341543
23 Alexa M. Hasenburg A. Battista M.J. The TCGA molecular classification of endometrial cancer and its possible impact on adjuvant treatment decisions Cancers 13 6 2021 1478 33806979
24 León-Castillo A. de Boer S.M. Powell M.E. Molecular classification of the PORTEC-3 trial for high-risk endometrial cancer: impact on prognosis and benefit from adjuvant therapy J Clin Oncol 38 29 2020 3388 3397 32749941
25 Reijnen C. Küsters-Vandevelde H.V.N. Prinsen C.F. Mismatch repair deficiency as a predictive marker for response to adjuvant radiotherapy in endometrial cancer Gynecol Oncol 154 1 2019 124 130 31103324
26 Di Dio C. Bogani G. Di Donato V. The role of immunotherapy in advanced and recurrent MMR deficient and proficient endometrial carcinoma Gynecol Oncol 169 2023 27 33 36493574
27 Baretti M. Le D.T. DNA mismatch repair in cancer Pharmacol Therap 189 2018 45 62 29669262
28 Le D.T. Durham J.N. Smith K.N. Mismatch repair deficiency predicts response of solid tumors to PD-1 blockade Science 357 6349 2017 409 413 28596308
29 O'Malley D.M. Bariani G.M. Cassier P.A. Pembrolizumab in patients with microsatellite instability-high advanced endometrial cancer: results from the KEYNOTE-158 study J Clin Oncol 40 7 2022 752 761 34990208
30 Oaknin A. Gilbert L. Tinker A.V. Safety and antitumor activity of dostarlimab in patients with advanced or recurrent DNA mismatch repair deficient/microsatellite instability-high (dMMR/MSI-H) or proficient/stable (MMRp/MSS) endometrial cancer: interim results from GARNET—a phase I, single-arm study J Immunother Cancer 10 1 2022 e003777
31 Antill Y. Kok P.S. Robledo K. Clinical activity of durvalumab for patients with advanced mismatch repair-deficient and repair-proficient endometrial cancer. A nonrandomized phase 2 clinical trial J Immunother Cancer 9 6 2021 e002255
32 Makker V. Colombo N. Casado Herráez A. Lenvatinib plus pembrolizumab for advanced endometrial cancer N Engl J Med 386 5 2022 437 448 35045221
33 Saeed O.A.M. Mann S.A. Luchini C. Evaluating mismatch repair deficiency for solid tumor immunotherapy eligibility: immunohistochemistry versus microsatellite molecular testing Hum Pathol 115 2021 10 18 34052294
34 McConechy M.K. Talhouk A. Li-Chang H.H. Detection of DNA mismatch repair (MMR) deficiencies by immunohistochemistry can effectively diagnose the microsatellite instability (MSI) phenotype in endometrial carcinomas Gynecol Oncol 137 2 2015 306 310 25636458
35 Streel S. Salmon A. Dheur A. Diagnostic performance of immunohistochemistry compared to molecular techniques for microsatellite instability and p53 mutation detection in endometrial cancer Int J Mol Sci 24 5 2023 4866 36902292
36 Li C. Song W. Xu Y. A one-stop approach to diagnosing hereditary colorectal cancer in the Chinese population J Gastroenterol Hepatol 38 11 2023 1980 1987 37749864
37 Yang M. Li D. Jiang W. Development and external validation of a novel nomogram for screening Chinese Lynch syndrome: based on a multicenter, population study Ther Adv Med Oncol 13 2021 17588359211023290
38 Thiel K.W. Devor E.J. Filiaci V.L. TP53 sequencing and p53 immunohistochemistry predict outcomes when bevacizumab is added to frontline chemotherapy in endometrial cancer: an NRG oncology/gynecologic oncology group study J Clin Oncol 40 28 2022 3289 3300 35658479
39 Vermij L. Léon-Castillo A. Singh N. p53 immunohistochemistry in endometrial cancer: clinical and molecular correlates in the PORTEC-3 trial Mod Pathol 35 10 2022 1475 1483 35752743
40 Singh N. Piskorz A.M. Bosse T. p53 immunohistochemistry is an accurate surrogate for TP53 mutational analysis in endometrial carcinoma biopsies J Pathol 250 3 2020 336 345 31829441
41 Asami Y. Kobayashi Kato M. Hiranuma K. Utility of molecular subtypes and genetic alterations for evaluating clinical outcomes in 1029 patients with endometrial cancer Br J Cancer 128 8 2023 1582 1591 36797358
