Article Text
Abstract
Objectives To investigate the pattern of first recurrence of disease in patients with endometrial cancer according to molecular classification, and to assess the independent role of molecular profiling in each type of failure.
Methods Retrospective single-center study including patients diagnosed with endometrial cancer stage I–IVB (International Federation of Gynecology and Obstetrics 2009) between December 1994 and May 2022, who underwent primary surgical treatment and had a complete molecular profile. First recurrence was classified as isolated or multiple, and as vaginal, pelvic, peritoneal, nodal, and distant according to its location. The log-rank test and univariate and multivariate adjusted Cox regression models were used for comparison between groups.
Results A total of 658 patients were included. Recurrence was observed in 122 patients (18.5%) with a recurrence rate of 12.4% among mismatch-repair deficient tumors, 14.5% among non-specific molecular profile, 2.1% among POLE-mutated, and 53.7% among p53-abnormal tumors. Recurrences were found to be isolated in 80 (65.6%) and multiple in 42 (34.4%) patients, with no differences in molecular subtype (p=0.92). Patients with p53-abnormal tumors had a recurrence mainly as distant (28.4%) and peritoneal (21.1%) disease, while patients with non-specific molecular profile tumors presented predominantly with distant failures (10.3%), and mismatch-repair deficient tumors with locoregional recurrences (9.4%).
On multivariate analysis, p53-abnormal molecular profile was the only independent risk factor for peritoneal failure (OR=8.54, 95% CI 2.0 to 36.3). Vaginal recurrence was independently associated with p53-abnormal molecular profile (OR=6.51, 95% CI 1.1 to 37.4) and lymphovascular space invasion. p53-abnormal and non-specific molecular profiles were independent predictors for distant recurrence (OR=3.13, 95% CI 1.1 to 8.7 and OR=2.35, 95% CI 1.1 to 5.0, respectively), along with lymphovascular space invasion and high-grade tumors. Molecular profile was not independently associated with pelvic and nodal recurrences.
Conclusions Endometrial cancer featured different patterns of recurrence depending on the molecular profile. p53-abnormal molecular profiling was the only independent risk factor for peritoneal relapse, while non-specific molecular profile showed a strong association with distant failures.
- Endometrial Neoplasms
- Pathology
- Surgical Oncology
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Molecular profile of endometrial cancer is a strong independent prognostic factor and predicts the response to adjuvant treatment. However, its influence on the pattern of recurrence is unknown.
WHAT THIS STUDY ADDS
Molecular subgroups of endometrial cancer present distinctive recurrence patterns. Molecular profile is a stronger independent predictor for vaginal, peritoneal, and distant recurrence than classic histologic factors. p53-abnormal is the sole independent risk factor for peritoneal relapse, and non-specific molecular profile independently predicts distant recurrence.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
In this study, molecular profiling emerges as a more robust predictor of recurrence patterns compared with traditional histologic factors in certain locations. This suggests the potential of molecular classification in personalizing follow-up protocols for patients with endometrial cancer, potentially enhancing detection of early recurrence.
INTRODUCTION
Endometrial cancer is the most common gynecological cancer in developed countries1 and has an overall good prognosis, with an 80% 5-year survival rate.2 However, approximately 15–20% of patients will experience a recurrence,3 4 resulting in poor oncological outcomes, with a median survival that ranges from 10 to 21 months.4 5 The prognosis of recurrent endometrial cancer is associated with the pattern of recurrence. The survival is generally worse in cases of distant, peritoneal, or multiple recurrences, compared with locoregional or oligometastatic relapse, since those may be suitable for local treatments.3 5–7 Understanding the factors that influence the pattern of recurrence is essential to optimize patient management and follow-up.
Traditional histopathological features, such as histological type and grade, International Federation of Gynecology and Obstetrics (FIGO) stage, and lymphovascular space invasion, have been associated with different failure patterns in previous studies. Specifically, distant recurrence is associated with high-grade and non-endometrioid tumors,6 8 9 advanced FIGO stages,9 and lymphovascular space invasion.9–11 Nodal recurrence is related to lymphovascular space invasion and nodal disease at diagnosis,11–13 and peritoneal recurrence to non-endometrioid histology.12
In 2013, The Cancer Genome Atlas Research Network14 identified four subgroups of endometrial cancer with distinct genomic expression and clinical outcomes. In the last years, the use of surrogate markers has facilitated its application in clinical practice by combining POLE mutational analysis with immunohistochemistry analysis of p53 and mismatch repair proteins.15–17 Therefore, patients with endometrial cancer can be classified into four molecular subgroups: POLE-mutated (POLE-mut), mismatch repair deficient (MMR-d), non-specific molecular profile (NSMP), and p53-abnormal (p53-abn). The independent prognostic role of the molecular profile has been consistently demonstrated,18 leading to the integration of molecular subgroups in the current classification system19 and treatment guidelines,20 with a shift towards a model based on molecular and histologic features. However, there is a lack of studies assessing the association between molecular profile and the pattern of recurrence in endometrial cancer.
In this study, we aimed to analyze the pattern of first recurrence of patients with endometrial cancer according to their molecular classification, and to assess if molecular profile plays an independent role on each pattern of first failure, compared with pathology, surgical and adjuvant treatment.
METHODS
Patient Characteristics
This is a retrospective single-center observational study including patients diagnosed with endometrial cancer stage I–IVB (FIGO 2009)21 and who underwent surgical treatment at Hospital Vall d’Hebron in Barcelona between December 1994 and May 2022. Patients diagnosed with histologies other than adenocarcinoma, who did not undergo surgical primary treatment, without tumor tissue sample available to perform molecular profile, or with incomplete medical records were excluded from the analysis. Clinical information was collected for each patient, including diagnostic information, surgical and adjuvant treatment, clinical follow-up, and in the case of recurrence, location and date of recurrence.
Only first recurrences were considered for statistical analysis. Recurrences were classified according to their location as vaginal, pelvic (pelvic tissues other than vagina), peritoneal, nodal (only if pelvic and/or aortic nodal relapse), and distant (recurrence in extra-abdominal areas such as lungs, brain, liver, bones, as well as lymph nodes outside the pelvic or para-aortic regions). We recorded whether the first recurrence was isolated or multiple, and patients who presented with simultaneous recurrences in different locations were accounted for in each of the corresponding recurrence groups.
Surgical and adjuvant treatment was based on current international recommendations at the time of diagnosis, and molecular profile did not influence the therapeutic management of the patients included in the study. Tumor staging was determined based on final pathology reports and reclassified according to FIGO 2009 guidelines.21 Clinicopathological data were collected from institutional records, and codified information was registered into a RedCap database. This study was approved by the hospital’s institutional review board (Vall d’Hebron University Hospital Research Ethics Committee) and codified as PRAMI276-2018. Authorization for research was granted for the patients included in the present study.
Molecular Classification
The molecular profiling of all patients was performed in the pre-operative endometrial biopsy or hysterectomy specimen using immunohistochemical analysis of mismatch repair (PMS2, MLH1, MSH2, MSH6) and p53 proteins, as well as the evaluation of somatic POLE mutation by genotyping using KASP technology to identify five of its most common hotspot mutations sites (P286R, S297F, V411L, A456P, S459F).22 Patients were classified as POLE-mut, MMR-d, NSMP or p53-abn, as per the European Society of Gynaecological Oncology/European Society for Radiotherapy and Oncology/European Society of Pathology (ESGO/ESTRO/ESP) guidelines.20 Patients harboring more than one molecular feature were classified according to the guideline’s recommendations.
Statistical Analysis
Continuous variables were expressed as median values with IQR and were compared using Student’s t-test, Welch’s t test, or the Mann-Whitney test, as appropriate. Categorical variables were expressed as frequencies with percentages and compared using the χ2 test or the Fisher exact test.
Oncologic outcomes were analyzed with the Kaplan-Meier method, whereas the log-rank test and univariate and multivariate adjusted Cox regression models were used for the comparison between groups. For the construction of the multivariate model, a selection method based on maximum likelihood estimation and Akaike information criterion was used, considering all relevant variables related to the primary end point. Disease-free survival was defined as the time from the date of surgery to diagnosis of the recurrence. Overall survival was defined as the time from the date of surgery to death or last follow-up for those patients who were not deceased. Patients were censored at the time of the last follow-up if they did not experience an event. Disease-free survival and overall survival were calculated for the whole cohort of patients.
The STATA statistical program (version 14.2) was used for data analysis.
RESULTS
Clinicopathological and Survival Characteristics
We included 658 patients diagnosed with endometrial adenocarcinoma who underwent surgery and had a complete molecular profile. Median age at diagnosis was 64 years (IQR 56–73). Most patients were diagnosed at early stages (86.2%, n=567), with endometrioid carcinoma (82.1%, n=540) and low-grade tumors (61.7%, n=406). Lymphovascular space invasion was observed in 20.2% (n=133) of cases. Overall, 42.8% (n=282) of tumors were NSMP, 35.5% (n=234) MMR-d, 14.4% (n=95) p53-abn, and 7.1% (n=47) POLE-mut.
In comparison with the other groups, patients with p53-abn tumors exhibited a significantly higher incidence of non-endometrioid (83.2%, p<0.001) and high-grade tumors (92.6%, p<0.001) and presented the highest rate of lymphovascular space invasion (37.9%, p<0.001). Up to 31.5% of these patients were diagnosed at advanced stages, and the vast majority (77.9%, p=0.01) received adjuvant treatment, with almost half of patients being treated with adjuvant chemotherapy.
Patients with POLE-mut tumors were the youngest at diagnosis among the entire cohort (p=0.043). They presented more non-endometrioid (10.6%) and high-grade tumors (38.3%) than patients with MMR-d and NSMP tumors (p<0.001). All POLE-mut patients but one were diagnosed at initial FIGO stages. These patients received significantly less adjuvant treatment than the other groups, particularly with chemotherapy (4.3%, p<0.01)
Patients with NSMP tumors presented the highest incidence of endometrioid (94.7%) and low-grade tumors (78.3%, p<0.001) among the entire cohort. NSMP and MMR-d tumors were diagnosed in 89% and 90.6% of cases in early stages and exhibited similar rates of adjuvant treatment with radiotherapy and chemotherapy (Table 1)
After a median follow-up of 5.33 years (IQR 2–9.4), the average 5-year overall survival and disease-free survival for the entire cohort was 85.5% and 74.6%, respectively. Among all patients, those with p53-abn tumors presented the worst 5-year overall survival and disease-free survival (54.1% and 33.1%, respectively), while patients with POLE-mut tumors exhibited excellent prognostic outcomes, with an overall survival and disease-free survival of 100% and 97.8%, respectively (p<0.001). No death related to endometrial cancer was recorded among patients harboring POLE-mut tumors. Patients with MMR-d and NSMP tumors had a similar intermediate prognosis (Table 1, Figure 1).
Recurrence Pattern According to Molecular Profile
Of the 658 patients, 122 (18.5%) experienced a recurrence. The most frequent location of recurrence was distant (10.8%) followed by peritoneal and nodal (5.3% and 5.2%, respectively) and lastly by vaginal and pelvic recurrences (3.2% and 2.3%, respectively). Recurrences were isolated in 65.6% of cases and multiple in 34.4% of cases, with no differences in this distribution regarding molecular subtype (p=0.92).
The location of the recurrence was significantly influenced by the molecular profile. The p53-abn group presented the highest rate of relapse (53.7%, p<0.001), and disease in these patients recurred more frequently in the form of distant and peritoneal recurrences (28.4% and 21.1%, respectively, p<0.001). NSMP and MMR-d groups showed similar rates of total recurrences (14.5% and 12.4%), although NSMP presented with distant failures in a higher proportion (10.3% vs 6.4%, p<0.001) and MMR-d with locoregional (vaginal, pelvic, and nodal) relapses (9.4%). The POLE-mut group experienced the lowest rate of relapse (2.1%), corresponding to a unique patient with aortic nodal recurrence treated locally, who was free of disease at the end of follow-up (Table 2).
Recurrence According to Clinicopathologic and Molecular Features
In multivariate analysis, p53-abn molecular profile showed the strongest association with recurrence (OR=20.06, 95% CI 2.4 to 166.5). FIGO stage, lymphovascular space invasion, and high-grade tumors were also independent risk factors for recurrence. When assessing risk factors according to the location of relapse, we observed that p53-abn was the strongest predictor for vaginal recurrence (OR=6.51, 95% CI 1.1 to 37.4). Lymphovascular space invasion also showed an independent association (OR=4.30, 95% CI 1.4 to 12.7), while adjuvant radiotherapy remained a protective factor (OR=0.13, 95% CI 0.04 to 0.39).
Regarding pelvic and nodal recurrences, the molecular profile did not show an independent association in the multivariate analysis. Only the administration of radiotherapy was associated with a lower risk of pelvic recurrences (OR=0.19, 95% CI 0.05 to 0.80), whereas high-grade tumors showed a strong association with nodal recurrence (OR=10.91, 95% CI 3.2 to 37.5). Multivariate analysis confirmed that p53-abn remained the only independent risk factor for peritoneal recurrence (OR=8.54, 95% CI 2.0 to 36.3), and no independent association with other histopathologic features, such as non-endometrioid histology or grade, was observed. Both NSMP and p53-abn groups were independent predictors for distant recurrence, along with lymphovascular space invasion and high-grade tumors, whereas adjuvant chemotherapy or non-endometrioid histology were not found to be independently associated with distant failure (Table 3).
DISCUSSION
Summary of Main Results
Our findings revealed distinct recurrence patterns among molecular subgroups, with molecular profile emerging as an independent risk factor for vaginal, distant, and peritoneal recurrences, and showing no impact on pelvic and nodal failures. Notably, the p53-abn group was the sole independent predictor for peritoneal recurrence, and, along with NSMP, independently predicted distant failures.
Results in the Context of Published Literature
To date, only one study has analyzed recurrence patterns based on molecular profile in endometrial cancer,23 including only first recurrences classified as locoregional, peritoneal, and distant recurrences. Patients with multiple recurrence sites were allocated into the most aggressive recurrence location group. In this study, patients with p53-abn tumors presented predominantly with peritoneal and distant recurrence, while NSMP recurred mainly distantly and MMR-d locoregionally. These findings are consistent with our results, which may suggest a reproducibility in the recurrence pattern for each molecular profile.
The poor prognosis of the p53-abn group observed in our cohort, and consistent with previous reports,14–18 could be attributed to its high rate of peritoneal and distant recurrences, which are associated with the worst survival outcomes.3 4 11 However, Siegenthaler et al described that p53-abn tumors harbored the poorest prognosis regardless of the recurrence location, highlighting the independent prognostic signature of p53-abn molecular profile.23 We also observed that p53-abn was the only independent risk factor for peritoneal recurrence, while non-endometrioid histology and other histopathologic factors classically associated with peritoneal disease10 24 did not exhibit an independent association. These findings suggest that p53-abn molecular profile is the sole determinant factor for peritoneal recurrence, exhibiting a biological behavior like that of high-grade serous ovarian cancer, which frequently features p53 mutations.25
Despite its excellent histopathologic prognostic features at diagnosis, NSMP group recurred mostly with distant disease. Additionally, NSMP group was an independent predictor for distant recurrence, along with p53-abn group. These findings are consistent with those published by Stelloo et al,26 reporting 39% of distant failures for NSMP tumors, and with those of Siegenthaler et al and Luzarraga et al.23 27 However, the NSMP group exhibits a variable prognosis, and new molecular factors, such as the presence of L1CAM or the absence of estrogen receptors, have been recently described as predictors of worst prognosis.28–31 A recent study29 identified the presence of estrogen receptors in NSMP as an independent protective factor against recurrence, highlighting the potential utility of this biomarker for better understanding the behavior of this molecular subgroup.
Our findings also revealed that patients with MMR-d tumors experienced more locoregional recurrences, consistent with the report of Kim et al,32 which found that MMR-d tumors were more likely to recur in retroperitoneal lymph nodes. Moreover, in our cohort MMR-d tumors showed a non-significant trend towards vaginal relapse, which may suggest a predisposition for local failures in these tumors. Conversely, radiotherapy demonstrated a protective effect for both vaginal and pelvic recurrences, in line with previous clinical trials highlighting its role in reducing locoregional recurrences.33–35
Within the POLE-mut group, only one patient exhibited a recurrence (an aortic nodal relapse treated locally) and remained disease-free at the end of follow-up. This low recurrence rate, along with the excellent post-relapse prognosis is in accordance with the results of the meta-analysis published by McAlpine et al36 evaluating 294 patients with POLE pathogenic mutations, describing only 11 (3.7%) patients with recurrence, of whom 8 (72.7%) are alive and without disease after treatment.
In our cohort, almost half of the patients of the POLE-mut group received adjuvant treatment, mostly with radiotherapy. Looking at the good prognosis of this group, with only one recurrence, a de-escalation of the adjuvant treatment seems reasonable to avoid toxicities without compromising survival, in accordance with current recommendations.18 20 Conversely, p53-abn group received 81% of adjuvant treatment, of which 45% was with adjuvant chemotherapy. According to previous evidence, this group would benefit most from the addition of chemotherapy to radiotherapy, and these rates of adjuvant chemotherapy would be higher according to the current recommendations. Current clinical trials37 will help to provide the evidence to change our indications of adjuvant treatment.
Finally, we found no differences in the number of recurrence sites among the four molecular subgroups, with single-site recurrence being the most frequent pattern, comparable to that published by Bricou et al.4 This novel discovery suggests that molecular profile does not correlate with the number of recurrence sites at first relapse.
Strengths and Weaknesses
To our knowledge, this is the first study evaluating the independent role of molecular profile in predicting different locations of relapse. The major strength of the present study is the long follow-up, which gave us the opportunity to evaluate late recurrences, as well as the large sample size. Nevertheless, the retrospective design of the study introduces limitations and potential bias. Furthermore, molecular profiling was performed retrospectively either on the pre-operative biopsy or hysterectomy specimen, depending on tissue availability at the time of molecular analysis. However, this limitation is mitigated by the well-established high reproducibility of molecular profile between pre-operative and post-operative samples.38 39
Implications for Practice and Future Research
In this study, molecular profiling emerges as a more robust predictor of recurrence patterns compared with traditional histologic factors in certain locations. This suggests the potential of molecular classification for personalizing follow-up protocols for patients with endometrial cancer, potentially enhancing early recurrence detection. However, further research is needed to translate these findings into clinical practice and provide additional insights for a more precise clinical decision-making.
CONCLUSIONS
Molecular subgroups of endometrial cancer exhibit distinctive recurrence patterns, with molecular profile emerging as a stronger independent predictor for vaginal, peritoneal, and distant relapse than classic histologic factors. The p53-abn group acts as the sole independent predictor for peritoneal recurrence in patients with endometrial cancer .
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by Vall D’Hebron University Hospital research ethics committee with medicines and research project commission: PRAMI276-2018. The need for informed consent was waived by the ethics committee.
References
Footnotes
EC, AG-M and SC contributed equally.
Contributors ALA, VB, and SC conceived and planned the study. VB performed the statistical analysis. ALA carried out the manuscript preparation. SC, AG-M and EC are the senior members of the project and supervised the whole study. All authors (ALA, VB, CL-G, LS, JC, BV-M, EC, AG-M, SC) gave final approval of the version to be published. ALA is responsible for the overall content of the study as the guarantor.
Funding ISCIII grants (PI20/00664, PI20/01566), Miguel Servet grant (CP22_00147), RETOS Colaboración (CPP2021-008440), Fundación Científica AECC (GCTRA1804MATI), Biomedical Research Center Network CIBERONC (CB16/12/00328) and Generalitat de Catalunya (2021SGR11757). As well, this project was supported by AECC and ISCIII (PERME212443COLA and AC21_2/00030), under the frame of ERA PerMed.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Linked Articles
- Editorial
- Letter
- Letter