Survival benefit of surgical treatment for patients with stage IVB endometrial cancer: a propensity score-matched SEER database analysis

Abstract The study aimed to evaluate the value of cancer-directed surgery (CDS) in improving the prognosis of patients with stage IVB endometrial cancer (EC) and under which kind of conditions could maximise its value. The Surveillance, Epidemiology, and End Results database was used to analyse patients diagnosed with stage IVB EC who received chemotherapy between 2004 and 2016. Among 1978 patients were enrolled following propensity score matching (PSM). We found that CDS was closely associated with prolonged overall survival. Moreover, CDS can effectively improve the survival rate of patients with protective or unfavourable factors and should be considered in a range of circumstances. Almost all patients (96.15%) who received surgery were operated on primary tumours of the reproductive organs and obtained favourable surgical outcomes. In conclusion, surgery can improve the survival of patients with stage IVB EC, palliative hysterectomy is worth considering in such patients. IMPACT STATEMENT What is already known on this subject? Patients with stage IVB EC account for a small proportion, so previous researches usually had an insufficient sample size. And it is still controversial whether to perform surgery on patients with stage IVB EC. What do the results of this study add? We verified the value of CDS in improving prognosis of patients with stage IVB EC. We also found that surgery outcomes were better in patients aged ˂ 60 years, and with T1 and T2 invasion. Moreover, resection of the primary site played an important role in prolonging survival time. What are the implications of these findings for clinical practice and/or further research? Surgical treatment can prolong the overall survival of patients with stage IVB EC, even if only primary site resection is performed. Surgery should be more aggressive in patients aged ˂ 60 years, and with lesions confined in the pelvis (with T1 and T2 invasion). The survival rate of patients with brain metastasis may also be improved by surgery. However, because of the small sample size, the surgical benefit needs further confirmation.


Introduction
Adenocarcinoma of the endometrium (known as endometrial cancer, EC) is a common malignancy among gynecological tumours, which is the most common genital tumour in women in the United States, according to the National Comprehensive Cancer Network (NCCN) Guidelines (NCCN 2022).Most cases are accompanied by vaginal bleeding, which can help in early detection leading to favourable prognosis.However, the mortality rates of EC have increased more than the incidence rates in recent years, possibly related to an increased proportion of patients with an advanced stage, a high-risk pathological type, and an older age of onset (NCCN 2022).Patients with stage IVB disease account for a relatively small proportion, but they contribute to the majority of mortality.Stage IV disease constitutes 3-13% of all cases, with a 5-year survival rate of 10-20% (Ayhan et al. 2002).The liver and lungs are the primary sites of distant metastasis, with the brain and bones being the less common sites (Fujimoto et al. 2009, Wang et al. 2022).
Due to the limited number of cases, there is no reliable clinical evidence for the treatment of advanced EC (Eto et al. 2013, Ribeiro et al. 2020).Stage IVB EC usually has a higher proportion of high-risk pathological types (serous carcinoma, clear cell carcinoma, and others), and cytoreductive surgery is the conventional treatment, as for ovarian cancer (Albright et al. 2021).Retrospective studies have investigated the benefits of surgery in patients with stage IVB EC (Ueda et al. 2010, Shih et al. 2011, Eto et al. 2013, Rajkumar et al. 2019).Given these advances, cancer-directed surgery (CDS) for patients with stage IVB EC has been applied in clinical practice and performed on selected patients.Approximately 66% of stage IVB EC undergo surgery (Eto et al. 2013).There was a nonnegligible difference in the clinicopathological characteristics or physician bias for surgical options in previous studies (surgical tolerance, patient preference, number of distant metastases, and possibility of achieving satisfactory R0 resection; Eggemann et al. 2017, Straubhar et al. 2021, Conway et al. 2019).Simultaneously, there have been no previous studies with sufficient sample sizes.
Herein, we used the Surveillance, Epidemiology, and End Results (SEER) database to increase the sample size and verify whether surgical treatment improves the overall survival (OS) of patients with stage IVB EC.We also compared the subgroups that could benefit from surgery.Propensity score matching (PSM) was used to control for potential confounders between the CDS and non-CDS groups.

Materials and methods
The principles and methodological framework of this study followed the Enhancing the Quality and Transparency of Health Research (EQUATOR) Network guidelines.

Data source
The SEER database is an official source of cancer information and survival outcomes in the United States (http://seer.cancer.gov).Patient demographics, clinical characteristics, and survival status were collected from the SEER database using SEER � Stat software (version 8.4.0.1).

Study population
Our study was limited to a sample that included patients who were diagnosed with EC with distant metastasis (stage IVB) and those who underwent chemotherapy between 1 January 2004 and 31 December 2016 (Figure 1).The study excluded patients (1) who did not undergo surgery for other reasons or with unknown surgical history (2) whose diagnosis was not confirmed microscopically, and (3) whose metastasis information was unknown.Based on the surgical history, patients were divided into two groups: the CDS group and the non-CDS group.

Statistical analysis
Statistical analyses were performed using R software (version 4.2.1, R Core Team, Vienna, Austria).Categorical variables are presented as frequencies.Demographic and clinical characteristics were compared both pre-and post-matching using Pearson chi-square test and Fisher's exact test.Stratified on the basis of CDS, patients were matched through nearest neighbour matching by propensity score using a calliper width of 0.3 in a 1:1 ratio to mitigate selection bias between the CDS and non-CDS groups.
Univariate and multivariate Cox regression analyses were performed to determine independent risk factors for OS in patients with stage IVB EC.OS rates were calculated using Kaplan-Meier curves, and the log-rank test was used to compare differences between the groups.We conducted subgroup analysis to explore the value of surgery in each subgroup.Forest plots were constructed using GraphPad Prism 9. Statistical significance was set at p < 0.05.

Results
Of 5403 patients with microscopically confirmed EC and distant metastases who received chemotherapy were extracted from the database, 4072 (75.4%) underwent CDS of the primary tumour.The demographic and clinical characteristics of the patients with stage IVB EC are summarised in Table 1.Before PSM, there were significant distribution differences between the CDS and non-CDS groups in some subgroups.After 1:1 matching, the differences were balanced, and 1978 patients met the matching requirements (989 patients in both the CDS and non-CDS groups; Table 1).
The OS rate of the CDS group was higher than that of the non-CDS group, both before and after PSM (Supplementary figure 1a and 1b).After PSM, the median overall survival times were 21 and 8 months in the CDS and non-CDS group, respectively.Moreover, the estimated 1-year OS rate was greater for CDS patients (CDS [95% confidence interval (CI)]: 0.739 [0.725-0.752];non-CDS [95% CI]: 0.354 [0.329-0.380]),while the 3-year and 5-year survival rates were an order of higher magnitude ( In the univariate regression model; race; histology; grade; T stage; N stage; radiotherapy; CDS; and metastasis to the brain, lungs, and liver were independent prognostic factors for OS.Significant variables from the univariate Cox regression analysis were applied to the multivariate Cox regression model, and a forest plot was constructed for further analysis.These obtained estimates implied that patients who underwent CDS had significantly higher OS than those who did not.Radiotherapy was also a protective factor; lymph node metastasis, liver metastasis, non-endometrioid cancer, poor differentiation of tumour (grade III/IV), and tumour invasion were independent risk factors that negatively impacted OS (Table 2).Based on the results, CDS is considered an important predictor of OS (as seen from the multifactor forest plot in Figure 2).Therefore, we used subgroups to analyse the CDS values and calculated the HR and median survival time in each group.Because few patients with brain metastases underwent CDS, they were not included in the subgroup analysis.CDS improved survival outcomes in each subgroup, and the differences were statistically significant (p < 0.05).We selected variables that deviated from the 95% CI for the entire sample (Figure 3).These deviations indicate that the patients who underwent CDS had better OS rates than that of those in the total sample who were below 60 years of age, with T1 and T2 invasion.However, OS was worse in the subgroups that included people of colour; patients who were divorced; patients with disease staging of T3, N2, Nx/Blank(s); and patients with distant metastasis to liver, lungs, and bones.A forest plot (Supplementary figure 2) was used to show the median survival time of each subgroup.Among the population undergoing surgery, the median survival time for patients below 60 years of age (median survival time ¼ 21months, 95%CI:17-25 months) and with T1 (median survival time ¼ 26 months, 95%CI:21-38 months) and T2 (median survival time ¼ 27months, 95% CI:17-29 months) was higher than that of the total sample, this was shorter in  1e).Finally, we collected the specific surgical modalities of the primary site:804 of the 989 patients who underwent surgical treatment were performed hysterectomy with or without bilateral salpingo-oophorectomy.A total of 127 patients underwent radical hysterectomy; 20 pelvic exenteration, 20 local tumour resection, and 19 unknown surgical methods.Surgery was not performed in the distant metastatic region in our cohort, and almost all (96.15%) patients with distant metastasis underwent hysterectomy for the primary lesion.

Discussion
A comparison of matched cohorts from the SEER database indicated that patients with stage IVB EC who received CDS and chemotherapy had higher OS rates than that of those who received chemotherapy alone.In the liver, lung, and bone metastasis subgroups, CDS substantially improved the prognosis.Consistent with the findings of previous studies, we found that tumour invasion, poor differentiation, nonendometrioid cancer, and liver and lymph node metastasis were independent predictors of poor outcomes (Tanioka et al. 2010, Guo et al. 2020, NCCN 2022).
Radiotherapy is frequently used as an adjuvant treatment to prevent local recurrence in patients with intermediate or high-risk factors for EC (Klopp et al. 2014, Brooks et al. 2019, Sorolla et al. 2020, Crosbie et al. 2022).Nevertheless, vaginal brachytherapy can achieve vaginal control similar to pelvic external beam radiotherapy with fewer gastrointestinal side effects (Nout et al. 2010).Radiotherapy in combination with chemotherapy has been shown to control local and remote lesions in patients with advanced EC (NCCN 2022, Tung et al. 2022, Winarto et al. 2022).This study included all patients who received chemotherapy, possibly explaining why radiotherapy was identified as a protective factor, even in the presence of distant metastases.Although radiotherapy was associated with improved survival, CDS as a primary therapy is a more important determinant of good outcomes than any other type of treatment, raising the topic of which subgroups might obtain the largest survival benefit after surgical treatment.
The above subgroup analysis results showed that it is better to perform CDS when patients are below 60 years of age, and have T1 and T2 invasion.In addition, CDS is less effective in patients of colour, those who were divorced; those with a disease staging of T3, N2, Nx/Blank(s); and those with metastases to the liver, lungs, and bones.
Although elderly patients with EC may not be at an increased risk of postoperative complications, they are more likely to experience recurrence, have more comorbidities, and have a greater modified fragility index (mFI) (Giannini et al. 2020).Moreover, elderly patients with EC exhibit a more aggressive tumour phenotype and a higher proportion of mutant p53 expression (Alektiar et al. 2003, Gonzalez-Rodilla et al. 2012), which can affect surgical effectiveness and postoperative survival time.
Consistent with the findings of previous studies, lesions with T1 and T2 invasion were still confined within the pelvis, and surgical treatment had a greater probability of achieving satisfactory R0 resection for better surgical outcomes (Eggemann et al. 2017, Conway et al. 2019, Straubhar et al. 2021).The poor surgical outcomes of T3 and N2 disease and metastases to the liver, lungs, and bones are well explained because these variables are universally regarded as features of high-risk endometrial carcinoma.
The conclusion of the Nx/Blank (s) subgroup is specific, with a pronounced deviation of the HR from the overall cohort, but the survival time did not deviate significantly.This is most likely due to the heterogeneity of survival time; patients with lymph node metastasis survived for a shorter period than did those without.Recent studies have stated controversial findings as regards systematic lymph node dissection to improve survival (Havrilesky et al. 2005, Todo et al. 2010, Kitchener et al. 2009, Multinu et al. 2019, Casarin et al. 2020).Moreover, extensive lymph node dissection may lead to surgical complications such as haemorrhage and lymphedema.Sentinel lymph node identification by preoperative mapping may be considered for appropriate patients to assess lymph node status (Rossi et al. 2017, Sozzi et al. 2020, Restaino et al. 2022, Capozzi et al. 2022).Thus, lymph node Figure 3. Hazard ratios ± 95% confidence intervals for patients in the CDS and non-CDS groups.The sample size, hazard ratios and 95% confidence intervals for each subgroup are provided in the table.Dashed lines and shaded areas cover the hazard ratios and confidence intervals for the total sample.Statistical methods used for univariate cox regression, and p-value for all subgroups were less than 0.05.Because few patients with brain metastasis were treated with CDS and possible existence of bias, they were removed from the subgroup analysis.Factors are highlighted if the confidence intervals for their hazard ratios are distinct from the total.Blue is used on the left (indicating better surgical outcomes in this group than the total) and red is used on the right (indicating worse surgical outcomes in this group than the total).evaluation is still recommended, and sentinel lymph node dissection can be chosen to minimise the complications of systematic lymph node dissection while also effectively assessing lymph node status and predicting patient prognosis.In summary, CDS can effectively improve the survival rate of patients with protective or unfavourable factors and should be considered in a range of circumstances.
Based on the results of previous studies, the question of whether surgery should be performed in patients with EC and distant metastases remains controversial.According to the 2022 NCCN guidelines, surgical treatment and specific operative procedures should be performed under different circumstances.Preoperative adjuvant chemotherapy should be considered, and no measurable residual lesions should be pursued.For unresectable diseases, comprehensive treatments including chemotherapy, radiotherapy, hormone therapy, and palliative hysterectomy should be considered (NCCN 2022).
Surgery has been gradually accepted and used by clinicians as an essential treatment for advanced EC.Several retrospective studies have indicated that optimal cytoreduction is vital for patients with advanced EC to achieve a higher survival rate (Ayhan et al. 2002, Vandenput et al. 2009, Ueda et al. 2010).Ueda et al. (2010) studied 18 patients with stage IVB EC and suggested, for the first time, that aggressive cytoreductive surgery improved survival.Shih et al. (2011) studied 58 patients with stage IV endometrioid EC and concluded that surgical cytoreduction with no gross residual disease in selected patients was associated with improved survival.In their study, the median overall survival was 42.2 months for patients without gross residual disease, 19 months for patients with residual disease, and 2.2 months for patients who did not undergo cytoreduction (p < 0.001).
Similar conclusions were drawn in this study.Surgical treatment was the most important independent predictor of the OS.We speculate that even with unresectable extraperitoneal metastasis, surgical excision of the primary tumour would prolong the survival of patients with stage IVB EC.
Surgical treatment in each subgroup can improved survival time.In the liver, lung, and brain metastasis subgroups, the hazard ratio and median survival time indicated that CDS improved prognosis.There were 40 patients with brain metastasis and subgroup analysis results suggested that surgical treatment might also improve the survival of these patients (p < 0.05).Simultaneously, the median survival time of this subgroup was shorter than that of other subgroups with distant metastasis.However, because of the small sample size, the calculation of the HR and median survival time may have been biased; therefore, they were not included in the subgroup comparison.Brain metastases rarely occur in endometrial cancer, and the survival outcome is usually poor.Treatments include surgical resection, radiotherapy, chemotherapy, and other comprehensive treatments (Monaco et al. 2008, Uccella et al. 2016, Bhambhvani et al. 2021).According to a previous study (Guo et al. 2020), surgical treatment did not improve the survival of patients with EC brain metastasis.However, PSM was not performed in that study, and this deviation may affect the results, necessitating further research.
This study had some limitations.First, in recent years, for patients with surgically unresectable endometrial carcinoma, interval surgery following neoadjuvant chemotherapy can achieve the same effect as primary cytoreductive surgery, and there is less possibility of residual lesions after surgery (Vandenput et al. 2009, Ueda et al. 2010, Shih et al. 2011, Rajkumar et al. 2019).However, in this study, chemotherapy was not classified as neoadjuvant or postoperative adjuvant.Therapeutic responses after chemotherapy and specific chemotherapy schemes were not recorded in the SEER database.Second, this study only included information on primary tumour resection.The size of the residual lesions and surgical details of the distant metastatic sites were unclear.It is not clear whether patients with stage IVB EC benefitted from the resection of metastatic lesions, and this requires further research.Moreover, differences in the clinical behaviour of EC can be partly attributed to different genetic and molecular profiles, including polymerase epsilon (POLE) mutations, microsatellite instability-high (MSI-H), low copy number (CN), and high CN.However, this classification system was not included as a possible predictor in this cohort study.This classification is a prerequisite for precise treatments, such as chemotherapy, radiotherapy and immunotherapy (Morice et al. 2016, Conway et al. 2019, NCCN 2022, Cuccu et al. 2023).The application of genetic and molecular profile for therapy selection in patients with early-stage EC is supported by clinical evidence and practice, but it is still being gradually explored in patients with advanced EC.The MSI-H subclass is attributed to mismatch repair deficiency (dMMR) in the DNA repair pathway, which occurs in 15-30% of EC and may be higher in advanced or recurrent disease (Nagle et al. 2018).This subclass of tumours with a higher neoantigen load generate strong CD3þ and CD8þ T-cell responses determining the sensitivity of immunotherapy (Roudko et al. 2020).The efficacy of programmed death ligand 1 (PD-L1) and programmed death 1 (PD-1) immune checkpoint inhibitors have been demonstrated in clinical trials, particularly in patients with MSI-H (Antill et al. 2021, Cuccu et al. 2023).Co-inhibition of vascular endothelial growth factor (VEGF) and PD-1 signalling to evoke immune activation is also an effective antitumour strategy for patients with advanced or recurrent microsatellite stable (MSS) EC (Makker et al. 2019, Cuccu et al. 2023).
Finally, tumour markers such as CA125, performance status, preoperative comorbidities, and postoperative complications were not included in this study because of the limitations of the SEER database.
Although this study provides a suitable option for treating advanced EC, new technologies and approaches that enable early identification of EC could improve survival times and reduce medical expenditure.These methods offer intriguing perspectives and research directions, including innovative cancer biomarkers (i.e.micro RNAs, DNA methylation, and histones modifications) or combination with machine learning (Wang et al. 2016), statistical models, and radiomics (Bogani et al. 2022).In conclusion, we confirmed the value of surgical treatment in improving the OS of patients with stage IVB EC; however, but the possibility of benefits from a therapeutic process involving CDS and tumour biology should be fully considered in practice.We obtained reliable estimates using PSM to balance the bias from various variables.We found that CDS was most likely to improve survival in patients younger than 60 years, in T1, and T2 invasion subgroups.Compared with this, it is less likely to benefit black people, divorced individuals, those with disease staging of T3, N2, Nx/Blank(s); and those with metastasis to the liver, lungs, and bones.Receiving CDS has also been demonstrated to improve the prognosis of individuals with brain metastases; however, further verification requires a larger sample size.

Figure 1 .
Figure 1.Flow chart depicting the patient selection process.EC: endometrial cancer, CDS: cancer-directed surgery.

Figure 2 .
Figure2.Multivariate cox regression hazard ratios with 95% confidence intervals.Sorted according to hazard ratios and the dashed line indicated equivalent hazard ratio (HR ¼ 1).Transparency of the confidence interval estimates represents the range of the p-value, where p < 0.05 is the solid black line, as the p-value increases, the more transparent the line is, the less likely it is to affect survival.

Table 1 .
Baseline characteristics before and after propensity score matching between CDS and non-CDS groups.

Table 2 .
Univariate and multivariate analysis for overall survival.Bolded p-values in the table are less than 0.05.