Gender Differences in Prognosis After Primary Resection for Retroperitoneal Liposarcoma

Background Current evidence regarding gender difference in retroperitoneal liposarcoma (RLPS) is scarce, so we sought to investigate whether gender may affect prognosis after primary resection of RLPS. Methods We used the Surveillance, Epidemiology, and End Results (SEER) database to identify RLPS patients from January 1973 to December 2015. Multivariate cox proportional hazard analysis was adopted to generate adjusted hazard ratio (AHR) and 95% confidence intervals (CIs) of survival outcomes. Results In total, 2108 RLPS patients, including 971 women and 1137 men, were identified, with a median follow-up of 45.0 (17.0-92.0) months. The 5-year and 10-year overall survival rates were 50.5% and 31.5% for men and 60.4% and 42.5% for women. The 5-year and 10-year disease-specific survival rates for men and women were 71.5%, 57.3% and 76.3%, 62.1%, respectively. We found men were associated with an increased risk of all-cause mortality (AHR 1.3, 95% CI 1.0-1.6, P = .017) but not disease-specific mortality (AHR 1.2, 95% CI .9-1.6, P = .246). The subgroup analyses revealed that men were associated with an increased risk of all-cause mortality in patients with low-grade tumors (AHR 1.8, 95% CI 1.3-2.5) or patients who received non-radical resection (AHR 1.6, 95% CI 1.2-2.1). In the subgroup of low-grade tumors, men were also associated with an increased risk of disease-specific mortality (AHR 2.0, 95% CI 1.2-3.3). Conclusion Men may have worse survival after primary resection of RLPS compared with women, especially in patients with low-grade tumors or patients who received non-radical resection. Gender-based disparities may deserve more attention in patients with RLPS.


Introduction
Liposarcoma is the most common type of soft tissue sarcoma, accounting for approximately 20% of all mesenchymal neoplasms in adults. 1 Retroperitoneal liposarcoma (RLPS) constituted 10%-15% of all LPSs, also representing the most common (33%) primary sarcoma to occur in the retroperitoneum. 1,2This disease is characterized by various types of adipocytic differentiation, 3 which can be classified into 4 major subtypes: well-differentiated LPS (WDLPS), dedifferentiated LPS (DDLPS), myxoid LPS (MLPS), and pleomorphic LPS (PLPS).The prognoses of LPSs differ according to the primary sites, among which RLPS has the worst prognosis with a 12-year diseasespecific survival of 32% compared to 77%-87% in other parts of the body. 4Therefore, identifying prognostic factors specifically for RLPS is necessary.
In the recent decade, gender-based differences in tumor biology and prognosis have received increasing attention.In 2013, research in mouse models found gender may distinguish tumor behavior and prognosis in certain genetic alterations, 5 which suggested gender may play a subtle role in tumor suppression and oncogenesis, as well as the development of specific subsets of human tumors. 6In addition, other researchers also observed large gender-based differences in both tumor mutational burden and T-cell inflammation of the cancer microenvironment, which revealed the relationship between genderdimorphism and anticancer immune response. 7Given increasingly accumulating evidence in gender-disparities in cancer, the 2019 consensus of European Society for Medical Oncology (ESMO) workshop recommended men and women with non-gender-related cancers should be considered as biologically distinct groups of patients, for whom specific management strategies merit consideration, especially in diseases with significant differences in outcomes. 8Therefore, investigation of gender disparities in disease prognosis is fundamental for a better understanding of cancer biology.
In terms of soft tissue sarcomas, only a few studies explored potential gender differences from cell and molecular perspectives, 5,9 and there is no evidence showing if there are any gender differences in clinical prognosis, especially in RLPS, the most common subtypes of sarcoma with the poorest prognosis.The present study aimed to conduct a national population-based retrospective cohort study to identify whether gender has an impact on the survival in patients who underwent primary surgery for RLPS.We present the following study in accordance with the STROBE reporting checklist.

Data Sources
We retrieved data from the Surveillance, Epidemiology, and End Results (SEER) database for analysis.SEER is a network of population-based cancer registries from geographically different regions in the United States maintained by the National Cancer Institute (NCI), covering 28% of the US population. 10The dataset was extracted from SEER Ã Stat Database: "Incidence -SEER 18 Regs Custom Data (with additional treatment fields), based on the November 2017 submission (1973-2015 varying)."

Variables and Outcome Measures
The primary exposure variable in this study was gender (men vs women).Other independent variables involved in multivariable regression were age, race, tumor size (diameter larger than 150 mm or not), histological behaviors (RLPS without specific subtypes, well-differentiated LPS, dedifferentiated LPS, pleomorphic LPS, myxoid/round cell LPS, and mixed liposarcoma), summary stages (localized, regional, and distant), grade of histology (low grade vs high grade, "Well differentiated; Grade I" and "Moderately differentiated; Grade II" were categorized to low-grade, and "Poorly differentiated; Grade III" and "Undifferentiated; anaplastic; Grade IV" were categorized to high-grade), surgical procedures (radical surgery vs non-radical surgery), chemotherapy (yes or not), and radiation therapy (yes or not).In the SEER database, a radical resection was defined as a "partial or total removal of the primary site WITH an en bloc resection (partial or total removal) of other organs."Partial removal of the primary site can be considered as a radical surgery if vital organs are involved.The primary outcome was all-cause mortality, and secondary outcome was disease-specific mortality.Survival time was counted from the date of diagnosis of LPS to the date of death or right-censored at the end of follow-up period.

Statistical Analysis
Continuous variables were described as medians (interquartile ranges), and categorical variables were described as numbers (percentages).The distribution of demographics between the 2 groups was compared using logistic regression models.For univariate analysis, Kaplan-Meier method and log-rank test were adopted to assess survival outcomes.For multivariate analysis, Cox proportional hazard regression models were used to generate adjusted hazard ratio (AHR) and 95% confidential intervals (CIs).Variables included in the multivariate analyses were identified by clinical significance and its impact on the outcome measures in the univariate analysis, that is, changing the HR by at least 10% when added into the model.
To address the effect of gender in specific population, we conducted subgroup analysis by age (<70 years and ≥70 years), tumor size (≤150 mm and >150 mm), tumor grade (low grade vs high grade), histological behaviors (well-differentiated LPS, dedifferentiated LPS, pleomorphic LPS, and myxoid/round cell LPS), summary stages (localized, regional, and distant), and surgical

576
The American Surgeon™ 90( 4) procedures (radical resection vs non-radical resection).Pvalues of interactions between gender and each subgroup variables were evaluated using likelihood ratio test by including the interaction terms in the Cox regression model.All statistical analyses were performed using R studio Version 1.2.1335 (http://www.R-project.org) and Empower(R) (www.empowerstats.com,X&Y solutions, Inc., Boston, MA).All P-values were 2-sided and considered significance at a level of .05.

Baseline Characteristics
After selection of study population, a total of 2108 RLPS patients involving 1137 (53.9%) men and 971 (46.1%) women were included in our study.The flow diagram is presented in Figure 1.Among the included patients, demographics including race and summary stage were similar between men and women.Therapeutic strategies were also comparable between the 2 groups regarding surgical procedures, lymph node dissection, perioperative radiation, and chemotherapy.However, men were older than women (65 (55-72) years vs 60 (51-70) years).In addition, men had higher proportion of high-grade tumors (39.3% vs 30.7%) but less proportion of tumors larger than 150 mm (94.8% vs 97.1%).Detailed baseline characteristics are displayed in Table 1.

Univariate Analyses for Survival Outcomes
The median follow-up time for men and women was 42.0 (14.0-82.0)months and 51.0 (20.5-106.5)months, respectively.The 5-year and 10-year overall survival rates were 50.5% and 31.5% for men and 60.4% and 42.5% for women.The 5-year and 10-year disease-specific survival rates for men and women were 71.5%, 57.3% and 76.3%, 62.1%, respectively.The survival curves of all-cause mortality and disease-specific mortality are shown in Figures 2 and 3. Univariate analysis suggested men were associated with a significantly increased risk of all-cause mortality (HR 1.4, 95% CI 1.2-1.6,P < .001)and diseasespecific mortality (HR 1.2, 95% CI 1.1-1.5,P = .008).Furthermore, age, summary stage, grade of histology, lymph node dissection, chemotherapy, and numbers of malignant tumors were also found to be associated with survival outcomes.The survival curves stratified by different subgroups are displayed in Figures 2 and 3.

Multivariate Analyses for Survival Outcomes
After variable selection, age, histology subtypes, summary stages, tumor sizes, grade of histology, surgical procedures, radiotherapy, and chemotherapy were included in the multivariate Cox regression model.In adjusted model I (age, histology subtypes, and summary stages), men were associated with a significantly increased risk of all-cause mortality (AHR 1.2, 95% CI 1.1-1.4,P < .001)but not associated with disease-specific mortality (AHR 1.2, 95% CI 1.0-1.4,P = .086).In the fully adjusted model II, the results were similar with the adjusted model I, and men were also associated with a significantly increased risk of all-cause mortality (AHR 1.3, 95% CI 1.0-1.6,P = .017)but not associated with diseasespecific mortality (AHR 1.2, 95% CI .9-1.6, P = .246).The results of multivariate analyses are shown in Table 2.

Subgroup Analysis by Age
To test whether the effect of gender may differ in the elderly population older than 70 years, we performed subgroup and interaction analysis by age in the

578
The American Surgeon™ 90(4) multivariate model adjusting for histology behavior, summary stages, grade of histology, tumor size, surgical procedures, and chemotherapy.While no gender-based difference in survival was found in the subgroup of patients older than 70 years, we found men were associated with an increased risk of all-cause mortality (AHR 1.4, 95% CI 1.0-1.8,P = .024)but not disease-specific mortality (AHR 1.2, 95% CI .9-1.7, P = .266)in patients younger than 70 years.

Subgroup Analysis by Tumor Grade
In addition, to address the potential interaction effect of histology grades with gender on survival, we further performed subgroup and interaction analysis regarding grade of histology in the multivariate adjusted model.After adjusting for age, histology behavior, tumor size, summary stages, surgical procedures, and chemotherapy, we did not observe significant association between men and all-cause mortality (AHR 1.1, 95% CI .8-1.5, P = .494)or disease-specific mortality (AHR 1.1, 95% CI .8-1.6, P = .447)in the subgroup of high-grade tumors.However, in terms of patients with low-grade tumors, we found men were associated with significantly increased risks with both all-cause mortality (AHR 1.8, 95% CI 1.3-2.5,P < .001)and diseasespecific mortality (AHR 2.1, 95% CI 1.2-3.4,P = .005).
Additionally, significant interaction effect between grade of histology and gender on all-cause mortality was found (P for interaction = .04).

Subgroup Analysis by Tumor Size
In the multivariate model adjusted for age, histology behavior, grade of histology, summary stages, surgical procedures, and chemotherapy, we performed subgroup analyses stratified by tumor size.The results revealed that men were associated with both increased risks of all-cause mortality (AHR 1.3, 95% CI 1.0-1.7,P = .028)and diseasespecific mortality (AHR 1.5, 95% CI 1.1-2.1,P = .015)in patients with tumor larger than 150 mm.By comparison, no association between gender and survival outcomes was found in patients with tumor smaller than 150 mm.

Subgroup Analysis by Surgical Procedure
We also performed subgroup analysis stratified by surgical procedure in the multivariate model adjusted for age, histology behavior, grade of histology, tumor size, summary stages, and chemotherapy.Men were found to be associated with increased risks of all-cause mortality (AHR 1.6, 95% CI 1.2-2.2,P = .001)and disease-specific mortality (AHR 1.7, 95% CI 1.1-2.5,P = .011)in patients who underwent non-radical resection.As for patients received radical tumor resection, no association between gender and all-cause mortality, as well as disease-specific mortality, was observed.A test of interaction between surgical procedures and gender on all-cause mortality was statistically significant (P = .029).The results of other subgroup and interaction analyses are summarized in Figure 4.

Discussion
In this study, we analyzed SEER data to investigate the effect of gender on survival of patients who underwent primary surgical resection of RLPS.Our results revealed that men were associated with an increased risk of allcause mortality, but not disease-specific mortality, after primary resection of RLPS in overall population.However, in patients with low-grade tumors, men were associated with increased risks of both all-cause mortality and disease-specific mortality after surgery.And these gender differences were more pronounced in patients who underwent non-radical surgery.
Given the rarity of the disease, no specific guideline for RLPS has been established yet.The most recent consensus was built by Trans-Atlantic RPS Working Group in 2015, focusing on the management of primary retroperitoneal sarcoma (RPS) in the adult. 11The consensus provided detailed perioperative management strategies of RPS but did not cover the prognostic values of basic characteristics of the patients.Though increasing numbers of studies have suggested gender disparities may play important roles in cancer biology and prognosis, only few studies have examined the role of gender in the prognosis of RLPS.Moore Dalal et al built a nomogram to predict survival outcomes in 801 LPS patients, and gender was identified as a predicting factor for overall LPS.However, the nomogram also suggested the primary site of LPS strongly affected the prognosis, and RLPS represented the worst subtypes.Therefore, identification of the role of gender specifically in the prognosis of RLPS is necessary.The following studies only performed nomograms for patients with primarily resected or recurrent retroperitoneal sarcomas, and gender was not identified as one of the predictors of survival. 12,13The results of our study fill the gap concerning the impact of gender disparities on the prognosis of RLPS.
In the overall population, we found men had a worse overall survival than women after primary resection of RLPS.][16][17][18] According to the SEER data, we found male LPS patients were generally older and had a higher proportion of highgrade tumors, which may lead to worse overall survival of male patients in the unadjusted analyses.However, we came to the same conclusion after we adjusted for age, histology, summary stages, tumor sizes, grade, and perioperative interventions.Regarding the potential reasons for survival benefits in women, a recent study found that the age-standardized rates for the majority of cancer types were greater in men than women in the 50 populations studied, which might suggest that at any given chronological age, the female tissues are physiologically younger than male tissues. 19Other studies indicated that women's benefits in overall survival may be attributed to women's more responsive inflammatory functioning and men's cardiovascular weakness. 20,21uture studies are necessary to further explore the underlying reason for the gender difference in the prognosis of RLPS.In addition, several noteworthy findings emerged in subgroup analyses.First, as age is usually identified as an important predictor for survival, we tested whether the advantage of better overall survival still existed in elderly female RPLS patients.Our results revealed that the benefit of overall survival in women disappeared in the subgroup older than 70 years, which might be explained by the equal susceptibility to surgical risks in the elderly patients.Second, surgical procedures seemed to be another interactive factor with gender in the prognosis of RLPS.We found the gender disparity of overall survival was more evident in the patients who underwent non-radical resection, while the survival benefit in women disappeared after radical surgery.This interaction effect may be caused by the strong effect of radical surgery on overall survival. 22Furthermore, this may primarily be attributed to the fact that radical resections were still advantageous in male patients after accounting for the effects of age, histology, grade, tumor size, and radiotherapy. 23,24Additionally, this gender disparity can be attributed to inherent anatomical differences in the pelvic region between males and females, as radical resection often involves the removal of a greater number of organs in female patients than male patients.Evidence suggests a negative correlation between the extent of organ resection and long-term overall survival rates. 25Third, though no gender difference in disease-specific mortality was observed in overall population, subgroup analysis stratified by tumor grade indicated that men had a higher risk of disease-specific mortality than women in the subgroup of low-grade tumors with significant interactive effect.Recent STRASS/ STREXIT study has demonstrated the potential impact of radiotherapy on low-grade tumors, 26 which may contribute to the gender differences in our study.These findings may affect clinical practice by suggesting male RLPS patients with low-grade tumors should be treated and followed with a personalized strategy.
Some limitations existed in our study.First, our study was a retrospective study, subjective to information and selection biases.However, our study had a large sample size with a long period of follow-up, which could provide sufficient information.Second, RLPS is characterized by a high rate of local recurrence after primary resection, but the SEER database did not present details of recurrence during follow-up.As a potential substitute, the results of disease-specific death in this study might shed light on the tumorrelated adverse outcomes.Disease-specific death (DSD) is often used in research to assess the clinical outcomes of retroperitoneal Liposarcoma, 27 and studies have shown that histological type/subtype is the most significant independent predictor of DSD in retroperitoneal Liposarcoma. 28Third, SEER database did not report surgical margins, which prevented us from evaluating the potential confounding effect of resection completeness on survival outcomes.Fourth, French Federation of Cancer Centers Sarcoma Group (FNCLCC) grading criteria for soft tissue sarcoma was not used due to lack of data, instead our study adopted histology grade of SEER database, which involved 4 categories: "Well differentiated; Grade I," "Moderately differentiated; Grade II," :Poorly differentiated; Grade III," and "Undifferentiated; anaplastic; Grade IV." Though not as elaborate as FNCLCC grading criteria, our study could reveal an overall trend in the interaction between pathological information and gender on the prognosis of RLPS.

Conclusions
In conclusion, this population-based cohort study revealed that men had an increased risk of all-cause mortality after primary resection of RLPS compared with women, which is more evident in patients with low-grade tumors or patients who received non-radical resection.No gender difference in disease-specific mortality was found in overall population; however, in patients with low-grade tumors, men also have an increased risk of disease-specific mortality.In future clinical practice, gender-based disparities in the prognosis of RLPS should receive more attention, and individualized therapeutic and surveillance plan are warranted, particularly in patients with low-grade tumors or patients who received non-radical resection.

Figure 1 .
Figure 1.Flow diagram of patient selection.

Figure 2 .
Figure 2. Kaplan-Meier curves of overall survival (A) and disease-specific survival (B) in patients who underwent primary resection of retroperitoneal liposarcoma, stratified by histological grade of tumor.

Figure 3 .
Figure 3. Kaplan-Meier curves of overall survival (A) and disease-specific survival (B) in patients who underwent primary resection of retroperitoneal liposarcoma, stratified by surgical procedures.

Figure 4 .
Figure 4. Forest plots showing results of multivariate subgroup analyses and interaction analyses for overall survival (A) and diseasespecific survival (B).

Table 1 .
Baseline Characteristics of Retroperitoneal Liposarcoma Patients.
Continuous data was expressed as median (interquartile range); NOS, not otherwise specified.

Table 2 .
Multivariate Cox Regression Analyses for the Effect of Gender on the Survival Outcomes of Different Population.Adjusted for age, histological type, summary stage, tumor grade, tumor size, surgical procedures, radiotherapy, and chemotherapy.b Adjusted for histological type, summary stage, tumor grade, tumor size, surgical procedures, and chemotherapy.Adjusted for age, histological type, summary stage, tumor size, surgical procedures, and chemotherapy.d Adjusted for age, summary stage, tumor grade, tumor size, surgical procedures, and chemotherapy.e Adjusted for age, histological type, tumor grade, tumor size, surgical procedures, and chemotherapy.f Adjusted for age, histological type, summary stage, tumor grade, surgical procedures, and chemotherapy.g Adjusted for age, histological type, summary stage, tumor grade, tumor size, and chemotherapy.
a c