Prognostic value of positive peritoneal cytology in FIGO 2009 stage IA grade 1 endometrioid endometrial cancer.

A total of 1151 patients with FIGO 2009 stage IA grade 1 endometrioid endometrial carcinoma met inclusion criteria. Of these, 1101 (95.7%) had NPC and 50 (4.3%) had PPC. For a detailed CONSORT-style diagram, see Supplemental Figure S1 . The median age at surgery was 58 years (range, 24–92). Demographic characteristics were similar between the PPC and NPC groups ( Table 1 ). Preoperative hysteroscopy was performed in 286 patients (24.8%), including 276 of 1101 patients (25.1%) with NPC and 10 of 50 patients (20.0%) with PPC ( P =0.5). The median time from hysteroscopy to definitive surgery was 1.2 months (range, 0.1–43.6) and did not differ between groups ( P =0.96). Of 1151 patients, 32 (2.8%) were treated with progestins prior to surgery, all of whom had NPC. Surgery was performed using a minimally invasive technique (including robotic-assisted surgery) in 1094 cases (95.0%); a uterine manipulator was used in 1061 (97.0%) of these cases and equally between groups ( P =0.4). Among these 1061 cases, a majority used the VCare ® (70.8%) or ZUMI ® (23.7%) manipulator, with fewer cases utilizing the HUMI ® manipulator (0.9%) or a Hulka clamp (3.8%) ( Supplemental Table S1 ). Distribution of manipulator type did not differ between the PPC and NPC groups ( P =0.29). All patients underwent lymph node evaluation; sentinel lymph node mapping was performed in 1097 patients (95.3%) and equally between groups ( P >0.99). The median uterine weight was 116.0 grams (range, 17.0–2186.5) in patients with NPC compared to 99.0 grams (range, 41.0–410.0) in those with PPC ( P =0.04). Transvaginal surgical specimen extraction was successful in all but 5 cases with NPC, which required a mini laparotomy due to enlarged uterine size. Vaginal tears from transvaginal uterine extraction were documented in 40 patients (3.5%), 37 (3.4%) in the NPC group and 3 (6.0%) in the PPC group ( P =0.25). Competing risk analysis showed the 5-year risk of vaginal recurrence was 1.8% (95% CI: 1.1% to 2.8%) for patients without vaginal tears, compared to 8.4% (95% CI: 2.1% to 20.4%) in patients who developed vaginal tears ( Supplemental Table S2 ). On final pathology, 953 tumors (82.8%) were found to be limited to the endometrium; 198 (17.2%) exhibited superficial invasion (<50% myometrial thickness), including 184 of 1101 tumors (16.7%) in the NPC group and 14 of 50 (28.0%) in the PPC group ( P =0.05). Microcystic elongated and fragmented (MELF) pattern of invasion was found in 36 of 1151 tumors (3.1%), including 35 of 1101 tumors (3.2%) in the NPC group and 1 of 50 (2.0%) in the PPC group ( P >0.99). Mismatch repair (MMR) protein IHC was performed in 794 of 1151 tumors (69.0%), including 755 of 1101 (68.6%) in the NPC group and 39 of 50 (78.0%) in the PPC group ( P =0.21). Of the 794 tumors with MMR IHC, 164 (20.7%) were MMR-deficient, including 152 of 745 tumors (20.4%) in the NPC and 12 of 39 tumors (30.8%) in the PPC group; 10 tumors had equivocal immunohistochemical staining ( P =0.13). IHC staining for p53 was available for 562 of 1151 tumors (48.8%), including 535 of 1101 tumors (48.6%) in the NPC group and 27 of 50 (54.0%) in the PPC group. Of the 562 tumors with p53 staining, 7 (1.2%) had aberrant p53 staining, all in the NPC group. On logistic regression analysis, larger uterine size was associated with decreased odds of PPC (odds ratio [OR]: 0.96, 95% CI: 0.91 to 1.00) ( Supplemental Table S3 ). Molecular subtype, depth of myometrial invasion, preoperative hysteroscopy, and MELF pattern of invasion were not associated with PPC. The median follow-up duration of the study was 54.3 months (range, 0.5–169.5), during which time 31 recurrences and 36 deaths without documented recurrence occurred. Of the 31 documented recurrences, 22 were vaginal cuff recurrences (18 in the NPC group and 4 in the PPC group) and 9 were non-vaginal recurrences (8 in the NPC group and 1 in the PPC group). No patients received postoperative treatment (including vaginal brachytherapy). The 5-year PFS rate for the NPC group was 95.2% (95% CI: 93.6% to 96.5%), compared to 88.8% (95% CI: 75.0% to 95.2%) for the PPC group ( P =0.02) ( Figure 1 ). A univariable Cox regression analysis revealed that PPC, depth of myometrial invasion, vaginal tears from uterine extraction, age at surgery, and a history of secondary primary malignancy were predictors of worse PFS ( Table 2 ). In a multivariable model, PPC remained significantly associated with worse PFS, with an adjusted hazard ratio (HR) of 2.63 (95% CI: 1.19 to 5.80; P =0.02) ( Table 3 ). Of 1151 tumors, 259 (22.5%) underwent molecular testing, of which 43 were excluded for not meeting quality criteria ( Supplemental Figure S1 ). Overall, 194 tumors in the NPC and 22 in the PPC group were molecularly characterized. Molecular subtyping revealed that 139 tumors (64.4%) were of copy number-low/no specific molecular profile subtype, 47 (21.8%) were microsatellite instability-high, 9 (4.2%) were copy number-high/ TP53 abnormal, and 21 (9.7%) were of POLE molecular subtype. Among the 22 molecularly profiled tumors in the PPC group, only 1 recurrence was documented, precluding subgroup survival analysis. The most common recurrent gene alterations were found in PTEN (81%), PIK3CA (63%), and ARID1A (48%). A comparative mutational analysis between groups revealed that tumors in the PPC group were enriched in alterations affecting the chromatin-remodeling gene ARID1A (82% vs. 44%, P =0.001), as shown in the OncoPrint in Figure 2 . Among the 47 tumors with microsatellite instability-high profile, 4 of 5 (80.0%) in the PPC group were enriched in PIK3R1 alterations compared to 10 of 42 (23.8%) in the NPC group ( P =0.023). Among the 139 tumors of no specific molecular subtype, 10 of 13 (76.9%) in the PPC group were enriched in ARID1A alterations compared to 46 of 126 (36.5%) in the NPC group ( P =0.007). No genes were differentially altered in tumors of POLE subtype across the 2 groups. ER IHC was performed on all 6 tumors that recurred in the PPC group. All 6 retained ER expression, with ≥1% nuclear staining consistent with ER positivity based on ASCO/CAP breast cancer criteria.

This single-institution retrospective cohort study was approved by the Institutional Review Board of Memorial Sloan Kettering Cancer Center (MSK). Patients who were diagnosed with FIGO 2009 stage IA grade 1 endometrioid endometrial cancer and underwent surgical staging, including bilateral pathological evaluation of pelvic lymph nodes and peritoneal cytology sampling, between January 01, 2008, and August 01, 2021, were included. Patients with lymphovascular space invasion, synchronous tumors, or positive lymph nodes (including isolated tumor cells), those without surgical nodal evaluation, and those who received any form of adjuvant therapy (including vaginal brachytherapy), were excluded. Patients with suspicious or atypical peritoneal cytology were also excluded from analysis. Patients were grouped according to peritoneal cytology status (negative vs. positive). Data on demographics, disease characteristics, and outcomes were extracted from electronic health records and compiled in a centralized database. Surgical staging included hysterectomy, bilateral salpingo-oophorectomy with nodal evaluation, and peritoneal cytology sampling. The method of nodal evaluation was at the physician’s discretion. Following surgery, patients were monitored every 3 to 6 months for the first 2 to 3 years, then every 6 months for the following 2 to 3 years, with clinical exams. Imaging was at the physician’s discretion and was based on onset of new symptoms or concerns for recurrence. Descriptive statistics were provided. Comparisons between PPC and negative peritoneal cytology (NPC) were performed using the Wilcoxon rank sum test for continuous variables, and the Fisher exact test for categorical variables. The primary outcome, progression-free survival (PFS), was calculated from date of surgery to date of recurrence, death, or last follow-up. Five-year PFS and median PFS survival time were estimated using the Kaplan-Meier method. To examine the prediction ability of clinical variables for PFS in the univariate setting, P values were obtained by applying log-rank test or log-rank test with permutation for categorical variables and the Wald test based on the Cox proportional hazards model for continuous variables. A multivariable Cox regression model was built based on variables that were statistically significant in univariable analysis. An exploratory analysis was conducted to evaluate factors associated with vaginal recurrences using competing risk analysis with the Gray test.[ 7 ] Logistic regression analysis was conducted to evaluate any significant associations between clinicopathological factors and the occurrence of PPC. All P values were calculated as 2-sided, and P values <0.05 were considered statistically significant. All analyses were performed in R 4.4.2 ( https://www.R-project.org/ ). A subgroup analysis that included patients whose tumors underwent clinical paired tumor-normal panel DNA sequencing (MSK-IMPACT [MSK-Integrated Mutation Profiling of Actionable Cancer Targets]) of 410–505 genes was performed. Molecular subtyping was performed using an integrated immunohistochemical-molecular approach, as previously described.[ 8 ] At our institution, during the study period, molecular testing was ordered at the discretion of the treating physician, frequently in the context of clinical trial enrollment or institutional sequencing initiatives. Molecular profiles and the mutational landscapes of primary tumors from patients with PPC and NPC were compared.[ 8 ] To explore potential biological differences in tumors that recurred within the PPC group, estrogen receptor (ER) immunohistochemistry (IHC) was performed on all recurrent tumors in this cohort. Although ER status is not routinely reported for this histology in our clinical practice, this exploratory analysis was undertaken to assess whether loss of ER expression could be a confounding factor contributing to the higher recurrence rate observed in the PPC group. Staining interpretation was conducted by expert gynecologic pathologists. Tumors with ≥1% nuclear staining were considered ER positive, in accordance with American Society of Clinical Oncology/College of American Pathologists (ASCO/CAP) breast cancer guidelines.[ 9 ] We acknowledge this cutoff is extrapolated and not validated in endometrial cancer, but the clinical relevance in this patient population was demonstrated in a recent publication by our team.[ 10 ]

In this large retrospective cohort of patients with surgically staged FIGO 2009 stage IA grade 1 endometrioid endometrial cancer, PPC was independently associated with worse PFS, even within this traditionally low-risk population. Although prior studies have shown cytology is prognostically relevant in broader or higher-grade cohorts[ 3 , 11 – 16 ], data specific to low-grade, early-stage disease are limited and inconsistent.[ 3 , 11 , 12 ] Fumagalli et al. recently demonstrated that PPC is an independent predictor of recurrence in patients with low-risk endometrial cancer, despite having no impact on overall survival. Their analysis, however, did not incorporate molecular profiling, leaving unanswered questions regarding the biological underpinnings of this association.[ 17 ] Our study addresses this gap by examining a carefully curated cohort devoid of deep myometrial invasion, lymphovascular space invasion, and high-grade histology, allowing for a focused evaluation of cytology’s prognostic role in the absence of traditional risk factors. After adjusting for age at surgery, prior malignancy, myometrial invasion, and vaginal tears at the time of uterine extraction, cytology remained an independent predictor of adverse outcome. Importantly, in tumors with available molecular data, we found no association between molecular subtype and recurrence, suggesting that cytology may capture prognostic information not reflected by existing molecular classifications. We did, however, detect a possible enrichment of ARID1A mutations in tumors from patients with PPC, suggesting a possible molecular mechanism for peritoneal dissemination. We also found no link between preoperative hysteroscopy or intrauterine manipulator use and PPC, mitigating concerns that procedural factors artificially inflate detection rates. Taken together, our findings suggest that peritoneal cytology may provide additional prognostic information, even in low-grade disease; however, due to the low observed recurrence rate and the excellent overall prognosis in this cohort, the clinical implications of PPC alone remain uncertain. By integrating detailed clinical variables and molecular data, our study provides compelling evidence against the notion that PPC in low-risk endometrial cancer is merely incidental or artifactual. Several factors have been associated with the presence of PPC, including deep myometrial invasion, high tumor grade, and lymphovascular space invasion.[ 12 , 15 ] The use of intrauterine manipulators during laparoscopic hysterectomy has also been debated.[ 18 , 19 ] A nationwide retrospective cohort study by Yoshida et al. reported that manipulator use increased the odds of malignant peritoneal cytology by 77% (adjusted OR: 1.77, 95% CI: 1.29 to 2.31). In our cohort, manipulator use was prevalent (97%), and manipulator types did not differ between groups ( P =0.29). Additionally, there was no association between manipulator use and PFS (HR: 1.07, 95% CI: 0.26 to 4.37). Regarding surgical approach, our findings align with those of Nasioudis et al., who reported no significant difference in overall survival between minimally invasive and open surgical approaches, despite higher cytology positivity rates with the former.[ 20 ] Concerns regarding hysteroscopy-induced dissemination of cancer cells have also been raised.[ 21 , 22 ] Our results show no association between hysteroscopy and increased cytology positivity (OR: 0.75, 95% CI: 0.35 to 1.46) or adverse oncologic outcomes (HR: 0.55, 95% CI: 0.28 to 1.07). These findings are consistent with prior meta-analyses concluding that hysteroscopy does not compromise long-term survival, even when liquid distention media are used.[ 22 , 23 ] National Comprehensive Cancer Network guidelines support the use of hysteroscopy as a diagnostic modality,[ 24 ] and our findings further reinforce its safety when used judiciously in the evaluation of endometrial cancer. Interestingly, we observed an association between smaller uterine weight – a surrogate measure that we used for tumor size – and PPC. This finding contrasts with recent studies linking larger tumor size to extrauterine tumor spread and worse oncological outcomes.[ 25 ] One possible explanation is that uterine weight and tumor size are not tightly correlated, as uterine weight may be influenced by factors such as leiomyomas, adenomyosis, or other benign pathology, and may not accurately reflect tumor burden. Future studies should prospectively evaluate both tumor size and uterine weight to clarify their respective roles in predicting PPC. A National Cancer Database study analyzing 1518 patients with stage IA endometrioid endometrial cancer reported a 4.8% prevalence of PPC, similar to our findings.[ 16 ] It also reported that PPC was predictive of decreased survival, with a 4-year overall survival rate of 79.5% compared to 92.2% in patients with NPC (HR: 1.85, 95% CI: 1.54 to 2.21). Conversely, Scott et al. found no significant association between PPC and disease-free or overall survival, possibly due to the low prevalence of PPC (2.2%) in their cohort and, by extension, low event counts over the follow-up period.[ 26 ] In our cohort, 4 of 5 recurrences in the PPC group were isolated to the vaginal cuff. Isolated vaginal recurrence is often detected by clinical examination alone and can be definitively treated with radiotherapy. [ 26 ] In light of this, the impact of PPC on clinical management may be modest. Positive cytology may help to inform surveillance strategies, such as tailoring follow-up intervals or considering early imaging in select patients. In our analysis, vaginal tears during uterine extraction were associated with increased risk of vaginal cuff recurrence in competing risk analysis. While intriguing, these findings were based on a small number of events. Further studies are needed to assess whether mechanical disruption influences recurrence patterns. Depth of myometrial invasion is a well-established predictor of recurrence in endometrial cancer. In our analysis, we noted a higher risk of PFS with superficial invasion compared to no invasion (HR: 1.8, 95% CI: 1.06 to 3.07); however, this association did not persist in multivariable models. While evidence suggests a clear relationship between depth of myometrial invasion and oncologic outcome in low-grade, early-stage endometrioid endometrial cancer,[ 27 ] the absolute difference in recurrence rates between no invasion and superficial invasion remains small, potentially explaining the loss of significance in multivariable analysis. On the other hand, a recent paper by Jamieson et al., found no difference in oncological outcomes between non-invasive and superficially invasive tumors in p53-abnormal endometrial cancers.[ 28 ] This finding hints that the prognostic impact of superficial invasion may depend on context, modulated by underlying molecular subtype or other tumor biology. From a molecular standpoint, among the 216 tumors with available profiling, we observed no association between The Cancer Genome Atlas (TCGA) molecular subtype and either peritoneal cytology status or recurrence risk. At the genomic level, however, tumors from patients with PPC demonstrated a significantly higher frequency of ARID1A mutations compared to those from patients with NPC (82% vs. 44%, P =0.001). This is a compelling finding, but it is based on a limited number of PPC cases (n=22) with differential testing frequency between groups, and therefore should be interpreted with caution. ARID1A inactivation is known to impair E-cadherin–mediated cell adhesion and promote epithelial-mesenchymal transition, facilitating tumor cell detachment and migration. Prior studies have linked ARID1A loss to enhanced tumor motility and a more permissive microenvironment for dissemination.[ 29 ] Although we could not demonstrate a direct association between ARID1A mutation and recurrence due to a low number of events, this genomic enrichment may help explain the presence of malignant cells in the peritoneal cavity independent of iatrogenic factors. Notably, emerging data indicate that inhibition of histone deacetylase 6 (HDAC6)—a cytoplasmic enzyme that plays a critical role in cell motility—can counteract the enhanced migration and invasiveness driven by ARID1A deficiency. This observation provides a mechanistic rationale for ongoing clinical trials of HDAC6 inhibitors in this molecular context.[ 30 ] These findings are hypothesis-generating and require validation in larger prospective cohorts. Strengths of our study include its large, homogeneously staged cohort; exclusion of key confounding factors; and integration of molecular profiling. Limitations include its retrospective, single-center design and the limited number of patients with positive cytology, which restricts power for subgroup analyses. Future directions include integration of peritoneal cytology with emerging liquid biopsy techniques and panel-based sequencing of peritoneal washings, which may refine recurrence risk stratification beyond the current molecular classification framework.

In this large, retrospective cohort study of surgically staged FIGO 2009 stage IA grade 1 endometrioid endometrial cancer, PPC was an independent predictor of worse PFS, in the absence of other established high-risk features such as lymphovascular space invasion or deep myometrial invasion. While PPC appears to be associated with increased risk of recurrence in low-grade, early-stage disease, its role as an independent biomarker remains hypothesis-generating. Prospective studies are needed to determine whether PPC should inform surveillance intensity or treatment decisions in otherwise low-risk patients. Importantly, molecular characterization of tumors revealed a possible enrichment of ARID1A mutations in patients with PPC, suggesting a potential mechanistic link between ARID1A alterations and peritoneal dissemination.

Peritoneal cytology has long occupied a controversial position in the staging and management of endometrial cancer. Initially incorporated into the International Federation of Gynecology and Obstetrics (FIGO) 1988 staging system, its inclusion was based on early assumptions linking the presence of malignant cells in peritoneal washings to worse oncologic outcomes. Subsequent retrospective analyses, however, revealed that patients with uterine-confined disease and positive cytology often experience far better prognoses than those with true extrauterine spread, prompting FIGO to eliminate cytologic status from the 2009 staging revision.[ 1 , 2 ] Yet, the biological and clinical implications of positive peritoneal cytology (PPC) remain unresolved. Contemporary studies continue to report associations between PPC and worse survival outcomes, particularly in high-grade or non-endometrioid histologies.[ 3 – 5 ] In contrast, the prognostic relevance of PPC in low-grade, early-stage endometrioid endometrial cancer remains ambiguous. Some data suggest minimal impact on survival after adjustment for confounders, raising the possibility that PPC in this context may be clinically inert—or, alternatively, may represent an underrecognized marker of dissemination in a biologically indolent tumor type.[ 6 ] To address this uncertainty, we conducted a retrospective cohort study focusing exclusively on patients with surgically staged FIGO 2009 stage IA grade 1 endometrioid carcinoma, excluding known adverse risk factors. Our primary objective was to evaluate the prognostic impact of PPC in this rigorously selected, low-risk population. We also sought to explore potential contributing clinical and molecular features that may underlie the presence of PPC or influence recurrence patterns.