This study included 351 patients with endometrial cancer who underwent surgical resection between 2010 and 2022 and had available formalin-fixed, paraffin-embedded blocks. Data retrieved from electronic medical records encompassed patient demographics, including age, body mass index (BMI), menopausal status, parity, and survival outcomes. This study was approved by the institutional review board of the Gachon University Gil Medical Center (Incheon, Republic of Korea), which waived patient consent (Approval Number GBIRB 2022-236).

Information regarding the primary diagnosis and tumor size was obtained from surgical pathology reports. Hematoxylin and eosin (H&E)-stained slides underwent a thorough review to reassess various pathologic features, such as diagnosis, International Federation of Gynaecology and Obstetrics (FIGO) grade, FIGO stage, lymphovascular invasion, lymph node metastasis, MMR status, and p53 status.

Immunohistochemical staining was conducted using tissue microarray (TMA) blocks, with a single 2-mm core representing an individual patient. The blocks were subsequently sectioned into 4-μm-thick slices for staining. Tissue sections were deparaffinized at 60°C and rehydrated through sequential immersion in xylene and graded ethanol series. Endogenous peroxidase activity was blocked by incubation in 3% hydrogen peroxide for 10 min, followed by heat-induced antigen retrieval. Tissue sections were incubated overnight at 4°C with primary antibodies, including anti-NLRP3 (1:150, AG-20B-0014-C100, Adipogen, San Diego, CA, United States), anti-cleaved caspase-1 p20 (Asp296) (1:100, PA5-99390, Thermo Fisher, Waltham, MA, United States), anti-cleaved gasdermin D (Asp275) (1:500, 36425, Cell Signaling, Danvers, MA, United States), and anti-CHMP4B (1:20, polyclonal, Invitrogen, Waltham, MA, United States). Following overnight incubation, counterstaining with hematoxylin was conducted for 3 min. Subsequently, the sections were dehydrated through immersion in ethanol and xylene, followed by mounting.

The intensities of NLRP3, cleaved caspase-1 p20, cleaved gasdermin D, and CHMP4B immunostaining slides were assessed based on the following scale: negative (0), weak positive (1+), moderate positive (2+), and strong positive (3+). Subsequently, binary classification into low and high expressions was performed. For NLRP3 and cleaved gasdermin D, negative was categorized as low expression, whereas weak to strong positive were classified as high expression. For cleaved caspase-1 p20 and CHMP4B, negative to moderate positive were designated as low expression, whereas strong positive was considered high expression. Representative images of high and low expression for each marker are illustrated in Figure 1 .

The primary outcomes were recurrence-free survival (RFS) and OS. RFS was defined as the duration from surgical resection to the first recurrence or last follow-up, with patients with synchronous distant metastasis being excluded from RFS analysis. OS was defined as the duration from surgical resection to death from any cause or last follow-up.

Continuous variables were compared using either the independent two-sample t-test or Mann–Whitney U test, whereas categorical variables were assessed using the chi-square test or Fisher exact test. Survival curves were generated using the Kaplan–Meier method and compared using the log-rank test. A univariate Cox regression analysis was conducted to identify clinicopathologic characteristics associated with RFS and OS. Additionally, multivariate Cox regression analysis was performed to identify independent prognostic significance. All statistical tests were performed using SPSS version 21 (IBM Corp., Armonk, NY, United States), with a p < 0.05 indicating statistical significance.

Table 1 summarizes the baseline clinical and pathologic characteristics of the included patients. Notably, the included patients had a mean age of 54.1 ± 9.8 years (range, 23–82 years), with 115 (32.8%) and 236 (67.2%) aged ≤50 and >50 years, and a mean BMI of 26.3 ± 4.9 kg/m², with 154 (43.9%) and 197 (56.1%) having a BMI of ≤25 kg/m² and >25 kg/m², respectively. Moreover, 205 (60.8%) and 132 (39.2%) patients had a postmenopausal and premenopausal status, respectively, whereas 52 (14.8%) and 299 (85.2%) patients had a nulliparous and parous status, respectively. The diagnosis included endometrioid carcinoma in 323 patients (92.0%) and nonendometrioid carcinoma in 28 patients (8.0%), which comprised 9 serous carcinomas, 4 clear cell carcinomas, 12 mixed carcinomas, 2 undifferentiated carcinomas, and 1 carcinosarcoma. Among the included patients, 135 (38.5%), 160 (45.6%), and 56 (16.0%) had FIGO grades 1, 2, and 3, respectively, whereas 283 (80.6%), 19 (5.4%), 39 (11.1%), and 10 (2.8%) were in FIGO stages 1, 2, 3, and 4, respectively. The mean tumor diameter was 3.7 ± 2.4 cm (0.1–13.0 cm). Lymphovascular invasion was present and absent in 120 (34.2%) and 231 (65.8%) patients, respectively, whereas lymph node metastasis was observed and absent in 35 (90.0%) and 316 (10.0%) patients, respectively. Defective MMR, indicated by the absence of nuclear staining in the four MMR proteins (MLH1, MSH2, MSH6, and PMS2), was detected in 74 cases (21.1%), whereas normal MMR was observed in 274 cases (78.1%). P53 status, determined through p53 immunohistochemistry, exhibited a mutant pattern (including a null-mutant pattern) in 52 (14.8%) and a wild-type pattern in 292 (83.2%) patients. The median follow-up duration was 53 months (range, 1–78 months).

Immunohistochemistry revealed high and low NLRP3 expression in 235 (67.0%) and 115 (32.8%) patients, high and low cleaved caspase-1 p20 expression in 11 (3.1%) and 333 (94.9%) patients, high and low cleaved gasdermin D expression in 49 (14.0%) and 300 (85.5%) patients, and high and low CHMP4B expression in 9 (2.6%) and 340 (96.9%) patients, respectively.

The correlation between clinicopathologic characteristics and the expression of the four pyroptosis-related markers is presented in Table 2 . Notably, our findings showed that high NLRP3 expression was associated with age ≤ 50 years (p = 0.022) and premenopausal status (p = 0.012); high cleaved caspase-1 p20 expression was associated with nonendometrioid carcinoma (p = 0.008), FIGO grade 3 disease (p = 0.020), and a p53 mutant pattern (p = 0.003); high cleaved gasdermin D expression was associated with BMI > 25 kg/m² (p = 0.049), FIGO grades 1–2 (p = 0.014), early FIGO stage (I–II) (p = 0.030), and absence of lymph node metastasis (p = 0.042); and high CHMP4B expression was associated with nonendometrioid carcinoma (p = 0.028).

The correlation between the expression of the four pyroptosis-related markers is presented in Table 3 . Accordingly, our results showed that high NLRP3 expression was associated with high cleaved caspase-1 p20 expression (p = 0.019) and that high cleaved caspase-1 p20 expression was associated with high CHMP4B expression (p = 0.030).

To evaluate the consequences of cleaved gasdermin D-mediated pyroptosis without CHMP4B-mediated membrane repair, we analyzed the clinicopathologic characteristics of cleaved gasdermin D-low/CHMP4B-high and cleaved gasdermin D-high/CHMP4B-low endometrial cancer ( Supplementary Table 1 ). Notably, cleaved gasdermin D-high/CHMP4B-low endometrial cancer was associated with endometrioid carcinoma (p = 0.010) and FIGO grade 1–2 (p = 0.023), whereas cleaved gasdermin D-high/CHMP4B-low endometrial cancer was significantly associated with endometrioid carcinoma (p = 0.010) and FIGO grades 1–2 (p = 0.023).

After evaluating the correlation between pyroptosis-related marker expression and survival (RFS and OS), we found that high NLRP3 expression was not associated with RFS (HR, 1.176; 95% CI, 0.514–2.691; log-rank p = 0.070) and OS (HR, 1.091; 95% CI, 0.383–3.108; log-rank p = 0.871) ( Table 4 , Figures 2A, B ). However, high cleaved caspase-1 p20 expression was associated with worse RFS (HR, 11.359; 95% CI, 4.253–30.334; log-rank p < 0.001) and OS (HR, 1.091; 95% CI, 0.383–3.108; log-rank p < 0.001) ( Table 4 , Figures 2C, D ). Moreover, high cleaved gasdermin D expression was not associated with RFS (HR, 0.442; 95% CI, 0.105–1.865; log-rank p = 0.252) and OS (HR, 0.359; 95% CI, 0.048–2.708; log-rank p = 0.299) ( Table 4 , Figures 2E, F ). High CHMP4B expression was associated with worse RFS (HR, 4.220; 95% CI, 0.999–17.827; log-rank p = 0.033) but not with OS (HR, 3.260; 95% CI, 0.432–24.623; log-rank p = 0.225) ( Table 4 , Figures 2G, H ).

To evaluate the impact of cleaved gasdermin D-mediated pyroptosis unhindered by CHMP4B, we compared survival outcomes between cleaved gasdermin D-low/CHMP4B-high and cleaved gasdermin D-high/CHMP4B-low endometrial cancers. Accordingly, we found that cleaved gasdermin D-high/CHMP4B-low cancer was significantly associated with favorable RFS (log-rank p = 0.021) and tended to have a favorable OS (log-rank p = 0.128) ( Figures 2I, J ).

We then performed subgroup analysis among patients with endometrial cancer who had advanced disease stages (FIGO III–IV). Notably, among those with advanced stage, high NLRP3 was associated with adverse RFS (log-rank p = 0.009) but not OS (log-rank p = 0.357) ( Supplementary Figures 1A, B ), high cleaved caspase-1 p20 expression was associated with worse RFS (log-rank p < 0.001) and OS (log-rank p < 0.001) ( Supplementary Figures 1C, D ), high cleaved gasdermin D expression was not associated with RFS (log-rank p = 0.464) and OS (log-rank p = 0.580) ( Supplementary Figures 1E, F ), and high CHMP4B was not associated with RFS (log-rank p = 0.164) and OS (log-rank p = 0.057) ( Supplementary Figures 1G, H ).

Other clinicopathologic factors found to be associated with RFS included age > 50 years (HR, 3.925; 95% CI, 1.182–13.038), postmenopausal status (HR, 3.487; 95% CI, 1.197–10.160), nonendometrioid carcinoma (HR, 7.964; 95% CI, 3.568–17.776), FIGO grade 3 (HR, 5.472; 95% CI, 2.571–11.646), advanced stage (FIGO stages III–IV) (HR, 5.069; 95% CI, 2.318–11.084), lymphovascular invasion (HR, 3.280; 95% CI, 1.522–7.071), lymph node metastasis (HR, 2.837; 95% CI, 1.143–7.042), and p53 mutant pattern (HR, 4.628; 95% CI, 2.136–10.024) ( Table 4 ). Other clinicopathologic factors found to be associated with OS included nonendometrioid carcinoma (HR, 7.250; 95% CI, 2.669–19.693), FIGO grade 3 (HR, 3.678; 95% CI, 1.399–9.669), advanced stage (FIGO stages III–IV) (HR, 8.356; 95% CI, 3.212–21.740), lymph node metastasis (HR, 4.718; 95% CI, 1.728–12.882), and p53 mutant pattern (HR, 3.814; 95% CI, 1.401–10.379) ( Table 4 ).

Given that high cleaved caspase-1 p20 expression was significantly associated with RFS and OS on univariate Cox regression, it was incorporated into the multivariate Cox regression model. Other clinicopathologic characteristics significantly associated with RFS and OS on univariate analysis were adjusted during multivariate analysis. Notably, high cleaved caspase-1 p20 expression (HR, 7.017; 95% CI, 2.054–23.975) and lymphovascular invasion (HR, 3.067; 95% CI, 1.161–8.103) were identified as independent prognostic factors for adverse RFS ( Table 5 ), whereas high cleaved caspase-1 p20 expression (HR, 5.289; 95% CI, 1.270–22.033) and advanced stage (FIGO stages III–IV) (HR, 6.525;95% CI, 1.586–26.838) were identified as independent prognostic factors for adverse OS ( Table 5 ).