This prospective cohort study utilized serum samples that were prospectively collected from 73 patients with histologically confirmed EC and 76 controls with benign gynecological conditions. The study was conducted at the Department of Obstetrics and Gynecology, Division of Gynecologic Oncology, Istanbul Faculty of Medicine, Istanbul University, in collaboration with the Department of Biochemistry. Patient enrollment and serum collection took place between 2018 and 2019. The initial approval from the Institutional Review Board and Ethics Committee was obtained in 2018 (Approval No: 2018/1609). As the scientific scope of the research expanded to include additional ncRNAs, further approval was granted in 2023 (Approval No: 2023/119), supported by the Istanbul University Scientific Research Projects Unit (Project ID: TTU-2023-39847). This study was initiated prior to the requirement for registration of prospective cohort studies. All procedures were conducted in accordance with the Declaration of Helsinki, and written informed consent obtained at baseline included permission for long-term storage and future use of biospecimens.

The study population included women with histologically confirmed EC, all diagnosed by endometrial biopsy and subsequently managed surgically. The control group consisted of women who presented to the general gynecology outpatient clinic for routine gynecological evaluation due to non-malignant indications and were confirmed to have no evidence of uterine or adnexal pathology on clinical, ultrasonographic, or laboratory assessment. To minimize potential confounding factors, women with uterine (cervical, myometrial, or endometrial) and/or ovarian/tubal pathologies—including benign gynecologic conditions such as uterine fibroids, ovarian cysts, endometrial polyps, and endometrial hyperplasia—were excluded. In particular, patients diagnosed with atypical and/or complex endometrial hyperplasia were explicitly excluded to avoid potential overlap with premalignant endometrial conditions. Exclusion criteria for the EC group included final histopathology not confirming EC, the presence of synchronous malignancies, recurrent disease, or prior oncology treatment (chemotherapy, radiotherapy, or immunotherapy).

All patients with EC were routinely followed up every three months after their initial surgery according to institutional protocol. Each follow-up visit included gynecological examination, transvaginal ultrasonography, and magnetic resonance imaging (MRI) to evaluate possible disease recurrence. Recurrence was defined as radiologically or histopathologically confirmed evidence of disease reappearance after initial treatment. In cases where MRI findings were unequivocally consistent with recurrent disease, diagnosis and treatment planning were based on radiological confirmation. For patients with indeterminate or atypical radiological or clinical findings, histopathological verification was obtained before initiating treatment. For recurrence-related analyses, only patients with complete recurrence data were included. Two patients had incomplete follow-up; their recurrence status was recorded as missing. We did not perform any imputation for missing data. A complete-case (available-case) analysis was therefore used for recurrence outcomes.

Both EC patients and controls were recruited consecutively during the predefined study period. No prior sample size calculation was performed; however, based on the total sample size, a post hoc power analysis was conducted using the G*Power statistical software (version 3.1.9.7; Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany). For a two-tailed t-test with an alpha level of 0.05 and a moderate effect size (d = 0.5), the calculated statistical power of the study was 88%, indicating adequate power to detect clinically meaningful differences between groups. A p-value 0.05 was considered statistically significant. This approach allows a post hoc assessment of whether the sample size was sufficient for the observed effect size.

Venous blood samples were obtained from patients with EC at the time of diagnosis and from controls during outpatient visits. All samples were transferred to the Department of Biochemistry within 15 minutes of collection, centrifuged, aliquoted, and immediately stored at –80 °C in RNase-free tubes to avoid repeated freeze–thaw cycles and minimize RNA degradation. The storage duration before RNA extraction ranged from 38 to 50 months. All EC and control samples were processed and analyzed simultaneously to minimize batch effects. Previous studies have demonstrated that circulating miRNAs remain stable for years when stored at ultra-low temperatures [22, 23, 24], and current biospecimen practice guidelines indicate that serum or plasma aliquots stored at $\le$–80 °C are suitable for cf-miRNA analysis over extended periods [25]. Because all specimens were collected, processed, and stored under identical conditions, any potential effect of storage duration on RNA integrity was expected to be consistent across groups. Total RNA was extracted from serum using the miRNeasy Serum/Plasma Advanced Kit (Cat. no. 217204; Qiagen, Germantown, MD, USA) according to the manufacturer’s instructions. Complementary DNA (cDNA) synthesis was performed using the miRCURY LNA™ RT Kit (Cat. no. 339346; Qiagen, Germantown, MD, USA) and the RT² PreAMP cDNA Synthesis Kit (Cat. no. 330451; Qiagen, Germantown, MD, USA).

Expression levels of miR-182-5p and lncRNA HOTAIRM1 were quantified using a Rotor-Gene Q real-time PCR system (Qiagen, Germantown, MD, USA). Primers were purchased from Qiagen Technologies (Germantown, MD, USA): miRCURY LNA™ miRNA PCR Assay U6 (Cat. no. YP02119464), miRCURY LNA™ miRNA PCR Assay hsa-miR-182-5p (Cat. no. YP00206070), RT² lncRNA qPCR Assay (HOTAIRM1) (Cat. no. 330701LPH10483A), and RT² lncRNA qPCR Assay (GAPDH) (Cat. no. 330001PPH00150F). Relative expression levels were calculated using the 2^{-Δ⁢Δ⁢CT} method, normalizing to U6 (for miRNA) and GAPDH (for lncRNA). All qRT-PCR reactions were performed in triplicate (technical replicates) using the same instrument and reagent lots to ensure reproducibility. Negative (no-template) controls were included in every run. The average coefficient of variation (CV) for cycle threshold (Ct) values among replicates was below 2%, indicating high intra-assay precision. Moreover, all patient and control samples were analyzed within a single batch under identical reaction conditions to eliminate inter-assay variability.

Data analysis was performed using NCSS (Number Cruncher Statistical System) 2007 software (Kaysville, UT, USA). Descriptive statistics were expressed as mean $\pm$ SD, median (interquartile range [IQR]), frequency, percentage, minimum, and maximum values. The normality of continuous variables was assessed using the Shapiro-Wilk test. Data with a normal distribution were presented as mean $\pm$ SD, while non-normally distributed data were expressed as median (IQR). For comparisons between two groups, the Student’s t-test or Mann-Whitney U test was used, depending on data distribution. The Shapiro-Wilk test revealed that serum lncRNA HOTAIRM1 and miR-182-5p expression levels were not normally distributed (p 0.05); therefore, comparisons between groups were performed using the Mann-Whitney U test. For comparisons involving more than two groups, the one-way ANOVA or Kruskal-Wallis test was applied, followed by appropriate post-hoc analyses when significant differences were found. Associations between categorical variables were examined using the chi-square test or Fisher’s exact test, as appropriate. Correlations between quantitative variables were analyzed using Spearman’s rank correlation coefficient. Receiver operating characteristic (ROC) curve analysis was performed to evaluate diagnostic performance, including the determination of optimal sensitivity, specificity, and cut-off values. The optimal cut-off for serum lncRNA HOTAIRM1 was derived using the Youden index (J = sensitivity + specificity – 1), which identifies the maximum point of sensitivity and specificity on the ROC curve. Survival analyses were conducted using the Kaplan–Meier method, and all statistical tests were two-sided with a p-value 0.05 considered statistically significant.

A total of 73 patients with EC and 76 control subjects were included in the study. Age (mean $\pm$ SD: 57.84 $\pm$ 11.19 vs. 55.44 $\pm$ 12.09 years, p = 0.171) and body mass index (BMI) (33.20 $\pm$ 6.64 vs. 33.75 $\pm$ 6.38 kg/m², p = 0.563) did not differ significantly between groups (Table 1). Similarly, no significant differences were observed regarding obesity (p = 0.841), smoking status (p = 0.210), hypertension (p = 0.124), or diabetes mellitus (p = 0.099), indicating that the groups were demographically comparable (Table 1).

Among patients with EC (n = 73), preoperative serum CA-125 levels ranged from 5 to 745 U/mL (mean $\pm$ SD: 42.44 $\pm$ 131.34), with elevated values detected in 4 patients (5.5%). The majority were postmenopausal (n = 58, 79.5%) and parous (n = 55, 75.3%). According to the FIGO classification, 58 patients (79.5%) had early-stage disease (stages IA–IB), whereas 15 (20.5%) presented with advanced-stage disease (stages II–IV). Histologically, 58 patients (79.5%) had low-grade endometrioid adenocarcinoma (grades 1–2), and 15 (20.5%) had high-grade tumors (grade 3 endometrioid or non-endometrioid), including 9 grade 3 endometrioid adenocarcinomas, 2 serous carcinomas, 1 mixed serous and clear cell carcinoma, and 3 carcinosarcomas.

Myometrial invasion was 50% in 49 cases (67.1%) and $\ge$50% in 24 (32.9%). Lymphovascular space invasion (LVSI) was present in 21 patients (28.8%), and microcystic, elongated and fragmented (MELF)-type invasion was observed in 13 (17.8%). Cervical and isthmic involvement occurred in 7 (9.6%) and 13 (17.8%) cases, respectively. LN metastasis was detected in 7 (9.6%), and distant metastasis in 2 (2.7%) patients.

Regarding surgical management, 9 patients (12.3%) underwent no LN dissection (LND), 14 (19.2%) had pelvic–paraaortic LND, 50 (68.5%) had pelvic or sentinel LND, and 31 (42.5%) underwent omental sampling or omentectomy. Among surgically staged patients, LN counts ranged from 0 to 51 (mean $\pm$ SD: 13.74 $\pm$ 11.14). Postoperatively, 40 patients (54.8%) received no adjuvant therapy, 12 (16.4%) underwent brachytherapy, 7 (9.6%) received pelvic radiotherapy (RT), and 14 (19.2%) received combined chemotherapy (CT) and pelvic RT.

Recurrence data were available for 71 of the 73 patients; two patients were excluded from recurrence analysis due to incomplete follow-up. Among the evaluable cohort, recurrence was identified in 9 patients (12.7%) based on radiological or histopathological confirmation—managed with CT in 5 cases, surgery with hyperthermic intraperitoneal chemotherapy (HIPEC) plus CT in 1, and combined CT and RT in 3.

Serum lncRNA HOTAIRM1 expression levels were significantly higher in EC patients compared to controls (median [IQR]: 1.36 [0.74–3.37] vs. 1.03 [0.58–2.03]; p = 0.043) (Fig. 1 and Table 2). In contrast, serum miR-182-5p expression levels did not differ significantly between groups (1.26 [0.09–4.39] vs. 1.71 [0.21–3.58]; p = 0.327) (Fig. 2 and Table 2).

Fig. 1.

Comparison of serum lncRNA HOTAIRM1 expression between EC patients and controls. * indicates statistical significance, defined as p 0.05. lncRNA, long non-coding RNA; HOTAIRM1, HOXA transcript antisense RNA myeloid-specific 1.

Fig. 2.

Comparison of serum miR-182-5p expression between EC patients and controls. ns denotes not statistically significant.

Spearman’s correlation analysis demonstrated no significant relationship between serum lncRNA HOTAIRM1 and miR-182-5p expression in the EC groups (p $>$ 0.05) (Fig. 3).

Fig. 3.

Correlation between serum lncRNA HOTAIRM1 and miR-182-5p expression in patients with EC.

ROC curve analysis demonstrated that serum lncRNA HOTAIRM1 had an area under the curve (AUC) of 0.606 (95% CI: 0.507–0.706; p = 0.037), indicating a statistically significant yet modest discriminative ability to distinguish EC patients from controls (Fig. 4 and Table 3). The optimal diagnostic cut-off value was 3.06, corresponding to 29.6% sensitivity and 93.2% specificity (Table 3). These findings indicate that, although serum lncRNA HOTAIRM1 has high specificity—reflecting a strong capacity to correctly identify non-EC cases—its low sensitivity limits its ability to detect all true positives. Therefore, serum lncRNA HOTAIRM1 alone may not be sufficient as a screening biomarker; however, its high specificity supports its potential utility as a confirmatory diagnostic or prognostic adjunct when used in combination with other molecular or clinicopathological parameters.

Fig. 4.

ROC curve of serum lncRNA HOTAIRM1 for the diagnosis of EC. ROC, receiver operating characteristic.

Overall survival (OS) was defined as the interval from diagnosis to death from any cause, and disease-free survival (DFS) was defined as the interval from surgery to disease progression or recurrence. In the study cohort, the follow-up duration for DFS ranged from 0 to 75 months (mean $\pm$ SD: 47.58 $\pm$ 17.69), and the follow-up duration for OS ranged from 1 to 75 months (mean $\pm$ SD: 48.49 $\pm$ 15.42). Disease-related mortality was documented in 8 patients (10.9%). In survival analysis, neither serum lncRNA HOTAIRM1 nor miR-182-5p expression showed a significant association with DFS or OS (p $>$ 0.05). For prognostic subgroup analyses, patients were stratified into “high” and “low” expression groups according to the optimal cut-off value (3.06) derived from the ROC curve using the Youden index (J = sensitivity + specificity – 1), as described previously. Kaplan-Meier survival analysis comparing patients with high vs. low serum lncRNA HOTAIRM1 expression revealed no statistically significant difference in OS (Log-rank p $>$ 0.05) (Fig. 5), and a similar non-significant trend was observed for DFS during the study period. These findings suggest that while lncRNA HOTAIRM1 and miR-182-5p may have limited prognostic value when considered individually, traditional survival endpoints such as DFS and OS continue to serve as the fundamental outcome measures in EC research. Given the relatively small sample size, multivariate analyses adjusting for clinicopathological parameters (such as age, FIGO stage, or adjuvant therapy) were not performed to prevent model overfitting. Future large-scale studies incorporating these factors are needed to confirm the independent prognostic value of circulating lncRNA HOTAIRM1 and miR-182-5p. This underscores the importance of integrating molecular biomarkers with established clinicopathological parameters to achieve a more comprehensive and reliable prognostic assessment.

Fig. 5.

Kaplan-Meier survival analysis of serum lncRNA HOTAIRM1 expression in EC.

Subgroup analyses were performed to examine potential associations between serum lncRNA HOTAIRM1 and miR-182-5p expression levels and key clinicopathological parameters, including FIGO stage, tumor grade, recurrence status, disease-related mortality, LN involvement, deep myometrial invasion, LVSI, MELF pattern, cervical stromal or isthmic involvement, and distant metastasis.

When stratified by FIGO stage, no significant differences in serum lncRNA HOTAIRM1 or miR-182-5p expression were observed between early-stage (IA–IB; n = 58) and advanced-stage (II–IV; n = 15) groups (p = 0.968 and p = 0.180, respectively). Similarly, when categorized by tumor grade, expression levels did not differ significantly between low-grade (grades 1–2 endometrioid; n = 58) and high-grade (grade 3 endometrioid or non-endometrioid; n = 15) tumors (lncRNA HOTAIRM1: median [IQR] 1.32 [0.74–3.44] vs. 2.01 [0.71–3.70]; miR-182-5p: 0.76 [0.09–4.36] vs. 1.64 [0.13–5.87]; p = 0.606 and p = 0.728, respectively).

Although median miR-182-5p levels were higher in patients with advanced-stage disease (2.99 [0.17–19.12] vs. 0.76 [0.09–3.82]), LN involvement (4.46 [0.03–23.40] vs. 1.01 [0.10–3.99]), and recurrence (1.55 [0.39–4.46] vs. 0.75 [0.09–4.34]), these differences were not statistically significant (p = 0.180, p = 0.616, and p = 0.595, respectively). Similarly, lncRNA HOTAIRM1 expression showed a non-significant increase in patients with recurrence (3.44 [0.85–7.04] vs. 1.28 [0.73–2.83]; p = 0.087).

Additional subgroup analyses, including disease-related mortality, deep myometrial invasion, LVSI, MELF pattern, cervical stromal or isthmic involvement, and distant metastasis, also revealed no statistically significant differences in serum lncRNA HOTAIRM1 or miR-182-5p expression levels.

Overall, serum lncRNA HOTAIRM1 and miR-182-5p expression levels showed variability across clinical and pathological subgroups; however, none of the observed differences reached statistical significance. Given the limited sample sizes within certain subgroups and the absence of confidence interval calculation due to small subgroup counts, these findings should be interpreted with caution. Taken together, the results suggest that although these biomarkers may hold potential clinical relevance, their current utility in predicting disease progression, recurrence, or prognosis in EC remains limited. Detailed subgroup data for serum lncRNA HOTAIRM1 and miR-182-5p expression are presented in Tables 4A and 4B.

This study has several limitations that should be acknowledged. First, the sample size was relatively small and derived from a single institution, which may restrict the generalizability of the results. Second, despite standardized sample handling, potential pre-analytical variations—such as storage duration, RNA stability, or variability in internal reference gene expression—might have influenced measured expression levels. Although U6 and GAPDH were selected as internal controls based on prior validation in serum-based ncRNA studies and demonstrated consistent amplification efficiency in our dataset, cohort-specific stability testing was not performed. Third, external validation in an independent cohort and functional in vitro assays were beyond the scope of this investigation. Lastly, although the control group consisted of women with non-malignant gynecological conditions and no uterine or adnexal pathology, they may not fully represent healthy individuals, introducing minimal background variability.