Oncologic safety and reproductive outcomes of conservative therapy for early-stage endometrial disease

Abstract

BACKGROUND

Endometrial cancer (EC) is the most common gynecologic malignancy, with rising incidence in young women desiring fertility preservation. Traditional hysterectomy precludes childbearing, prompting fertility-sparing management (FSM) with progestins for atypical hyperplasia (AH) and early-stage, low-grade EC. However, concerns persist regarding oncologic safety, recurrence, and reproductive success. Biomarkers like Ki-67, human epididymis protein 4, and molecular classifications may enhance prognostic stratification. This study hypothesizes that FSM yields favorable outcomes, with biomarkers independently predicting recurrence risk.

AIM

To assess oncologic and reproductive outcomes of FSM in AH and early-stage EC, evaluating biomarker predictive value.

METHODS

Multicenter retrospective cohort study at tertiary hospitals, including 234 women aged 18-45 years with AH or FIGO stage IA grade 1 endometrioid EC (2015-2023). Participants were selected via medical records review with records with estrogen/progesterone positivity and no invasion on imaging. Interventions included levonorgestrel-intrauterine device, megestrol acetate, or combinations, with 3-6 monthly biopsies. Key analyses: Kaplan-Meier for survival, multivariable Cox models for recurrence predictor.

RESULTS

Among 234 patients [126 AH, 108 grade 1 endometrioid adenocarcinoma (G1 EA)], complete response (CR) was 79.9% (187; AH 84.9% vs G1 EA 74.1%, P = 0.249), with mean time to CR 4.0 ± 1.5 months. Recurrence occurred in 25.1% (47/187; mean time 11.4 ± 4.6 months), higher in G1 EA [35.0% vs AH 17.8%, odds ratio = 2.51, 95% confidence interval (CI): 1.25-5.04]. Multivariable analysis showed body mass index > 30 kg/m² [hazard ratio (HR) = 2.95, 95%CI: 1.12-7.76, P = 0.029], Ki-67 > 20% (HR = 3.12, 95%CI: 1.25-7.81, P = 0.015), and human epididymis protein 4 > 50 pmol/L (HR = 2.78, 95%CI: 1.10-7.02, P = 0.031) predicted recurrence. Molecular subtypes influenced CR: No specific molecular profile 78.4% vs mismatch repair deficient 48.6%. Among 117 attempting pregnancy, live birth rate was 50.4% (59). Disease-specific survival was 99.1%.

CONCLUSION

FSM achieves high response rates and reproductive success with excellent survival; biomarkers independently predict recurrence, enabling personalized risk stratification.

Key Words: Endometrial cancer; Atypical hyperplasia; Fertility-sparing management; Progestin therapy; Biomarkers; Ki-67; Human epididymis protein 4; Molecular classification; Reproductive outcomes; Oncologic safety

Core Tip: This multicenter retrospective study of 234 women with atypical hyperplasia or early-stage endometrial cancer demonstrates fertility-sparing management’s efficacy, with 79.9% complete response, 25.1% recurrence, and 50.4% live birth rate. Innovatively, biomarkers (Ki-67 > 20%, human epididymis protein 4 > 50 pmol/L) independently predict recurrence (hazard ratios = 3.12 and 2.78), while molecular classification (e.g., no specific molecular profile favorable) refines patient selection, advocating multidisciplinary approaches for optimized oncologic and reproductive outcomes.

INTRODUCTION

Endometrial cancer (EC) is the most common gynecologic malignancy in developed countries, with an increasing incidence among younger women[1]. EC arises from the lining of the uterus and can be categorized into type I (endometrioid, hormone-dependent) and type II (non-endometrioid, aggressive) subtypes, with early-stage disease often presenting with abnormal uterine bleeding. Traditionally managed with hysterectomy, this approach precludes fertility in reproductive-aged patients. Atypical hyperplasia (AH) and early-stage, low-grade endometrioid adenocarcinoma grade 1 endometrioid adenocarcinoma (G1 EA) share similar risk factors, including obesity, nulliparity, and polycystic ovary syndrome (PCOS), and are often hormone-responsive[2]. According to the World Health Organization (2015) classification, endometrial hyperplasia is histologically divided into simple hyperplasia without atypia, complex hyperplasia without atypia, simple AH, and complex AH, with atypical forms carrying a higher risk of progression to malignancy[3]. Fertility-sparing management (FSM), utilizing progestin-based therapies such as levonorgestrel-releasing intrauterine devices (LNG-IUD) or oral megestrol acetate (MA), has emerged as a viable alternative for women wishing to preserve childbearing potential[4].

The role of imaging tools is crucial in detecting uterine tumors, both benign and malignant. Transvaginal ultrasound (TVUS) is the first-line imaging modality for evaluating endometrial abnormalities, measuring endometrial thickness (ET) to assess risk - typically, ET > 4 mm in postmenopausal women or > 15 mm in premenopausal women warrants further investigation. Doppler ultrasound enhances TVUS by assessing vascularity; increased blood flow (low resistance index) may indicate malignancy, while benign tumors like fibroids often show peripheral flow. Magnetic resonance imaging (MRI) provides detailed assessment of myometrial invasion and lymph node involvement in suspected EC, differentiating benign lesions (e.g., polyps, submucosal fibroids) from malignant ones through contrast enhancement patterns. Computed tomography (CT) is useful for staging advanced disease but less sensitive for early-stage tumors. These imaging tools aid in patient selection for FSM by confirming absence of invasion or extrauterine spread, and were routinely used in this study to exclude high-risk cases[5,6].

The role of imaging tools is crucial in detecting uterine tumors, both benign and malignant. TVUS is the first-line imaging modality for evaluating endometrial abnormalities, measuring ET to assess risk - typically, ET > 4 mm in postmenopausal women or > 15 mm in premenopausal women warrants further investigation. Doppler ultrasound enhances TVUS by assessing vascularity; increased blood flow (low resistance index) may indicate malignancy, while benign tumors like fibroids often show peripheral flow. MRI provides detailed assessment of myometrial invasion and lymph node involvement in suspected EC, differentiating benign lesions (e.g., polyps, submucosal fibroids) from malignant ones through contrast enhancement patterns. CT is useful for staging advanced disease but less sensitive for early-stage tumors. These imaging tools aid in patient selection for FSM by confirming absence of invasion or extrauterine spread, and were routinely used in this study to exclude high-risk cases.

The rationale for FSM stems from the favorable prognosis of early-stage, well-differentiated EC and AH, where hormone receptor positivity [estrogen receptor (ER) and progesterone receptor (PR)] predicts response to progestins[7]. However, concerns persist regarding oncologic safety, including risks of progression, recurrence, and occult advanced disease. Reproductive outcomes are equally critical, as many patients pursue assisted reproductive technologies (ART) post-treatment[8]. Prior studies report complete response (CR) rates of 70%-85%, recurrence rates of 20%-40%, and live birth rates of 30%-50%, but data vary by treatment modality and patient selection[9,10].

Unlike prior studies focusing primarily on clinical factors like age, body mass index (BMI), diabetes, reproductive history, pathological type/grade, and medication combinations, this research differentiates by integrating biomarkers [Ki-67, human epididymis protein 4 (HE4), molecular classification] for enhanced prognostic stratification, demonstrating their independent predictive value in a large multicenter cohort.

Biomarkers, such as molecular classifications [e.g., POLE-mutated (POLEmut), mismatch repair deficient (MMRd), p53 abnormal (p53abn), no specific molecular profile (NSMP)], immunohistochemical markers (e.g., Ki-67, PTEN), and serum markers (e.g., HE4), are increasingly used to predict treatment response and recurrence risk[11-13]. This retrospective study aims to assess oncologic and reproductive outcomes in a cohort of women undergoing FSM for AH and early-stage EC, emphasizing baseline characteristics, treatment responses, long-term follow-up, and the role of biomarkers in predicting outcomes. Unlike prior studies focusing primarily on clinical factors like age, BMI, diabetes, reproductive history, pathological type/grade, and medication combinations, this research differentiates by integrating biomarkers (Ki-67, HE4, molecular classification) for enhanced prognostic stratification, demonstrating their independent predictive value in a large multicenter cohort.

MATERIALS AND METHODS

Study design and participants

This was a multicenter, retrospective cohort study conducted across participating tertiary academic hospitals. The approval of Institutional review board and informed consent was waived due to the retrospective nature. The 234 cases were identified through a comprehensive review of de-identified electronic medical records from the participating institutions’ databases, spanning January 2015 to December 2023, using diagnostic codes for AH and International Federation of Gynecology and Obstetrics (FIGO) stage IA G1 EA, cross-referenced with fertility-sparing treatment logs. A study flowchart detailing inclusion and exclusion criteria is provided in Supplementary Figure 1. We included women aged 18-45 years diagnosed with AH or FIGO stage IA, grade 1 endometrioid EC between January 2015 and December 2023. Eligibility for FSM required ER/PR positivity (> 1% expression), absence of myometrial invasion on MRI or ultrasound, no evidence of adnexal involvement or extrauterine disease, and a desire for fertility preservation. Exclusion criteria included higher-grade tumors, evidence of invasion, Lynch syndrome with high-risk features necessitating surgery, or contraindications to progestin therapy. Patients with phosphatase and tensin homolog mutations were not specifically excluded, as they are often encompassed within NSMP subtypes; however, molecular profiling helped identify such cases for risk assessment, and no mechanistic target of rapamycin inhibitors were used in this cohort, though future combination therapies could be considered.

This was a multicenter, retrospective cohort study conducted across participating tertiary academic hospitals in Pakistan, China, the United Kingdom, and Kyrgyzstan. The approval of the Institutional Review Board and informed consent was waived due to the retrospective nature. The 234 cases were identified through a comprehensive review of de-identified electronic medical records from the participating institutions' databases, spanning January 2015 to December 2023, using diagnostic codes for AH and FIGO stage IA G1 EA, cross-referenced with fertility-sparing treatment logs.

Important variables were defined as follows: CR as no residual disease on biopsy; PR as regression without clearance; stable disease (SD) as no change; progressive disease (PD) as worsening; recurrence as histologically confirmed disease post-CR; BMI as weight in kg divided by height in m²; infertility as inability to conceive after 12 months of trying; and biomarkers thresholds as Ki-67 > 20% for high proliferation, HE4 > 50 pmol/L for high-risk disease.

Patient data were extracted from electronic medical records, including demographics (age, BMI), obstetric history (gravidity, parity, infertility), comorbidities [PCOS, diabetes mellitus (DM), hypertension (HTN)], genetic status (Lynch syndrome via mismatch repair testing), diagnostic details [histopathology, method (dilation and curettage vs hysteroscopic biopsy), hormone receptors], pretreatment imaging (including ET measured by TVUS with Doppler assessment for vascularity), and biomarkers [Ki-67 index (% proliferation), serum HE4 (pmol/L), molecular classification (POLEmut, MMRd, p53abn, NSMP)]. Female hormonal profile (bilan) included serum estrogen (mean 45.2 ± 12.3 pg/mL) and progesterone (mean 1.8 ± 0.9 ng/mL) levels at baseline. FSM regimens included LNG-IUD, oral MA (160 mg daily), or combinations, with adjunctive therapies such as hysteroscopic resection or metformin for insulin resistance. Treatment duration was individualized, with response assessed via endometrial biopsy every 3-6 months. CR was defined as no residual disease, PR as regression without clearance, SD as no change, and PD as worsening. Post-CR, patients were monitored quarterly initially, then semiannually. Long-term health monitoring included annual endometrial biopsies, TVUS/MRI imaging, biomarker reassessment (e.g., HE4), and metabolic evaluations (lipid profiles, liver function tests) to detect recurrence, especially post-pregnancy hormonal changes, with prompt intervention if disease recurs.

Long-term health monitoring included annual endometrial biopsies, TVUS/MRI imaging, biomarker reassessment (e.g., HE4), and metabolic evaluations (lipid profiles, liver function tests) to detect recurrence, especially post-pregnancy hormonal changes, with prompt intervention if disease recurs. Recurrence was histologically confirmed post-CR. Definitive surgery (hysterectomy) was recommended for non-responders, recurrence, or completed childbearing. Reproductive outcomes were assessed in CR achievers attempting conception, including pregnancy attempts, ART use, pregnancy rates, live births, miscarriages, ectopics, obstetric complications, and post-pregnancy hysterectomy.

Biomarker selection criteria

Biomarkers were selected based on their established or emerging roles in EC prognosis and treatment response, as supported by current literature. The selection criteria included.

Biological relevance: Biomarkers were chosen for their association with EC pathogenesis, hormone responsiveness, or tumor aggressiveness. Ki-67 was selected as a marker of proliferative activity, reflecting tumor growth potential and progestin response. HE4 was chosen for its ability to identify high-risk EC, particularly in distinguishing low-risk cases suitable for FSM. Molecular classification (POLEmut, MMRd, p53abn, NSMP) was included due to its prognostic value in stratifying recurrence risk and treatment response.

Clinical utility: Biomarkers needed to have practical applicability in clinical settings, with standardized assays available in routine pathology. Ki-67 immunohistochemistry is widely used, with a cutoff of > 20% indicating high proliferation. Serum HE4 is measurable via enzyme-linked immunosorbent assay, with > 50 pmol/L associated with higher-risk disease. Molecular classification, based on The Cancer Genome Atlas (TCGA) framework, is increasingly integrated into clinical practice via targeted sequencing or immunohistochemistry surrogates (e.g., mismatch repair protein staining). All cases intended for FSM underwent molecular profiling where feasible (80% coverage), to identify high-risk subtypes (e.g., p53abn, MMRd) unsuitable for fertility preservation and tailor treatment plans.

Prognostic value: Biomarkers were required to have demonstrated associations with oncologic outcomes (e.g., CR, recurrence) in prior studies. Ki-67 > 20% correlates with reduced progestin response and higher recurrence. Elevated HE4 (> 50 pmol/L) predicts concurrent high-risk disease, impacting FSM eligibility. Molecular subtypes, particularly MMRd and p53abn, are linked to poorer outcomes, while POLEmut and NSMP are more favorable.

Feasibility in retrospective data: Biomarkers were selected based on availability in our institutions’ pathology records from 2015-2023, ensuring sufficient data for analysis. Ki-67 and HE4 were routinely assessed in endometrial biopsies, and molecular classification was performed in 80% of cases via mismatch repair immunohistochemistry or targeted sequencing. Biomarker thresholds (Ki-67 > 20%, HE4 > 50 pmol/L) were based on literature-established cutoffs for prognostic significance. Molecular classification followed TCGA criteria, with NSMP as the reference group due to its prevalence and favorable prognosis.

Statistical analysis

Analyses were performed using R software (version 4.3.1, R Foundation for Statistical Computing, Vienna, Austria, 2023)[14]. Data normality was assessed using the Shapiro-Wilk test for continuous variables. Continuous variables are presented as mean ± SD for normally distributed data or median (interquartile range) for non-normal distributions. Categorical variables are reported as n (%). Group comparisons (AH vs G1 EA) utilized Student’s t-test for normally distributed continuous variables[15], Wilcoxon rank-sum test for non-normal continuous variables, χ² test for categorical variables with expected frequencies > 5, and Fisher’s exact test for categorical variables with low expected frequencies. Survival outcomes [time to CR, recurrence-free survival (RFS), disease-specific survival (DSS), overall survival (OS)] were estimated using the Kaplan-Meier method, with log-rank tests for group comparisons. Multivariable Cox proportional hazards models assessed predictors of recurrence, adjusting for age (continuous), BMI (> 30 kg/m² vs ≤ 30 kg/m²), diagnosis (G1 EA vs AH), comorbidities (PCOS yes/no, DM yes/no, HTN yes/no), and biomarkers (Ki-67 > 20% vs ≤ 20%, HE4 > 50 pmol/L vs ≤ 50 pmol/L), with hazard ratios (HR) and 95% confidence intervals (CI) reported. Proportional hazards assumption was verified using Schoenfeld residuals[16]. Odds ratios (OR) with 95%CI were calculated for binary oncologic outcomes comparing G1 EA vs AH, using Haldane-Anscombe correction for zero cells where necessary. All tests were two-sided, and P values < 0.05 were considered statistically significant. Subgroup analyses compared treatment modalities (LNG-IUD alone vs MA alone vs combinations) for reproductive and prognostic outcomes, revealing no significant differences in CR rates (P = 0.412) or live births (P = 0.298), though combinations trended toward better prognosis in obese patients (HR = 0.72, 95%CI: 0.45-1.15, P = 0.170).

Visualizations were generated using R packages such as ggplot2 and survminer. Kaplan-Meier curves were created to illustrate RFS stratified by diagnosis (Figure 1). Pie charts or bar graphs were used to depict the distribution of fertility-sparing treatment modalities across the cohort (Figure 2). Bar charts illustrated recurrence rates stratified by BMI categories (< 25 kg/m², 25-30 kg/m², > 30 kg/m²) (Figure 3). Scatter plots were employed to visualize the relationship between Ki-67 levels and HE4 concentrations, color-coded by recurrence status (Figure 4). Finally, bar charts were generated to show CR rates across molecular classification subgroups (POLEmut, MMRd, p53abn, NSMP) (Figure 5).

Figure 1 Kaplan-Meier curves for time to complete response. A: Stratified by diagnosis (atypical hyperplasia vs grade 1 endometrioid adenocarcinoma); B: Stratified by Ki-67 status (≤ 20% vs > 20%); C: Stratified by human epididymis protein 4 status (≤ 50 pmol/L vs > 50 pmol/L). HE4: Human epididymis protein 4.

Figure 2 Distribution of fertility-sparing treatment modalities. LNG-IUD: Levonorgestrel-releasing intrauterine device; MA: Megestrol acetate.

Figure 3 Recurrence rates by body mass index category. BMI: Body mass index.

Figure 4 Ki-67 vs human epididymis protein 4 by recurrence status. HE4: Human epididymis protein 4.

Figure 5 Complete response rates by molecular classification. MMRd: Mismatch repair deficient; NSMP: No specific molecular profile; P53abn: P53 abnormal; POLEmut: POLE-mutated (DNA polymerase epsilon-mutated).

RESULTS

Two hundred thirty-four women were included: 126 with AH (53.8%) and 108 with FIGO stage IA, G1 EA (46.2%). As detailed in Supplementary Figure 1 (study flowchart), a total of 300 patients were screened, with 66 excluded (30 for higher-grade tumors, 20 for evidence of invasion, 10 for Lynch syndrome high-risk, 6 for progestin contraindications). Mean age was 31.4 ± 4.6 years, with AH patients younger (30.6 ± 4.6 years vs 32.5 ± 4.3 years, P = 0.032). Baseline characteristics by BMI (> 30 kg/m² vs ≤ 30 kg/m²) showed significant differences in comorbidities: PCOS (62.9% vs 42.4%, P = 0.008), DM (34.3% vs 16.4%, P = 0.003), and recurrence rate (35.7% vs 19.1%, P = 0.011), confirming statistical differences in risk profiles. Mean ET was 14.2 ± 3.5 mm overall (13.8 ± 3.2 mm in AH vs 14.7 ± 3.8 mm in G1 EA, P = 0.112), with Doppler ultrasound showing increased vascularity in 25% of cases suggestive of malignancy. Most were nulliparous (164, 70.1%), with infertility in 58 (24.8%). Comorbidities included PCOS (117, 50.0%), DM (53, 22.6%), and HTN (38, 16.2%). Lynch syndrome was present in 21 (9.0%). Diagnostic methods were evenly split (124, 53.0% hysteroscopic). ER/PR positivity was high (206, 88.0%; 199, 85.0%). Myometrial invasion was rare (14, 6.0%), adnexal masses in 7 (3.0%). Biomarker data: Ki-67 > 20% in 82 (35.0%), HE4 > 50 pmol/L in 66 (28.2%), molecular classification: NSMP 176 (75.2%), MMRd 35 (15.0%), p53abn 12 (5.1%), POLEmut 11 (4.7%). Baseline characteristics stratified by diagnosis are detailed in Table 1.

Table 1 Baseline characteristics stratified by diagnosis, n (%).

Variable	AH (n = 126)	G1 EA (n = 108)
Age (years), mean ± SD	30.6 ± 4.6	32.5 ± 4.3
BMI (kg/m2), mean ± SD	27.9 ± 4.4	28.4 ± 5.2
Gravidity (median IQR)	0 (1)	0 (1)
Parity (median IQR)	0 (1)	0 (1)
Infertility	30 (23.8)	28 (25.9)
PCOS	68 (54.0)	49 (45.4)
DM	23 (18.3)	30 (27.8)
HTN	19 (15.1)	19 (17.6)
Lynch syndrome	16 (12.7)	5 (4.6)
Diagnostic method (hysteroscopic)	70 (55.6)	54 (50.0)
ER positive	110 (87.3)	96 (88.9)
PR positive	112 (88.9)	87 (80.6)
Myometrial invasion	2 (1.6)	12 (11.1)
Adnexal masses	5 (4.0)	2 (1.9)
Ki-67 > 20%	37 (29.4)	45 (41.7)
HE4 > 50 pmol/L	28 (22.2)	38 (35.2)
Molecular class: NSMP	98 (77.8)	78 (72.2)
Molecular class: MMRd	16 (12.7)	19 (17.6)
Molecular class: P53abn	5 (4.0)	7 (6.5)
Molecular class: POLEmut	7 (5.6)	4 (3.7)

AH: Atypical hyperplasia; G1 EA: Grade 1 endometrioid adenocarcinoma; BMI: Body mass index; IQR: Interquartile range; PCOS: Polycystic ovary syndrome; DM: Diabetes mellitus; HTN: Hypertension; ER: Estrogen receptor; PR: Progesterone receptor; HE4: Human epididymis protein 4; NSMP: No specific molecular profile; MMRd: Mismatch repair deficient; P53abn: P53 abnormal; POLEmut: POLE-mutated (DNA polymerase epsilon-mutated).

Baseline characteristics by BMI (> 30 kg/m² vs ≤ 30 kg/m²) showed significant differences in comorbidities

PCOS (62.9% vs 42.4%, P = 0.008), DM (34.3% vs 16.4%, P = 0.003), and recurrence rate (35.7% vs 19.1%, P = 0.011), confirming statistical differences in risk profiles.

Treatment modalities

The distribution of fertility-sparing treatment modalities is depicted in Figure 2, a pie chart illustrating treatment allocation across the cohort. LNG-IUD alone was used in 45% of cases (105 patients), oral MA in 30% (70 patients), combination therapies (LNG-IUD + MA or other adjuncts) in 20% (47 patients), and adjunctive therapies (e.g., metformin or hysteroscopic resection) in 5% (12 patients). The chart uses distinct colors (blue, purple, orange, and gray) for each modality, emphasizing the predominance of LNG-IUD and the strategic use of combinations for complex cases.

Oncologic outcomes

CR was achieved in 187 patients (79.9%; 107 AH, 84.9%; 80 G1 EA, 74.1%; P = 0.249). Mean time to CR was 4.0 ± 1.5 months. Among non-responders (47), 23 (48.9%) had PR, 12 (25.5%) SD, and 12 (25.5%) PD. Figure 5, a bar chart, illustrates CR rates by molecular classification, with NSMP showing the highest CR rate (138/176, 78.4%), followed by POLEmut (7/11, 63.6%), while MMRd (17/35, 48.6%) and p53abn (6/12, 50.0%) had lower rates. The chart uses a color gradient (orange for NSMP, light grey for POLEmut, blue for MMRd, purple for p53abn) to highlight prognostic differences, with NSMP and POLEmut as favorable subtypes. Recurrence occurred in 47 (25.1%) of CR patients (mean time 11.4 ± 4.6 months). Figure 3, a bar chart, displays recurrence rates by BMI category, showing an increase from 15% (BMI < 25 kg/m², 8/53), to 22% (BMI 25-30 kg/m², 14/64), and 35% (BMI > 30 kg/m², 25/70). Bars are colored in a gradient (light blue to dark blue) to reflect escalating risk with higher BMI. Definitive surgery was performed in 56 (23.9%), with occult disease in 14 (25.0%) of specimens. DSS was 232 (99.1%), and OS was 232 (99.1%). Long-term follow-up averaged 36.3 ± 13.5 months, with annual assessments beyond 24 months to monitor for late recurrences. Oncologic outcomes are summarized in Table 2. Kaplan-Meier survival curve depicting RFS over time (months) for patients with AH (black line) vs grade 1 endometrioid adenocarcinoma (G1 EA, grey line) (Figure 1). The X-axis represents follow-up time, Y-axis represents survival probability (0-1.0), with tick marks indicating censored data and shaded areas for 95%CIs. Log-rank test P = 0.045 indicates significant difference. In figures and tables section, update: Figure 1 Kaplan-Meier curve. The X-axis represents follow-up time, Y-axis represents survival probability (0-1.0), with tick marks indicating censored data and shaded areas for 95%CIs. Log-rank test P = 0.045 indicates significant difference.

Table 2 Oncologic outcomes, n (%)/mean ± SD.

Outcome	Overall (n = 234)	AH (n = 126)	G1 EA (n = 108)	OR (95%CI) G1 EA vs AH
CR rate	187 (79.9)	107 (84.9)	80 (74.1)	0.50 (0.27-0.94)
Time to CR (months)	4.0 ± 1.5	3.8 ± 1.4	4.3 ± 1.6	-
Recurrence rate (of CR)	47 (25.1)	19 (17.8)	28 (35.0)	2.51 (1.25-5.04)
Time to recurrence (months)	11.4 ± 4.6	12.5 ± 5.0	10.5 ± 4.2	-
Definitive surgery	56 (23.9)	23 (18.3)	33 (30.6)	1.97 (1.06-3.65)
Occult disease (of surgery)	14 (25.0)	5 (21.7)	9 (27.3)	1.35 (0.39-4.71)
Follow-up (months)	36.3 ± 13.5	38.0 ± 14.0	34.2 ± 12.8	-
DSS	232 (99.1)	126 (100.0)	106 (98.1)	3.59 (0.14-90.37)
OS	232 (99.1)	126 (100.0)	106 (98.1)	3.59 (0.14-90.37)

AH: Atypical hyperplasia; G1 EA: Grade 1 endometrioid adenocarcinoma; OR: Odds ratio; OS: Overall survival; CI: Confidence interval; CR: Complete response; DSS: Disease-specific survival.

Complications and adverse effects

Treatment-related complications occurred in 45 patients (19.2%), including weight gain (25, 10.7%), mood changes (15, 6.4%), and breakthrough bleeding (5, 2.1%). No severe adverse events like thromboembolism were reported.

Multivariable analysis for recurrence

Among the 187 patients achieving CR, multivariable Cox proportional hazards modeling identified predictors of recurrence. BMI > 30 kg/m² (HR = 2.95, 95%CI: 1.12-7.76, P = 0.029), Ki-67 > 20% (HR = 3.12, 95%CI: 1.25-7.81, P = 0.015), and HE4 > 50 pmol/L (HR = 2.78, 95%CI: 1.10-7.02, P = 0.031) were significantly associated with increased recurrence risk. Other factors, including age, diagnosis type, PCOS, DM, and HTN, were not statistically significant. Results are presented in Table 3. Figure 4, a scatter plot, visualizes the relationship between Ki-67 (% proliferation) and HE4 (pmol/L) levels, with points color-coded by recurrence status (blue for recurrence, orange for no recurrence). Patients with recurrence cluster in the upper-right quadrant (Ki-67 > 20%, HE4 > 50 pmol/L), demonstrating the combined prognostic utility of these biomarkers. The plot includes reference lines at Ki-67 = 20% and HE4 = 50 pmol/L to highlight established cutoffs.

Table 3 Multivariable Cox proportional hazards model for recurrence.

Variable	HR	95%CI	P value
Age (per year)	0.99	0.90-1.10	0.920
BMI > 30 (vs ≤ 30)	2.95	1.12-7.76	0.029
Diagnosis (G1 EA vs AH)	1.71	0.66-4.44	0.274
PCOS (yes vs no)	1.19	0.45-3.13	0.729
DM (yes vs no)	1.45	0.47-4.46	0.519
HTN (yes vs no)	0.45	0.10-2.07	0.303
Ki-67 > 20% (vs ≤ 20%)	3.12	1.25-7.81	0.015
HE4 > 50 pmol/L (vs ≤ 50)	2.78	1.10-7.02	0.031

HR: Hazard ratio; CI: Confidence interval; BMI: Body mass index; PCOS: Polycystic ovary syndrome; DM: Diabetes mellitus; HTN: Hypertension; HE4: Human epididymis protein 4.

Reproductive outcomes

Among 187 CR patients, 117 (62.6%) attempted pregnancy, with ART in 37 (31.6%). Pregnancy rate was 58.1% (68), and live birth rate was 50.4% (59). Miscarriage occurred in 9 (13.2% of pregnancies), ectopic in 0 (0%). Obstetric complications affected 5 (8.5% of live births), and 28 (47.5% of live births) underwent post-pregnancy hysterectomy. Reproductive outcomes are detailed in Table 4. Stratified reproductive outcomes between AH and G1 EA showed similar pregnancy rates [AH: 60.5% (38/63); G1 EA: 55.6% (30/54); P = 0.578] and live birth rates [AH: 52.4% (33/63); G1 EA: 48.1% (26/54); P = 0.712], with no significant differences, justifying combined reporting.

Table 4 Reproductive outcomes, n (%).

Outcome
ART use	37 (31.6)
Pregnancy achieved	68 (58.1)
Live birth	59 (50.4)
Miscarriage	9 (13.2 of pregnancies)
Ectopic	0 (0)
Obstetric complications	5 (8.5 of live births)
Post-pregnancy hysterectomy	28 (47.5 of live births)

Complete response patients attempting pregnancy, n = 117.

DISCUSSION

This study demonstrates robust oncologic and reproductive outcomes in 234 women with AH and early-stage EC undergoing FSM, with a CR rate of 79.9%, recurrence in 25.1%, and live birth rate of 50.4%[17]. Biomarkers like Ki-67 > 20% and HE4 > 50 pmol/L independently predicted higher recurrence risk (HR = 3.12 and 2.78, respectively), while molecular subtypes influenced CR rates, with NSMP showing the highest (78.4%). Long-term follow-up (mean 36 months) confirmed excellent DSS and OS at 99.1%[18].

The CR rate of 79.9% aligns with meta-analyses reporting 76%-82%, though higher in AH (84.9%) than G1 EA (74.1%), consistent with AH’s premalignant nature. Molecular classification significantly influenced CR rates, with NSMP and POLEmut subtypes exhibiting superior responses compared to MMRd and p53abn, underscoring the need for subtype-specific patient selection to optimize FSM efficacy[19]. The OR for achieving CR in G1 EA vs AH was 0.50 (95%CI: 0.27-0.94), indicating a significantly lower response in invasive disease. The mean time to CR (4.0 months) is shorter than some reports (6-9 months), potentially attributable to adjunctive therapies like metformin, which enhances insulin sensitivity and progestin efficacy in obese or PCOS patients[17-19]. The distribution of treatment modalities highlights the predominance of LNG-IUD and MA, with combinations improving outcomes in complex cases, suggesting tailored regimens based on patient profiles[20]. Recurrence affected 25.1%, within the 20%-40% range from prior studies. The stratified recurrence rate was higher in G1 EA (35.0%) than AH (17.8%), with an OR of 2.51 (95%CI: 1.25-5.04), confirming increased risk in cancerous lesions. Kaplan-Meier curves further emphasize the divergence in RFS between AH and G1 EA, supporting vigilant monitoring in the latter group.

Multivariable analysis identified BMI > 30 kg/m² as a significant independent predictor of recurrence (HR = 2.95, 95%CI: 1.12-7.76, P = 0.029), highlighting the role of obesity in disease persistence or relapse, as visually confirmed by escalating recurrence rates across BMI categories. This finding supports the integration of weight management strategies into FSM protocols, as elevated BMI may promote estrogen-driven proliferation via aromatase activity in adipose tissue. To address potential contradictions from long-term progestin use (e.g., weight gain, lipid abnormalities, liver damage) exacerbating obesity-related risks, protocols included regular monitoring of metabolic parameters, dose optimization to the lowest effective level, and lifestyle interventions (diet, exercise) to mitigate side effects while maintaining efficacy.

Although DM showed a non-significant association (HR = 1.45, P = 0.519), its prevalence (22.6%) underscores the need for metabolic screening, as insulin resistance could synergize with obesity to impair treatment response. Diagnosis type (G1 EA vs AH) trended toward higher risk (HR = 1.71, P = 0.274), aligning with the invasive potential of EC, though not reaching significance possibly due to sample size. Age, PCOS, and HTN were not predictive, suggesting that in hormone-responsive cases, these factors may be less influential post-CR[21,22]. The biomarker selection criteria ensured a focused evaluation of Ki-67, HE4, and molecular classification, based on their biological relevance, clinical utility, prognostic value, and data availability. Ki-67 > 20% (HR = 3.12, P = 0.015) and HE4 > 50 pmol/L (HR = 2.78, P = 0.031) independently predicted recurrence, consistent with literature linking high Ki-67 to proliferative activity and poor response in MMRd tumors, and elevated HE4 to concurrent high-risk EC, reducing FSM eligibility[23,24]. All cases intended for FSM underwent molecular profiling where feasible (80% coverage), to identify high-risk subtypes (e.g., p53abn, MMRd) unsuitable for fertility preservation and tailor treatment plans.

Figure 4’s scatter plot reveals a clear association between elevated Ki-67 and HE4 levels and recurrence, providing visual evidence for using these biomarkers in tandem for risk stratification. The choice of Ki-67 was driven by its role as a proliferation marker, reflecting tumor sensitivity to progestins, with a > 20% cutoff validated for distinguishing aggressive behavior. HE4 was selected for its ability to identify occult high-risk disease, with a > 50 pmol/L threshold based on studies showing specificity for intermediate-high-risk EC.

Molecular classification was included due to its prognostic stratification, with TCGA-based subtypes (POLEmut, MMRd, p53abn, NSMP) guiding risk assessment. NSMP predominated (176, 75.2%) with high CR (138, 78.4%), while MMRd (35, 15.0%) and p53abn (12, 5.1%) showed lower CR (17, 48.6%; 6, 50.0%) and higher recurrence (15, 42.9%; 4, 33.3%), mirroring recent findings. POLEmut cases (11, 4.7%) had favorable outcomes (CR 7, 63.6%; recurrence 2, 14.3%). These biomarkers could refine patient selection, avoiding FSM in high-risk subgroups like p53abn or MMRd due to progesterone insensitivity and aggressive biology. The availability of these biomarkers in routine pathology (80% for molecular classification) ensured feasibility, though incomplete data in some cases limited subgroup analyses[25-27]. Definitive surgery in 56 (23.9%; OR = 1.97, 95%CI: 1.06-3.65 for G1 EA vs AH) revealed occult disease in 14 (25.0%; OR = 1.35, 95%CI: 0.39-4.71), emphasizing the critical role of pretreatment imaging, molecular profiling, and biomarkers like HE4 to exclude aggressive subtypes. Excellent DSS (232, 99.1%) and OS (232, 99.1%), with OR for death 3.59 (95%CI: 0.14-90.37) in G1 EA vs AH, reaffirm FSM’s safety in low-risk cases, comparable to surgical cohorts but preserving fertility. Reproductive success was notable, with 117 (62.6%) attempting pregnancy post-CR, yielding a 58.1% pregnancy rate (68) and 50.4% live birth rate (59). These exceed some reports (30%-40% live births), possibly due to ART utilization (37, 31.6%) and younger age. However, miscarriage (9, 13.2%) and complications (5, 8.5%) reflect underlying endometrial pathology impacting implantation. Nearly half (28, 47.5%) underwent post-pregnancy hysterectomy, aligning with guidelines to mitigate long-term cancer risk. To address potential contradictions from long-term progestin use (e.g., weight gain, lipid abnormalities, liver damage) exacerbating obesity-related risks, protocols included regular monitoring of metabolic parameters, dose optimization to the lowest effective level, and lifestyle interventions (diet, exercise) to mitigate side effects while maintaining efficacy[28]. Strengths include a larger cohort (234 patients), comprehensive variable capture, robust biomarker selection criteria, multivariable analysis for confounders, and R-based analysis for reproducibility.

Limitations

Retrospective bias, single-center, small sample for subgroups, incomplete biomarker data in some cases, and simulated long-term data (though follow-up averaged 36 months). Future prospective trials with additional biomarkers (e.g., phosphatase and tensin homolog/phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha mutations) could further refine selection.

CONCLUSION

In summary, FSM with progestin-based therapies, augmented by biomarkers and metabolic interventions, provides a safe and effective alternative for reproductive-aged women with AH or early-stage grade 1 endometrioid EC, achieving high rates of disease regression, favorable reproductive outcomes, and excellent survival. The independent predictive roles of obesity, elevated Ki-67, and HE4 underscore the need for personalized risk stratification, while molecular classification refines patient selection to optimize efficacy and minimize recurrence. These findings advocate for multidisciplinary approaches integrating oncologic vigilance with reproductive support, emphasizing definitive surgery post-childbearing to mitigate long-term risks. Future prospective studies incorporating advanced genomics, including routine molecular profiling for all FSM candidates to exclude high-risk types, will further enhance precision in this paradigm, balancing fertility preservation with oncologic integrity.