Efficacy of fertility-sparing treatment with LNG-IUS is associated with different ProMisE subtypes of endometrial carcinoma or atypical endometrial hyperplasia

Abstract

Objective

To determine whether proactive molecular risk classifier for endometrial cancer (ProMisE) could be used to assess the prognosis of patients with atypical endometrial hyperplasia (AEH) or early-stage endometrial cancer (EC) treated with levonorgestrel-releasing intrauterine system (LNG-IUS).

Methods

A retrospective cohort study was conducted among 93 AEH or early-stage EC patients who received LNG-IUS to preserve fertility . By immunohistochemistry and gene sequencing, 4 subtypes of ProMisE were identified (p53 wild type [p53 wt], mismatch repair-deficient [MMRd], p53-abnormal, and POLE-mutated). The primary outcome was the time to complete response (CR) after LNG-IUS therapy. Secondary outcomes included the recurrence rate after CR and success rate of conception.

Results

Among the 93 patients, 15 (16.1%) were classified as MMRd, 6 (6.5%) as POLE-mutated, 5 (5.4%) as p53-abnormal, and 67 (72.0%) as p53 wt. Comparison of serum cancer antigen 125, family history of tumor, and positive rates of programmed cell death 1 ligand 1 protein and Ki67 protein in 4 groups showed statistically significant differences (p<0.05). Patients with the p53-abnormal subtype had the lowest overall CR rate (40%) and the highest recurrence rate (2/2). Patients with POLE-mutated subtype had the best prognosis, and all 6 patients achieved CR. When patients achieved complete remission, assisted reproductive technology was more likely to help them conceive than natural conception (p<0.05).

Conclusion

Patients with early-stage EC or AEH who are more likely to benefit from fertility-sparing treatment can be identified using ProMisE classifier. Patients with POLE-mutated are suitable for fertility-sparing treatment with LNG-IUS.

Synopsis

Patients with early-stage endometrial cancer or atypical endometrial hyperplasia who are more likely to benefit from fertility-sparing treatment can be identified using proactive molecular risk classifier for endometrial cancer classifier.

Graphical Abstract

INTRODUCTION

Endometrial cancer (EC) is one of the most common malignant tumors of the female reproductive system, and its incidence has been increasing in the past 2 decades [1]. Despite the fact that EC is more common in postmenopausal women, it still affects 3% to 14% of women younger than 40 years old [2]. Fertility-sparing treatment for patients with EC has become the focus of gynecological tumor research. The pathological types of young patients are mostly well-differentiated EC, which is confined to the endometrium, and the 5-year disease-free survival rate is as high as 99.2% [3,4]. Atypical endometrial hyperplasia (AEH) is a precancerous lesion of EC. Approximately 15% to 75% of AEH will progress to EC, and AEH and EC may coexist [5].

For patients with AEH or EC, the standard treatment is hysterectomy, however, this treatment is not optimal for those with reproductive needs. In recent years, fertility-sparing treatment for EC has proven safe and effective, including high-dose progesterone, levonorgestrel-releasing intrauterine system (LNG-IUS), and gonadotropin releasing hormone agonist [6,7,8]. Compared with oral progesterone, the intrauterine placement of LNG-IUS has a higher rate of CR and reduces systemic adverse effects in patients with AEH [9]. It was found that LNG-IUS treatment was effective in treating AEH as well as EC with remission rates of 80% and 67%, respectively [10]. LNG-IUS is now included in National Comprehensive Cancer Network guidelines as a fertility preservation option for EC patients since 2014.

The pathological type of fertility preservation therapy for patients with EC should be endometrioid adenocarcinoma, G1. Endometrial carcinomas are traditionally classified as type I and type II: type I is estrogen-dependent and characterized by endometrioid adenocarcinoma, while type II is mainly characterized by p53 mutation and includes serous adenocarcinoma. The identification of histopathological types of EC is mainly performed by pathologists, and some types are difficult to distinguish under the microscope. There is only 62.5% diagnostic concordance among 3 pathologists in a study of EC [11]. It was found that the histological type and grade of tumors were moderately to poorly correlated between observers in Hoang et al. [12]. Therefore, it is not enough to use the traditional classification method to stratify the fertility treatment population, and further improvement is needed.

In 2013, the molecular classification of EC was proposed by The Cancer Genome Atlas, which provides important information for patient classification, prognosis prediction, and individualized precision treatment [13]. And then the researchers proposed the proactive molecular risk classifier for endometrial cancer (ProMisE) classification, which divided EC into 4 molecular subtypes: mismatch repair-deficient (MMR-d), POLE-mutated, p53 wild type (p53 wt), and p53 abnormal (p53abn) [14]. ProMisE classification could be conducted in formalin-fixed paraffin embedded (FFPE) samples, which would facilitate clinical collection and operation. At present, only a few small sample studies have applied the ProMisE classifier in fertility-sparing treatment for EC, and the conclusions are controversial [15,16,17,18]. The study by Puechl et al. [16] was the first to report the feasibility of molecular classification in endometrial intraepithelial neoplasia patients treated with LNG-IUS.

However, no studies have reported whether molecular classification can predict the outcomes of EC or AEH treated with LNG-IUS. The objective of this study was specifically to determine the effectiveness of LNG-IUS therapy in patients with different ProMisE subtypes of EC or AEH. We conducted this study in accordance with the STROBE checklist.

MATERIALS AND METHODS

A 2-year retrospective cohort study was conducted in patients with EC or AEH receiving LNG-IUS at Tongji Hospital, Tongji Medical College, Huazhong University of Science & Technology, between February 2018 and February 2021. The present study was supported by the Ethics Committee of Tongji Hospital in accordance with the ethical standards in the Declaration of Helsinki. Patients were identified and followed up by searching the database of Tongji Hospital. Inclusion criteria are as follows: no more than 40 years old; had a strong desire to preserve fertility function and signed an informed consent, clearly stating that fertility preservation was not a standard treatment and required regular follow-up and may require surgery in the future; hysteroscopy and curettage proved to be AEH or endometrioid adenocarcinoma G1 by pathology; magnetic resonance imaging indicated that the lesion was confined to the endometrium; no contraindication of progesterone therapy. The exclusion criteria are as follows: had taken hormone drugs regularly in the past 3 months; important data were incomplete and complete follow-up information could not be obtained; patients refused to use LNG-IUS; patients had contraindications to pregnancy.

Placement of LNG-IUS (Mirena; Bayer, Shanghai, China) was performed by the same gynecologist for all enrolled patients. The patient was instructed to empty the bladder before treatment. After obtaining the bladder lithotomy position, cleaning and disinfection of the perineum and cervix, the patient was ready for surgery. The depth of the uterine cavity of the patient was measured, and the tail fillamentum of the LNG-IUS ring was pulled down to make both arms of the ring reach the cannula, and the lower edge of the locator was moved to the measured position. Thirty seconds later, the arms of the LNG-IUS were opened and placed into the fundus of the patient. Using a color Doppler ultrasound, the LNG-IUS location was confirmed. The cannula was removed and the ring tail was cut approximately 3 cm from the cervical opening. The efficacy of the treatment was evaluated every 3 months using vaginal ultrasound and hysteroscopy with curettage. All endometrial tissues were sent to the Department of Pathology of Tongji Hospital for pathological diagnosis. The initial treatment phase can be continued for 6–9 months, but patients who do not respond to LNG-IUS therapy beyond 6 months need to be treated conservatively for another 3–6 months after the progression or metastasis of the disease is excluded, but close follow-up is required to avoid delaying the disease. For patients who fail to obtain a pathological complete response (CR) after 12 months of conservative treatment, the reasons should be analyzed and the change of treatment plan or abandonment of fertility conservation therapy should be considered and replaced with standard surgical treatment to avoid delaying the disease and affecting the prognosis. If early-stage EC persists for 6 to 12 months, complete hysterectomy + double adnexectomy + surgical staging should be performed, and relevant imaging examinations should be performed before surgery. If the disease is completely resolved after 6 months, the patient is encouraged to conceive, and it is still necessary to continue monitoring every 6 months before pregnancy.

During our outpatient appointments, we collected information on patients, including their age, pregnancy history, infertility, height, weight, history of oral progesterone therapy, endometrial pathology, family history, and pregnancy outcome of infertility patients. Molecular typing was performed using formaldehyde-fixed paraffin-embedded endometrial tissue [19,20]. We used the NovaSeq 6000 (Illumina, San Diego, CA, USA) second generation sequencing platform to sequence POLE. Immunohistochemistry was used to determine the expression level of programmed cell death 1 ligand 1 (PD-L1) in paraffin-embedded sections. The proportion of PD-L1 protein positive cells ≥1% is defined as PD-L1 positive. The primary outcome was the time to CR after LNG-IUS therapy. Secondary outcomes included the recurrence rate after the CR and success rate of conception. Efficacy was evaluated according to previous studies [21]: CR, endometrial lesions disappeared completely, no cancer tissue and atypical hyperplasia tissue; partial response, pathological examination showed that the endometrial glands were less crowded, but the papillary and ethmoid structures were still present; SD, stable disease, compared with the previous pathological examination, there was no change in the lesion; PD, progression of disease, the pathological grade or stage of the tumor increased, and the cell atypia increased; Recurrence: After CR, EC or AEH lesions reappear in the endometrial tissue.

1. Statistical analysis

The statistical analysis was performed using SPSS 22.0 software (IBM Corp., Armonk, NY, USA) and R software (R3.6; R Core Team, Vienna, Austria). Counting data were expressed as percentage or number, and analyzed by Fisher’s exact probability and Pearson χ² tests. The measurement data were presented as mean ± standard deviation, analysis of variance was used for comparison between multiple groups. The Bonferroni method was used for pairwise comparison after adjusting p-values. Kaplan-Meier analyses was used to draw the survival curve, and log-rank was used to test the effect of molecular typing on prognosis. It was considered statistically significant if the p-value was less than 0.05.

RESULTS

As shown in Fig. 1, a total of 93 EC or AEH patients treated with LNG-IUS were enrolled in this study. They were then divided into 4 subgroups by immunohistochemistry and single gene sequencing. A total of 15 (16.1%, one case was Lynch syndrome) patients had MMRd, 6 (6.5%) in total had POLE gene mutations, and 5 (5.4%) had mutations in the p53 gene. There were 67 (72.0%) patients in the p53 wt group, accounting for the largest proportion. Among all participants, 25 cases (26.9%) were complicated with polycystic ovary syndrome (PCOS), 18 cases (19.4%) were diabetic, 70 cases (75.3%) were not pregnant before, and 27 cases (29.0%) had a family history of tumor. It was found that patients in the p53 abn subgroup were elderly than those in the p53 wt subgroup (p<0.05). However, there was no significant difference in age, body mass index, pregnancy history, PCOS or diabetes among the 4 molecular types (p>0.05). There were statistical differences in serum cancer antigen 125 (CA125) level and family history of tumor in patients with 4 molecular classifications. The serum CA125 level of p53 wt patients was the highest among the 4 molecular subtypes, which was 33.6±17.8U/mL. MMRd patients had the highest proportion of family history of tumor (9/15) (Table 1).

Fig. 1. Flow diagram of this study.

AEH, atypical endometrial hyperplasia; EC, endometrial cancer; LNG-IUS, levonorgestrel-releasing intrauterine system; MMR, mismatch repair; MMRd, mismatch repair-deficient; ProMisE, proactive molecular risk classifier for endometrial cancer.

Table 1. Baseline characteristics of patients.

Characteristics		Total	MMRd	POLE	p53 abn	p53 wt	p
Total		93	15	6	5	67
Age (yr)
	<35	61	11	2	2	46
	≥35	32	4	4	3	21	0.182
BMI (kg/m2)		93	28.5±6.5	27.2±2.3	27.4±4.3	29.2±5.1	0.680
PCOS		25	2	1	2	20	0.511
CA125 (U/mL)		93	26.9±11.4	17.3±6.2	22.3±8.7	33.6±17.8	0.042
Family history of cancer		27	9	2	1	15	0.029
Diabetes		18	3	2	2	11	0.303
No. of pregnancies
	0	70	13	4	2	51	0.175
	>0	23	2	2	3	16

BMI, body mass index; CA125, cancer antigen 125; MMRd, mismatch repair-deficient; PCOS, polycystic ovary syndrome; p53 abn, p53 abnormal; p53 wt, p53 wild type.

As a result of immunohistochemical analysis, the expression rates of estrogen receptor (ER) and progesterone receptor (PR) of the 4 molecular types were no statistical differences (p>0.05). Pairwise comparison showed that the positive rate of PR was higher in p53 wt group than that in p53 abn group and POLE-mutated group (p<0.05). There were statistically significant differences between the 4 groups in the expression of Ki67 (Fig. 2) and PD-L1 (Table 2), among which the expression rates of PD-L1 (9/15) and Ki67 (50.11%±5.42%) in MMRd subtype were the highest. Furthermore, Kaplan-Meier method was used to calculate the cumulative CR rate and draw the survival curve. The results showed that patients with p53 abn subtype had the lowest cumulative CR rate (Fig. 3).

Fig. 2. The expression levels of Ki67 in the 4 groups.

MMRd, mismatch repair-deficient; p53 abn, p53 abnormal; p53 wt, p53 wild type.

Table 2. Relationship between 4 molecular classifications and pathological features.

Characteristics	MMRd	POLE	p53 abn	p53 wt	p
ER expression
Positive	12	4	4	50
Negative	3	2	1	17	0.929
PR expression
Positive	10	3	2	45
Negative	5	3	3	22	0.527
PD-L1 expression
Positive	9	1	2	16
Negative	6	5	3	51	0.036

ER, estrogen receptor; MMRd, mismatch repair-deficient; PD-L1, programmed cell death 1 ligand 1; PR, progesterone receptor; p53 abn, p53 abnormal; p53 wt, p53 wild type.

Fig. 3. Comparison of cumulative CR rates across molecular types.

CR, complete response; MMRd, mismatch repair-deficient; p53 abn, p53 abnormal; p53 wt, p53 wild type.

After 76 patients achieved CR, 13 (17.11%) patients experienced disease recurrence. A significant difference was found among the groups (p<0.01), with the p53 abn subgroup recording the highest recurrence rate (2/2) and PD rate (2/5) (Table 3). The success rate of 40 patients who chose in vitro fertilization-embryo transfer was 47.50%, and the success rate of 23 patients who chose natural pregnancy after ring extraction was 21.74%. There was a statistical difference between the 2 methods of conception (p=0.043). During the 2-year follow-up period, all recurrence and PD patients eventually underwent hysterectomy. Nine patients with PR and SD were trying other treatments to achieve their goal of conception.

Table 3. The efficacy of fertility-sparing treatment in 4 molecular subtypes.

Characteristics	Total	MMRd	POLE	p53 abn	p53 wt	p
IVF-ET	40	4	4	0	32
Success	19 (47.50%)	1	2	0	16
Failure	21	3	2	0	16	0.845
Natural conception	23	3	2	0	18
Success	8 (21.74%)	0	1	0	7
Failure	15	3	1	0	11	0.578
CR	76	11	6	2	57	0.048
PR	5	0	0	1	4	0.325
SD	4	0	0	0	4	1.000
PD	8	4	0	2	2	0.002
Recurrence	13	4	0	2	7	0.009

CR, complete response; IVF-ET, in vitro fertilization-embryo transfer; MMRd, mismatch repair-deficient; PD, progression of disease; PR, progesterone receptor; SD, stable disease; p53 abn, p53 abnormal; p53 wt, p53 wild type.

DISCUSSION

Over the past few years, EC fertility-preserving treatment has made great progress as a result of increasing treatment experience and evidence-based medical evidence. However, not all patients with EC respond to progesterone therapy, and the recurrence rate of patients who achieve complete remission is 21.0%–42.4% [22]. Therefore, it is crucial to explore biomarkers to predict responses after progesterone therapy for screening fertility conservation candidates. The proposed ProMisE classification system provides more accurate information for the clinical treatment and prognosis of patients with EC.

There is no consensus on whether there are differences in efficacy and prognosis among different ProMisE subtypes in patients with EC or precancerous lesions that preserve fertility function. Ran et al. [15] reported the outcomes of 13 EC patients in their institution and found no difference in CR rates between the MMRd subtype and the p53 wt subtype after fertility conservation treatment. In a study by Chung et al. [18], 57 EC patients treated with progesterone were molecular classified, and the results showed MMRd patients had a 44.4% CR rate, significantly lower than p53 wt patients. Puechl et al. [16] classified EC or AEH patients with ProMisE through FFPE specimens and found that p53 abn subtype had the shortest time for PD (5.7 months). Patients in the POLE-mutated subgroup had the longest median time to PD (21.4 months). To our knowledge, our study is the first study to use the ProMisE classifier to assess the prognosis of EC or AEH patients treated with LNG-IUS.

MMR proteins maintain genomic integrity by recognizing and correcting base mismatches during DNA replication. Defects in this system can lead to mutations in cancer-related genes that increase the risk of cancer. About 16%–40% of patients with EC have been reported to exhibit MMRd status [23,24]. In our study, there were 15 cases of MMRd type, accounting for 16.1%. The CR rate and recurrence rate of the MMRd subtype were 73.33% and 36.36%, respectively. It was found that 9/15 patients with MMRd type were positive for PD-L1, suggesting that immune therapy may have a greater benefit in this group of patients. In addition, since the proportion of MMRd patients with a family history of tumor is significantly higher than that of other molecular types, immediate family members of MMRd patients are recommended to enter the relevant genetic counseling and genetic testing process.

POLE-mutated subtype has a good prognosis and can avoid adjuvant therapy [25]. These data are based on survival data of patients treated with surgery, and there is limited data on response of POLE-mutated subtype to hormonal therapy. In our study, all 6 patients with POLE mutations achieved CRs. At present, the mechanism of better prognosis in POLE-mutated EC patients remains unclear. It may be caused by the high tumor mutation burden in POLE mutant patients, which stimulates the production of a large number of tumor antigens and induces anti-tumor immune response [26]. For POLE mutant patients with good prognosis, whether the indications of fertility preservation therapy can be relaxed is worth further investigation.

In this study, the fertility preservation effect of p53 abn subtype was the worst among the 4 types, and the patients were usually older. Given the small number of p53 abn subtypes in this study, we need to conduct prospective studies with larger sample sizes. Patients with p53 wt subtype usually have higher expression of ER and PR, which are considered to be independent prognostic risk factors for EC. However, not all patients with p53 wt subtype respond well to progesterone. Considering that patients with the p53 wt subtype have highly heterogeneous clinical manifestations and prognoses, it is possible to refine and distinguish the p53 wt subtype by combining with other individualized molecular markers, so as to provide better guidance for patient prognosis and treatment.

In conclusion, patients in p53 abn subtype have poor response to progesterone therapy and are prone to relapse, so fertility preservation therapy is not recommended. The POLE-mutated subgroup has a good prognosis, and it is recommended that assisted reproduction be performed as soon as possible after CR. Patients in the p53 wt subgroups should be further stratified in combination with other biomarkers.