Efficacy and safety of different subsequent therapies after fertility preserving surgery for endometriosis: A systematic review and network meta-analysis

Dienogest and LNG-IUS are superior to the other 5 interventions in alleviating EMT related symptoms and reducing the recurrence rate of EMT in patients undergoing fertile-preserving surgery, with good safety. For those who want to have children after surgery, dienogest has a better pregnancy rate than the other 6 interventions.

Endometriosis (EMT) is an estrogen-dependent, progesterone-resistant gynecologic condition characterized by the presence of ectopic endometrial-like tissue outside the uterine cavity. [ 1 ] It is common in women of childbearing age, and many patients look forward to restoring their reproductive function. For the treatment of EMT, it is recommended to start empiric medication as early as possible based on clinical diagnosis. Evidence suggests that long-term outcomes and the risk of complications are more favorable with medication than surgery for both EMT and deep lesions.Hormonal therapy improves pain symptoms in around 2-thirds of women with EMT. [ 2 ] However, Patients affected by posterior deep infiltrating EMT usually have severe painful symptoms (dysmenorrhea, deep dyspareunia, chronic pelvic pain and dyschezia) that severely reduce quality of life. [ 3 ] Surgical excision of EMT is required in the case of pain symptoms resistant to hormonal therapy, complicated disease (bowel or urinary obstruction) or infertility after several assisted reproductive technology cycles. [ 4 ] The lesions removed or destroyed by fertility-preserving surgery are only macroscopic lesions, while small and atypical lesions have a high chance of residual. The recurrence of EMT after surgery is an urgent problem to be solved. It has been reported that the recurrence rate of endometrioma after laparoscopic surgery is 19% to 50%, and the second surgery undoubtedly increases the medical cost and the physical and mental harm to the patient. [ 5 ] At present, the commonly used drugs after fertility-preserving surgery in EMT include non-steroidal anti-inflammatory drugs, oral contraceptives, progestins, progesterone receptor antagonists, gestrinone, danazol, gonadotropin-releasing hormone agonist (GnRH-a), and levonorgestrel intrauterine system (LNG-IUS) and dienogest. With the development of evidence-based medicine, many research teams at home and abroad have carried out randomized controlled trials (RCTs) on drug treatment after fertility-preserving surgery in EMT. However, in different RCTs, the types of drugs are different, and the outcome indicators to evaluate the efficacy and safety of drugs are not the same. At the same time, there is a lack of RCTs studying the differences between 3 or more drugs. This network meta-analysis compared the efficacy and safety of commonly used drugs (dienogest, leuprolide, danazol, gestrinone, mifepristone, and LNG-IUS) after fertile-preserving surgery in EMT from 4 aspects: effective rate, recurrence rate, pregnancy rate, and adverse reaction rate. By comparing the mixture of different drugs and ranking the efficacy of drugs, the most likely to be the best postoperative treatment was explored to provide evidence-based basis for clinical decision-making.

Conceptualization: Yu Zheng, Ruiheng Zhao. Data curation: Yu Zheng, Ruyue Ma, Hong Xu. Formal analysis: Yu Zheng, Ruyue Ma. Funding acquisition: Yu Zheng. Investigation: Yu Zheng, Ruyue Ma. Methodology: Yu Zheng, Ruyue Ma, Ruiheng Zhao. Project administration: Ruiheng Zhao. Software: Yu Zheng, Ruyue Ma, Lian Wang. Supervision: Ruiheng Zhao. Validation: Ruiheng Zhao. Writing – original draft: Yu Zheng, Ruyue Ma. Writing – review & editing: Lei Zhang, Huiqun Mao, Ruiheng Zhao.

This study review protocol has been registered and certified in the International Prospective Registry for Systematic Reviews (Registration number CRD42023393627). RCTs on the application of dienogest, leuprolide, danazol, gestrinone, mifepristone and LNG-IUS in the treatment of EMT patients after fertile-preserving surgery have been published. Patients of reproductive age who underwent fertility-preserving surgery (removal of all macroscopic lesions, releasing adhesions, preservation of the uterus, and preservation of one/both/at least part of the ovaries) and were histologically diagnosed as EMT after surgery were limited to 20 to 40 years old, regardless of their nationality, course of disease, and fertility requirements. The experimental group was treated with dienogest, leuprolide, danazol, gestrinone, mifepristone or LNG-IUS after operation as subsequent drug therapy. The control group received placebo or no medication. Effective rate: the number of cases of remission (no symptoms, no pelvic mass formation) and improvement (relief of symptoms, no positive signs, no pelvic mass formation)/the total number of cases; Recurrence rate: number of cases/total number of cases of recurrent pelvic mass or secondary dysmenorrhea; Pregnancy rate: the number of pregnant patients with fertility requirements/the number of patients with fertility requirements; Adverse reaction rate: the number of cases with adverse drug reaction symptoms/the total number of cases. Patients with adenomyosis or without fertility-preserving surgery; Basic research; the type of research was not RCT; duplicate published studies; Abstracts, meeting minutes, case reports, etc; incomplete data/unable to extract detailed data. PubMed, Cochrane Library, Web of Science, EMBase, China National Knowledge Internet, VIP database, China Biology Medicine disc and WanFang Data were searched. RCTs of dienogest, leuprolide, danazol, gestrinone, mifepristone and LNG-IUS used in fertility-preserving surgery of EMT were collected from the establishment of the database to February 2023. All databases were searched using subject headings combined with free words, and the search terms included: EMT, dienogest, leuprolide, danazol, gestrinone, mifepristone, LNG-IUS, RCTs, RCT; EMT, ovarian EMT cyst, chocolate cyst, dienogest, leuprolide, danazol, gestrinone, mifepristone, LNG-IUS, RCTs, etc. The literature was imported into Endnote X8 software to exclude duplicate literature. Two researchers independently scanned the titles and abstracts of all the literatures. The articles that met the exclusion criteria were excluded; The remaining articles that might meet the inclusion criteria were searched and read to determine whether they were included in this study. According to the pre-designed data extraction table, the data of the included studies were extracted, and appropriate remarks were made. The quality of the included studies was assessed by 2 reviewers independently according to the risk of bias tool [ 6 ] for RCTs recommended by Cochrane Handbook of Systematic Reviews. During the above process, if the 2 researchers had differences and could not reach a consensus, a third party was requested to assist in the adjudication. The network meta-analysis was performed based on the frequency method in STATA 15.0 software. The effective rate, recurrence rate, pregnancy rate and adverse reaction rate were used as the outcome indicators to evaluate the efficacy and safety of the drug. The network relationships among various interventions were shown by the network meta-analysis maps. The publication bias of the included studies was evaluated qualitatively by drawing the corrected comparison funnel plot. In Stata 15.0 software, the “mvmeta” command and the “network meta” group command were run. The odds ratio (OR) and the associated 95% confidence interval (95% CI) of the outcome measures were used to describe the efficacy and safety. The SUCRA and the average rank were calculated by drawing the “surface under the cumulative ranking (SUCRA)” to illustrate the ranking probabilities of different interventions on different outcomes.

PubMed, Cochrane Library, Web of Science, EMBase, China National Knowledge Infrastructure, VIP database, China Biology Medicine disc and WanFang Data were searched by computer, and 11424 articles were obtained. One article was obtained by reading relevant systematic reviews, and 3893 articles were excluded. Literature titles and abstracts were read, and 7470 articles that were obviously irrelevant, basic experiments, case reports, and conference proceedings were excluded. The remaining 62 full-text articles that may meet the inclusion criteria were read, and 18 articles were excluded again for non-RCTs, subjects did not undergo surgical treatment, the full text could not be obtained, the data could not be extracted, and repeated reports. Finally, 44 RCTs were selected for this network meta-analysis. The literature search process and results are shown in Figure 1 . Guidelines flow diagram. A total of 44 studies were included in this study, including 4576 patients with EMT, with an average age of 32.5 years. The basic characteristics of the included studies are shown in Table 1 , and the quality assessment, or risk of bias assessment, is shown in Figure 2 . Basic characteristics of the included studies. ①: effective rate; ②: recurrence rate; ③: pregnancy rate; ④ adverse reaction rate. biw = twice a week, im = intramuscular injection, LNG-IUS = levonorgestrel intrauterine system, PBO = placebo, po = oral, qd = daily, qiw = once every 4 weeks, rAFS = revised classification of the American Fertility Society. The result of the risk of bias assessment. See Figure 3 for the “network of evidence” between the 7 interventions. Nodes in the figure represent interventions, and the node size is positively correlated with the sample size involved in the intervention. Thicker lines between nodes indicate more studies on 2 interventions. Network meta-analysis maps. The results of publication bias evaluation of the included studies were shown by the “Inverted funnel plot” (Fig. 4 ). In the funnel plot of effective rate, pregnancy rate, and adverse reaction rate, the distribution of scatter points on both sides of X = 0 was roughly symmetrical, and the possibility of publication bias was low. In the corrected comparison funnel plot of recurrence rate, the distribution of scatter points on both sides of X = 0 was asymmetrical, suggesting that publication bias might exist. Inverted funnel plot. Note: (A) placebo (B) dienogest (C) leuprolide (D) danazol (E) gestrinone (F) mifepristone (G) LNG-IUS = levonorgestrel intrauterine system. Network meta-analysis mixed comparison results: Compared with placebo, postoperative dienogest (OR = 8.88, 95%CI = [2.67, 29.52]), leuprolide (OR = 5.17, 95%CI = [2.35, 11.31]), gestrinone (OR = 2.83, 95%CI = [1.23, 6.49]), LNG-IUS (OR = 8.97, 95%CI = [3.04, 25.48]) were more effective in relieving EMT-related symptoms than placebo, the difference was statistically significant. Compared with dienogest (OR = 5.39, 95%CI = [1.03, 28.29]) and mifepristone (OR = 4.15, 95%CI = [1.16, 14.82]), the effective rate of dienogest was better than mifepristone and danazol, and the difference was statistically significant. In addition, intrauterine placement of LNG-IUS was superior to oral danazol (OR = 0.18, 95%CI = [0.04, 0.86]). There was no significant difference in the other interventions (Table 2 ). Network meta-analysis matrix of results. Bold indicates the difference was statistically significant. 95% CI = 95% confidence interval, LNG-IUS = levonorgestrel intrauterine system, OR = odds ratio, PBO = placebo. According to the area under the curve in SUCRA (Fig. 5 ), the area under the curve of LNG-IUS after operation was the largest, indicating that its efficiency ranked first and its performance was the largest. Dienogest, leuprolide, gestrinone, mifepristone, danazol and placebo were in the next order. The surface under the cumulative ranking curve (SUCRA). Network meta-analysis mixed comparison results: Compared with placebo, dienogest (OR = 0.18, 95%CI = [0.09, 0.38]), leuprolide (OR = 0.29, 95%CI = [0.18, 0.47]), gestrinone (OR = 0.46, 95%CI = [0.26, 0.79]), mifepristone (OR = 0.45, 95%CI = [0.24, 0.86]), LNG-IUS (OR = 0.21, 95%CI = [0.12, 0.36]) was significantly lower than that of placebo, the difference was statistically significant. The use of dienogest after surgery was more effective than danazol (OR = 0.28, 95%CI = [0.09, 0.93]), gestrinone (OR = 0.40, 95%CI = [0.18, 0.93]) and mifepristone (OR = 0.41, 95%CI = [0.18, 0.94]). The difference was statistically significant. Compared with LNG-IUS implantation, danazol (OR = 3.15, 95%CI = [1.05, 9.45]), gestrinone (OR = 2.22, 95%CI = [1.13, 4.34]), mifepristone (OR = 2.20, 95%CI = [1.01, 4.77]) had a higher recurrence rate, and the difference was statistically significant. The pairwise comparisons of the remaining interventions were not statistically significant (Table 2 ). According to the area under the curve in SUCRA (Fig. 5 ), the area under the curve of dienogest after surgery was the smallest, and the recurrence rate of dienogest ranked the last with the greatest performance. LNG-IUS, leuprolide, mifepristone, gestrinone, danazol and placebo ranked higher than dienogest in order. Network meta-analysis mixed comparison results: Postoperative use of dienogest versus danazol (OR = 2.81, 95%CI = [1.27, 6.21]), gestrinone (OR = 2.60, 95%CI = [1.45, 4.66]), mifepristone (OR = 2.93, 95%CI = [1.65, 5.19]), placebo (OR = 3.30, 95%CI = [1.94, 5.62]), the pregnancy rate was significantly higher in patients with fertility requirements, and the difference was statistically significant. In addition, leuprolide was significantly associated with Dana (OR = 2.57, 95%CI = [1.28, 5.16]), gestrinone (OR = 2.38, 95%CI = [1.53, 3.71]), mifepristone (OR = 2.68, 95%CI = [1.60, 4.50]), placebo (OR = 3.02, 95%CI = [2.02, 4.53]), and the pregnancy rate of leuprolide was significantly higher. The differences between any pairwise comparisons of the remaining interventions were not statistically significant (Table 2 ). According to the area under the curve in SUCRA (Fig. 5 ), the area under the curve of dienogest was the largest, that is, it had the greatest ability to rank the first in the pregnancy rate of patients with fertility requirements, followed by leuprolide, gestrinone, mifepristone, danazol, and placebo in order. The results of network meta-analysis mixed comparison showed that postoperative application of leuprolide (OR = 2.83, 95%CI = [1.02, 7.88]) had a significantly higher incidence of adverse reactions than intrauterine LNG-IUS. Compared with placebo, after surgery, dinorgestrel (OR = 3.76, 95%CI = [1.24, 11.44]), leprerelin (OR = 11.79, 95%CI = [4.56, 30.46]), danazole (OR = 12.26, 95%CI = [2.87, 52.29]), pregnrienone (OR = 9.25, 95%CI = [3.26, 26.26]), mifepristone (OR = 6.37, 95%CI = [2.04, 19.88]) and LNG-IUS (OR = 4.17, 95%CI = [1.56, 11.16]) were higher than those of placebo, and the difference was statistically significant. The pairwise comparisons of the remaining interventions were not statistically significant (Table 2 ). According to the area under the curve in SUCRA (Fig. 5 ), the area under the curve of LNG-IUS after surgery was the smallest except for placebo, that is, the adverse reaction rate of LNG-IUS after surgery ranked last, except for placebo, and the performance was the largest. Dienogest, mifepristone, gestrinone, leuprolide and danazol ranked higher than LNG-IUS.

LNG-IUS is one of the intrauterine devices, which slowly releases levonorgestrel into the uterine cavity at a rate of 20 µg/day for 5 years, making the endometrium lose its sensitivity to estradiol, weakening the activity of the endometrium and inhibiting ectopic growth. Due to the thinning of the endometrium, the menstrual cycle is prolonged, the menstrual period is shortened, and the menstrual volume is reduced. At the same time, LNG-IUS can also prevent the production of prostaglandins and thromboxane by the endometrium, so it can reduce the spasmatic contraction of uterine smooth muscle and play a role in relieving dysmenorrhea. [ 51 ] Dienogest is a progestogen derived from a 19-nortestosterone source, but unlike other progestins of the same source, it has a cyanomethyl group instead of an acetylene group at position 17α. Therefore, Dienogestrel combines the advantages of a 19-nortestosterone derivative, such as a short half-life, potent endometrial inhibition, and high bioavailability, with the advantages of progesterone, such as antiandrogenic activity and moderate suppression of LH and FSH. [ 52 ] Dienogest activates the apoptotic pathway mainly by inhibiting the nuclear transcriptional signaling pathway of endometrium. In addition, dienogest reduces the activity of anti-apoptotic factors such as B lymphocytes (Bcl-2, Bcl-XL), and further promotes the apoptosis of ectopic endometrial cells. [ 53 ] Klipping showed that dienogest showed a strong endometrial transformation effect, with a baseline thickness of 10mm thinning to between 4.0 and 4.5mm after dienogest treatment, and different doses of dienogest (even 0.5 mg/d) had similar effects during the treatment. [ 54 ] Other traditional progestins require high doses to effectively treat EMT, and dienogest is 17 to 34 times more potent than other synthetic progestins in transforming the endometrium. One of the unsolved problems of fertility-preserving surgery in EMT is postoperative disease recurrence. Sesti showed no difference in recurrence rates between GnRH-a and placebo after surgery. [ 55 ] Park showed that patients in the dienogest group had significant improvements in pelvic pain VAS scores and EMT recurrence in the long-term observation of dienogest. [ 56 ] Similar to the network meta-analysis results, Takaesu conducted a comparison of the efficacy of dienogest and GnRH-a after fertilization-preserving surgery, and the results showed that compared with GnRH-a drugs, dienogest was more effective in preventing the recurrence of EMT. [ 57 ] LNG-IUS delivers levonorgestrel directly into the uterine cavity and can release the drug slowly and stably for a long time, which prevents the ectopic endometrium from synthesizing E- 2 receptors, so it has a strong ability to prevent the growth of ectopic endometrium. LNG-IUS can reduce the levels of sex hormones, Carbohydrate antigen-125, high-sensitivity C-reactive protein and vascular endothelial growth factor, effectively inhibit the inflammatory response, and play a positive role in preventing the recurrence of EMT. [ 58 ] The long-term treatment effect is good. While other anti-progestin drugs can inhibit the transcription, translation, protein synthesis and release of progesterone receptor gene during medication, and inhibit endometrial hyperplasia by binding to sex hormone receptors. However, the inhibitory effect of anti-progestins is significantly reduced after drug withdrawal, resulting in unsatisfactory long-term efficacy. In addition, it is worth discussing that complete surgical removal of EMT lesions is not always feasible because some implants may be very small or hidden. Removing as many lesions as possible during surgery has become a goal of surgeons. Giuseppe Vizzielli et al compared the visual detection rate of endometriotic lesions using near-infrared radiation imaging after intravenous injection of indocyanine green in laparoscopic/robot operating. They find out the use of near-infrared radiation imaging after intravenous injection of indocyanine green imaging technology was of value in identifying “occult” EMT, which could be a valid tool to complement the standard vision technique, in order to better identify and eradicate otherwise unseen EMT. In future, the use of this technique might allow the complete excision of endometriotic lesions without compromising surrounding healthy tissue, in addition to avoiding damage to ovaries and ureters, so preventing infertility and urological dysfunction. [ 59 ] Surgical removal of EMT lesions, release of tissue adhesion around the adnexa, and repeated pelvic irrigation can improve the pelvic microenvironment and promote the recovery of reproductive function. Lin Jinfang et al said that the pregnancy rate in the first year after EMT fertility-preserving surgery was 76%, indicating that the short term after surgery is a good time for pregnancy. [ 60 ] Therefore, it is necessary to give positive guidance to patients during the early postoperative period. A study of dienogest versus GnRH-a showed that menstruation returned about 50 days after completion of GnRH-a treatment. Even with continuous administration of dienogest, menstruation can resume on average 29 days after the completion of dienogest treatment, making it possible to achieve ovulation and pregnancy within a short period after treatment. [ 61 ] In this study, 44% of patients who wanted to conceive conceived naturally within 2 years, including 1 patient who ovulated and had an intrauterine pregnancy within a month after the end of dienogest treatment. In this study, the pregnancy rate of leuprolide ranked second. Hu Yan et al showed that the application of GnRH-a on the basis of laparoscopic treatment effectively improved the pregnancy rate and fertility rate of patients, and the short-term efficacy was obvious. [ 62 ] In addition, Bian et al also showed that for EMT patients with infertility, the application of GnRH-a after surgery had the most significant improvement in pregnancy rate within 1 year. [ 63 ] The main adverse events of mifepristone and gestrinone were nausea, vaginal bleeding, liver function damage, etc. The androgenic effect of danazol can cause hirsutism, acne, breast reduction, and irreversible liver damage, which greatly affects patient compliance. The common adverse reactions of GnRH-a include hot flashes, vaginal dryness, bone loss, etc. However, estrogen reverse addition therapy can reduce the occurrence of the above adverse reactions after 3 to 6 months of medication. The local effect of LNG-IUS has no significant effect on the body hormone levels and ovarian function, and greatly reduces the incidence of systemic adverse reactions. Abnormal uterine bleeding is the most common adverse effect of LNG-IUS, which is often manifested as endless irregular bleeding. The reason may be related to the dilatation of endometrial vessels, the thinning of the wall, the reduction of the supporting tissue around the tube, and the increase of vascular fragility and contraction weakness. [ 64 ] Dienogest is a well-tolerated agent that inhibits estradiol at a therapeutic window concentration of 30 to 50 pg/mL in EMT. This concentration range of estradiol levels is not sufficient to stimulate ectopic endometrial growth without causing perimenopausal symptoms and bone loss to achieve a reduction in adverse effects without compromising treatment efficacy, which makes the discontinuation rate of dienogest <5%, and therefore can be used for a long time. [ 65 ] Combined with the results of the current study, oral dienogest and intrauterine placement of LNG-IUS can effectively relieve symptoms, prevent recurrence, and have mild adverse reactions in patients without fertility requirements after fertile-preserving surgery. For patients with fertility requirements after surgery, dienogest has good efficacy and safety, and shows a higher pregnancy rate, which is a new choice.

This work was supported by the Suzhou Ninth People Hospital, Suzhou Ninth Hospital Affiliated to Soochow University (grant number: YK202230; YK202202).