The effectiveness of different down-regulating protocols on in vitro fertilization-embryo transfer in endometriosis: a meta-analysis

Endometriosis, a common clinical gynecological disease in women of childbearing age, refers to the presence of functional endometrial tissue (gland and stroma) in parts of the body outside the uterus. Although the disease has a benign manifestation in morphology, it has clinical behaviors characterized by similar malignant tumors, such as planting, invasion, distant metastasis, and recurrence. The main symptoms of endometriosis are lower abdominal pain and dysmenorrhea, sexual discomfort, and infertility. Studies have shown that about 30–50% of women with endometriosis have infertility, and about 20–50% of infertility patients have endometriosis [1]. Endometriosis with infertility is thought to be multifactorial and impairs fertility directly by destroying the normal anatomy of the fallopian tube and ovary or indirectly through inflammatory reaction and oxidative damage to degrade the quality of oocytes [2‐5]. Of course, male factors, such as sperm quality, are also important factors affecting women’s fertility [6]. With the development of assisted reproductive technology, IVF-ET has gradually become an important treatment for patients with infertility from endometriosis. Pituitary down-regulation is a key link in the IVF-ET process. Gonadotropin-releasing hormone agonist (GnRH-a) can play a competitive role in the pituitary gland and block its release of GnRH, thereby inhibiting the secretion of related hormones in the ovary and achieving the effect of pituitary down-regulation. In addition, GnRH-a can effectively prevent premature luteinization of follicles and improve the synchronization of follicular growth and development [7]. Furthermore, it can reduce the degree of inflammatory reaction, improve the pelvic microenvironment, and obtain high-quality eggs and embryos [7, 8].

With the continuous exploration of the down-regulating protocol, research into adopting different down-regulation protocols to implement IVF-ET-assisted pregnancy in infertility patients with endometriosis and improve the success rate already exists at home and abroad, but the clinical outcomes are still controversial. One study showed that the 3–6 month GnRH-a therapy might increase their clinical pregnancy rate by four times [9], but another study found that ultra-long protocol and long protocol provide no significant difference in improving clinical outcomes in patients with endometriosis [10]. Therefore, this paper systematically evaluated, through meta-analysis, the clinical studies of the effects of GnRH-a ultra-long, long-term, and short-term protocols in the treatment of IVF-ET in infertile patients with endometriosis. The aim was to screen out the optimal down-regulation protocol for patients with infertility from endometriosis.

This study indicated that compared with the GnRH-a long protocol, the GnRH-a ultra-long protocol improved the clinical pregnancy rate or implantation rate of infertile patients, which might improve the clinical outcomes of patients with endometriosis. A previous study reported that GnRH-a pretreatment can improve the microenvironment in the ovary and the quality of oocytes [7]. In addition, it could enhance the expression of endometrial αγβ3 integrin as a hallmark of endometrial receptivity, increase the number of pinopodes that are important for embryo transfer [32], and increase the apoptotic activity of endometrial cells in patients with endometriosis and repair its reduced sensitivity thus increasing its implantation rate [33]. Pretreatment with GnRH-a could also reduce the concentration of interleukin-1, tumor necrosis factor, and nitric oxide that are toxic to embryos in the peritoneal fluid [8, 34] it could make pituitary cells unresponsive to endogenous GnRH-a to achieve desensitization, reduce the secretion of FSH and luteinizing hormone (LH), inhibit ovarian activity, reduce the level of estradiol, and promote atrophy of ectopic foci to favor embryo implantation [35]. However, the dose and time of the drug were extended, and the ultra-long protocol had too deep an inhibition on the pituitary gland, which led to a decrease of ovarian reactivity to gonadotropin. To solve this problem, studies have shown that the level of serum estradiol (E2) after GnRH-a down-regulation was measured to determine whether the patient’s ovaries recovered responsiveness, and gonadotropin was initiated when the E2 value began to rise. Compared with the conventional ultra-long protocol, it could significantly reduce the dosage of gonadotropin used without affecting pregnancy outcomes [36]. This meta-analysis found that among RCTs comparing the GnRH-a long protocol and the GnRH-a ultra-long protocol, the GnRH-a ultra-long protocol mainly improved the clinical pregnancy rate. However, among non-RCTs, the ultra-long protocol mainly improved the implantation rate. The observed differences between the results within RCTs and non-RCTs demonstrate that reliance on neither should be absolute; this difference might be due to analytical methods and confounding factors. Similar to the results of RCTs combining with non-RCTs in this study, a recent Cochrane review (8 RCTs included) indicated that the long-term GnRH agonist therapy had uncertain impacts on the pregnancy outcomes [37]. However, it contradicted with the results of RCTs (7 RCTs included) in this study, which was due to different inclusion criteria and different GnRH-a down-regulation protocols (minimum 3 months versus 1 to 3 months mostly) on infertility patients with endometriosis.

Each of these two types of study designs had strengths and limitations. Although randomized controlled trials could reduce or eliminate the imbalance between the treatment and control groups by random, double-blind, or other methods, RCT studies had strict entry criteria, required greater adherence, lacked universality, and were more expensive. Therefore, the sample sizes of the RCT studies included in this study were small. However, observational studies might be more universal and less affected by selection bias, allowing for larger sample sizes to be collected. As such, both types of data were useful in assessing the effectiveness of various protocols in the treatment of IVF-ET among infertility patients with endometriosis.

The study also further analyzed the clinical pregnancy rate based on the ASRM classification, and the stage I and II were classified according to the ASRM criteria that belonged to a domaine of surgery. The findings were as follows. In RCTs, the GnRH-a ultra-long protocol significantly improved the clinical pregnancy rate among endometriosis patients at stages III–IV compared with the long-term GnRH-a protocol but had no effect on patients with endometriosis at stages I–II. This might be because, compared with patients at stages I–II, changes in pelvic anatomy, the level of inflammatory response in the pelvic cavity, and the degree of the local micro-environment imbalance in the endometrium were all heavier in patients in phases III–IV; therefore, it took a long time and a large dose of down-regulation to inhibit the inflammatory response and improve the pelvic microenvironment. However, the ultra-long protocol also had some potential risks, such as hyper-inhibition of the pituitary and long-term low levels of endogenous FSH, which might lead to fewer numbers of follicles and smaller follicular diameter, which might increase the amount of gonadotropin and reduce the number of retrieved oocytes [38]. For endometriosis patients at stages I–II, the conventional long protocol had been able to achieve the same clinical outcome as the ultra-long protocol; thus, the long protocol might be a cost-effective controlled ovarian hyperstimulation (COH) protocol for patients with stages I–II endometriosis. For patients with severe endometriosis stage III-IV, the GnRH-a ultra-long protocol could achieve better pregnancy outcomes, and this was similar to previous literature reports [23, 39]. Although it has been confirmed by a lot of literatures which suggested that the GnRH-a ultra-long protocol could improve pregnancy outcomes with stages III-IV endometriosis, most of the published RCT or non-RCT studies were performed in a small number of patients. Thus, the systematic meta-analysis of RCT with larger number of patients in this study would be required to obtain more reliable results with higher confidence levels.

In the non-RCTs, two studies (n = 150) found that the GnRH-a ultra-long protocol improved the fertilization rate compared to the GnRH-a short protocol, which was similar to the results in previous literature [23]. Analysis of three other studies, including 328 patients, showed that the ultra-long protocol could increase the number of retrieved oocytes. In addition, five studies indicated that the ultra-long protocol could also increase the clinical pregnancy rate, but the numbers of stimulation days were longer. Because the five included studies were retrospective cohort studies with smaller sample sizes, this research conclusion only represented current evidence, and the reliability of the result requires prospective and randomized studies with a larger sample size.

The GnRH-a ultra-long protocol, compared to the long protocol, and the results of the RCT studies and the non-RCT studies were all consistent in terms of fertilization rate, basal FSH, and the number of retrieved oocytes, none of which were statistically significant. The ultra-long and the long protocol had no effects on these three aspects. In this study, female age (Additional file 5: Figure S1), BMI (Additional file 6: Figure S2), antral follicle count (AFC) (Additional file 7: Figure S3), basal E2 levels (Additional file 8: Figure S4), and basal LH levels (Additional file 9: Figure S5) were also not significantly different between the GnRH-a ultra-long protocol and the GnRH-a long/short protocol in RCTs and/or non-RCTs. Serum CA125 level was correlated with the severity of endometriosis and treatment effect. In this data, there was statistically significant difference in CA125 levels before down-regulation between the GnRH-a ultra-long protocol and the GnRH-a long protocol in non-RCTs. However, there was no significant difference between the two groups after down-regulation (P >0.05) (Additional file 10: Figure S6). This result might be due to the small sample size and the large inter-group heterogeneity.

Nevertheless, our meta-analysis still had several shortcomings. Even if strict inclusion and exclusion criteria were used, the conclusions drawn from this data were still subject to the limitations of the original studies themselves. Only seven RCTs were included in our meta-analysis, three of which were low-quality due to unclear methods of randomized allocation, and the remaining 14 articles were cohort studies. Although we performed subgroup analyses based on different types of research, heterogeneity was still quite apparent between studies because of the GnRH-a dose, the treatment time, and the starting time of gonadotropin, which were not completely the same in each protocol. Therefore, large-sample, multicenter, randomized controlled trials are still needed to obtain more reliable results.

This systematic meta-analysis firstly reports the GnRH-a ultra-long protocol can improve the clinical pregnancy rate of infertile patients in RCT studies, especially in patients with stages III–IV endometriosis. However, the results of subgroup analysis suggest that different down-regulation protocols provide no significant differences in improving clinical outcomes in patients with endometriosis in the non-RCT studies. Although it is generally believed that the results of RCT are more reliable, the conclusions of non-RCT studies cannot be easily neglect, which let us drawn conclusions more cautious. Furthermore, the drugs produced by different manufacturers, dose of gonadotropin, duration of ovarian stimulation, starting time of gonadotropin and the ethnic groups were not completely the same in each study. The non-RCT studies had defects in the control of several confounding factors mentioned above, which might lead to bias, and these factors might be precisely the key to affecting the conclusions.