Background
Assisted reproductive technology is one of the therapeutic options offered for managing endometriosis-associated infertility. Yet, published data on assisted reproductive technology outcome in women affected by endometriosis are conflicting and the determinant factors for pregnancy chances unclear.
Objective
We sought to evaluate assisted reproductive technology outcomes in a series of 359 endometriosis patients, to identify prognostic factors and determine if there is an impact of the endometriosis phenotype.
Study Design
This was a retrospective observational cohort study, including 359 consecutive endometriosis patients undergoing in vitro fertilization or intracytoplasmic sperm injection, from June 2005 through February 2013 at a university hospital. Endometriotic lesions were classified into 3 phenotypes–superficial peritoneal endometriosis, endometrioma, or deep infiltrating endometriosis–based on imaging criteria (transvaginal ultrasound, magnetic resonance imaging); histological proof confirmed the diagnosis in women with a history of surgery for endometriosis. Main outcome measures were clinical pregnancy rates and live birth rates per cycle and per embryo transfer. Prognostic factors of assisted reproductive technology outcome were identified by comparing women who became pregnant and those who did not, using univariate and adjusted multiple logistic regression models.
Results
In all, 359 endometriosis patients underwent 720 assisted reproductive technology cycles. In all, 158 (44%) patients became pregnant, and 114 (31.8%) had a live birth. The clinical pregnancy rate and the live birth rate per embryo transfer were 36.4% and 22.8%, respectively. The endometriosis phenotype (superficial endometriosis, endometrioma, or deep infiltrating endometriosis) had no impact on assisted reproductive technology outcomes. After multivariate analysis, history of surgery for endometriosis (odds ratio, 0.14; 95% confidence ratio, 0.06–0.38) or past surgery for endometrioma (odds ratio, 0.39; 95% confidence ratio, 0.18–0.84) were independent factors associated with lower pregnancy rates. Anti-müllerian hormone levels <2 ng/mL (odds ratio, 0.51; 95% confidence ratio, 0.28–0.91) and antral follicle count <10 (odds ratio, 0.27; 95% confidence ratio, 0.14–0.53) were also associated with negative assisted reproductive technology outcomes.
Conclusion
The endometriosis phenotype seems to have no impact on assisted reproductive technology results. An altered ovarian reserve and a previous surgery for endometriosis and/or endometrioma are associated with decreased pregnancy rates.
Introduction
Endometriosis is a benign chronic gynecological disease, defined as the presence of endometrial tissue outside the uterine cavity. Prevalence has been estimated to reach 10-15% of reproductive-aged women, and 25-50% of infertile women. It is widely accepted that endometriosis alters fertility, but the exact pathophysiology of this effect remains unclear. Current views suggest multifactorial mechanisms, including inflammatory changes in peritoneal fluid altering sperm-oocyte interaction, reduced functional ovarian tissue, and hampered endometrial receptivity.
Endometriosis is heterogeneous in nature with lesions having 3 distinct phenotypes: (1) superficial peritoneal endometriosis (SUP), (2) ovarian endometrioma (OMA), and (3) deeply infiltrating endometriosis (DIE). Moreover, endometriosis is frequently associated with adenomyosis, which has detrimental effects on fertility of its own.
Assisted reproductive technologies (ART) are commonly offered for managing endometriosis-related infertility. ART results, however, vary according to reports, with some showing identical outcome as in endometriosis-free counterparts, and others describing lower pregnancy rates. In this context of discordant ART results in endometriosis, there is no consensus about the possible impact of the endometriosis phenotype on ART outcome. A recent meta-analysis from Rossi and Prefumo, comparing 980 endometriosis patients to controls, pointed at decreased clinical pregnancy rates and live birth rates (LBR) in stages III-IV endometriosis. Conversely, Barbosa et al reviewing 3930 endometriosis cases showed no difference in outcome according to disease stage. These discrepancies could result from important heterogeneity in the published series including, first, uncertainty about the exact endometriosis phenotype. Previous ART results published in endometriotic patients relied on the revised American Society for Reproductive Medicine (ASRM) classification of the disease, which does not take into account the phenotype of lesions, therefore, regrouping different forms of the disease under a same ASRM stage. Hence, precise phenotyping of lesions is needed for clarifying the impact of endometriosis on ART outcome. Secondly, confounding factors such as history of surgery for OMA and/or associated adenomyosis, which likely impacts on ART results, are rarely taken into account in retrospective studies.
To assess the impact of lesion phenotypes on ART outcome, we conducted an observational cohort study on a consecutive series of endometriosis patients undergoing ART whose lesions were prospectively phenotyped. The study objective was to analyze the possible role of endometriosis lesion phenotype (SUP/OMA/DIE) and to identify other potential prognostic factors of ART outcomes.
Materials and Methods
Study protocol
The local ethics committee (Comité Consultatif de Protection des Personnes dans la Recherche Biomédicale) of our institution approved the study protocol. The study population consisted of a continuous series of 359 phenotyped endometriosis patients who underwent in vitro fertilization (IVF)/intracytoplasmic sperm injection (ICSI) treatment in a tertiary care center, from June 2005 through February 2013. Endometriotic lesions were classified according to their phenotype as SUP, OMA, or DIE. All patients underwent an appropriate workup to precisely diagnose and stage endometriosis. For DIE and OMA phenotypes, diagnosis and staging of endometriosis was based on previously published imaging criteria using transvaginal sonography (TVS) or magnetic resonance imaging (MRI) (359 [100%] patients had TVS; 210 [58.5%] also had MRI). In addition, in women who had history of surgery for endometriosis, the diagnosis was confirmed histologically. In SUP, pre-ART imaging workup showing neither OMA nor DIE lesions, the diagnosis was solely based on previous histologically proved SUP lesion. These phenotypes being frequently combined, patients were assigned to the group corresponding to the most severe lesion, according to a previously published classification, going from the least to most severe: SUP, OMA, DIE.
All patients of OMA and DIE groups had superficial lesions. Women in the OMA group could not have DIE lesions, whereas some patients in the DIE group had associated OMA lesions. In case of DIE, the severity was assessed on the basis of 2 parameters : the number and anatomic location of DIE lesions. In cases of multiple DIE sites, patients were classified according to the worst finding (least to most severe: uterosacral ligament(s), vagina, bladder, intestine, and ureter ). Associated adenomyosis was diagnosed using imaging criteria based on TVS and MRI.
General characteristics
The study analysis used a prospectively managed database. For each patient, personal history data and results of fertility investigations were collected before ART treatment. The following data were recorded: age, height, weight, body mass index, parity, gravidity, duration of infertility, results of hysterosalpingography, cycle day-3 levels (follicle-stimulating hormone [FSH], luteinizing hormone, and estradiol), anti-müllerian hormone (AMH), antral follicle count (AFC) score, and semen analysis as per World Health Organization manual. For each patient, history of surgery for endometriosis was recorded. History of surgery was defined as excision of superficial lesions, deep lesions excision, bowel resection, or ovarian cystectomy. Women were then classified into 2 groups: previous surgery for OMA if they had a history of ovarian cystectomy with or without resection of SUP and/or DIE lesions, and previous surgery for endometriosis without OMA if they had a history of surgery for endometriosis (SUP and/or DIE) without associated ovarian cystectomy.
Controlled ovarian stimulation regimen
Patients were stimulated either by a long gonadotropin-releasing hormone (GnRH) agonist, a short agonist, or an antagonist protocol. In long GnRH-agonist protocol, ovarian stimulation was started following pituitary desensitization with doses of gonadotropins ranging from 150-450 IU/d, depending on individual patient characteristics. In case of antagonist protocol (n = 167/720 cycles, 23.2%), GnRH antagonist was arbitrarily initiated on controlled ovarian stimulation day 6. Both long and antagonist protocols were initiated following timely use of oral contraceptive pill (OCP) (ethinyl estradiol 0.03 mg; levonorgestrel 0.125 mg). In all cases, controlled ovarian stimulation was initiated 6 days after discontinuing OCP, using a mix of FSH and human menopausal gonadotropin preparation for palliating at the luteinizing hormone suppressing effects of OCP. Oocyte pickup was canceled in case of poor response, defined by the presence of <3 follicles measuring ≥17 mm and/or estradiol levels <750 pg/mL at the time of the triggering decision. Transvaginal oocyte retrieval was scheduled 36 hours after human chorionic gonadotropin administration and embryo transfer was performed 2-3 days later. The luteal phase was supported by vaginal administration of micronized progesterone (600 mg/d) from the day of oocyte retrieval to the day of the first ultrasound at the fifth gestational week. Pregnancies were diagnosed by increasing concentrations of serum human chorionic gonadotropin, 14 days after oocyte retrieval. ART results were assessed by analyzing the following outcomes: (1) clinical pregnancies, (2) live births, and (3) early miscarriages. These outcome parameters were studied in the whole population, and according to endometriosis phenotypes. Patient characteristics were then compared between women who conceived and those who did not, looking for prognostic factors affecting ART outcome.
Statistical analysis
Data were analyzed using software (SPSS, Version 12.0; IBM Corp, Armonk, NY). Continuous data were presented as mean and SD; categorical data, as number and percentages. Patient characteristics and ART outcome parameters were compared according to endometriosis phenotype, using a Pearson χ 2 test for qualitative variables, and Kruskal-Wallis test for quantitative variables. If statistical significance was reached ( P < .05), variables were compared 2-by-2 using Dunn nonparametric comparison post-hoc test for quantitative variables, and Pearson χ 2 posttest for qualitative variables; P < .05 was then considered to be significant. The Kaplan-Meier method was used to estimate the cumulative pregnancy rate.
When analyzing patients who became pregnant and those who did not, we used a Pearson χ 2 test for qualitative and Student t test for quantitative variables. Subsequently, the variables associated with pregnancy at the threshold of P < .15 in univariate analysis were tested in a multiple logistic regression model taking into account several interactions between variables. When 2 variables were highly correlated, we introduced in the model only one of them and suppressed the other, as for the number of DIE lesions ≥2 and intestinal DIE, the latter having been suppressed. After studying the interactions between all variables, we held back 2 significant interactions (previous surgery for endometriosis without OMA and AFC; previous surgery for OMA and number of DIE lesions ≥2) and introduced the terms of these interactions in the model, as well as all variables selected as previously described. Odds ratios (OR) and their 95% confidence intervals (CI) were calculated from the model’s coefficients and their SD. The final model was built using all selected significant variables in the multivariate analysis.
Materials and Methods
Study protocol
The local ethics committee (Comité Consultatif de Protection des Personnes dans la Recherche Biomédicale) of our institution approved the study protocol. The study population consisted of a continuous series of 359 phenotyped endometriosis patients who underwent in vitro fertilization (IVF)/intracytoplasmic sperm injection (ICSI) treatment in a tertiary care center, from June 2005 through February 2013. Endometriotic lesions were classified according to their phenotype as SUP, OMA, or DIE. All patients underwent an appropriate workup to precisely diagnose and stage endometriosis. For DIE and OMA phenotypes, diagnosis and staging of endometriosis was based on previously published imaging criteria using transvaginal sonography (TVS) or magnetic resonance imaging (MRI) (359 [100%] patients had TVS; 210 [58.5%] also had MRI). In addition, in women who had history of surgery for endometriosis, the diagnosis was confirmed histologically. In SUP, pre-ART imaging workup showing neither OMA nor DIE lesions, the diagnosis was solely based on previous histologically proved SUP lesion. These phenotypes being frequently combined, patients were assigned to the group corresponding to the most severe lesion, according to a previously published classification, going from the least to most severe: SUP, OMA, DIE.
All patients of OMA and DIE groups had superficial lesions. Women in the OMA group could not have DIE lesions, whereas some patients in the DIE group had associated OMA lesions. In case of DIE, the severity was assessed on the basis of 2 parameters : the number and anatomic location of DIE lesions. In cases of multiple DIE sites, patients were classified according to the worst finding (least to most severe: uterosacral ligament(s), vagina, bladder, intestine, and ureter ). Associated adenomyosis was diagnosed using imaging criteria based on TVS and MRI.
General characteristics
The study analysis used a prospectively managed database. For each patient, personal history data and results of fertility investigations were collected before ART treatment. The following data were recorded: age, height, weight, body mass index, parity, gravidity, duration of infertility, results of hysterosalpingography, cycle day-3 levels (follicle-stimulating hormone [FSH], luteinizing hormone, and estradiol), anti-müllerian hormone (AMH), antral follicle count (AFC) score, and semen analysis as per World Health Organization manual. For each patient, history of surgery for endometriosis was recorded. History of surgery was defined as excision of superficial lesions, deep lesions excision, bowel resection, or ovarian cystectomy. Women were then classified into 2 groups: previous surgery for OMA if they had a history of ovarian cystectomy with or without resection of SUP and/or DIE lesions, and previous surgery for endometriosis without OMA if they had a history of surgery for endometriosis (SUP and/or DIE) without associated ovarian cystectomy.
Controlled ovarian stimulation regimen
Patients were stimulated either by a long gonadotropin-releasing hormone (GnRH) agonist, a short agonist, or an antagonist protocol. In long GnRH-agonist protocol, ovarian stimulation was started following pituitary desensitization with doses of gonadotropins ranging from 150-450 IU/d, depending on individual patient characteristics. In case of antagonist protocol (n = 167/720 cycles, 23.2%), GnRH antagonist was arbitrarily initiated on controlled ovarian stimulation day 6. Both long and antagonist protocols were initiated following timely use of oral contraceptive pill (OCP) (ethinyl estradiol 0.03 mg; levonorgestrel 0.125 mg). In all cases, controlled ovarian stimulation was initiated 6 days after discontinuing OCP, using a mix of FSH and human menopausal gonadotropin preparation for palliating at the luteinizing hormone suppressing effects of OCP. Oocyte pickup was canceled in case of poor response, defined by the presence of <3 follicles measuring ≥17 mm and/or estradiol levels <750 pg/mL at the time of the triggering decision. Transvaginal oocyte retrieval was scheduled 36 hours after human chorionic gonadotropin administration and embryo transfer was performed 2-3 days later. The luteal phase was supported by vaginal administration of micronized progesterone (600 mg/d) from the day of oocyte retrieval to the day of the first ultrasound at the fifth gestational week. Pregnancies were diagnosed by increasing concentrations of serum human chorionic gonadotropin, 14 days after oocyte retrieval. ART results were assessed by analyzing the following outcomes: (1) clinical pregnancies, (2) live births, and (3) early miscarriages. These outcome parameters were studied in the whole population, and according to endometriosis phenotypes. Patient characteristics were then compared between women who conceived and those who did not, looking for prognostic factors affecting ART outcome.
Statistical analysis
Data were analyzed using software (SPSS, Version 12.0; IBM Corp, Armonk, NY). Continuous data were presented as mean and SD; categorical data, as number and percentages. Patient characteristics and ART outcome parameters were compared according to endometriosis phenotype, using a Pearson χ 2 test for qualitative variables, and Kruskal-Wallis test for quantitative variables. If statistical significance was reached ( P < .05), variables were compared 2-by-2 using Dunn nonparametric comparison post-hoc test for quantitative variables, and Pearson χ 2 posttest for qualitative variables; P < .05 was then considered to be significant. The Kaplan-Meier method was used to estimate the cumulative pregnancy rate.
When analyzing patients who became pregnant and those who did not, we used a Pearson χ 2 test for qualitative and Student t test for quantitative variables. Subsequently, the variables associated with pregnancy at the threshold of P < .15 in univariate analysis were tested in a multiple logistic regression model taking into account several interactions between variables. When 2 variables were highly correlated, we introduced in the model only one of them and suppressed the other, as for the number of DIE lesions ≥2 and intestinal DIE, the latter having been suppressed. After studying the interactions between all variables, we held back 2 significant interactions (previous surgery for endometriosis without OMA and AFC; previous surgery for OMA and number of DIE lesions ≥2) and introduced the terms of these interactions in the model, as well as all variables selected as previously described. Odds ratios (OR) and their 95% confidence intervals (CI) were calculated from the model’s coefficients and their SD. The final model was built using all selected significant variables in the multivariate analysis.
Results
Study population
From June 2005 through February 2013, 359 phenotyped endometriosis patients underwent 720 ART cycles at our tertiary care center. Demographic data and clinical characteristics of the study population are summarized in Table 1 . The mean age of the population was 33.4 ± 4 years, and the mean duration of infertility was 4 ± 2.2 (range 1-12) years. In all, 277 (77.2%) patients had primary infertility. Mean AMH level was 2.8 ± 2.2 (range 0.1-15) ng/mL, and mean AFC was 11 ± 6.7 (range 2-56).
| Characteristics | Values |
|---|---|
| Age, y | 33.4 ± 4.0 |
| Body mass index, kg/m 2 | 22.5 ± 3.5 |
| Duration of prior infertility, y | 4.0 ± 2.2 |
| Gravidity | |
| 0 | 262 (73%) |
| 1 | 69 (19.2%) |
| 2 | 17 (4.7%) |
| ≥3 | 11 (3.1%) |
| Parity | |
| 0 | 318 (88.6%) |
| 1 | 35 (9.7%) |
| ≥2 | 6 (1.7%) |
| Type of infertility | |
| Primary | 277 (77.2%) |
| Secondary | 82 (22.8%) |
| Associated male factor | 67 (18.7%) |
| Associated tubal factor | 38 (10.6%) |
| Endometriosis phenotype | |
| SUP | 49 (13.6%) |
| OMA | 98 (27.3%) |
| DIE | 212 (59.1%) |
| Without associated OMA lesions | 69 (32.5%) |
| With associated OMA lesions | 143 (67.5%) |
| Previous surgery for endometriosis | 282 (78.6%) |
| No. of prior surgeries | 1.3 ± 0.7 |
| Previous surgery for OMA | 170 (47.4%) |
| No. of prior surgeries | 1.2 ± 0.6 |
| Right | 47 (27.7%) |
| Left | 66 (38.8%) |
| Bilateral | 57 (33.5%) |
| Complete surgical exeresis of endometriotic lesions | 83 (23.1%) |
| Ovarian reserve: | |
| Day-3 FSH, IU/L | 7.7 ± 4.2 |
| Day-3 LH, IU/L | 5.3 ± 2.8 |
| Day-3 estradiol, pg/mL | 45.0 ± 26.3 |
| AFC | 11.0 ± 6.7 |
| AMH, ng/mL | 2.8 ± 2.2 |
| ART outcomes | Numerator/denominator (%) |
| ART cycles | 720 |
| 1 | 359 |
| 2 | 220 |
| 3 | 103 |
| 4 | 38 |
| Embryo transfers | 500 |
| Cancellation rate | 220/720 (30.6%) |
| Pregnancies | 158 (44%) |
| Live births | 114 (31.8%) |
| Clinical pregnancy rate per cycle | 182/720 (25.3%) |
| Clinical pregnancy rate per embryo transfer | 182/500 (36.4%) |
| Implantation rate a | 208/918 (22.7%) |
| Abortion rate b | 68/182 (37.4%) |
| Live birth rate per cycle | 114/720 (15.8%) |
| Live birth rate per embryo transfer | 114/500 (22.8%) |
a No. of gestational sacs/no. of embryos transferred
Endometriosis phenotype
The endometriosis phenotype was as follows: SUP, 49 (13.6%) patients; OMA, 98 (27.3%) patients; and DIE, 212 (59.1%) patients. In 143 (67.5%) cases, DIE was associated with OMA lesions. The patient distribution according to their worst DIE lesion was as follows: uterosacral ligament(s), 67 (31.6%) patients; vagina, 10 (4.7%) patients; bladder, 11 (5.2%) patients; intestine, 118 (55.7%) patients; and ureter, 6 (2.8%) patients. The mean number of DIE nodules per patient was 2.1 ± 0.9 (range 0-4).
In all, 282 (78.6%) women had history of surgery for endometriosis, with a mean number of previous surgeries of 1.3 ± 0.7 (range 0-6). Of them, 170 (47.4%) had surgery for OMA. Complete exeresis of the lesions was achieved for 83 (23.1%) patients. Moreover, 145 (40.4%) women of the study population had associated adenomyosis.
ART outcomes
In total, 720 ART cycles associated with 500 embryo transfers were analyzed. ART outcome in the general population are shown in Table 1 . Overall, 158 (44%) women became pregnant and 114 (31.8%) had a live birth. Clinical pregnancy rate and LBR per embryo transfer were 36.4% and 22.8%, respectively. Cumulative pregnancy rates were 18.2% after 1 ICSI-IVF cycle, 36.4% after 2 ICSI-IVF cycles, 50.2% after 3 ICSI-IVF cycles, and 65.8% after 4 ICSI-IVF cycles. Cancellation rate was 30.6%: 190 (86.4%) cycles were cancelled for poor response, 20 (9.1%) for absence of oocyte or fertilization failure, and 10 (4.5%) for poor embryo quality.
Patient characteristics and IVF results according to the endometriosis phenotype are presented in Table 2 . All patients in the SUP group had a history of surgery, compared to the OMA (80.6%) and DIE (72.6%) groups ( P < .001). DIE patients had a significantly lower AFC (11 ± 6.4) than SUP patients (14 ± 8) ( P = .013). The prevalence of associated adenomyosis was higher in the DIE (56.6%) group, than in the SUP (14.3%) and OMA (18.4%) groups ( P < .001). Clinical pregnancy rates and LBR did not differ among the 3 groups. Cancellation rate was higher in DIE (33.6%) patients, compared to SUP (18.9%) and OMA (29.5%) patients ( P = .018). Miscarriage rate was higher in the SUP (55.2%) group, than in the OMA (40%) and DIE (30.6%) groups ( P = .049).
