Objective
We sought to assess the association between polycystic ovary syndrome (PCOS) and ectopic pregnancy after in vitro fertilization–embryo transfer (ET).
Study Design
In this retrospective cohort study, we included 5339 women who had clinical pregnancies after in vitro fertilization treatment (PCOS, 205 women; non-PCOS, 5134 women) at Nanjing Medical University (China) between 2007 and 2011. Fresh and cryo-thawed ET cycles were analyzed respectively. The primary outcome measure was the occurrence of ectopic pregnancy. Multivariate logistic regression analysis was used to adjust for important confounders.
Results
In fresh ET cycles of women who were undergoing controlled ovarian hyperstimulation (COH; n = 3303), women with PCOS had 3.06 times higher risk of ectopic pregnancy compared with those without PCOS (7.0% vs 2.4%; adjusted odds ratio [aOR], 3.06; 95% confidence interval [CI], 1.34–6.96). In the stratified analysis, for women without PCOS, the high estradiol group (>4085 pg/mL) had higher ectopic pregnancy rates compared with the low estradiol group (≤4085 pg/mL; 3.4% vs 2.0%; aOR, 1.99; 95% CI, 1.19–3.35); however, for women with PCOS, both high and low estradiol groups had high ectopic pregnancy rates (5.6% vs 7.7%; aOR, 0.92; 95% CI, 0.15–5.67). In cryo-thawed ET cycles without COH (n = 2036), the ectopic rates between women with and without PCOS were similar (2.2% vs 2.0%; aOR, 0.94; 95% CI, 0.22–4.07).
Conclusion
PCOS was associated with an increased risk of ectopic pregnancy after COH in fresh ET cycles, but not in cryo-thawed ET cycles. A possible explanation is that, compared with women without PCOS, women with PCOS appear to hold a lower threshold of hyperphysiologic estradiol level that triggers the occurrence of ectopic pregnancy after COH.
Ectopic pregnancy (EP) is a known complication of in vitro fertilization (IVF). The reported incidence of EPs after IVF varies from 2.1-8.6% of all clinical pregnancies, which is higher than the estimated EP rate of 1.97 per 100 pregnancies among the general population in the United States and the 1.0 per 100 pregnancies incidence in China. Abnormal estradiol and/or progesterone milieu has been proposed as a risk factor for EP ; during IVF treatment, hyperphysiologic hormonal levels, which are associated with controlled ovarian hyperstimulation (COH), have been linked to an increased risk of EP.
Polycystic ovary syndrome (PCOS) is a common disorder that is associated with infertility in women who undergo IVF treatment and is characterized by ovulatory dysfunction and endocrine disorders such as hyperandrogenism. In IVF cycles, women with PCOS are highly sensitive to medication during COH and exhibit possibly higher estradiol and progesterone levels on the day of human chorionic gonadotropin (hCG) administration compared with women without PCOS, which suggests that they might be at an increased risk of EP.
Abundant literature has reported an increased risk of pregnancy complications in women with PCOS after both natural conception and assisted reproductive technologies, which include miscarriage, preeclampsia, very preterm birth, gestational diabetes mellitus, meconium aspiration, and large for gestational age independent of assisted reproductive technologies. However, with regard to EP, the clear association between PCOS and EP after IVF has not been reported yet, although some studies mentioned EP rates of women with PCOS who conceived after different assisted reproductive technologies treatment modalities. The reported rates are inconsistent and vary from 0-10% because of different treatment and grouping methods that have been used.
The purpose of this study was to analyze the possible association between PCOS and EP among clinical pregnancies that are achieved by fresh and cryo-thawed IVF cycles. If the association exists, we also aimed to assess its possible cause.
Materials and Methods
This study was a retrospective cohort analysis of women who achieved clinical pregnancies after IVF-ET in the Clinical Center for Reproductive Medicine, First Affiliated Hospital of Nanjing Medical University between January 2007 and December 2011. Data for individuals who started IVF treatment in our clinical center have been captured routinely in the Database of Clinical Center of Reproductive Medicine since 2006. The database includes the detailed information on patients’ demographic and baseline characteristics, medical history, infertility-associated disorders such as PCOS, data of IVF procedure, and follow-up pregnancy outcomes. With access to the database with a unique treatment identification number for each patient, trained nurses collected the patient’s medical and treatment information continuously along with the IVF procedure, until each embryo transfer (ET) procedure ended with a follow-up pregnancy outcome.
PCOS was diagnosed based on the Rotterdam criteria, which means the fulfilment of 2 of the 3 diagnosis criteria (presence of oligoovulation/anovulation, polycystic ovaries, and excess adrogen activity) and excluding other diseases that would represent these.
During the study period, 15,924 ET cycles were performed. The overall implantation rate, clinical pregnancy rate, and live birth rate were 27.9% (8263/29,638), 41.9% (6675/15,924) and 35.2% (5613/15,924) per ET, respectively. Of these ET cycles, 6675 resulted in pregnancies. To ensure no duplication of data from the same patient who underwent multiple treatment cycles, only cycles that resulted in the first clinical pregnancy during the study period were included. Because embryo implantation potential affected EP occurrence, pregnancies that resulted from donated oocytes or sperm were excluded from the analysis. Moreover, cycles that involved preimplantation genetic diagnosis or in vitro maturation were excluded. All patients who were included were Chinese. Fresh and cryo-thawed ET cycles were analyzed separately. For fresh ET cycles, to avoid interference by cryopreservation, cycles with cryopreserved oocytes or sperm were excluded. Additionally, to eliminate the bias caused by treatment, natural cycles (n = 58) and mild stimulation cycles (n = 166) were excluded from the fresh ET cycles. Thus, the final study population consisted of 5339 women who achieved clinical pregnancies. These women were grouped by diagnosis of PCOS into PCOS and non-PCOS cohorts in fresh and frozen-thawed cycles respectively.
The primary outcome measure was the occurrence of EP, which was diagnosed by ultrasound examination or laparoscopic surgery visualizing ≥1 gestational sacs outside the uterus or by abnormally increasing serum hCG level without sonographic visualization and the absence of chorionic villi inside the uterus after uterine curettage, which was treated by methotrexate. A clinical pregnancy was defined as the ultrasound observation of ≥1 gestational sac at 6 weeks after ET or diagnosis of an EP. A clinical intrauterine pregnancy was diagnosed as a pregnancy with ≥1 gestational sacs detected within the uterus cavity. Heterotopic pregnancy was defined when fetal heart beat was found at both intrauterine and ectopic sites. All EPs were followed until a urine hCG test was negative after treatment. In our study, either ectopic or heterotopic pregnancy was categorized as EP. The EP rate was calculated by dividing EPs by the total number of clinical pregnancies (sum of ectopic and clinical intrauterine pregnancies).
In fresh ET cycles, all cycles were treated with COH, under which 3 standard regimens applied: gonadotropin – releasing hormone – agonist long protocol (73%), gonadotropin – releasing hormone – agonist short protocol (20%), or gonadotropin – releasing hormone – antagonist protocol (7%). In frozen – thawed ET cycles, COH was not performed; instead, 3 methods of endometrial preparation were carried out that included natural (44%), artificial (19%), and mild stimulation (37%) protocol. Mild stimulation refers to ovarian stimulation with the use of clomiphene citrate or letrozole and subsequently the administration of human menopausal gonadotropin or not, depending on follicle growth. Good – quality embryos were defined as embryos with a normal cleavage rate and ≤10% fragmentation. The embryos were transferred with the Cook Sydney IVF ET catheter (K – Jets-7019-SIVF; Cook IVF, Eight Miles Plains, Queensland, Australia) and placed 1.5 – 2 cm below the fundus under ultrasound guidance.
Approval from the institutional review board was not required because our center has been licensed by the Ministry of Public Health of People’s Republic of China since 2001; this was a retrospective study with no specific clinical intervention on IVF treatment.
For univariate analysis of clinical variables and the outcome measures, 2-sample t tests, 2-sample Wilcoxon tests, χ 2 , or Fisher exact test were used where appropriate. To estimate the independent association between PCOS and EP with the adjustment of important confounders, multivariate logistic regression models were used. Considering different features in fresh and frozen – thawed ET cycles, 2 independent regression models were constructed. Based on published findings, the confounding variables included previous EPs, tubal factor infertility, endometriosis, male factor infertility, other infertility factors, number of embryos transferred, endometrial thickness, days of stimulation, gonadotropins dosage, ovarian stimulation protocol, and estradiol and progesterone levels on hCG day. We also performed univariate logistic regression analysis to select potential EP risk factors that had a probability value of ≤ .05 from the variables in baseline and cycle characteristics among the patients who were recruited. The EP risk factors that were distributed unevenly in PCOS and non-PCOS groups were also included in the logistic regression model as confounders. It turned out that the selected EP risk factors were a subset of the confounding variables mentioned earlier. The independent risk effect of PCOS on EP was represented by adjusted odds ratio (OR) and 95% confidence interval (CI).
To identify a possible explanation on how PCOS affected EP development, variables that were observed as risk factors for EP were analyzed. Joint effect analysis of PCOS and tubal factor infertility in logistic regression model was performed. Estradiol on the day of hCG administration was categorized by percentile analysis. With the use of the published literature, estradiol concentrations below or above the 75th percentile (≤4085; >4085 pg/mL) were defined as the low and high estradiol group, respectively. We stratified our anaylsis by having PCOS or non-PCOS in the logistic regression model to estimate the possible effect modification of the variable on the association between PCOS and EP.
The data were analyzed with the SAS statistical system (version 9.2; SAS Institute Inc, Cary, NC). A probability value of ≤ .05 was considered to be statistically significant.
Results
A total of 5339 women who achieved clinical pregnancies were recruited in this study, among whom 123 (2.3%) had EPs.
Among the 5339 women, 3303 women conceived after fresh ET, and 2036 women conceived after frozen-thawed ET. The ectopic rates were similar in fresh (2.5%; 83/3303 women) and frozen-thawed ET cycles (2.0%; 40/2036 women; OR, 0.78; 95% CI, 0.53 – 1.14; P = .195). However, when the study population was grouped as women with and without PCOS, the overall EP rate was significantly higher in women with PCOS (4.9%; 10/205 women) compared with those without PCOS (2.2%; 113/5134 women; OR, 2.28; 95% CI, 1.18–4.42; P = .027).
Fresh ET cycles
The total of 3303 women who achieved clinical pregnancies after fresh ET were divided into 2 groups: women with PCOS (n = 114) and without PCOS (n = 3189).
Table 1 shows the baseline characteristics for women in the 2 studied groups. Women with PCOS were characterized by fewer previous conceptions and previous EPs, higher body mass index, luteinizing hormone level, and lower follicle-stimulating hormone level. The proportions of infertility factors were comparable ( Table 1 ). Table 2 shows the fresh cycle characteristics. Cycle features of women with PCOS were characterized by more numbers of previous IVF cycles, longer duration of ovarian stimulation, more oocytes, fertilized oocytes, cryopreserved embryos obtained, and more cycles having surplus embryos cryopreserved. Antagonist stimulation protocol was used more commonly for women with PCOS. Regarding hormonal profiles on the day of hCG administration, estradiol concentration was not significantly increased; progesterone concentration was significantly lower, and progesterone/estradiol was significantly higher in women with PCOS than in women without PCOS ( Table 2 ).
Variable | Women with PCOS (n = 114) | Women without PCOS (n = 3189) | P value |
---|---|---|---|
Maternal age, y a | 29.5 ± 3.2 | 30.0 ± 3.8 | .240 b |
Type of infertility, n (%) | < .001 c | ||
Primary infertility | 92 (80.7) | 1745 (54.7) | |
Secondary infertility | 22 (19.3) | 1444 (45.3) | |
Parity, n (%) | .108 c | ||
0 | 109 (95.6) | 2913 (91.3) | |
≥1 | 5 (4.4) | 276 (8.7) | |
Previous spontaneous abortions, n (%) | .906 c | ||
0 | 104 (91.2) | 2899 (90.9) | |
≥1 | 10 (8.8) | 290 (9.1) | |
Previous ectopic pregnancies, n (%) | < .001 c | ||
0 | 109 (95.6) | 2524 (79.1) | |
≥1 | 5 (4.4) | 665 (20.9) | |
Body mass index, kg/m 2 a | 23.0 ± 3.0 | 21.7 ± 2.6 | < .001 b |
Basal: days 2-4, IU/L a | |||
Follicle-stimulating hormone level | 6.8 ± 1.5 | 7.6 ± 2.8 | < .001 b |
Luteinizing hormone level | 7.6 ± 4.9 | 4.4 ± 4.4 | < .001 b |
Male factor, n (%) | 42 (36.8) | 1445 (45.3) | .074 c |
Tubal factor, n (%) | 40 (35.1) | 1402 (44.0) | .060 c |
Endometriosis, n (%) | 6 (5.3) | 301 (9.4) | .076 c |
Salpingectomies, n (%) | 4 (3.5) | 230 (7.2) | .130 c |
Other infertility factors, n (%) e | 1 (0.9) | 59 (1.9) | .723 d |
a Values are given as mean ± SD
e Other infertility factors: infertility related to immunologic factors, diminished ovarian reserve, maternal chromosome abnormality disorders, or other chronic diseases.
Variable | Women with PCOS (n = 114) | Women without PCOS (n = 3189) | P value |
---|---|---|---|
Year of in vitro fertilization procedures, n (%) | .176 b | ||
2007 | 18 (15.8) | 370 (11.6) | |
2008 | 21 (18.4) | 665 (20.9) | |
2009 | 37 (32.5) | 806 (25.3) | |
2010 | 18 (15.8) | 609 (19.1) | |
2011 | 20 (17.5) | 739 (23.2) | |
Previous in vitro fertilization cycles, n a | 1.4 ± 1.0 | 1.2 ± 0.7 | .042 c |
Days of stimulation, n a | 9.8 ± 2.8 | 8.7 ± 1.6 | < .001 d |
Gonadotropins dosage, IU a | 1736.9 ± 584.2 | 1733.8 ± 545.0 | .956 d |
Ovarian stimulation protocol, n (%) | < .001 b | ||
Long protocol | 82 (71.9) | 2321 (72.8) | |
Short protocol | 0 (0) | 659 (20.7) | |
Antagonist protocol, n (%) | 32 (28.1) | 209 (6.6) | |
Estradiol, on hCG day, pg/mL a | 3528.1 ± 2025.9 | 3185.8 ± 1759.4 | .135 c |
Progesterone on hCG day, ng/mL | 1.3 ± 0.6 | 1.4 ± 0.6 | .005 c |
Progesterone/estradiol on hCG day ×1000, n a | 0.48 ± 0.44 | 0.54 ± 0.37 | < .001 c |
Endometrial thickness on embryo transfer day, mm a | 11.2 ± 1.9 | 11.0 ± 2.1 | .308 d |
Oocytes retrieved, n a | 13.7 ± 5.6 | 10.9 ± 5.5 | < .001 d |
Follicles ≥14 mm on hCG day, n a | 11.9 ± 5.3 | 9.0 ± 4.1 | < .001 d |
Metaphase II oocytes, n a | 12.6 ± 4.9 | 10.1 ± 4.8 | < .001 d |
Fertilized oocytes: 2 pronuclear, n a | 9.0 ± 4.4 | 7.1 ± 3.7 | < .001 d |
Fertilization rate, % a | 71.6 ± 18.0 | 71.8 ± 19.3 | .891 d |
Good-quality embryos rate, % a | 64.7 ± 28.3 | 60.0 ± 29.2 | .066 d |
Embryo transferred, n (%) | .624 c | ||
1 | 14 (12.3) | 230 (7.2) | |
2 | 91 (79.8) | 2822 (88.5) | |
3 | 9 (7.9) | 137 (4.3) | |
Day of embryo transfer, n (%) | .297 e | ||
2-4 | 113 (99.1) | 3180 (99.7) | |
5-6 | 1 (0.9) | 9 (0.3) | |
Cycles with assisted hatching, n (%) | 1 (0.9) | 113 (3.5) | .185 e |
Cryopreserved embryos, n (%) a | 4.9 ± 3.4 | 3.4 ± 3.0 | < .001 d |
Cycles with surplus embryos cryopreserved, n (%) | 102 (89.5) | 2445 (76.7) | .001 b |
Ectopic pregnancy, n (%) | 8 (7.0) | 75 (2.4) | .007 e |
a Values are given as mean ± SD
The EP rate in women with PCOS was significantly higher compared with that in women without PCOS (7.0% [8/114] vs 2.4% [75/3189]; OR, 3.13; 95% CI, 1.47–6.66; P = .007). To assess the independent impact of PCOS on the risk of EP, a multivariate logistic regression analysis was performed that controlled for the following important confounders together with PCOS: previous EPs, tubal factor infertility, endometriosis, male factor infertility, other infertility factors, number of embryo transferred, days of stimulation, gonadotropins dosage, ovarian stimulation protocol, estradiol levels on hCG day, progesterone levels on hCG day, and endometrial thickness. The results showed that women with PCOS had a significantly higher risk of EP, even after adjustment for these confounders (adjusted OR, 3.06; 95% CI, 1.34 – 6.96; P = .008).
Among variables in baseline and cycle characteristics, tubal factor infertility and estradiol level on hCG day were associated significantly with EP. Women with tubal factor infertility had a higher risk of EP compared with women without tubal factor (adjusted OR, 1.90; 95% CI, 1.17–3.08; P = .010). The joint effect of PCOS and tubal factor infertility was explored by comparison with women without PCOS and tubal factor, which showed a significantly higher joint risk effect (adjusted OR, 4.57; 95% CI, 1.26–16.57; P = .021). In addition, women with both PCOS and tubal factor tended to have the highest risk of the development of EP ( Table 3 ).
PCOS | Tubal factor | Women (n = 83) with ectopic pregnancy, n (%) | Women (n = 3220) with intrauterine pregnancy, n (%) | Adjusted OR (95% CI) a | P value a |
---|---|---|---|---|---|
No | No | 29 (34.9) | 1758 (54.6) | 1.00 | – |
Yes | No | 5 (6.0) | 70 (2.2) | 3.94 (1.40–11.12) | .010 |
No | Yes | 46 (55.4) | 1356 (42.1) | 2.02 (1.21–3.37) | .007 |
Yes | Yes | 3 (3.6) | 36 (1.1) | 4.57 (1.26–16.57) | .021 |
a Adjusted for previous ectopic pregnancies, endometriosis, male factor infertility, other infertility factors, number of embryo transferred, days of stimulation, gonadotropins dosage, ovarian stimulation protocol, estradiol levels on hCG day, progesterone levels on hCG day, and endometrial thickness.
Women in the high estradiol group had a significantly higher risk of EP compared with those women in the low estradiol group (3.5% [29/825] vs 2.2% [54/2478]; adjusted OR, 1.83; 95% CI, 1.12–3.01; P = .016). Specifically, in the stratified analysis, for women without PCOS, the high estradiol group had higher EP rates than the low estradiol group (3.4% [27/789] vs 2.0% [48/2400]; adjusted OR, 1.99; 95% CI, 1.19–3.35; P = .009); for women with PCOS, high EP rates were observed in both the groups (5.6% [2/36] vs 7.7% [6/78]; adjusted OR, 0.92; 95% CI, 0.15–5.67; P = .928; Table 4 ).
Variable | Women with PCOS, n (%) | Women without PCOS, n (%) | ||||||
---|---|---|---|---|---|---|---|---|
Women with ectopic pregnancy (n = 8) | Women with intrauterine pregnancy (n = 106) | Adjusted OR (95% CI) a | P value a | Women with ectopic pregnancy (n = 75) | Women with intrauterine pregnancy (n = 3114) | Adjusted OR (95% CI) a | P value a | |
Estradiol on hCG day ≤4085, pg/mL | 6 (7.7) | 72 (92.3) | 1.00 | .928 | 48 (2.0) | 2352 (98.0) | 1.00 | .009 |
Estradiol on hCG day >4085, pg/mL | 2 (5.6) | 34 (94.4) | 0.92 (0.15–5.67) | 27 (3.4) | 762 (96.6) | 1.99 (1.19–3.35) |
a Adjusted for previous ectopic pregnancies, endometriosis, male factor infertility, other infertility factors, tubal factor, number of embryo transferred, days of stimulation, gonadotropins dosage, ovarian stimulation protocol, progesterone levels on hCG day and endometrial thickness.
Cryo-thawed ET cycles
The total of 2036 women who achieved clinical pregnancies after cryo-thawed ET were divided into 2 groups: women with PCOS (n = 91) and women without PCOS (n = 1945).
Tables 5 and 6 show the baseline and cycle characteristics for women who underwent frozen-thawed treatment. Women with PCOS were characterized by fewer previous conceptions, higher luteinizing hormone level, lower follicle-stimulating hormone level, more cryopreserved embryos, and cycles having surplus embryos cryopreserved. No natural cryo-thawed ET cycles applied to women with PCOS. Less male factor infertility occurred in women with PCOS; proportions of other infertility factors were similar.