Effect of Frozen Embryo Transfer and Progestin-primed Ovary Stimulation on IVF outcomes in women with high body mass index

Among women undergoing IVF, high BMI negatively affects pregnancy outcomes when using the conventional ovary stimulating protocols combined with fresh embryo transfer. Therefore, finding a proper treatment for these high BMI women is more important and urgent when obesity is prevalent. In our study, we reported a retrospective study of 4457 women who were divided into normal BMI group (18.5 kg/m2–24.9 kg/m2) and high BMI group (≥25 kg/m2) undergoing 4611 IVF treatment cycles with frozen embryo transfer (FET). We found the high BMI group originally had the poor oocytes performance, but after FET they got the similar pregnancy outcomes as the normal BMI group. Then under FET we analyzed the IVF outcomes of our new progestin-primed ovary stimulation (PPOS) protocol, indicating that the hMG + MPA (4 or 10 mg/d) groups had the obvious better pregnancy results than the conventional short group in the high BMI group, and binary logistic regression analysis showed the hMG + MPA (10 mg/d) group was associated with better pregnancy outcomes than the hMG + MPA (4 mg/d) group. These results indicate PPOS - hMG + MPA (10 mg/d) combined with FET might be a new potential treatment choice for the high BMI women undergoing IVF treatments.


Results
Clinical characteristics and IVF outcomes in the two BMI groups under the FET technology. Table 1 shows the baseline characteristics and IVF outcomes by using FET technology in the patients of two BMI groups. There were 2212 normal BMI women and 320 high BMI women (21.28 vs. 27.13; p < 0.001) enrolled. There were no differences in the age, duration of infertility, and numbers of antral follicles. Consistent with the previous research [26][27][28][29] , the basal FSH (5.00 vs. 5.66, p < 0.001), E2 (28.90 vs. 33.85, p < 0.001), P (0.22 vs. 0.29, p < 0.001) level in the high BMI group were lower than that in the normal BMI group. Although the mean ovarian hMG stimulation duration (9.4 days vs. 8 15.8%) comparing with the normal BMI group in the Fig. 1. In our results, the high BMI group had poor outcomes of oocyte performance than the normal BMI group as previous research reported, including the oocytes retrieved, MII oocytes, fertilized oocytes, top-quality embryos and viable embryos. Interestingly, this adverse tendency in the oocyte performance was changed after using FET, and the high BMI group got the similar clinical outcomes as the normal BMI group. As we know, FET permits the longer endometrium recovery time from the disturbance caused by ovary stimulation compared with ET. These unexpected and satisfactory IVF clinical outcomes indicated that FET might be a preferential choice for these high BMI women.
Cycle characteristics and oocytes outcomes in three ovarian stimulation groups according to female BMIs. Combined with the FET technology, we then analyzed the IVF outcomes of our three different ovary stimulation groups: conventional short, hMG + MPA (4 mg/d), and hMG + MPA (10 mg/d) separately in the normal BMI (18.5-24.9 kg/m 2 ) and high BMI (≥25 kg/m 2 ) women groups. Baseline characteristics, hormonal profiles and cycle characteristics of patients were presented in Table 2. Firstly, we analyzed the IVF outcomes of the same BMI women using three different ovary stimulation groups in Table 2. For the normal BMI women group, the duration of hMG was significantly longer in the hMG + MPA (10 mg/d) group than that Figure 1. The normal BMI and high BMI groups did not have difference in the implantation rate (A), clinical pregnancy rate (B), live-birth rate (C) and early miscarriage rate (D) after FET. The raw data were shown in the figures and analyzed by Chi-square test. The differences were considered statistically significant when the p-value was less than 0.05.
in conventional short and hMG + MPA (4 mg/d) group (8.94 vs. 8.14 vs. 6.87, p < 0.05 for any two groups); the dose of hMG in the hMG + MPA (4 mg/d) group was more than that in the hMG + MPA (10 mg/d) and conventional short groups (2002 vs. 1939 Table 2. Cycle's characteristics and oocytes outcomes in the three ovarian stimulation groups according to female BMIs. One-way ANOVA analysis, the nonparametric test, student's t-tests or Mann-Whitney U-test were applied. Data are presented as mean ± standard error of the mean. The differences were considered statistically significant when the p-value was less than 0.05. When compared with the p-value of subgroups in the same BMI group, the different superscript alphabets of a, b, c stand for a significant difference (p < 0.05). When compared with the p-value of the normal BMI group and high BMI group in the same ovary stimulation group, the symbols *, # , ※ separately stands for a significant difference (p < 0.05) in the conventional short, HMG + MPA (4 mg) and HMG + MPA (10 mg) ovary stimulation group. Pregnancy outcomes in FET cycles from three ovarian stimulation groups according to female BMIs. Table 3 describes the FET characteristics of three ovarian stimulation groups separately in the two different BMI women groups. Similarly, we first analyze the pregnancy and live birth outcomes of the same BMI women using three different ovary stimulation treatments in Fig. 2 Logistic regression of three ovarian stimulation treatment groups on the pregnancy outcomes in the two BMI groups.  Table 3. Women and cycle's characteristics of frozen-thawed embryos originating from three ovarian stimulation groups according to female BMIs. One-way ANOVA analysis, the nonparametric test, student's t-tests or Mann-Whitney U-test were applied. Data are presented as mean ± standard error of the mean. The differences were considered statistically significant when the p-value was less than 0.05. When compared with the p-value of subgroups in the same BMI group, the different superscript alphabets of a, b, c stand for a significant difference (p < 0.05). When compared with the p-value of the normal BMI group and high BMI group in the same ovary stimulation group, the symbols *, # , ※ separately stands for a significant difference (p < 0.05) in the conventional short, HMG + MPA (4 mg) and HMG + MPA (10 mg) ovary stimulation group.

Discussion
Our IVF outcomes indicate that under control standard short term protocol or our initially used PPOS protocol (4 or 10 mg/day), the high BMI women group got the poorer oocyte performance than the normal BMI women group, but after FET this adverse situation in the oocyte performance of high BMI women was changed and achieved the similar pregnancy and live birth outcomes as the normal BMI women. Then within FET, we further analyzed the IVF outcomes of these three different ovary stimulation groups separately in the normal and high BMI women group. We found for the normal BMI women, the hMG + MPA (4 or 10 mg/d) groups got similar or significantly better oocytes, pregnancy and live birth results than the conventional short group, for the High BMI women the hMG + MPA (10 mg/d) group had the best oocytes, pregnancy and live birth outcomes in the three groups. Finally, we used the binary logistic regression to analyze the IVF outcomes of these three different ovary stimulation groups, and found compared with the conventional short protocol (ref), for the Normal BMI women the hMG + MPA (4 or 10 mg/d) groups had the similar and higher odds of pregnancy and live birth, but for the high BMI women the hMG + MPA (10 mg/d) group had the highest odds of pregnancy and live birth in three of them. These results indicate that combined with FET, the hMG + MPA (4 or 10 mg/d) ovarian stimulation treatment groups had the better pregnancy and live birth results than the conventional short group, regardless of in the normal or high BMI group. Moreover, the advantage in the pregnancy and live birth outcomes of hMG + MPA (10 mg/d) treatment is more obvious and prominent for the high BMI women, showing it may be a potential and more appropriate treatment for the high BMI women. The obese women had lower pregnancy and live birth outcomes when ET was used in ART cycles, which was concluded from huge database derived from the multicenter trials 26,27,30 . Therefore, the debate between oocyte quality and endometrium receptivity, which is the main reason of that in the high BMI woman remains as before. However, high BMI does not affect IVF outcomes in obese women using donor oocytes, which indicates oocyte quality rather than endometrial receptivity may be the most important factor in obese women 10 . Our results showed that under three ovary stimulation treatments, the high BMI group (BMI ≥ 25) originally had the decreased retrieved oocytes, MII oocytes, fertilized oocytes, top-quality embryos and viable embryos than the normal BMI group, but after FET this adverse situation of oocyte performance in high BMI women was changed and achieved the similar pregnancy outcomes as the normal BMI women. Using FET might reduce the endometrium disturbance induced by the ovary simulation.
Observational studies have shown that the FET group had significantly greater clinical pregnancy rate per transfer than in the ET group. The results of women with the PCOS showed that FET was associated with a higher rate of live birth, a lower risk of the ovarian hyper-stimulation syndrome after the first transfer than fresh-embryo transfer. These results strongly showed that it impaired the endometrial receptivity in fresh ET cycles after ovarian stimulation, comparing to FET cycles with artificial endometrial preparation 11,31 . FET not only allows the ovary and exposed endometrium to recover from the ovarian stimulation, but also avoids the adverse effect of high estradiol levels on the implantation. The increased risks of pregnancy loss 32 have been associated with higher estradiol levels after IVF, which are much higher in fresh cycles than that in frozen cycles. Higher live-birth rates have been reported with the use of letrozole, which would result in lower mid luteal estrogen levels, as an ovulation stimulating agent than with clomiphene 33 . Thus it can be seen that the FET cycles could improve the endometrial status and shied the adverse impact of high estrogen levels inducing by the ovary stimulation, consequently had higher clinical pregnancy rate, live birth and lower pregnancy loss.
Then we further analyzed the IVF outcomes of these three different ovary stimulation groups with FET separately in the normal and high BMI women. We found that hMG + MPA (4 or 10 mg/d) treatment groups got the similar or significantly better oocytes, pregnancy and live birth results than the conventional short group in the normal BMI women; and for the high BMI women, the hMG + MPA (10 mg/d) group had the best oocytes, pregnancy and live birth outcomes in three of all. In spite of oral contraceptives mainly consisted of progesterone could inhibit gonadotropin secretion by negative feedback 34 and may have a remnant adverse effect on the subsequent fresh embryo transfer cycle used for menses induction, which was associated with the lower level of LH, estradiol and decreased endometrium thickness 35 , but rates of pregnancy and live birth after FET were not affected by oral contraceptives used for scheduling endometrial preparation. Even a meta-analysis showed in GnRH agonist cycles that progestin pretreatment was associated with higher clinical pregnancy rate than placebo or no pretreatment 36 . Moreover, luteal-phase ovarian stimulation group had a significantly increased cumulative pregnancy rate per initiated stimulation cycle than mild treatment in patients with a normal ovarian reserve 22 , as well MPA group had slightly higher implantation and live-birth rate 17 . Especially for the PCOS patients who are in a progesterone deficiency state, the Utrogesten or Medroxyprogesterone Acetate with hMG protocols could significantly improve the pregnant outcomes, such as the biochemical pregnancy, clinical pregnancy rate and implantation rate 24,25 . The high BMI women (BMI ≥ 25) also have the lower progesterone level 37, 38 like the PCOS women, so the hMG + MPA (4 or 10 mg/d) protocol may just reverse this deficiency of progesterone and significantly improve pregnancy and live birth outcomes than the conventional short group. Moreover, the hMG + MPA  Table 4. Logistic regression of three ovarian stimulation groups combined with FET on the pregnancy outcomes according to female BMIs. After adjusting for female age, duration of infertility, basic serum FSH level, the E2, P level and endometrial thickness on the FET day, and the number of embryos transferred, binary logistic regression was performed. The differences were considered statistically significant when the p-value was less than 0.05.
ScieNTific RepoRts | 7: 7447 | DOI:10.1038/s41598-017-07773-w (10 mg/d) group had better clinical pregnancy and live birth outcomes than the hMG + MPA (4 mg/d) group in the high BMI women group, not the normal BMI women group. That is may due to the hMG + MPA (10 mg/d) group could rescue the progesterone deficiency close to the physiological state in the high BMI women group compared with hMG + MPA (4 mg/d) group, which may not occur in the normal BMI women group. Futhermore, we should consider that the endometrial preparation is significantly different in the normal BMI group, which is might be a potential covariate to the pregnancy and live birth outcomes in the logistic regression analysis. On the other hand, we consider that the hMG + MPA (4 or 10 mg/d) treatments may partly improve the oocyte developmental competence, which is defined as the ability of oocyte to compete meiosis, undergo fertilization, embryo genesis and term development 39 . Although the effects of progesterone on oocytes and embryo development have been assessed by some investigations, and always have the contradictory results both in vitro and in vivo. As a dominant steroid in follicular fluid at approximately 18 hours after the luteinizing hormone (LH) surge may have a critical role in maturation of oocytes at the germinal stage 40 , progesterone could influence the oocyte nuclear maturation, cytoplasmic maturation and development, in order to improve oocyte developmental competence 41 . Firstly progesterone plays an important role in the resumption of meiosis and progression to MII in oocytes, so it is strongly associated with oocyte quality and maturity 42 . And in rhesus monkeys, the improvement of in vitro oocyte development was demonstrated in the presence of progesterone and E2 43 . Secondly, progesterone could affect the relationship between cumulus cells and oocytes, which is important for oocyte cytoplasmic maturation and subsequent developmental competence 44 . It would appear that the progesterone, PGR and Ptgs2 are active participants in a positive loop of sustained EGFR induced-ERK signaling required of cumulus cell expansion and oocyte maturation during the ovulation process 45 . Thirdly, studies in periovulatory granulosa cells show P4 could protect the oocyte from the onset of apoptosis, mainly mediated by the classical PGR 46,47 , and lower rates of apoptosis in cumulus cells have been correlated with developmental potential in both human and cattle oocytes [48][49][50] . In addition, progesterone could increase subsequent sperm head decondensation, male pronuclei formation 51 and fertilization 52 when it increased in pocine oocyte maturation medium. Finally, progesterone influences the cytoplasmic maturation of porcine oocytes, at least partially, by decreasing their polyadenylation, thereby altering maternal gene expression 53 .
There are also limitations of our study. First, this is a retrospective design of study, and need to be confirmed by randomized controlled trial in the future. Second, the high BMI group in our study just had one group (BMI ≥ 25), which is better when divided more accurately if have enough cycles, such as into overweight (BMI 25.0-29.0), obese (BMI 30.0-49.9) and superobese (BMI > 50). Whereas the two hMG + MPA groups both improved the clinical pregnant outcomes in this one high BMI group, and the more usage of MPA, the better IVF outcomes. These two hMG + MPA (4 or 10 mg/day) groups has the consistent and dose-dependent therapeutic effects in the high BMI group, which helps confirm the effect of MPA in the high BMI women. Thirdly, we did not have ET control in the high BMI group, which could help us to know the improvement of pregnancy outcomes in the high BMI women who originally had the poor oocyte performance than the normal BMI women is directly induced by FET or other factors. However, so many former researches show when using ET, the high BMI women would pursue this adverse performance in the pregnancy as in the oocyte.
In conclusion, it can be assumed that hMG + MPA (4 or 10 mg/d) ovary stimulation protocol combined with FET technology could be an efficient and acceptable treatment for high BMI (≥25) women comparing with the conventional short protocol, and for the high BMI women the hMG + MPA (10 mg/d) is the most effective treatment strategy in three of them. The results suggest that hMG + MPA (4 or 10 mg/d) protocol could be better effectiveness, with many theoretical advantages, such as FET could allow the ovary and exposed endometrial to be better recovered from the ovarian stimulation, avoiding of the adverse effect of high estradiol levels on the implantation, and MPA could partly improve the oocyte developmental competence, furthermore MPA (10 mg) greatly complement the deficiency of progesterone in the high BMI women and have higher probability in success of IVF.

Methods
We performed a retrospective cohort study using the data from the Department of Assisted Reproduction of the Ninth People's Hospital of Shanghai Jiaotong University School of Medicine database of 2006-2016. This study was approved by the Ethics Committee (Institutional Review Board) of the Ninth People's Hospital of Shanghai. And the study was conducted according to the Declaration of Helsinki for medical research and informed consent was obtained. The characteristics and cycle parameters of the patients were recorded in the database of our center. We screened the database according to the following search criteria. Inclusion criteria included 1) age no older than 40; 2) basal serum FSH concentration of no more than 10 IU/L; 3) antral follicle count (AFC) of more than 3 on menstrual cycle day 2-3; and 4) regular menstrual cycles over the previous 3-month period (25-35 days in duration); and 5) women with infertility diagnoses of male, unexplained, and tubal factors. The exclusion criteria were as follows: 1) receipt of hormone treatments within the previous 3-month period; 2) diagnosis of polycystic ovarian syndrome; 3) presence of a functional ovarian cyst with E 2 > 100 pg/mL; 4) preimplantation genetic diagnosis, in vitro maturation (IVM), and donor sperm were used; 5) any contraindications to ovarian stimulation treatment; 6) presence of fresh embryo transplantation.
Body Mass Index Assessment. Weight and height were used to calculate BMI according to the standard formula: BMI = weight/height 2 (kg/m 2 ). Appropriate study subjects were divided into two groups depending on the BMI of patients and based on World Health Organization standards: normal BMI group (18.5 kg/m 2 ≤ BMI < 25 kg/m 2 ) and high BMI group (BMI ≥ 25 kg/m 2 ) 1 . And BMI data was based on measured height and weight information in each patient's medical record.
ScieNTific RepoRts | 7: 7447 | DOI:10.1038/s41598-017-07773-w Ovarian Stimulation Protocols. Conventional short ovarian stimulation protocol. A standard short term protocol was used as the control group. Patients were administered 0.1 mg of triptorelin daily from menstrual cycle day 2 (MC2) to the time of human chorionic gonadotropin (HCG) administration, and exogenous hMG (150-225 IU) was administered daily beginning on MC3. After 4-5 days of hMG administration, a transvaginal ultrasound examination and serum hormone level tests were performed and the dose of hMG was optimally adjusted based on the number and size of developing follicles. The final stage of oocyte maturation was induced with an intramuscular injection of hCG (Lizhu Pharmaceutical Trading Co., Zhuhai, China) after the dominant follicles had reached 18 mm in diameter. If a patient had more than three dominant follicles, 2,000 IU of hCG was administered, while for patients with no more than three dominant follicles, 5,000 IU of hCG was used for triggering. Transvaginal ultrasound-guided oocyte retrieval was conducted 34-36 h later.
PPOS protocol. A detailed description of the hMG + MPA procedure has been presented in our previous publication 17 . Briefly, ovarian stimulation was initiated for patients from MC3 onward, via the injection of 150-225 IU of hMG (Anhui Fengyuan Pharmaceutical Co. Ltd, Hefei, China) and MPA (4 mg/d or 10 mg/d) 17 . The initiating dose of 150 IU/day was used for patients with high AFC counts greater than 20 or slightly elevated basal FSH (7-10 IU/L), while 225 IU/day was used for all other patients. Prior contraception studies indicated that 10 mg MPA can be used to inhibit ovulation, while 5 mg MPA failed to inhibit ovulation 54 , it is likely that 10 mg MPA may be better to prevent premature LH surge in the milieu of controlled ovarian stimulation. In this trial, to explore the optimal MPA dose, we chose a dose of 4 mg and 10 mg MPA for further dose-effectiveness research. When three dominant follicles reached 18 mm in diameter, the final stage of oocyte maturation was cotriggered with SC injections of triptorelin (0.1 mg, Decapeptyl, Ferring Pharmaceuticals) and hCG (1,000 IU; Lizhu Pharmaceutical Trading Co.). Oocytes were retrieved via transvaginal aspiration 34-36 h after ovulation was triggered.
IVF outcomes assessment. Patients underwent IVF treatment, which was performed according to the routine laboratory procedures on the day of oocyte retrieval as previously described 55,56 . Briefly, Flushing Medium (Origio Medical Company, Denmark) was used for oocyte retrieval, and Human Tubal Fluid (HTF; Irvine Scientific, CA, USA) with 10% Serum Substitute Supplements (SSS; Irvine Scientific, CA, USA) was used as the oocyte collection and insemination medium. The embryos were cultured in 10%-SSS supplemented Continuous Single Culture medium (CSC; Irvine Scientific, CA, USA).
The day-3 embryos from each IVF treatment cycle were examined for the number and regularity of blastomeres and the degree of embryonic fragmentation according to Cummins' criteria 57 . Grade I and II embryos, which were regarded as top-quality, were cryopreserved; Grade III and IV embryos were placed in extended culture until they reached the blastocyst stage, and only morphologically good blastocysts were frozen on day 5 or day 6. The vitrification procedure for freezing cleavage-stage embryos and blastocysts was performed using the Cryotopcarrier system (Kitazato Biopharma Co.) as previously reported 58 . For vitrification, a cryotop carrier system was utilized, with a mixture of 15% (v/v) ethylene glycol (EG), 15% (v/v) dimethyl sulfoxide (DMSO), and 0.5 M sucrose as the cryoprotectant solution. For thawing, solutions of 1, 0.5, and 0 M sucrose were used for the step-by-step cryoprotectant dilution.

Endometrial Preparation and Frozen Embryo Transfer.
Endometrial preparation for every transfer cycle was performed in a natural cycle, a mild stimulation cycle or a hormone replacement therapy as described in our previous article 19,21 . Our method of embryo and endometrium synchronizationin FET was as follows: for natural FET cycles, follicular growth was monitored by serum hormones and ultrasound from cycle day 10. When endometrial thickness was >8 mm and the diameter of the dominant follicle was >16 mm, with E2 > 150 pg/mL and P < 1.0 ng/mL, one of two procedures was performed depending upon the LH value. If LH was <20 IU/L, we injected hCG 10,000 IU at night (21:00) to trigger ovulation and arranged the transfer of 3-day-oldembryos for 5 days later. If the LH value was >20 IU/L, hCG 10,000 IU was administrated the same afternoon and the transfer time was conducted 4 days later. The transfer of blastocysts was arranged for the 6 or 7 days later depending on serum hormones and ultrasound results. Dydrogesterone (Duphaston; Abbott Biologicals B.V., America) 40 mg/day was used for luteal phase support beginning on the 3rd day after hCG injection.
For some patients with irregular menstrual cycles after ovarian stimulation who originally had the regular menstrual cycles before the first ovarian stimulation, we used letrozole 2.5-5 mg administered from cycle day 3 to 7 to stimulate mono follicular growth. Follicle growth was monitored from cycle day 10. In some women, hMG at a dose of 75 IU/d was used to stimulate follicular and endometrial lining growth. Administration of 10,000 IU of hCG and the timing of FET were performed according to the above criteria.
For patients with thin endometria during either natural cycles or stimulation cycles, hormone therapy was recommended for endometrial preparation, oral E2 (ethinylestradiol; Shanghai Xinyi Pharma) 75 mg/day was used from cycle day 3 onwards. Once the endometrial thickness was >8 mm, Femoston (Solvay Pharmaceuticals B.V.) 8 mg/day was started. Embryo transfer was determined on the third day after Femoston administration.
In all FET cycles, one or two thawed embryos were transferred according to the patient's intention. Once pregnancy was achieved, the exogenous estrogen and progesterone supplement were continued until 8-10 weeks of gestation.
Statistical Analysis. All statistical analyses were performed with the use of the Statistical Package for the Social Sciences software (SPSS, version 17.0, SPSS Inc., Chicago, IL, USA) or GraphPad Prism 5 (GraphPad software Inc., La Jolla, CA 92037 USA). The outcome measures for this study included the number of oocytes retrieved, the number of mature oocytes, the number of fertilized oocytes, the number of viable embryos, ScieNTific RepoRts | 7: 7447 | DOI:10.1038/s41598-017-07773-w number of top-quality embryos, and pregnancy and live birth outcomes from FETs. The implantation rate was defined as the number of gestational sacs divided by the number of embryos transferred. Biochemical pregnancy rate was defined as the proportion of patients with positive β-human chorionic gonadotrophin (β-HCG) in blood or urine. Clinical pregnancy was defined as the presence of a gestational sac with or without fetal heart activity under ultrasound examination 7 weeks after FET. The early miscarriage rate was defined as the proportion of patients with spontaneous pregnancy termination before the gestational age of 12 weeks. The live-birth rate was defined as the proportion of patients with live birth among all transfered cycles. In the study, the data were presented as the mean ± SEM. The one-way ANOVA analysis was used for the comparison of data in three groups if the normality assumption was true; otherwise, the nonparametric test was applied. Comparison between two groups, the number data were analyzed using Student's t-tests if the normality assumption was true; otherwise, the Mann-Whitney U-test was applied. Proportions were compared using the chi-square test when appropriate. Logistic regression was performed on the clinical outcomes between three ovary stimulation groups separately in two BMIs groups. The differences were considered statistically significant when the P-value was less than 0.05.