In this retrospective study, a total of 269 women were reviewed. A total of 139 women were included. The participants had undergone ART cycles (including fresh cycles and freeze–thaw cycles) and were diagnosed with POI according to the 2016 ESHRE guidelines, namely: (i) 4 months of persistent oligomenorrhea or amenorrhea and (ii) two tests with follicle-stimulating hormone (FSH) levels above 25 (at an interval of at least 4 weeks) (10). Participants who (1) had endometriosis, chromosomal abnormalities, and other endocrine diseases and (2) had ovarian surgery or gynecological tumors were excluded ( Figure 1 ).

Firstly, the 139 female patients were split into two groups based on oocyte origin: using their own oocytes (OG group) or accepting oocyte donations (OD I group). In the OG group, nine patients decided to use others after a failure of attempting to use their own oocytes, and this population was the oocyte donation II group (OD II group). Each of the three groups was then divided into pregnant or non-pregnant group.

Endometrial preparation in all of the patients involved hormone replacement therapy. The detailed endometrial preparation protocol for freeze–thaw cycles has been described in a previous article, including the classification of endometrial types and thickness measurement methods (11). For estrogen–progesterone (EP) cycles, oral estradiol (Progynova, Bayer, Germany) administration began on days 2 to 3 of the target cycle and lasted for about 2 weeks. When the thickness of the endometrium reaches 8 mm and above, the patient is asked to add oil-based progesterone (60 mg). On the same day, the thickness of the endometrium was recorded using transvaginal ultrasound examination. To avoid cavity fluid and other unfavorable conditions, the patients were hospitalized and re-measurement of endometrial thickness was done in the morning of the transplantation day. Luteal supplement was altered to vaginal progesterone gel (90 mg, Crinone 8%; Merck Serono) and oral dydrogesterone (20 mg Duphaston; Abbott) after embryo implantation. Clinical pregnancy was confirmed by ultrasound observation after embryo transfer.

The cleaved embryos were incubated in an incubator with 6% CO₂ at 37°C and individually in microdrops (50 μL) containing G1 (Vitrolife) + 5% HSA (Vitrolife). Embryo quality was assessed on day 3 based on the Peter cleavage stage embryo scoring system. Vitrification was used for embryo cryopreservation and thawing.

IBM SPSS, 21.0 (IBM Corp., Armonk, NY, USA) was employed. Numerical data were shown as mean ± standard deviation (SD), while categorical variables were shown as % (n/N). Mann–Whitney test and chi-square test were utilized for continuous and categorical variables, respectively. The threshold was set as two-tailed P <0.05.

There were 88 (63.3%) patients who used their own oocytes, while 51 (36.7%) patients accepted oocyte donations ( Table 1 ). No significant difference was found between the two groups in age (32.73 ± 0.51 vs. 31.27 ± 0.61) and body mass index (BMI) (22.04 ± 0.28 vs. 22.73 ± 0.40). The women in the OG group had lower basal serum FSH (43.98 ± 2.53 vs. 71.37 ± 4.18) and LH (22.45 ± 1.76 vs. 34.53 ± 2.17). The women in the OG group had higher serum E2 (44.24 ± 7.49 vs. 9.97 ± 1.09) and AMH (0.08 ± 0.01 vs. 0.06 ± 0.04). In addition, the women in the OG group had more AFC (1.15 ± 0.15 vs. 0.10 ± 0.06), fewer years of infertility (3.82 ± 0.30 vs. 5.04 ± 0.48), and more numbers of previous pregnancies (0.85 ± 0.14 vs. 0.24 ± 0.06).

There were nine patients who accepted oocyte donations after trying using their own oocytes ( Table 2 ). In the OG group, the results were comparable between pregnant patients and non-pregnant patients in age (33.40 ± 1.30 vs. 32.64 ± 0.55), BMI (22.58 ± 0.59 vs. 21.98 ± 0.30), and AFC (1.70 ± 0.58 vs. 1.08 ± 0.15), FSH(52.23 ± 11.91 vs. 42.93 ± 2.43, E2 (49.16 ± 26.43 vs. 43.61 ± 7.81), LH (20.00 ± 7.01 vs. 22.76 ± 1.78), and AMH (0.14 ± 0.06 vs. 0.07 ± 0.12). In the OD I group, no statistical difference was found between pregnant women and non-pregnant women in the number of oocyte donations (3.71 ± 0.11 vs. 3.89 ± 0.13), age (31.74 ± 0.73 vs. 30.25 ± 1.07), BMI (22.55 ± 0.48 vs. 23.12 ± 0.74), and AFC (0.09 ± 0.06 vs. 0.13 ± 0.13). Meanwhile, FSH (73.33 ± 5.48 vs. 67.06 ± 5.90), E2 (10.10 ± 1.42 vs. 9.68 ± 1.65), LH (35.15 ± 2.83 vs. 33.16 ± 3.21), and AMH (0.08 ± 0.06 vs. 0.01 ± 0.00) were not significantly different between the two groups. In the OD II group, data were comparable between pregnant patients and non-pregnant patients in the number of oocyte donations (3.83 ± 0.17 vs. 3.33 ± 0.33), age (31.17 ± 1.83 vs. 35.00 ± 2.52), BMI (21.57 ± 1.36 vs. 22.29 ± 2.23), and AFC (0.50 ± 0.22 vs. 0.33 ± 0.33). In the meantime, FSH (92.06 ± 11.95 vs. 50.16 ± 16.82), E2 (9.55 ± 4.672 vs. 18.46 ± 11.62), LH (52.30 ± 5.60 vs. 33.00 ± 11.12), and AMH (0.05 ± 0.03 vs. 0.04 ± 0.03) were not significantly different between the pregnant group and the non-pregnant group.

There were 10 pregnancies among 88 women in the OG group. The detailed clinical information of each patient is listed in Table 3 . Five patients were over 35 years old (patients 2, 3, 5, 6, and 10). Only four patients had pregnancies within one cycle (patients 2, 3, 4, and 6). Four pregnancies were obtained after the transfer of fresh embryos (patients 2, 3, 4, and 6) and six pregnancies after a frozen/thawed cycle. Additionally, eight patients achieved live births, including two twin births (patients 2 and 9). The basic serum FSH ranges from 25.37 to 138.4 IU/L, the basic serum E2 ranges from 1.36 to 73.1 pg/mL, the basic serum LH ranges from 6.39 to 68.33 IU/L, and the serum AMH ranges from 0.01 to 0.66 ng/mL. The number of AFC ranges from one to five.

For the 10 pregnancies who used their own oocytes, the ovarian stimulation characteristics of each patient are listed in Table 4 .

For patient 1, one embryo was frozen in the first PPOS cycle. One oocyte was obtained in the second PPOS cycle, while fertilization was abnormal. In the third PPOS cycle and fourth mild cycle, there were no oocytes retrieved. In the following E-P cycles, the woman got successfully pregnant and delivered.

For patient 2, six oocytes were obtained in the luteal-phase short-acting long protocol. After the transfer of two cleavage embryos, the patient obtained clinical pregnancy and delivered.

For patient 3, luteal-phase short-acting long protocol was administrated, and two oocytes were retrieved and then transferred.

For patient 4, similar to patient 3, she got five oocytes and then transferred two embryos and got pregnant.

For patient 5, ovarian stimulation was made by using three different protocols: GnRH antagonist protocol, mild stimulation protocol, and PPOS protocol. The first cycle was canceled because of follicle dysplasia, while one and two embryos were frozen in the latter two cycles, respectively. Finally, this patient got pregnant after two embryos were implanted in the GnRH-a and E-P cycle.

For patient 6, she got one embryo and got pregnant successfully in the luteal-phase short-acting long protocol after IVA therapy.

For patient 7, no oocyte was obtained in the first mild stimulation cycle. The second PPOS cycle got one oocyte but failed to fertilize, and the third PPOS cycle also got no oocytes. Two embryos were frozen in the fourth PPOS cycle, and she got pregnant after one was transferred in the second E-P cycle.

For patient 8, one embryo was frozen in the first mild stimulation cycle. Unfortunately, there was no mature follicle in the second follicular-phase long-acting protocol cycle, and no oocyte was obtained in the third PPOS cycle. She canceled the fourth cycle and required FET. After transferring one embryo in the following E-P cycle, she got pregnant.

For patient 9, mild stimulation was used; two embryos were frozen in the first and fourth cycles, and the third cycle also got one oocyte. However, normal fertilization failed. Early ovulation was detected in the other two cycles. Finally, this patient got pregnant after two embryos were implanted in the GnRH-a and E-P cycle.

For patient 10, one embryo was transferred in the first luteal-phase short-acting long protocol cycle, but she failed to conceive. The following two mild stimulation cycles obtained three embryos in total, and then two embryos were transferred in the natural cycle and she conceived.

In the patients with POI who had the chance of embryo transfer, we tried to find out the impact factor related to embryo transformation ( Table 5 ). In the OG group, the results were comparable between pregnant patients and non-pregnant patients in basal endometrial thickness (3.83 ± 0.70 vs. 4.53 ± 0.48), endometrial thickness on ET day (9.00 ± 0.58 vs. 9.16 ± 0.29), number of embryos transferred (1.50 ± 0.22 vs. 1.42 ± 0.12), and number of good-quality embryos (1.50 ± 0.22 vs. 1.21 ± 0.15). In the OD I group, similar results were obtained between the two groups in basal endometrial thickness (3.81 ± 0.57vs 3.08 ± 0.74), endometrial thickness on ET day (10.71 ± 0.39 vs. 9.50 ± 0.66), number of embryos transferred (1.76 ± 0.07 vs. 1.68 ± 0.11), and number of good-quality embryos (1.35 ± 0.11vs 1.16 ± 0.19). In the OD II group, no statistical difference was found between pregnant patients and non-pregnant patients in basal endometrial thickness (4.50 ± 2.08vs 2.57 ± 1.19), endometrial thickness on ET day (10.00 ± 0.82vs 11.71 ± 0.75), number of embryos transferred (1.83 ± 0.17vs 1 1.57 ± 0.20), and number of good-quality embryos (1.67 ± 0.21 vs. 1.43 ± 0.20).

To investigate the factors associated with pregnancy outcomes, we further analyze the baseline characteristics and hormone levels of patients with POI with different cycle results ( Table 6 ). The factors that resulted in non-pregnancy included no oocytes retrieved, follicle dysplasia, follicle ovulation in advance, no transferable embryo, and abnormal fertilization. A statistical difference was found between women with no oocytes retrieved and patients with follicle dyspepsia in serum E2 levels on HCG (p < 0.001, 323.68 ± 37.33 vs. 34.74 ± 10.85). Patients with follicle dysplasia and patients with follicle ovulation in advance showed a difference in AFC (p = 0.011, 0.58 ± 0.15 vs. 2.00 ± 0.72), serum E2 levels on HCG (p < 0.001, 34.74 ± 10.85 vs. 354.70 ± 46.72), and serum P4 levels on HCG (p = 0.028, 0.39 ± 0.06 vs. 0.66 ± 0.20). Patients with follicle dysplasia and patients with no transferable embryo also showed a difference in serum basal FSH level (p = 0.034, 39.23 ± 2.71 vs. 33.05 ± 0.36), serum E2 levels on HCG (p < 0.001, 34.74 ± 10.85 vs. 394.51 ± 102.25), and serum P4 levels on HCG (p = 0.013, 0.39 ± 0.06 vs. 0.62 ± 0.07). A significant difference was likewise identified between patients with follicle dysplasia and abnormal fertilization in serum E2 levels on HCG (p < 0.001, 34.74 ± 10.85 vs. 467.47 ± 97.76) and serum LH levels on HCG (p = 0.014, 14.85 ± 2.14 vs. 16.92 ± 6.11).