Effect of repeated controlled ovarian stimulation on pregnancy outcomes in fresh embryo transfer cycles: a retrospective cohort study

We assembled people who received one and multiple COS-IVF/ICSI fresh cycles at the First Affiliated Hospital of Zhengzhou University Reproductive Center from January 2015 to March 2021. Ultimately, we administered a retrospective study of medical records of 45,555 cycles from 39,641 patients according to the inclusion and exclusion criteria. This retrospective study was approved by the Ethics Committee of the First Affiliated Hospital of Zhengzhou University (reference number: 2022-KY-0841-002). The tailored inclusion criteria were as follows: fresh cycles of IVF-ET or ICSI-ET; patients with single cycle and multiple cycles; female COS age between 20 and 39 years; and the time interval between each COS treatment cycle > 2 months. Following were the criteria for exclusion: patients with infertility factors such as PCOS, endometriosis, uterine anomalies (e.g., unicornuate uterus, bicornuate uterus, etc.), and chromosomal abnormalities; FSH > 15 U/L; and antral follicle count (AFC) < 3. The patients were grouped based on the number of repeated cycles in the included population: Group A: No-repeat cycles (single cycle group, n  = 34,738); Group B: Patients who underwent 2 cycles of COS (Repeat cycle 2 group, n  = 4,138 patients, 8,276 cycles); Group C: Repeat cycle 3 group ( n  = 593 patients, 1,779 cycles); Group D: Repeat cycle 4 group ( n  = 116 patients, 464 cycles); Group E: Repeat cycle 5 group ( n  = 38 patients, 190 cycles). The patients in the repeated cycle groups (Groups B, C, D, and E) were further divided into subgroups based on the number of cycles: Group B: Cycle 1 and Cycle 2; Group C: Cycle 1, Cycle 2, and Cycle 3; Group D: Cycle 1, Cycle 2, Cycle 3, and Cycle 4; Group E: Cycle 1, Cycle 2, Cycle 3, Cycle 4, and Cycle 5. The screening and grouping of the study population are illustrated in Fig.  1 . Fig. 1 Patient selection flowchart Patient selection flowchart The primary observables translating ovarian reserve function were antral follicle count (AFC) and anti-Müllerian hormone (AMH), while the secondary observables were basal follicle-stimulating hormone (FSH) and basal luteinizing hormone (LH). We calculated clinical pregnancy rate (CPR), live birth rates (LBR), and early miscarriage rate (EMR) to symbolize clinical pregnancy outcomes. CPR was defined as the number of clinical pregnancies per 100 embryones transferred cycles. The LBR referred to the number of pregnancies that progressed to the delivery of a viable offspring per 100 embryo transplanted cycles. The EMR was calculated as cycles of a pregnancy loss before 28 weeks of gestation and all biochemical pregnancies/total number of clinical pregnancy cycles. Levels of hormones such as AMH, FSH, and LH were measured using electrochemiluminescence immunoassay (ECLIA) on days 2–3 of the menstrual cycle, with the Roche cobas 6000-e601 platform utilized for detection. The laboratory regularly conducts instrument calibration and quality control, using standard samples for quality assurance, and participates in inter-laboratory comparison and validation to ensure the consistency of results. AFC was assessed via transvaginal ultrasound on days 2–3 of the menstrual cycle, with measurements independently taken by two experienced sonographers and averaged. Endometrial thickness was measured via transvaginal ultrasound prior to embryo transfer, recording the thickest part of the endometrium. COS was accomplished with gonadotropin-releasing hormone (GnRH) agonist and antagonist protocols. The COS protocol and Gn initiation dose were determined according to the patient’s age, body mass index (BMI), follicle count, and medical history. The Gn dose during ovulation induction was adjusted according to endocrine levels and follicle development. The Center’s article described the details of these protocols [ 9 ]. When follicles ≥ 2 reached a diameter of ≥ 18 mm or more than 2/3 of follicles reached a diameter ≥ 16 mm, ovulation was triggered by Aizer (Merck Serono, Italy) or hCG (Lizhu Medicine). Transvaginal ultrasound-guided oocyte retrieval was conducted approximately 37 h after ovulation. IVF and/or ICSI were performed for fertilization. Depending on the patient’s condition and embryo development, one or two D3/D5 embryo(s) were selected for transfer. Luteal support was initiated on the day of oocyte retrieval with daily transvaginal progesterone (Xenotong, Merck Serono, Switzerland) and oral progesterone (Dupbaston, Abbott, Holland). The serum chorionic gonadotropin (β-hCG) levels > 50 IU/mL at 14 or 18 days after the embryo transfer were considered as biochemically pregnant. With the luteal phase support continued and abdominal ultrasound was performed 35 days after the transplant to determine if they were clinically pregnant. Patients with clinical pregnancy were then followed up periodically by telephone to obtain their obstetric results, and this information was recorded in the central archives. In our statistical analysis, we controlled for several confounders known to influence ovarian reserve function and pregnancy outcomes. These included age, BMI, basal hormone levels (FSH, LH, E2), AFC, AMH, treatment protocol, endometrial thickness, fertilization method, and the number of embryos transferred. Age and BMI are well-established factors affecting ovarian response and pregnancy success [ 10 , 11 ]. Basal hormone levels and ovarian reserve markers (AFC, AMH) are critical predictors of ovarian function and response to stimulation [ 12 , 13 ]. The treatment protocol (GnRH agonist vs. antagonist) and fertilization method (IVF vs. ICSI) significantly impact ovarian response and embryo quality [ 14 ]. Endometrial thickness and the number of embryos transferred are directly associated with implantation and live birth rates [ 15 ]. By controlling for these confounders, we aimed to minimize bias and ensure the validity of our findings. Data were analyzed utilizing SPSS version 25.0 software. Continuous numerical variables were represented by mean ± SD, and qualitative data were recorded as frequency and percentage (%). Comparisons among populations undergoing different numbers of COS cycles: Continuous variables with normal distribution were compared using Student’s t-test, while those with non-normal distribution were compared using the Wilcoxon rank-sum test. Qualitative data were analyzed using the Chi-square test or Fisher’s exact test. For self-control comparisons within populations undergoing multiple COS cycles: Baseline comparisons of repeated cycles for multi-cycle patients were performed using paired t-tests or ANOVA with repeated measures design. The Wilcoxon signed-rank test was applied for data with uneven variances or non-normal distributions. Multifactorial logistic regression analysis was employed to control for potential confounders. Adjusted advantage ratios (aOR) and 95% confidence intervals (CIs) were calculated for the included variables. Line and bar charts were employed to delineate increasing cycles trends in CPR, LBR, and EMR. p  < 0.05 was considered a statistically significant difference. For missing data, we included a table (Table  1 ) in the manuscript detailing the percentage of missing data. Since the proportion of missing data was low (< 5%) and completely random, we excluded incomplete records to minimize their impact on the statistical analysis. Table 1 Percentage of missing data for each variable Variable Missing Data (%) Age 0 BMI 0.14% Basal FSH 1.62% Basal LH 1.55% Basal E2 1.92% AMH 0.98% AFC 0 Endometrial Thickness 4.7% Protocol 0 Number of Oocytes Retrieved 1.87% Fertilization Method (IVF/ICSI) 1.87% Stage of embryo transferred 1.67% Number of Embryos Transferred 1.58% CPR 2.08% LBR 3.82% EMR 3.51% Percentage of missing data for each variable

After filtering through the inclusion and exclusion criteria, we ultimately included 45,555 COS-IVF/ICSI fresh cycles from 39,641 patients, which were divided into five groups (Groups A, B, C, D, and E) based on the number of repeated cycles. First, we conducted comparisons among populations undergoing different numbers of COS cycles, including Group A, Cycle 2 in Group B, Cycle 3 in Group C, Cycle 4 in Group D, and Cycle 5 in Group E. Baseline characteristics were compared, and regression analysis was performed to assess the impact on pregnancy outcomes. The second part involved self-control comparisons within the same population undergoing repeated COS cycles to further evaluate the effects of repeated COS on ovarian reserve and pregnancy outcomes. We analyzed the following variables: age, BMI, basal FSH, basal LH, basal E2, AMH, AFC, endometrial thickness, fertilization method, indicators reflecting ovarian reserve function, number of oocytes obtained, stage of embryo transferred, and number of embryos transferred. Comparisons among populations undergoing different numbers of COS cycles revealed significant differences in all variables ( P  < 0.05), as shown in Table  2 . Subsequently, we used line graphs to illustrate the trends in CPR, LBR, and EMR across different populations. The results demonstrated that as the number of cycles increased, CPR and LBR showed a gradual decline, while EMR exhibited a gradual increase, depicted in Fig.  2 . Table 2 Basic characteristics among populations undergoing different numbers of COS cycles Group A ( n  = 34,738) Cycle 2 in Group B ( n  = 4,138) Cycle 3 in Group C ( n  = 593) Cycle 4 in Group D( n  = 116) Cycle 5 in Group E( n  = 38) P Age (y) 30.05 ± 4.12 31.57 ± 4.27 33.06 ± 4.08 33.78 ± 4.02 34.95 ± 3.10 0 BMI (Kg/m 2 ) 22.89 ± 3.27 23.08 ± 3.15 22.97 ± 3.01 22.98 ± 2.95 23.20 ± 2.77 0.02 Basal FSH 6.52 ± 1.86 6.99 ± 2.27 7.34 ± 2.57 7.44 ± 2.41 8.20 ± 3.13 0 Basal LH 6.07 ± 12.81 5.45 ± 5.81 5.16 ± 6.52 5.32 ± 7.86 3.78 ± 1.82 0 Basal E2 125.69 ± 537.32 298.57 ± 914.60 467.17 ± 1127.38 373.13 ± 822.19 233.56 ± 479.35 0 AMH 4.22 ± 3.42 3.12 ± 2.78 2.26 ± 2.58 1.36 ± 1.19 1.12 ± 1.40 0 AFC 15.59 ± 6.41 12.90 ± 6.53 10.79 ± 6.31 8.66 ± 5.39 8.87 ± 6.50 0 Endometrial thickness 12.41 ± 2.67 11.68 ± 2.63 11.40 ± 2.49 11.08 ± 2.14 11.38 ± 3.29 0 Fertilization method 0 IVF 25,016 (72.01%) 1634 (39.50%) 185 (31.21%) 31 (26.36%) 8 (22.22%) ICSI 9,722 (27.99%) 2,504 (60.50% 408 (68.79%) 85 (73.64%) 30 (77.78%) Number of ocytes obtained 14.87 ± 7.32 12.00 ± 6.89 9.60 ± 7.25 7.00 ± 6.39 5.47 ± 4.78 0 Stage of embryo transferred 0 Day 3 embryo 26,361 (75.88%) 3,633 (87.79%) 553 (93.17%) 110 (95.00%) 33 (87.50%) Day 5 embryo 8,377 (24.12%) 505 (12.21%) 40(6.83%) 6(5.00%) 5(12.50%) Number of embryos transferred 0 1 10,930 (31.46%) 1025 (24.76%) 139 (23.41%) 32 (27.50%) 14 (37.50%) 2 23,808(68.54%) 3,113(75.24%) 454(76.59%) 84(72.50%) 24(62.50%) BMI, Body mass index; FSH, follicle-stimulating hormone; LH, luteinizing Hormone; AMH, Anti-Müllerian hormone; AFC, Antral Follicle Count Basic characteristics among populations undergoing different numbers of COS cycles BMI, Body mass index; FSH, follicle-stimulating hormone; LH, luteinizing Hormone; AMH, Anti-Müllerian hormone; AFC, Antral Follicle Count Fig. 2 CPR, LBR and EMR among the population with different cycles of COS CPR, clinical pregnancy rate; LBR, live birth rates; EMR, early miscarriage rate CPR, LBR and EMR among the population with different cycles of COS CPR, clinical pregnancy rate; LBR, live birth rates; EMR, early miscarriage rate Table 3 Odds ratios for clinical pregnancy, live birth and early miscarriage among populations undergoing different numbers of COS cycles clinical pregnancy live birth early miscarriage aOR 95% CI P  value aOR 95% CI P  value aOR 95% CI P  value Age (years) 0.95 0.94–0.96 0.00 0.94 0.93–0.95 0.00 1.07 1.06–1.09 0.00 BMI (Kg/m2) 1.01 0.998 − 1.02 0.09 1.00 0.99–1.01 0.58 1.05 1.03–1.07 0.00 AFC 1.01 1.01–1.02 0.00 1.00 1.00-1.01 0.38 1.01 1.00-1.03 0.07 Endometrial Thickness 1.05 1.04–1.07 0.00 1.06 1.04–1.07 0.00 0.98 0.96-1.00 0.09 Stage of embryo transferred 1.48 1.32–1.66 0.00 1.40 1.24–1.59 0.00 1.15 0.92–1.43 0.21 Number of embryos transferred 2.243 2.02–2.49 0.00 2.26 2.01–2.53 0.00 1.05 0.86–1.28 0.64 The number of cycles of COS 1 (Group A) Reference Reference Reference 2 (Cycle 2 in Group B) 0.77 0.57–1.24 0.20 0.69 0.26–1.31 0.14 1.133 0.67–1.91 0.63 3 (Cycle 3 in Group C) 0.72 0.46–1.22 0.30 0.55 0.34–1.53 0.24 1.344 0.47–3.85 0.58 4 (Cycle 4 in Group D) 0.74 0.38–1.44 0.14 0.67 0.35–1.36 0.28 1.132 0.14–9.31 0.90 5 (Cycle 5 in Group E) 0.26 0.05–1.34 0.27 0.19 0.23–1.61 0.13 --- --- --- Adjusted for basal FSH, basal LH, basal E2, AFC, AMH, number of eggs obtained, treatment protocol, total FSH days, total FSH dosage, fertilization method Odds ratios for clinical pregnancy, live birth and early miscarriage among populations undergoing different numbers of COS cycles Adjusted for basal FSH, basal LH, basal E2, AFC, AMH, number of eggs obtained, treatment protocol, total FSH days, total FSH dosage, fertilization method Furthermore, to explore the effects of different repeat cohorts on pregnancy outcomes, we included potential confounding variables such as age, BMI, basal FSH, basal LH, basal E2, AFC, AMH, the number of eggs retrieval, treatment regimen, total FSH days, total FSH dosage, endometrial thickness, fertilization method, stage of embryo transferred, and the number of embryos transferred, and performed logistic regression analysis. The results are visible in Table  3 . After adjusting for confounding factors, regression analyses were performed for clinical pregnancy, live birth, and miscarriage, respectively. It was found that the number of COS cycles had no significant impact on pregnancy outcomes ( p  > 0.05). The table also presents variables that significantly influenced pregnancy outcomes. Age was a major influencing factor; for each additional year of age, the odds of clinical pregnancy (aOR = 0.95; 95% CI, 0.94–0.96; P  = 0.00) and live birth (aOR = 0.94; 95% CI, 0.93–0.95; P  = 0.00) decreased by 5% and 6%, respectively, while the odds of miscarriage increased by 7% (aOR = 1.07; 95% CI, 1.06–1.09; P  = 0.00). BMI was a significant factor affecting miscarriage; for each unit increase in BMI, the odds of miscarriage increased by 5% (aOR = 1.05; 95% CI, 1.03–1.07; P  = 0.00). For each 1 mm increase in endometrial thickness, the odds of clinical pregnancy (aOR = 1.054; 95% CI, 1.04–1.07; P  = 0.00) and live birth (aOR = 1.06; 95% CI, 1.04–1.07; P  = 0.00) increased by 5.4% and 6%, respectively. The stage of embryo transferred and the number of embryos transferred also had significant effects on clinical pregnancy. As depicted in Tables  4 and 5 , there existed a considerable differences ( P  < 0.001) pertaining to age or BMI in the repeated cycle groups (Groups B, C, D, and E). Age and BMI in Group C and age in Group D were boosted with an escalating number of cycles ( P   0.05), while Cycle 5 BMI was reported as the highest. Cycles 1, 2, 3, and 4 in Group E exhibited lower age than Cycle 5 ( P   0.05). Among the indicators symbolizing ovarian reserve function, basal FSH, basal LH, AMH, and AFC exhibit no significant distinction ( P  > 0.05), but basal E2 was significantly different in Groups B, C, and D ( P  < 0.05). Two-by-two comparisons of basal E2 revealed all three cycles in Group C were significantly distinct, with the basal E2 value increasing in a cycle-dependent manner ( P  < 0.05). Only Cycle 4 and Cycle 5 were apparently higher than Cycle 1 in Group D ( P  < 0.05); Cycle 5 E2 was the highest in Group E ( P  < 0.05). In addition, CPR, LBR, and EMR were calculated and plotted as bar graphs for different population cycles with different numbers of repetitions (i.e., Groups B, C, D, and E). As disported in Fig.  3 A, Cycle 1 exhibited significantly higher CPR than Cycle 2 in Group B (19.24% vs. 54.88%, P  = 0.000); the highest CPR was shown by Cycle 3 (43.44%, P  = 0.000) and the lowest was reported for Cycle 2 (12.13%, P  = 0.000) in Group C; the difference in CPR between the first three cycles of Group D was not statistically significant, and Cycle 4 reported the highest CPR (52.50%, P  = 0.000); there was no difference in CPR in Group E. As displayed in Fig.  3 B, Cycle 1 was significantly higher than Cycle 2 in Group B (5.00% vs. 44.38%, P  = 0.000) in LBR; both Group C (35.94%, P  = 0.000) and Group D (37.50%, P  = 0.000) had the highest LBR in the last cycle; no difference in CPR in Group E. As shown in Fig.  3 C, both Groups B (13.38%, P  = 0.000) and C (14.39%, P  = 0.000) had the lowest EMR in the last cycle, and there was no statistical difference between cycles in Groups D and E. Table 4 Comparison of ovarian reserve function in COS-IVF/ICSI treatment Variable Group B: Repeat cycle 2 Group C: Repeat cycle 3 Cycle 1( n  = 4,138) Cycle 2( n  = 4,138) P  value Cycle 1( n  = 593) Cycle 2 ( n  = 593) Cycle 3 ( n  = 593) P  value Age (y) 30.77 ± 4.37 31.57 ± 4.27 0.000 31.70 ± 4.21 32.32 ± 4.17* 33.06 ± 4.08*γ 0.000 BMI(Kg/m 2 ) 22.82 ± 3.62 23.08 ± 3.15 0.000 22.69 ± 3.08 22.81 ± 3.07* 22.97 ± 3.01*γ 0.000 Basal FSH 7.00 ± 2.22 6.99 ± 2.27 0.283 7.32 ± 2.44 7.35 ± 2.53 7.34 ± 2.57 0.715 Basal LH 5.42 ± 4.28 5.45 ± 5.81 0.740 5.03 ± 3.14 5.16 ± 5.87 5.16 ± 6.52 0.843 Basal E2 93.55 ± 422.03 298.57 ± 914.60 0.000 66.68 ± 206.50 261.39 ± 805.34* 467.17 ± 1127.38*γ 0.000 AMH 3.13 ± 2.80 3.12 ± 2.78 0.081 2.28 ± 2.57 2.27 ± 2.58 2.26 ± 2.58 0.438 AFC 12.98 ± 6.52 12.90 ± 6.53 0.002 10.93 ± 6.27 10.85 ± 6.22 10.79 ± 6.31 0.083 BMI, Body mass index; FSH, follicle-stimulating hormone; LH, luteinizing Hormone; AMH, Anti-Müllerian hormone; AFC, Antral Follicle Count * Significantly different from cycle1 in Group C; γ Significantly different from cycle 2 in Group C Bonferroni correction has adjusted significance values for multiple tests Comparison of ovarian reserve function in COS-IVF/ICSI treatment BMI, Body mass index; FSH, follicle-stimulating hormone; LH, luteinizing Hormone; AMH, Anti-Müllerian hormone; AFC, Antral Follicle Count * Significantly different from cycle1 in Group C; γ Significantly different from cycle 2 in Group C Bonferroni correction has adjusted significance values for multiple tests Table 5 Comparison of ovarian reserve function in COS-IVF/ICSI treatment Variable Group D: Reapt cycle 4 Group E: Reapt cycle 5 Cycle 1( n  = 116) Cycle 2 ( n  = 116) Cycle 3 ( n  = 116) Cycle 4 ( n  = 116) P  value Cycle 1( n  = 38) Cycle 2( n  = 38) Cycle 3( n  = 38) Cycle 4( n  = 38) Cycle 5( n  = 38) P  value Age(y) 31.73 ± 4.34 32.25 ± 4.27# 33.05 ± 4.19#c 33.78 ± 4.02#cd 0.000 33.47 ± 3.24 33.84 ± 3.26 34.37 ± 3.19a 34.95 ± 3.10ab 35.53 ± 3.03abfg 0.000 BMI(Kg/m 2 ) 22.47 ± 3.02 22.61 ± 2.99# 22.86 ± 3.00# 22.98 ± 2.95#c 0.000 23.07 ± 2.87 23.17 ± 2.92 23.24 ± 2.89 23.20 ± 2.77 23.34 ± 2.76 0.248 Basal FSH 7.66 ± 2.64 7.63 ± 2.52 7.61 ± 2.58 7.44 ± 2.41 0.714 7.92 ± 2.99 7.96 ± 2.96 8.01 ± 3.10 8.20 ± 3.13 8.20 ± 3.02 0.903 Basal LH 5.74 ± 7.25 5.59 ± 7.40 5.04 ± 7.21 5.32 ± 7.86 0.261 3.94 ± 1.54 3.79 ± 1.56 4.15 ± 2.93 3.78 ± 1.82 3.92 ± 2.35 0.835 Basal E2 78.12 ± 140.05 216.41 ± 625.57 386.28 ± 984.63# 373.13 ± 822.19# 0.000 45.17 ± 33.59 198.13 ± 506.90 148.32 ± 314.80 233.56 ± 479.35 328.59 ± 549.10 a 0.024 AMH 1.35 ± 1.14 1.35 ± 1.13 1.33 ± 1.38 1.36 ± 1.19 0.746 1.33 ± 1.45 1.23 ± 1.42 1.09 ± 1.03 1.12 ± 1.40 1.20 ± 1.40 0.119 AFC 8.64 ± 4.77 8.59 ± 4.81 8.59 ± 4.98 8.66 ± 5.39 0.671 8.92 ± 6.06 8.68 ± 6.15 8.87 ± 6.50 8.87 ± 6.50 8.87 ± 6.50 0.736 BMI, Body mass index; FSH, follicle-stimulating hormone; LH, luteinizing Hormone; AMH, Anti-Müllerian hormone; AFC, Antral Follicle Count #, Significantly different from Cycle 1 in Group D; c, Significantly different from Cycle 2 in Group D; d, Significantly different from Cycle 3 in Group D a, Significantly different from Cycle 1 in Group E; b, Significantly different from Cycle 2 in Group E; f, Significantly different from Cycle 3 in Group E; g, Significantly different from Cycle 4 in Group E Bonferroni correction has adjusted significance values for multiple tests Comparison of ovarian reserve function in COS-IVF/ICSI treatment BMI, Body mass index; FSH, follicle-stimulating hormone; LH, luteinizing Hormone; AMH, Anti-Müllerian hormone; AFC, Antral Follicle Count #, Significantly different from Cycle 1 in Group D; c, Significantly different from Cycle 2 in Group D; d, Significantly different from Cycle 3 in Group D a, Significantly different from Cycle 1 in Group E; b, Significantly different from Cycle 2 in Group E; f, Significantly different from Cycle 3 in Group E; g, Significantly different from Cycle 4 in Group E Bonferroni correction has adjusted significance values for multiple tests Fig. 3 Histogram of CPR, LBR, and EMR for different cycles in repeated cohorts CPR, clinical pregnancy rate; LBR, live birth rates; EMR, early miscarriage rate a, significantly different from Cycle1 in Group B b, significantly different from Cycle1 in Group C; β, significantly different from Cycle2 in Group C c, significantly different from Cycle1 in Group D; α, significantly different from Cycle2 in Group D; e, significantly different from Cycle3 in Group D Histogram of CPR, LBR, and EMR for different cycles in repeated cohorts CPR, clinical pregnancy rate; LBR, live birth rates; EMR, early miscarriage rate a, significantly different from Cycle1 in Group B b, significantly different from Cycle1 in Group C; β, significantly different from Cycle2 in Group C c, significantly different from Cycle1 in Group D; α, significantly different from Cycle2 in Group D; e, significantly different from Cycle3 in Group D Table 6 Logistic regression within populations undergoing multiple COS cycles on CPR, LBR and EMR CPR LBR EMR aOR 95% CI P  value aOR 95% CI P  value aOR 95% CI P  value Group B Cycle1 Reference Reference Reference Cycle2 8.29 6.80-10.12 0.000 37.68 27.18–52.23 0.000 0.04 0.03–0.06 0.000 Group C Cycle1 Reference Reference Reference Cycle2 0.67 0.33–1.36 0.270 1.00 0.19–5.26 0.990 0.94 0.12–7.71 0.960 Cycle3 6.05 3.28–11.15 0.000 47.59 13.50-167.77 0.000 0.02 0.003-0.10 0.000 Group D Cycle1 Reference Reference Reference Cycle2 2.13 0.31–14.54 0.440 - - 1.000 2.79 0.12–66.77 0.530 Cycle3 4.37 0.74–25.97 0.110 - - 1.000 2.17 0.11–44.38 0.620 Cycle4 20.46 3.05-137.24 0.002 - - 1.000 2.62 0.09–73.12 0.570 Group E Cycle1 Reference Reference Reference Cycle2 0.66 0.03–13.71 0.789 - - - - - - Cycle3 3.02 0.21–44.42 0.418 - - - - - - Cycle4 1.38 0.08–22.88 0.832 - - - - - - Cycle5 8.76 0.47-164.95 0.147 - - - - - - Logistic regression within populations undergoing multiple COS cycles on CPR, LBR and EMR Adjusted for age, BMI, basal FSH, basal LH, basal E2, AFC, AMH, number of eggs obtained, treatment protocol, total FSH days, total FSH dosage, endometrial thickness, fertilization method, stage of embryo transferred, and number of embryos transferred. - No results for too little data. CPR, clinical pregnancy rate; LBR, live birth rates; EMR, early miscarriage rate. As observed in Table  6 , Cycle 2 (aOR = 8.29; 95% CI, 6.80-10.12; P  = 0.000) in Group B, Cycle 3 (aOR = 6.05; 95% CI, 3.28–11.15; P  = 0.000) in Group C, and Cycle 4 (aOR = 20.46; 95% CI, 3.05-137.24; P  = 0.002) in Group D had the highest CPR within each group; Cycle 2 in Group B (aOR = 37.68; 95% CI, 27.18–52.23; P  = 0.000 and aOR = 0.04; 95% CI, 0.03–0.06; P  = 0.000) and Cycle 3 in Group C (aOR = 47.59; 95% CI, 13.50- 167.77; P  = 0.000, aOR = 0.02; 95% CI, 0.003-0.10; P  = 0.000) had the highest LBR and lowest EMR within each group, the remaining groups did not reached the level of statistical significance.

Assisted reproductive technology (ART), especially in-vitro fertilization (IVF) and intracytoplasmic sperm injection (ICSI), has been fleetly evolved to aid infertile couples in achieving a pregnancy. The artificial induction of superovulation by gonadotropin injection, known as controlled ovarian stimulation (COS), is a necessary step in in-vitro fertilization-embryo transfer (IVF-ET). However, it is ambiguous whether the ovarian stimulation with supraphysiologic hormonal doses impacts ovarian functions or pregnancy outcomes. Moreover, most infertile women require multiple COS to increase the chances of clinical pregnancy [ 1 ]. The Ovarian reserve function symbolizes a woman’s fertility potential. Some studies have sought to clarify that high doses of exogenous gonadotropins may have depleted follicular reserve at an accelerated rate [ 2 ]. Information on the health effects of multiple COS in infertile women is scarce and needs a comprehensive debate over the existing gaps. Therefore, further thought and extensive research are required to analyze the effect of repeated COS on the health of infertile women. We hypothesize that repeated COS does not have a significant negative impact on pregnancy outcomes. Relevant retrospective studies are relatively scarce, but in recent years, several studies have explored the impact of repeated controlled ovarian stimulation (COS) on reproductive function. Xu et al. [ 3 ] demonstrated that up to four cycles of COS did not significantly affect ovarian reserve function or responsiveness, although the impact on pregnancy outcomes was not further investigated. Similarly, Paul et al. [ 4 ] found that repeated COS had no adverse effects on the number and quality of oocytes, fertilization rates, embryo quality, or implantation rates, but the study did not address clinical pregnancy rates (CPR) or live birth rates (LBR). A review indicated that repeated COS within the first three cycles did not significantly impair ovarian reserve function or CPR [ 1 ]. However, Martin-Johnston et al. [ 5 ] reported a declining trend in CPR with increasing cycle numbers, particularly after the third cycle. Additionally, animal studies have shown that repeated ovarian stimulation may lead to ovarian oxidative stress damage, mitochondrial DNA mutations, an increased proportion of oocyte spindle defects, and reduced oocyte quality [ 6 , 7 ]. Conversely, some studies have reported conflicting results, suggesting that repeated COS significantly improves CPR [ 8 ]. Therefore, conducting a large-scale, updated study to investigate the effects of repeated COS on ovarian reserve function and clinical pregnancy outcomes is inevitable. This study screened individuals who underwent repeated fresh IVF/ICSI transfer cycles and employed both inter-group comparisons among populations receiving different numbers of COS cycles and intra-group self-control comparisons within the repeated COS population. The aim was to robustly and reliably elucidate whether repeated COS alters ovarian reserve function and clinical pregnancy outcomes, including clinical pregnancy rate, miscarriage rate, and live birth rate. The findings suggest that, in inter-group comparisons among populations with varying numbers of COS cycles, the number of COS cycles may not have a negative impact on clinical pregnancy outcomes after controlling for confounding factors such as age and ovarian reserve. However, intra-group self-control comparisons within the same population revealed that multiple attempts at COS could increase the likelihood of clinical pregnancy rates and live birth rates. This study will provide scientific evidence regarding the safety and efficacy of repeated COS for infertile patients, assisting clinicians in optimizing treatment protocols.

In conclusion, the number of COS cycles did not significantly adversely affect pregnancy outcomes across different populations. Self-controlled comparisons within the same population indicated that ovarian reserve function may remain stable after repeated COS (≤ 5 cycles), while repeated COS (≤ 4 cycles) may have a cumulative positive effect on clinical pregnancy outcomes. These findings support the recommendation for patients to consider multiple COS cycles and embryo transfers to improve their chances of success. Decisions regarding 5 or more cycles should be made cautiously, based on individual circumstances and medical advice. Future prospective studies with larger sample sizes and extended follow-up periods are needed to validate these findings and provide more definitive clinical guidance.

This study, through regression analysis across populations with varying numbers of COS cycles, found that the number of COS cycles did not significantly negatively impact pregnancy outcomes, while age and embryo transfer parameters showed significant effects. Subsequent self-controlled comparisons within populations undergoing different numbers of COS cycles revealed that repeated COS may not adversely affect ovarian reserve. Regression analysis further indicated that cumulative COS over the first four cycles positively influenced pregnancy outcomes, whereas no significant effect was observed in populations undergoing five COS cycles. The present study revealed that the number of COS cycles among different populations did not significantly impact pregnancy outcomes ( p  > 0.05). This finding suggests that repeated ovarian stimulation, within the range studied, does not inherently compromise the chances of achieving a clinical pregnancy, live birth, or increase the risk of miscarriage. The differences in clinical pregnancy rates, live birth rates, and miscarriage rates across varying numbers of cycles are likely attributable to factors such as age, BMI, endometrial thickness, and embryo-related variables. Age was a major factor, with each additional year reducing clinical pregnancy and live birth odds by 5% and 6%, respectively, while increasing miscarriage odds by 7%. Higher BMI also raised miscarriage risk by 5% per unit increase, whereas greater endometrial thickness improved clinical pregnancy and live birth odds by 5.4% and 6%, respectively. Additionally, embryo transfer stage and number significantly influenced clinical pregnancy. Homburg et al. [ 16 ] reported a considerable decrease in CPR after three cycles, but this study did not perform a further regression analysis to obtain more reliable results. Likewise, another study strengthened that repeated COS did not negatively affect CPR and LBR in patients experiencing three cycles and less until after > 3 cycles [ 1 ]. A multicenter study by Meldrum et al. [ 17 ] illustrated a significant reduction in CPR for > 4 cycles and LBR for ≥ 4 cycles. It is reassuring that the present study came up with the similar results. Based on the self-controlled comparison among populations undergoing different numbers of COS cycles, this study suggests that repeated COS-IVF/ICSI may not significantly impact ovarian reserve function. Ovarian reserve, reflecting the number of oocytes in the ovary, is primarily influenced by genetics, age, and environmental factors [ 18 ]. To enhance reliability, a population with multiple COS cycles was selected for self-controlled comparison. AMH and AFC were identified as the most representative, sensitive, and superior predictors of ovarian function compared to day-3 FSH [ 18 , 19 ], with other indicators serving auxiliary roles. No significant differences in AMH, AFC, basal FSH, or basal LH levels were observed across cycles in Groups B, C, D, and E, while basal E2 levels increased with cycle number. Basal E2 was deemed unsuitable as an ovarian reserve indicator but useful for interpreting normal basal FSH values [ 19 ]. These findings align with prior studies. Luk et al. [ 1 ] found no negative impact of repeated COS on ovarian reserve within ≤ 3 cycles, while Xu et al. [ 3 ] reported no significant effect up to 4 cycles, though AMH comparisons between cycles were not addressed. In the self-controlled comparison among populations undergoing different numbers of COS cycles, this study exhibited that a raise in cycle number significantly increases CPR in the cohort of ≤ 4 repeated cycles, improves LBR, and reduces EMR in cohorts with ≤ 3 repeated cycles. Similarly, another study reported a substantial increase in fertilization and CPR in patients with repeated COS cycles compared to unrepeated cycles saving did not further differentiate repeated cycles by number [ 8 ]. The validity of advising patients to actively try 4 attempts in this study can be derived from the above study results, which enlighten hope and encouragement in infertile patients. The statistical analysis results of pregnancy outcomes after the fifth cycle of COS are not particularly satisfactory. Patients are advised to make careful choices based on their individual circumstances and financial situation; however, it is important to acknowledge the possibility of achieving successful pregnancies and deliveries after five or more COS cycles. We observed a significant increase in CPR and a notable decrease in EMR during the final cycle across several repeated cohorts. This phenomenon may be attributed to several factors: first, physicians may adjust the ovarian stimulation protocol (e.g., drug dosage, choice of GnRH agonist or antagonist) based on the patient’s response in previous cycles, thereby optimizing ovarian responsiveness and oocyte quality. There was a significant enhancement in oocyte number, fertilization outcome, and embryo quality upon increasing the number of cycles beyond 3 COS repeats in egg donors [ 4 ]. Second, adjustments to the luteal support regimen (e.g., progesterone dosage or administration method) may improve endometrial receptivity [ 16 ]. Third, the optimization of embryo selection strategies, along with the patient’s gradual adaptation to the treatment process, reduced psychological stress, and more stable physiological responses to medications, collectively contribute to improved pregnancy outcomes [ 20 ]. Additionally, the cumulative effect of repeated COS cycles significantly enhances pregnancy rates, particularly in the final cycle [ 21 ]. The cumulative effect of three complete IVF/ICSI cycles (hereafter referred to as multiple cycles) is considered by NICE to increase the chance of successful pregnancy to 45–53% in infertile patients under 40 years of age, which is the most cost-effective and clinically efficient option [ 22 ]. Therefore, patients are encouraged to undergo multiple treatment cycles to achieve better pregnancy outcomes [ 23 ], which aligns with the findings and implications of this study. The limited number of patients undergoing > 5 COS cycles complicates statistical analysis, prompting the use of animal models for further investigation. Studies in mice revealed that repeated gonadotropin stimulation (4–8 times) caused ultrastructural changes, such as reduced ciliated cells and mitochondrial degeneration, impacting fertilization and embryo transport [ 24 – 26 ]. Additionally, repeated stimulation increased oxidative stress and spindle defects in oocytes, reducing Oct4 expression and reproductive performance [ 27 ]. In rhesus monkeys, repeated COS over 5 years impaired mitochondrial function in granulosa cells and reduced StAR and P450arom expression, potentially affecting ovarian function, but no significant structural damage was observed in uterine and mammary tissues [ 28 , 29 ]. While these animal models highlight potential adverse effects of repeated COS, the lack of severe phenotypic damage in clinical studies suggests that significant functional changes may only emerge over longer periods, warranting further research with larger sample sizes and extended follow-ups. The strength of this study is the utility of self-control comparisons since it diminishes the influence of numerous uncontrollable factors such as genetics and environment, making the results and conclusions more reliable. Second, the study used univariate comparisons and multifactorial logistic regression analyses to compensate for potential confounding factors’ effects on clinical pregnancy outcomes. Third, strict exclusion criteria were implemented to preclude patients with PCOS, endometriosis, and genetic factors for infertility and patients with ovarian hyporesponsiveness with FSH > 15 U/L and AFC < 3. Fourth, the study explored the effect of repeat COS on clinical pregnancy outcomes, an aspect less frequently investigated in articles, and scant information are available in publications. Moreover, to our knowledge, this is the largest sample size, and the subgroups are more detailed, making the findings more convincing. This study is a retrospective cohort design and has several inherent limitations. Patients who underwent multiple COS cycles may have systematic differences in baseline characteristics compared to those who discontinued treatment early, potentially leading to selection bias and an overestimation of pregnancy rates among those who persisted with treatment. To mitigate the impact of selection bias, we employed multivariate logistic regression analysis, controlling for known confounding factors (e.g., age, BMI, baseline hormone levels, etc.). Additionally, we conducted multiple self-controlled analyses across different patient groups with varying numbers of repeated cycles, yielding consistent results and enhancing the robustness of our findings. Furthermore, data collected from medical records may be subject to information bias, particularly for variables such as baseline hormone levels and endometrial thickness, which may have inconsistent documentation. To address this, we implemented a standardized data collection process and cross-verified the data by two independent researchers. Moreover, although ultrasound examinations were performed by sonographers trained under a single-center protocol, some inter-individual variability may still exist. Unmeasured confounders, such as transfer fluid volume, transferring depth, lifestyle factors, and certain genetic factors, may have influenced the observed associations. Therefore, future multicenter studies or prospective designs incorporating more comprehensive variables are needed to further validate our findings.

Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1