Increasing dominant follicular proportion negatively associated with good clinical outcomes in GnRH-a prolonged protocol: a large-sample retrospective analysis

7 Background: Nowadays, there is no universal criteria for trigger time during controlled ovarian 8 hyperstimulation (COH). Particularly, in the so-called GnRH-a prolonged protocol, widely used in 9 China, the ideal time to trigger ovulation is not yet well defined. 10 Methods: This was a large-sample retrospective analysis. Between January 2016 and January 2020, 11 1,925 young patients who underwent their first in vitro fertilization (IVF)/intracytoplasmic sperm 12 injection (ICSI) cycles with normal ovarian response were divided into three groups based on their 13 dominant follicular proportions (DFP, defined as ≥ 18 mm follicles / ≥ 14 mm follicles; Group A: < 14 30%; Group B: 30% - 60%; Group C: ≥ 60%). Binary logistic regression and multivariate linear 15 regression were used to assessed whether DFP levels were related to clinical pregnancy, number of 16 blastocysts frozen, blastocyst formation rate and low blastocysts frozen. 17 Results: Binary logistics regression analysis showed that compared with Group A, the OR for 18 clinical pregnancy was 1.345 in Group B (P = 0.023); however, there was no statistical difference 19 between Group C and Group A (P = 0.216). On one hand, multivariate linear regression analysis 20 indicated that DFP was negatively associated with number of blastocysts frozen (β ± SE: Grou p B vs 21 Group A = -0.319 ± 0.115, P = 0.006; Group C vs Group A = -0.432 ± 0.154, P = 0.005) as well as 22 blastocyst formation rate (β ± SE: Group B vs Group A = -0.035 ± 0.016, P = 0.031; Group C vs 23 Group A = -0.039 ± 0.021, P = 0.067). On the other hand, compared with group A, the OR for low 24 blastocyst frozen was 1.312 in Group B (P = 0.039) and was 1.417 in Group C (P = 0.041). 25 Conclusions: Excessive delay of trigger in GnRH-a prolonged protocol might reduce the 26 developmental potential of oocytes and reduce the number of available blastocysts, which might 27 result in a lower cumulative pregnancy rate. But further confirmation by strict prospective randomized controlled study should be needed. Retrospectively registered.

blastocyst formation rate (β ± SE: Group B vs Group A = -0.035 ± 0.016, P = 0.031; Group C vs 23 Group A = -0.039 ± 0.021, P = 0.067). On the other hand, compared with group A, the OR for low 24 blastocyst frozen was 1.312 in Group B (P = 0.039) and was 1.417 in Group C (P = 0.041). 25 Conclusions: Excessive delay of trigger in GnRH-a prolonged protocol might reduce the 26 developmental potential of oocytes and reduce the number of available blastocysts, which might 27 result in a lower cumulative pregnancy rate. But further confirmation by strict prospective 28 randomized controlled study should be needed. 29 Trial registration: https://clinicaltrials.gov/; NCT03305510; Registered 08 October 2017 -30 Retrospectively registered. 31 Keywords: IVF/ICSI, controlled ovarian hyperstimulation, HCG trigger time, GnRH-a prolonged 32 protocol, dominant follicular proportion, clinical outcomes. 33 Background 34 COH is considered a key factor in the success of IVF/ICSI [1,2], because it induces the development 35 of multiple follicles and obtains as many high-quality oocytes as possible, thereby increasing the 36 numbers of available embryos for transfer and increasing pregnancy rates. A crucial step in 37 improving clinical pregnancy outcomes, however, is identifying an appropriate time for human 38 chorionic gonadotropin (HCG) trigger [2,3]. 39 5

Controlled ovarian stimulation protocols 83
For GnRH-a prolonged protocol, 3.75 mg of long-acting Ferring,84 Switzerland) was administered subcutaneously on day 2 of the menstrual cycle. Pituitary suppression 85 was evaluated 28 days after pituitary downregulation. The criteria for confirming the success of 86 downregulation were as follows: follicle diameter <5 mm, serum luteinizing hormone (LH) <5 87 mIU/ml, serum estradiol (E2) <50 pg/mL, and endometrial thickness < 5 mm. Then, daily injection 88 of rhFSH (Gonal-F; Merck-Serono, Geneva, Switzerland) ranging from 75 to 300 IU was given for 89 about 10 days. When at least three leading follicles ≥ 17 mm or two leading follicles ≥ 18 mm were 90 observed via transvaginal ultrasound, a dose of 0.25mg rHCG (Ovidrel; Merck-Serono, Geneva, 91 Switzerland) was administered to trigger ovulation. After 36 -37 hours, transvaginal ultrasound-92 guided oocyte retrieval was conducted. Fertilization was accomplished by standard IVF or ICSI. If 93 serum progesterone <1.5 ng/mL [17], the number of oocytes retrieved were <20, and serum E2 were 94 <7,000 pg/mL, the one or two best-quality Day 3 embryos were transferred; and the remaining 95 embryos were cultured to Day 5/6 until the blastocyst formed and then frozen. The luteal phase was 96 supported if embryo transfer was performed. 97

Definition of DFP Levels and Groups 98
Dominant follicular proportion (DFP) is defined as ≥18 mm follicles /≥14 mm follicles. DFP < 99 30% corresponds to at least two follicles ≥18 mm on HCG day, which is the most common trigger 100 time. Then, we set the groups as DFP <30% (Group A), 30% -60% (Group B), ≥ 60% (Group C) 101 corresponding to follicles ≥ 18 mm increasing gradually (2, 4, 6 follicles respectively). 102 Clinical outcomes 103 6 The number of mature oocytes (MII) was measured 3 -4 hours, and fertilization was assessed 16 -18 104 hours after IVF insemination or when ICSI was performed. Oocyte maturation rate was the 105 proportion of MII oocytes to the number of oocytes retrieved. Normal fertilization rate was equal to 106 2PN/number of oocytes retrieved (IVF) or 2PN/MII (ICSI). The blastocyst formation rate was equal 107 to the number of blastocysts frozen divided by the number of blastocysts continuously cultured. Low 108 blastocysts frozen was defined as the number of blastocysts frozen ≤ 1 (that is, less than half of the 109 average number of blastocysts frozen). Clinical pregnancy was identified with the presence of an 110 intrauterine gestational sac with fetal cardiac activity. The clinical pregnancy rate was the number of 111 clinical pregnancies in a given number of transplant cycles. Implantation rate reflects the number of 112 gestational sacs divided by the number of embryos transferred. 113

Statistical analysis 114
Pearson chi-square (χ2) test on categorical variables and analysis of variance (ANOVA) or  Wallis H test on continuous variables were performed appropriately. Multivariate linear regression 116 analyses were carried out for the predictive factors of number of blastocysts frozen and blastocyst 117 formation rate. Moreover, binary logistic regression analyze was carried out for the predictive factors 118 of clinical pregnancy and low blastocyst frozen. The results were given in terms of 95% confidence 119 intervals (CI) and P values. A two tailed P-value of < 0.05 indicated statistical significance. All 120 statistical analyses of the data were carried out using the Statistical Package for Social Science 121 version 25.0 (SPSS, Chicago, IL, USA). 122

123
A total of 1,925 consecutive IVF / ICSI cycles were included. As shown in Table 1, the clinical 124 characteristics, ovarian response characteristics, and reproductive outcomes of cycles were described. 125 The baseline characteristics and ovarian stimulation details were matched evenly in three groups, 126 7 including maternal age, BMI, basal FSH, dose of Gn, duration of Gn. Among the DFP groups, there 127 were no significant differences in terms of oocyte maturation rate (IVF/ICSI), normal fertilization 128 rate (IVF/ICSI), and clinical pregnancy rate. Group B had the highest implantation rate (Group A 129 48.67%, Group B 57.57%, Group C 50.71%, P = 0.031). Interestingly, Group  Multivariate linear regression analyses were carried out to evaluate the effect of type of infertility, 139 dose/duration of Gn, number of oocytes retrieved, type of fertilization and DFP levels on the number 140 of blastocysts frozen and blastocyst formation rate, respectively. As shown in Table 3, the results 141 indicated that DFP was negatively associated with number of blastocysts frozen (β ± SE: Group B vs 142 Group A = -0.319 ± 0.115, P = 0.006; Group C vs Group A = -0.432 ± 0.154, P = 0.005). As shown 143 in Table 4, the results indicated that DFP was also negatively associated with blastocyst formation 144 rate (β ± SE: Group B vs Group A = -0.035 ± 0.016, P = 0.031; Group C vs Group A = -0.039 ± 145 0.021, P = 0.067). 146 Binary logistic regression analysis was carried out to evaluate the effect of type of infertility, 147 dose/duration of Gn, number of oocytes retrieved, type of fertilization and DFP levels on low 148 blastocysts frozen ( Table 5). The results showed that compared with Group A, the OR for low 149 8 blastocyst frozen was 1.312 in Group B (95% CI = 1.014 -1.698, P = 0.039) and was 1.417 in Group 150 C (95% CI = 1.014 -1.979, P = 0.041). 151

152
LH plays an essential physiological role in follicle steroidogenesis and development, as well as 153 oocyte maturation [20,21]. Since the introduction of GnRH-a prolonged protocol, pituitary 154 desensitisation has usually been profound and endogenous LH level has been suppressed to be very 155 low (<1.0 IU/L). There is not any data available in published literature in regards to when to 156 administer HCG trigger in GnRH-a prolonged protocol. Some clinicians believe that the timing could 157 be determined using the same criteria as are used in short GnRH-a long protocol. However, the 158 traditional HCG trigger timing criteria are not strict, and the decision remains controversial [2 -5]. 159 The previous views were that follicular size was positively related to follicular maturity, fertilisation 160 and subsequent development [1, 2, 5, 12]. Oocytes derived from large follicles (14~21 mm; mean 161 diameter: 19.1 ± 2.1 mm) seem to be more inclined to form high-quality embryos in GnRH 162 antagonist protocol [13]. Therefore, in actual practice, the HCG trigger time is usually delayed [8]. 163 On the contrary, a previous study reported that oocytes in oversized follicles in the same protocol 164 might decrease in quality, and the recovery of oocytes [5]. However, our results showed that the DFP 165 groups did not differ in terms of oocyte maturation rate or normal fertilisation rate. This was 166 consistent with the views of previous studies: enlarging follicle size might not improve oocyte 167 maturation or fertilisation [4, 14 -16]. 168 In our study, the clinical pregnancy rate and the implantation rate seemed to decrease as DFP 169 increased (Group B vs Group C). This could be explained by the fact that a larger DFP might be 170 negatively associated with satisfactory pregnancy outcomes. Consistent with our results, one study 171 showed that high-quality embryo rate, pregnancy rate, and implantation rate were significantly higher 172 9 in the low proportion group (diameter ≥ 18 mm divided by the total number of follicles, low 173 proportion: <15%; middle proportion: 15 -27%; high proportion: >27% in short GnRH-a long 174 protocol) [8]. In another two randomised controlled trials of GnRH antagonist cycles, enlarging 175 follicle size by delaying HCG administration by one or two days after the time that three follicles had 176 reached 17 mm, corresponded with a decrease in ongoing pregnancy rates in the delayed group [4, 177 14]. Availability of surplus embryos for freezing was lower when delaying two days compared with 178 delaying one day, but this did not reach statistical significance [14]. Interestingly in our study, as 179 DFP increased, the number of blastocysts frozen and blastocysts formation rate decreased 180 significantly. Multiple linear regression results showed that the number of blastocysts frozen and 181 blastocysts formation rate were negatively associated to DFP. Furthermore, increasing DFP is a risk 182 factor for low blastocysts frozen. These results revealed that the overgrowth of dominant follicles 183 might lead to oocyte post-maturity, which in turn could have an inverse impact on the quality of 184 oocytes and ultimately lead to unsatisfying pregnant outcomes [8,18]. This phenomenon may be 185 related to the increased incidence of ultrastructural abnormalities in the oocytes, for example, the 186 appearance of degenerate organelles-smooth surface endoplasmic reticulum (sER) [22]. Embryos that 187 accumulate the sER may have a low rate of blastocyst formation and poor pregnancy outcomes [23]. 188 It is believed that sER aggregation is related to high E2 levels on HCG day and long-term Gn 189 stimulation [22]. From the results of clinical pregnancy and available blastocysts, it is speculated that 190 excessive delay of the trigger might negatively affect the cumulative pregnancy rate. But further 191 prospective study should be needed. 192 There are several advantages and limitations in our study. The most important innovation is the 193 discovery that the larger the DFP, the smaller the number of blastocysts frozen and blastocyst 194 formation rate. In other words, increasing DFP is a risk factor for low blastocysts frozen, and is also 195 negatively associated with good clinical outcomes in general. In addition, our study is the first to 196 10 analyse the GnRH-a prolonged protocol in terms of HCG trigger time. Finally, for the analysis of the 197 DFP groupings and clinical outcomes, we used a large panel of data, established multiple linear 198 regression and binary logistic regression, and conducted a thorough and comprehensive evaluation of 199 these relationships. To a certain extent, it would provide useful information for clinical decision-200 making. The limitations are as follows: First, due to its retrospective nature, some confounding bias 201 may exist. Second, we lack cumulative pregnancy rate results. It is hard to query and count the data 202 for such a large sample. In addition, we screened standardised young patients with normal ovarian 203 responses. The results may be not applicable to older patients or patients with low or high ovarian 204

responses. 205
In all, pursuing more and larger follicles may be of no benefit to clinical outcomes. This practice 206 might increase patient cost and time to a certain extent, without increasing the oocyte maturation rate, 207 normal fertilisation rate, or the number of embryos available for transfer. It might be preferable to 208 trigger as early as possible with patients who have not formed any available blastocysts (surprisingly, 209 these patients account for 1/5 calculated from our data), to avoid diminishing oocyte and embryo 210 quality, and avoid consequently poor IVF/ICSI outcomes. 211

223
The datasets used and/or analysed during the current study are available from the corresponding 224 author upon reasonable request. 225