Progesterone is essential for the development of a receptive endometrium for implantation, but high circulating progesterone levels at the end of the follicular phase in COH-IVF cycles accelerate the secretory transformation of the endometrium and negatively affect pregnancy rates [10]. An increased P level decreases chances of implantation by altering uterine-embryo synchrony. Current clinical practice focuses on a peak P level as a threshold-measure included in guidance for the recommendation for a “freeze-all” approach. There are likely some patients who are encouraged to defer fresh ET due to surpassing an established P threshold, who rather had a high baseline P and minimal progesterone rise, and may have benefited psychologically and financially from proceeding with fresh ET. Additionally, an absolute P threshold is fallible as progesterone assays vary between laboratories, and the threshold for an optimal serum P level may also vary between clinical locations as multiple threshold values have been studied in the literature. However, prior to this study, the predictive capacity of the relative change in P level from baseline to time of trigger (?P) in relation to pregnancy rate was not known. Furthermore, the optimal range for pregnancy outcome in regard to peak P in our study was wide (0.15–1.35 ng/mL), emphasizing the potential weakness of using a threshold peak P as opposed to a more individualized measure which considers a patient’s baseline P level. Our results suggest that ?P is more predictive than peak P. Specifically, based on our findings, a ?P of 0.7–0.85 ng/mL appears to be the “ideal” change in progesterone during COH-IVF cycles, which results in optimal clinical pregnancy rates. Therefore, the ?P level may have a higher utility in determining whether to proceed with fresh ET and can allow for individualized approach to decision making in these situations.

Moreover, the infertility specialist may intervene to prevent an unwanted rise in progesterone level by considering the ratio of gonadotropins used in the COH-IVF cycle. Our study highlights the importance of balanced use of hMG and rFSH, and specifically the incorporation of LH activity in COH regimens in regard to not only the subsequent progesterone rise but also pregnancy outcomes (CPR, LBR), which has not been previously described. Within the ovarian follicle, LH stimulates theca cells to produce androgens by the induction of genes involved in steroidogenesis, including cytochrome P450 CYP 17 hydroxylase and 17–20 lyase, which convert progesterone and pregnenolone to 17-hydroxylated products and androgens via the ?4 and ?5 pathways, respectively [11]. Further, FSH stimulates P synthesis within the granulosa cell without luteinization via up-regulation of 3β-hydroxysteroid dehydrogenase, converting pregnenolone to P [12]. Thus, a lack of LH in the late follicular phase of stimulated cycles likely allows for follicular progesterone production to exceed the limit of LH activity and contributes to a premature progesterone rise [13]. Indeed, a randomized trial comparing serum and follicular fluid levels of progesterone of patients treated with either rFSH or hMG found that the use of rFSH in the gonadotropin regimen as opposed to hMG results in higher progesterone levels, further supporting the theory that folliculogenesis in the absence of sufficient LH activity leads to premature granulosa cell luteinization [14]. Thus, the need for a balanced activity of both FSH and LH activity during COH-IVF cycles is intuitive. However, the “ideal” ratio of hMG:rFSH (a modifiable factor in COH-IVF cycles) specifically one which can achieve an ideal ?P of 0.7–0.85 and result in optimal pregnancy outcomes following fresh embryo transfer, has not been determined. While one prior study has evaluated gonadotropin ratios relative to progesterone rise and identified a ratio of 0.3–0.6 as optimal, in that study, their ratio was of ‘LH activity’ to ‘FSH activity’ as opposed to a ratio of units of dosed hMG to rFSH as in the present study [9]. In this study, for ease of use clinically, we used the gonadotropin regimen ratios of hMG (with 1:1 LH:FSH activity) and rFSH dosages and we show that an hMG:rFSH ratio of < 0.6 is associated with the greatest risk of premature progesterone rise. The finding that regimens with the lowest LH activity had the highest levels of unwanted progesterone rise is consistent with the results seen in the prior study, Werner et al. Unlike the prior study, we further identified ratios which are predicative of pregnancy and live birth rates, showing that an hMG:rFSH dosing ratio of between 0.3 and 0.6 is associated with the highest LBR; a smaller range of 0.3–0.4 is additionally associated with the highest CPR, which has not been reported previously. Thus, appropriate proportional use of both hMG and rFSH, specifically incorporating LH activity, should be considered in all fresh IVF cycles in order to achieve optimal outcomes. We acknowledge the limitations of our relatively small sample size which may explain the lack of overlap between the hMG:rFSH ratio minimizing ?P (> 0.6) and CPR (0.3–0.4). We emphasize that the ratio of 0.3–0.6 was predictive of LBR, which is somewhat consistent with the determined hMG:rFSH ratio of > 0.6 that best predicts a lower ?P. While CPR did not overlap, it not as clinically relevant clinical as LBR. A larger number of patients is needed to clarify the relationship or exact threshold.

A strength of our study’s design is its clinical utility, as its observational design demonstrated values determined from women who ultimately chose to undergo fresh ET. Given the average age of 35 with BMI 26 and 44% had tubal or male factor infertility, our cohort represents patients with a generally good prognosis. Of note, or sample did include a broad range of infertility diagnoses including 10% having DOR and 12% having PCOS. Thus, these findings are reflective of a representative patient sample who indeed elect for fresh ET, and may serve as a guide for clinical decision making for similar eligible patients who prefer to proceed with fresh ET. Additionally, our observational design and resulting exclusion of patients who did not undergo fresh ET resulted in a more conservative threshold for progesterone level, as our levels of peak and calculated ?P are likely lower than what would be seen across all patients who underwent COH-IVF cycles. There is unlikely to have been selection bias for fresh embryo transfers unique to our clinic, as our clinic performed approximately equivalent proportions of fresh and frozen embryo transfers in the time period of this study (2017–2018): 50.1% fresh ETs and 49.9% FETs.

A more robust, prospective study with an interventional design, assigning stratified ratios of gonadotropins and then having all participants proceed with fresh ET despite regardless of peak or ?P, would not be feasible in a typical infertility patient population due to the potential for adverse effects on pregnancy rates.

Moreover, previous studies have demonstrated that the adverse effects of progesterone rise are of most significance for women with low and intermediate ovarian response to COH as opposed to high response; these women are considered to be particularly susceptible to the negative effects on implantation from a premature progesterone rise [15]. Our study considered the contribution of this factor in our analysis, controlling for this variation in ?P as it relates to follicle and retrieved oocyte counts. Finally, while by its nature, our retrospective study does not delineate a causative relationship between ?P and pregnancy outcomes, such a relationship would be further supported by gene expression studies demonstrating changes in endometrial receptivity profiles, similar to those studies comparing endometrial samples of specific progesterone thresholds in women undergoing COH-IVF cycles [2,3,4,5].