Our analysis shows that, in patients undergoing COS with GnRHa trigger in a fertility preservation context, COS with letrozole supplementation (COSTLES) is associated to significantly lower progesterone levels the day after GnRHa trigger compared to COS with no letrozole. E2 levels were also significantly lower in case of letrozole administration compared to without.
To our knowledge, our study is the first to report data on progesterone levels in COSTLES and GnRHa trigger in a large cohort. A previous retrospective observational study reporting 3 cases of fertility preservation patients suggested that progesterone levels were higher after ovarian stimulation with letrozole in the middle luteal phase (21), but in one case, recombinant hCG had been used to trigger ovulation. Moreover, a long GnRHa following trigger (24 to 48h) had been administered, which may have induced a “flare-up” effect by stimulating progesterone production by the corpus luteum. A prospective observational study of 42 patients assessing progesterone levels on the day of hCG trigger, the day of oocyte retrieval, and 3 and 8 days after oocyte retrieval reported that progesterone levels were similar at all times between letrozole-associated COS and COS without letrozole administration (20). However, contrarily to our study, all patients had been triggered with recombinant hCG. Altogether, data on progesterone levels in this specific context are scarce and studies so far were performed on small effectives. Furthermore, despite the fact that GnRHa trigger is the most used and recommended in clinical practice in this context, studies so far were based on patients who had received hCG trigger and/or supplementary treatments during or after stimulation which could induce bias.
Lower progesterone levels after GnRHa trigger compared to hCG trigger may be explained by the fact that the use of GnRHa trigger removes the GnRH antagonist from the GnRH receptor and induces a release of endogenous LH and FSH. LH has a shorter half-life than exogenous hCG, thus resulting in a lower luteotropic stimulation with faster luteolysis and lower serum progesterone levels (23, 24). In COSTLES, lower progesterone levels might be further explained by an increase in progesterone metabolism due to the endogenous inhibition of E2 synthesis. Indeed, letrozole inhibits aromatase that converts androgens into E2, resulting in a decrease in E2 synthesis. In ovarian steroidogenesis, E2 inhibits two enzymes of cytochrome P450 C17 17 α-hydroxylase and 17,20 lyase that metabolizes progesterone to androgens(25). Hence, lower E2 levels may induce a lower inhibition of these two enzymes, resulting in lower progesterone levels (25). As such, patients with aromatase deficiency have signs of hyperandogenism (26). Finally, the higher LH surge in case of letrozole supplementation might stimulate the progesterone metabolism into androgens by its action on cytochrome P450 C17. Indeed, LH could inhibit TGF beta which is a powerful inhibitor of the P450 C17 enzyme that metabolizes progesterone in androgen (27–29).
As expected, E2 levels one day after GnRHa trigger were significantly lower in the COSTLES group. Thus, the higher LH surge following GnRHa administration in the COSTLES group may be associated to a weaker negative feed-back of E2 on the hypothalamic-pituitary axis. The mean number of oocytes recovered and vitrified at metaphase 2 stage were similar in both groups, as related by Oktay et al. (13). In our subgroup analysis of hormone profile according to stimulation starting phase (follicular or luteal), progesterone was significantly higher in case of luteal phase stimulation start in the Control group, with no difference in LH levels. These results are consistent with previous studies observing a correlation between progesterone levels at the end of the follicular phase and levels of progesterone after trigger by GnRHa (30). E2 levels are similar with results of Pereira et al.(31) comparing standard stimulations and random start, in which E2 levels after triggering were similar regardless of the early stage of the stimulation phase (1781 versus 1784 pg/mL). In the COSTLES group, progesterone and LH levels were not significantly
different depending on the stimulation start phase. Nevertheless, the E2 levels were significantly different with higher levels when COS was started in the follicular phase.
So far, safety data using COSTLES in BC patients are reassuring. Indeed, a prospective controlled study including 337 BC patients, of which 120 had COSTLES for FP prior to chemotherapy, reported no increased risk of recurrence after a mean follow-up of 5.0 years in case of COSTLES compared to BC patients in which no FP had been performed (32). Similarly, Azim et al.’s analysis of 215 BC patients, of which 79 had COSTLES, suggested that COSTLES did not adversely affect the risk of reccurrence nor survival outcomes compared to controls after a median follow-up of 23.4 months (33). Furthermore, the use of a GnRHa trigger in COSTLES cycles appears to have other advantages, as it was shown to significantly reduce the risk of ovarian hyperstimulation syndrome (OHSS) compared to hCG trigger in 129 BC patients undergoing COS for FP purposes (p = 0.032) (34). Consistently, Oktay et al. reported that the incidence rate of OHSS was lower in COSTLES cycles of BC patients triggered by GnRHa compared to hCG (35). In all, although there is no clinical evidence suggesting that the brief elevation of progesterone levels after hCG trigger impairs the survival of BC patients, it seems that the use of GnRHa trigger should be used in COSTLES cycles given its advantages and the reassuring safety data available so far.
Some limitations should be considered when interpreting results of our study. Progesterone levels were only assessed once, the morning after GnRHa trigger. Hence, our results show that progesterone levels are lower on the day after GnRHa trigger but do not contribute to whether progesterone levels remain consistently lower afterwards. Nevertheless, progesterone synthesis is known to increase in the early phase after triggering. Moreover, according to the study by Vuong et al. (36), progesterone levels assessed 12 hours after recombinant hCG are correlated with progesterone in the mid-luteal phase. Thus, we can assume that progesterone levels in our study might be extrapolated to progesterone levels during the following days in mid-luteal phase. Larger effectives are also required to confirm results of our analysis on hormonal levels depending on the phase during which COS was started.