The GnRH antagonist protocol is recommended for IVF ovulation stimulation due to its comparable efficacy and higher safety compared to GnRH agonist protocols (1). GnRH antagonists bind to specific receptors on the pituitary gland, inhibiting endogenous luteinizing hormone (LH). During ovarian stimulation, GnRH antagonists prevent an early rise in LH, thereby reducing the cycle cancellation rate.

LH, secreted by the pituitary gland, is essential for normal follicular development and oocyte maturation in the menstrual cycle (2). Exogenous LH supplementation is necessary for folliculogenesis in cases of hypogonadotropic hypogonadism (3–5). However, the role of LH in embryo implantation remains controversial, particularly in women undergoing GnRH antagonist protocols (6–8). Indeed, GnRH antagonists induce pronounced suppression of LH during ovarian stimulation (9). Some studies suggest that profound suppression of LH during the follicular phase may negatively affect IVF outcomes with fresh embryo transfer (10–12). However, there is no consensus on the LH threshold that would be deleterious for embryo implantation. Zhou et al. suggested canceling fresh embryo transfers in the < 25th percentile group of LH levels (<1.62 mIU/mL in normal responders, <2.25 mIU/mL in PCOS women, and 2.14 mIU/mL in poor responders) (12). Conversely, other studies have not found such negative impacts of LH levels during the follicular phase (6, 13, 14). The impact of LH levels on endometrial receptivity could be either direct, through LH’s effect on the endometrium, or indirect, through estradiol or progesterone levels at the end of the follicular phase. Another hypothesis would be the direct impact of Antagonists on IVF results through action on the endometrium by immunomodulation.

This study aimed to evaluate the impact of LH levels at the end of the follicular phase on embryo implantation and live birth rate.

A total of 552 IVF cycles with fresh ET were included in the study. Preovulatory LH level data were missing for 8 cycles, which were excluded from the analysis. Consequently, 544 IVF cycles with fresh ET were analyzed. The women included had a mean age of 35.3 ± 4.6 years, a mean BMI of 25.8 ± 5.2 kg/m², and 17% were smokers (n = 94).

The IVF cycles were categorized into four groups according to the interquartile range of preovulatory LH levels: Q1 = LH < 1.7 UI/L; Q2 = LH 1.7-2.9 UI/L; Q3 = LH 2.9-4.8 UI/L; Q4 = LH > 4.8 UI/L (Figure 1). The characteristics of the women in each of the four groups are detailed in Table 1. No significant differences were found among the groups regarding age, BMI, tobacco use, duration of infertility, type of infertility (primary or secondary), and cause of infertility. The causes of male and female infertility were comparable across all groups. The percentage of women with diminished ovarian reserve was 31.3% (n = 31) in Q1, 43% (n = 40) in Q2, 40% (n = 40) in Q3, and 37.4% (n = 34) in Q4 (p= 0,5). The proportion of poor responders was comparable across the four LH quartiles (31–43%), with no statistically significant difference between groups. The percentages of tubal anomalies were 16.2% (n = 16) in Q1, 5.4% (n = 5) in Q2, 12% (n = 12) in Q3, and 6.6% (n = 6) in Q4, with no statistical significance.

The mean level of basal LH before stimulation was significantly higher in Q4 compared to Q. Mean AMH levels were 2.7 ± 2.3 ng/mL in Q1, 2.3 ± 2 ng/mL in Q2, 2.4 ± 1.8 ng/mL in Q3, and 2.4 ± 1.7 ng/mL in Q4 (P = 0.3). Significant differences were observed between Q1 and Q4 regarding preovulatory estradiol levels (1837.5 ± 1025.6 pg/mL vs. 2159.9 ± 827.2 pg/mL; P = 0.025) and between Q2 and Q4 in terms of the duration of stimulation (11.1 ± 1.7 days vs. 10.5 ± 1.7 days; P = 0.026) and the total dose of gonadotropin (3303.8 ± 1310.3 UI vs. 2875.4 ± 1291.9 UI; P = 0.04). Other COS parameters, such as preovulatory progesterone levels, endometrial thickness, type of gonadotropin used (with or without LH activity), and the method of ART, were comparable across all groups (Table 2).

The number of oocytes retrieved, and the number of usable embryos were similar across all groups. IVF outcomes are presented in Table 3 and Figure 1. After fresh embryo transfer, there were no significant differences among the four groups regarding implantation rate, clinical pregnancy rate, early pregnancy loss, and live birth rate.

In our study of 544 GnRH antagonist cycles, the late follicular LH level did not influence the implantation rate or LBR after fresh ET. The increase in preovulatory estradiol levels was associated with the preovulatory LH level. Our outcome supports the conclusion from other literature that shows pre-ovulatory estradiol levels do not make a difference on IVF outcomes (17–19). Despite the correlation between LH and estradiol levels, Huang et al. found no impact of peri-implantation estradiol levels on IVF pregnancy outcomes (18).

Taken together, these findings suggest that although LH and estradiol levels are associated, their fluctuations do not appear to significantly affect clinical outcomes in IVF.

Some studies have suggested that low LH levels at the end of the follicular phase might negatively affect the endometrium and IVF outcomes with GnRH antagonists and fresh ET (10–12, 20). If confirmed, this hypothesis could make low LH levels in the late follicular phase an indication for embryo freeze-all.

Shoham et al. proposed an “LH window” necessary for follicular development, emphasizing the need for a minimum circulating endogenous LH level for adequate ovarian steroidogenesis (21). Adding “LH activity” during COS is recommended for women with a poor response to stimulation despite normal ovarian reserve. Conforti et al. found that women with a hypo-response to ovarian stimulation supplemented with recombinant LH (r-LH) had significantly higher clinical pregnancy rates, implantation rates, and number of oocytes retrieved than those who underwent r-FSH alone (22).

However, our data, together with most recent findings, suggest that low LH levels in antagonist cycles do not necessarily compromise oocyte quality or endometrial receptivity, and therefore should not automatically trigger a freeze-all approach.

This body of evidence highlights that while low LH levels may raise concerns, the clinical need for intervention such as embryo freeze-all remains debatable and context-dependent.

Regarding studies that observed an impact of LH levels on embryo implantation, the debate focuses on the LH threshold with conflicting results (10–12, 20). Few studies have investigated the effect of LH levels on LBR after fresh embryo transfer with the GnRH antagonist protocol. Our findings are consistent with those of Merviel et al., who found no significant difference in clinical outcomes between different LH levels on the day of hCG administration, with an arbitrary threshold of 0.5 IU/L (6). Similarly, Griesinger et al. reported no association between LH concentrations on day 8 of stimulation and ongoing pregnancy rates (13). Bosch et al. as well as Doody et al., also observed no differences in clinical pregnancy rates (23, 24).

In a study of 426 cycles, Eftekhar et al. reported no relationship between LH levels during GnRH antagonist cycles and pregnancy outcomes, though they did observe a higher implantation rate in the LH 2.60–4.60 IU/L group compared to the LH <1.49 IU/L group, but no significant difference between the LH <1.49 IU/L group and the LH >4.6 IU/L group (14).

These consistent findings reinforce our conclusion that there is no clear or universally applicable LH threshold predictive of live birth in antagonist cycles.

Conversely, Luo et al. showed that LH levels <4 IU/L significantly reduced LBR (38% vs 51.5%; P<0.05) and increased early pregnancy loss rate after fresh embryo transfer, without affecting the implantation rate. They included 1480 women undergoing COS with GnRH antagonist, who were arbitrarily divided into “low” and “high” LH groups with a cutoff of 4 IU/L (20). In the study of Benmachiche et al., 322 infertile women were included and underwent IVF with GnRH antagonist protocol and fresh ET. Above an LH threshold of 1.6 IU/L, early pregnancy loss rate decreased and the ongoing pregnancy rate and LBR increased. The absolute difference between the highest LH group (LH >1.6 IU/L) and the lowest LH group (LH<0.6 IU/L) was 13.4%, 12.1%, and 12% in ongoing pregnancy rate, LBR, and early pregnancy loss rate, respectively (P<0.05) (11). These discrepancies may be explained by differences in study design and patient populations. Luo et al. included a larger cohort with a higher proportion of PCOS women and used an arbitrary LH cutoff of 4 IU/L, whereas Benmachiche et al. applied a lower threshold of 1.6 IU/L. In addition, the LH assays and the timing and dosing of antagonist administration varied between studies, which may also account for heterogeneity in the reported outcomes.

These discrepancies highlight that the thresholds proposed in the literature can be considered specific to the study population rather than universal biological thresholds.

In a retrospective study, Chen et al. found that LH ≤0.8 mIU/ml during COS with GnRH antagonist was associated with a significantly higher early pregnancy loss rate (31.1%, n=19 vs 16.6%, n=36; P = 0.012) but no significant differences in LBR or implantation rate (10). There is variability in LH levels during COS with the antagonist protocol, with differences described between the first day of the cycle and after introduction of GnRH antagonists (25). For long-agonist protocol, Lahoud et al. observed that a drop in LH levels of >50% from early to mid-follicular phase resulted in a significantly lower live birth rate per cycle (22.2% vs 15.8%; P<0.05) (26). Despite the use of the same dose of GnRH antagonists, women do not have the same pituitary response. Kol et al. defined “normal” responders as those with LH levels greater than 50% of the pre-injection level 24 hours after the first GnRH antagonist injection and “over-suppressed” if less than 50%. About a quarter of women were “over-suppressed” in their study, though they did not investigate the impact on LBR (25). This variability in LH levels may be due to interindividual differences in pituitary response to GnRH antagonists, potentially explained by polymorphisms in the GnRH receptor (GnRHR). In a study of 269 women, Weng et al. found that genetic variants in the GnRHR gene could modulate LH release and affect ovarian stimulation outcomes (27).

LH receptor polymorphisms appear to have no impact on clinical outcomes or IVF LBR. In a study of 1183 women in whom genotyping of the FSH receptor and LH receptor polymorphisms was performed, Pirtea et al. found no association between gonadotropin receptor polymorphisms and LBR or implantation rate (28).

Altogether, these findings suggest that variability in LH response may be influenced by genetic and pharmacological factors, but these do not consistently translate into differences in live birth outcomes.

To understand the mechanism of action of GnRH antagonists, Murase et al. reported that prolonged GnRH antagonist treatment inhibits LH beta subunit mRNA and GnRH-R mRNA expression, like GnRH agonists (29). In a review, Maggi et al. reported that the GnRH/GnRH-R system is also expressed in female reproductive tissues, in addition to the pituitary gland. The expression of this system in the human endometrium supports its physiological role in the processes of embryo implantation (30). A comparative transcriptomic analysis study of the endometrium in 9 women with adenomyosis before and after GnRH agonist treatment suggested that GnRH agonists significantly alter immune system-associated signal transduction in the endometrium, a hypothesis that remains unexplored for GnRH antagonists (31).

These mechanistic data highlight the need for future research investigating how GnRH antagonists might influence endometrial gene expression and immune modulation, potentially mediating subtle effects on implantation.

Despite the retrospective and monocentric nature of our study, as well as the heterogeneity of the population, our sample was sufficiently large, even though we had few women with LH levels below 0.6 IU/L.

We acknowledge as a major limitation the retrospective cohort design, which may introduce selection bias and limits causal inference.

Overall, our study adds value by providing an analysis of a large sample of late follicular phase LH levels, with a focus on the results of fresh embryo transfers, which helps to guide clinical decision-making. Our findings suggest that low late follicular LH levels alone should not be used as a criterion to systematically opt for a freeze-all strategy. Instead, clinicians can continue to perform fresh transfers even in cases of low LH, provided that no other clinical or biological factors indicate otherwise. This can help avoid unnecessary delays in treatment and reduce patient burden.

However, as this is a descriptive study based on statistical associations, it cannot establish a causal relationship between late follicular LH levels and IVF outcomes. Therefore, our results should be interpreted with caution. Further prospective multicenter and mechanistic studies are needed to confirm these findings and to better understand the pathways underlying LH variability and its potential impact on endometrial receptivity.

In conclusion, our study indicates that preovulatory LH levels during cycles with GnRH antagonists are not correlated with IVF outcomes. Therefore, monitoring LH levels during antagonist protocols may be of limited use in deciding between fresh transfer or freeze all, and a low LH level at the end of the follicular phase does not necessarily require freezing all.