In recent decades, infertility has become a public health problem, with one out of six couples experiencing the absence of natural conception over a 1-year period (1). In 2019, more than 1 million treatment cycles, across 40 European countries were reported to the European IVF-monitoring (2). Infertility treatment success depends largely on a successful ovarian stimulation with increasing live birth rates after a fresh transfer obtained for up to 15 oocytes, reaching a plateau between 15 and 20 oocytes before decreasing, based on a large real-world cohort (3). Some complications can occur during fertility treatment, with the most reported complication being ovarian hyperstimulation syndrome (OHSS) (2). Thus, several methods have been recently discussed to limit the incidence and severity of OHSS, among them the personalization of the starting dose (4).

Since the early years of assisted reproductive technology (ART), patients have been classified as “low, good, or high” responders at the first attempt of treatment. Although these terms are widely used in clinical practice, there is still no consensus as to their exact definitions. Indeed, despite its importance to ART, the methodology for classifying high responder patients remains heterogeneous (5). Although the Bologna criteria have proven useful in identifying patients with low response, these criteria have been challenged because they carry the risk of grouping together women with very different biological characteristics, whose treatment calls for different strategies. Subsequently, the POSEIDON classification proposed a different classification including, besides ovarian reserve markers, two age categories and oocyte output at the previous attempt (6).

In routine practice, the classification of treatment-naive patients is based on the assessment of the ovarian reserve. The combined use of an ultrasound marker, antral follicle count (AFC) and a biochemical marker, anti-Müllerian hormone (AMH), is considered as a good predictor of ovarian response (7, 8). Other parameters such as age, weight, menstrual cycle length, basal follicle stimulating hormone (FSH), and estradiol (E2)/FSH at the beginning of a cycle, can be used to establish the stimulation protocol and doses.

However, limited data are provided to clinicians on the management of the various aspects of ovarian stimulation. In 2020, the European Society of Human Reproduction and Embryology (ESHRE) working group on ovarian stimulation therefore drew up guidelines based on the data available in the literature. The level of evidence was very often limited and of low quality: of the 61 recommendations developed, nine out of ten were backed by low or very low-quality evidence (8).

In this context, the aim of this study was to provide an experts’ consensus of best practices regarding personalization of the gonadotropin dose by classifying patients undergoing ART in three distinct ovarian response profiles.

This AMPLITUDE consensus brings new insights in clinical practices for the ovarian stimulation management. This study involved a large panel of fertility specialists practicing in France and Belgium. Despite the two rounds of voting, some propositions did not reach a consensus (agreement or disagreement). However, according to the literature, the optimal number of rounds seems to be 2 or 3, as a larger number of rounds will induce participant fatigue, questioning the relevance of obtained responses (10). Following this, a two-round Delphi consensus was conducted in this study. We can notice that several statements reached a consensus with a strong level of agreement or disagreement, improving the reliability of the results.

Firstly, this work strongly suggests that the initial dose of gonadotropins administered should be personalized for each patient profile. According to the literature, a daily dosage of 150 IU is considered as the standard dose. However, a subset of patients will have a low or a high response with this dosage. Thus, the objective is to obtain an optimal response, in order to improve the live birth rate and reduce the risk of hyperstimulation (11). Yet, the personalization of the initial gonadotropin dose implies a dose-response relationship between ovarian response and FSH dose. Despite limited data, the literature recommends personalizing the starting dose according to the patient profile (225–300 IU, 150–225 IU, and 100–150 IU in low, normal, and high responders, respectively) (11).

In this Delphi consensus, the panel of experts suggested that other parameters can be used concomitantly with AFC and AMH level for personalizing the gonadotropin dose: age, weight, BMI, nature of the cycle (ovulatory, dysovulatory, anovulatory), and the decision to adopt a fresh transfer or a freeze-all strategy, in all patient profiles. However, the experts do not recommend taking into account the patient history (gynecological and obstetrical histories reaching a disagreement consensus when no consensus being reached for general history, endometriosis, and parental project).

In the literature, none of the ovarian reserve tests is completely comprehensive in terms of sensitivity or precision but AFC and AMH are in the vast majority of cases good predictors of ovarian response (7, 8). There was a high agreement consensus for age as an important parameter for personalizing the gonadotropin dose. However, although age was also used to categorize responders’ profiles, especially the low responders (6, 12), age seems to be a better predictor of a successful pregnancy than of oocyte yield (7). Moreover, experts’ responses were aligned with the literature for weight and BMI that are classically described to influence fertility (13) and outcomes of infertility treatments (14, 15).

The absence of consensus can be explained for patients with endometriosis. Indeed, it is described that endometriosis alters ovarian function and can lead to a diminished ovarian reserve (16–18). Ovarian response after ovarian stimulation was also reported to be lower in women with ovarian endometriomas after adjusting for age, gonadotropin dose and AMH (17). However, as this impact depends on the localization of the lesions (ovarian lesions, especially), it can be assumed that some health professionals do not consider endometriosis when classifying their patients as hypo, normo-, or hyper-responders, as in some cases this pathology may have no effect on ovarian reserve.

In this Delphi we also questioned the relevance of a starting dose higher than 300 IU in low responders. Indeed, extremely high doses of gonadotropins have been used for decades in these patients. However, some studies described that FSH dosage beyond 300 IU do not improve ovarian response (19–21). Respondents agreed with this statement, suggesting that the maximum initial dose of gonadotropins for a low-responding patient should not exceed 300 IU. Several data in the literature support this result, including in first instance, the last ESHRE guidelines where a gonadotropin dose higher than 300 IU is not recommended for patients predicted as “poor responders” (8).

Respondents preferred a fresh transfer over freeze-all, when initiating ovarian stimulation in low and normal response patient profiles. Several studies evaluated the freeze-all benefits in patients undergoing IVF. The results showed that this strategy does not improve IVF outcomes in low responders (22). Data are controversial for normal responders: some authors described a positive effect on pregnancy rate in this population (22) when others concluded that freeze-all was not superior to fresh transfer for patients with normal ovarian response to stimulation (23). In high responders, data from the literature suggest beneficial effects of freeze-all, improving pregnancy rates (24).

Concerning the use of freeze-all when some events occur during stimulation, the panel of experts strongly recommended this option in case of high risk of OHSS, inadequate endometrium, new tubal or uterine cavity pathology, and premature elevation of progesterone. According to the literature, the freeze-all strategy was initially developed as a “rescue strategy” to counter the risk of OHSS in patients undergoing excessive ovarian response. Over the last decade, its indications were extended to other clinical conditions, i.e., endometrial, tubal, and uterine factors discovered during stimulation (thin endometrium, polyps, hydrosalpinx, endometritis, cervical anatomical features, cervical stenosis); slow-developing embryos or inadequate progesterone levels at the end of the follicular phase (however, various cut-offs have been reported, from 0.8 ng/ml to 1.8 ng/ml) (25–27).

This work then focused on the different trigger criteria, that appear to be consensual for the criterion of 3 follicles measuring ≥17 mm for a fresh transfer following an antagonist protocol according to the respondents. It is now well described that follicle size is linked to oocyte maturity. According to the literature, reaching the optimal follicular size on the day of oocyte retrieval would most likely yield a mature oocyte. However, although this optimal size range varies from one study to another: [12–19 mm]; [16–23 mm]; [23–28 mm] (28, 29), the 17 mm threshold seems to be the most commonly used criteria in studies and in daily practice, especially for antagonist protocols (29, 30).

Moreover, in the context of antagonist protocols, the use of GnRH agonist trigger was the preferred option if there are ≥18 follicles ≥11 mm as a predictor of severe OHSS or if the estradiol level is ≥5,000 ng/ml and this procedure should be systematically followed by a freeze-all according to respondents. These results agree with the literature where it is described that GnRH agonist trigger and a subsequent freeze-all strategy, when ≥18 follicles and E2 >5,000 ng/ml on the day of trigger, prevent the risk of OHSS (31–34).

According to the panel, the optimal oocyte target was 10–15 oocytes retrieved, at the first round of voting. However, another oocyte target reached a consensus at the second round: 15–20 oocytes. A possible explanation for this discordance may be due to the fact that the optimal oocyte target would be in the middle (around 15 oocytes) or to the lack of specification about the finality of the cycle, i.e., fresh or frozen transfer. Moreover, only two proposals that failed to reach consensus (5–10 and 15–20 oocytes) were put to the vote in the second round. It is therefore likely that respondents were encouraged to vote on these two proposals, in the absence of the more consensual proposal (10–15 oocytes).

In a Swedish study including 39,387 women undergoing an IVF procedure, the median number of oocytes retrieved was nine, with 0.5% OHSS. Authors also described that (1) live birth rate increased up to 11 oocytes retrieved and then levelled off and (2) the rate of severe OHSS increased with the number of oocytes retrieved, reaching 1% for 18 oocytes retrieved. These results suggest a shift in the balance between efficacy of treatment and patient safety regarding severe OHSS at approximately 18–20 oocytes (35–38). However, this refers to the live birth rate in a fresh transfer, while the cumulative live birth rate could be considered more relevant (39) both in a patient's and cost effectiveness perspective.

In some cases, an unexpected excessive ovarian response can occur. For the subsequent attempt, experts recommend decreasing the gonadotropin dose and to favor an antagonist protocol. Data from the literature are consistent with this result, as gonadotropin dose reduction is among the different strategies which should be considered for women with an unexpected high response (36). As well, antagonist protocols reduce the risk of severe OHSS by approximately 50%, compared to agonist protocols (34, 40).

In the context of hyper response, the necessity of performing a freeze-all approach without modification of the protocol is not consensual among the panel of respondents. Limited benefits of changing protocol after a first attempt have been described in the literature (41). Nevertheless, although freeze-all is the gold standard strategy to reduce OHSS for patients at risk of late OHSS, it does not completely prevent early-onset OHSS (42). Thus, a freeze-all strategy is not a fit to all strategy and a personalized approach should be preferred, based on patient characteristics, risk factors, and patient preference (43). For PCOS patients at high risk of OHSS, IVM could be considered as an alternative, but it is practiced in only a limited number of centers in France and Belgium (44).

Concerning patients with an unexpected suboptimal response, the consensus data suggest increasing the gonadotropin dose for the next treatment. According to the literature, different strategies can be set up in case of unexpected low ovarian response, including a higher FSH starting dose (37, 45, 46). The panel also agreed to modify the protocol in case of agonist protocol in the first attempt, but only in low and high response profiles. These results are in line with data from the literature, where a GnRH antagonist regimen is recommended in these populations (46, 47). The replacement of the gonadotropin does not seem to be an alternative to an inadequate response, but a LH/hCG supplementation can be suggested (48). Respondents are in favor of this strategy for unexpected insufficient ovarian response, but only in low and normal responders. In the literature, the benefits of LH supplementation are indeed more studied in low responders (45). The AMPLITUDE panel supported that unexpected hormonal levels (stagnant E2 or LH <1.2 ng/ml), hypogonadotropic hypogonadism, stagnant follicular growth, advanced age and suspected FSH receptor polymorphism, are additional indications for LH/hCG activity supplementation. This is consistent with literature findings. Indeed, in a systematic review of the literature, Hill et al., described that exogenous LH/hCG improved live birth rate. According to the authors, low responders and older patients (>35 years) are more likely to benefit from LH/hCG supplementation (47). Clinical evidence also largely supports LH/hCG supplementation in several patient types, including hypogonadotropic hypogonadism, patients older than 35 years old, low responders and unexpected low ovarian response, and profound suppression of LH serum concentrations (45, 47, 48).

FSH receptor (FSHR) polymorphisms can affect the level of FSH receptor expression in granulosa cells, that may lead to higher levels of FSH, lower ovarian sensitivity to FSH, and thus low ovarian response (49–51). In the literature, a systematic review suggested that LH supplementation in women presenting an impaired FSHR sensitivity to gonadotropins, may overcome the ovarian resistance (52).

It was important in this Delphi consensus to include the opinion of health professionals from different territories with their own particular socio-economical background. Sub-analyses by country were carried out but revealed no significant differences between the responses of healthcare professionals from France and those from Belgium. This Delphi was conceived and designed by and for Belgian and French health professionals, and as such it was not distributed on a wider scale to experts from other countries. This may constitute a non-negligible limitation of this study. However, there are no constraints with cost issues in the two countries, which could favor a higher diversity of applied treatment strategies to eventually identify the more clinically relevant protocol for different patients’ profiles.

We acknowledge that a consensus on the statements was not consistently achieved. This is a limitation off the study that may diminish the robustness of the decision-making process.

During the redaction phase, a particular attention was paid to ensure compliance with the guidelines for the reporting of consensus methods. Thus, the manuscript was drafted, reviewed, and adapted in accordance with the ACCORD guidelines (53).

In conclusion, data from this AMPLITUDE consensus support the idea of an optimized treatment for patients undergoing ART. In a context of lacking consensual data in the literature, this Delphi consensus provides interesting results, which hopefully will guide fertility specialists in their daily practice.