During the follicular phase of cycle, FSH is involved in recruitment, selection and dominance of follicle. It is involved in the recruitment of the cohort at an early follicular phase and stimulates the transcription of genes within the granulosa cells, initiating the synthesis of aromatase enzyme, inhibins and LH receptors that are involved in follicle differentiation and growth [5].

A certain amount of FSH secretion – ‘FSH threshold ‘ is required to induce follicle growth, and as this threshold is not identical for follicles even of the same cohort, the FSH dose for inducing multi-follicular development should cross the threshold of the least sensitive follicles. Thus the endogenous FSH is crucial to the cycle in inducing follicular recruitment. The other aspect of ‘FSH window’ is also crucial as this means that follicular growth is maintained as long as the FSH levels are above the threshold for the follicle. While in natural cycle the progressive decline in FSH secretion due to the feedback effect of ovarian hormones on the pituitary leads to dominance of a selected follicle and atresia of others, during COH the FSH levels are above the threshold and the window that remains open until the final stage of follicular development resulting in multiple follicles at the time of trigger. Thus FSH is the main therapeutic drug that controls the folliculogenesis in all cases except hypogonadotropic hypogonadism where LH supplementation is necessary for production of steroid hormones.

During COH both gonadotropins and GnRH analogue are used to attain multi-follicular development, but the effects of each on the levels of FSH are variable. With gonadotropins there is a plateau of plasma FSH levels due to the long elimination half-life of FSH molecules (30–35 h). This FSH accumulation lasts for 5 days, and despite cessation of exogenous FSH administration follicles continue to mature. Also plasma levels of FSH after an intramuscular or subcutaneous dose cause a transient (4–8 h) and modest rise in plasma FSH levels, which are further not reflective of the actual bioactivity of the molecule. For these reasons there appears little justification in measuring FSH levels to adjust FSH doses or to establish the threshold above which ovarian response can be observed. This was further supported by a study by Van Weissanbrunch et al., who measured serial FSH levels to determine the adequate threshold FSH dose [6]. FSH was administered in a pulsatile manner to adjust the daily requirement according to the plasma FSH levels. The study found poor correlation between plasma FSH levels and the FSH threshold, as there was an overlap of plasma FSH levels between the groups of patients who demonstrated follicular recruitment as against those who did not [6].

The effects of agonist on FSH levels depend on the preparation and duration of use. There is an initial rise or flare effect of agonist at the pituitary resulting in a significant rise in plasma FSH levels with resultant recruitment that is exploited in a short protocol. The amplitude of FSH response to GnRH-agonist is lower than LH [7], and the desensitizing effect of prolonged GnRH-a administration on FSH secretion is much less than LH [8, 9]. Further FSH bioactivity may not decrease during GnRH-a administration [10]. Therefore, measuring plasma FSH is unlikely to be of benefit in the course GnRH-a therapy. Even in antagonist cycle, the gonadotroph suppression was less marked for FSH than for LH as suggested in studies available [11]. Summing up from available evidence on FSH variations during ART treatment, there seems to be no contribution of FSH estimation in deciding gonadotropin doses or regimen in clinical practice.

Luteinizing hormone acts on the theca cells while producing androgens throughout the follicular phase. Androgens are the substrate for the production of oestradiol (E2) by granulosa cells. LH induces a dose-dependent production of E2, and this is foremost in endometrial preparation for embryo implantation [12]. There is a minimal amount of LH, described as the ‘LH threshold’, required for pregnancy; however, higher levels are detrimental with a negative impact on endometrium rather than on oocytes/embryos [13, 14]. Using high doses of LH has a negative influence on follicular development, a concept of ‘LH ceiling’ [15], wherein LH beyond a certain level suppresses granulosa cell proliferation and results in atresia of less mature follicles (11–15 mm) [15]. Substitution of LH in the later follicular phase with recombinant LH alone is known to cause reduction in size and number of large follicles as seen in both type I and type II WHO anovulation category of women [16]. LH therefore synergizes with FSH during the whole follicular phase of folliculogenesis.

As regards LH in ART cycles, urinary human menopausal gonadotropin (hMG) commonly used in earlier years contains LH, and this is cleared rapidly from circulation owing to a short half-life [17] approximately 12 h as compared to 30 h for FSH. Therefore there is little evidence of accumulation following hMG injection. The use of agonist in ART cycles is known for the initial flare effect, as used in ultra-short and short protocols where rise promotes early follicular recruitment. The endogenous FSH and LH rise within 24 h of GnRH agonist administration, the flare effect being more marked for LH than FSH [18]. This subsequently stimulates secretion of E2, which is considered a predictor of ovarian response to gonadotropins. Thus estimation of LH does not reflect adequacy of flare-up response.

Measurement of plasma LH is routinely done to confirm pituitary desensitization and confirm adequate down-regulation after long-term administration of GnRH agonist. Adequate down-regulation is suggested by LH levels <3 IU/L. The rate and extent of LH suppression is dependent on the type of agonist, its route and dose used [18]. As long as the GnRH-a is given, the hypophysis is refractory to endogenous GnRH and there are no pulsatile LH secretions. The extent of LH suppression is variable after desensitization as when there are profound falls in LH levels higher doses of gonadotropins are needed to achieve ovarian stimulation. Thus in GnRH agonist-based cycles, levels of LH may have a bearing on ART outcome. The endogenous LH levels are within the normal limits (1–4 IU/L) or are within the limits defined by LH ceiling and LH threshold as sufficient for steroidogenesis and folliculogenesis.

Though low LH levels are associated with little difference in birth rate, women with LH levels less than 1.2 IU/L require higher doses of gonadotropins during ovarian stimulation [19]. Thus in poor responders lower doses of GnRH-a are suggested for adequate response to gonadotropins. In keeping with the two-cell–two-gonadotropin theory, administration of pure or recombinant FSH should not be sufficient to stimulate E2 production. However, results from large studies suggest that FSH administration is sufficient to obtain adequate number of good-quality oocytes and embryos with high implantation rate, as there is still endogenous LH, despite profound down-regulation when LH levels are measured to be low [20]. Later studies evaluated outcome in ART cycles according to the plasma levels of LH at the time of desensitization or later in mid-follicular phase [21, 22]. Those patients with LH levels <0.5 IU/L had reduced E2 concentrations at the time of hCG trigger and lower number of oocytes and embryos when stimulated with FSH alone. However, the rate of blastocyst development was unaffected. Thus measurement of LH during the cycle fails to define a sub-group of women who would need additional LH to achieve ovarian stimulation. Therefore, the threshold of LH below which folliculogenesis may be impaired cannot be assessed by measuring LH after down-regulation. There is however no relation between endogenous LH levels and ovarian response, implantation rate and pregnancy rate when normogonadotropic women undergo IVF cycles as suggested by a later meta-analysis [23].

Plasma E2 levels are useful in assessing follicular maturity as the synthesis of E2 associates with dominant follicles in natural cycles. During ART cycles plasma E2 levels are performed combined with ultrasound to adjust doses of gonadotropins. E2 synthesis is related to follicle size, and a mature follicle has an output of approximately 200 pg/ml. In days before the use of GnRH analogues to prevent endogenous LH surge, serial E2 levels were estimated as they correlated well with cycle outcome [24]. With the availability of GnRH-a protocols the problem of premature LH surges is taken care of, but the estimations of E2 are still recommended to confirm pituitary desensitization. After effective desensitization plasma E2 levels must be lower than 50 pg/ml, that is, after 2 weeks of GnRH-a, when the initiation of gonadotropins for ovarian stimulation can be given. While plasma LH levels are also estimated to confirm desensitization, they cannot adequately reflect the same for reasons discussed in previous sections. Mid-luteal start of GnRH-a and use of long-acting preparations of agonist are associated with profound and immediate desensitization [25, 26]. Whether the prompt desensitization is associated with ovarian refractoriness and the need for higher doses of gonadotropins is debatable [25]; ovarian stimulation with FSH alone should be initiated only once ovarian activity is suppressed. During the ovarian stimulation, levels of E2 guide in determining the optimal response. Plasma E2 levels closely follow stages of development of growing follicles. After 6 days of gonadotropins, an increase in plasma E2 levels is defined as optimal response; however, due to extreme diversity of protocols, the ideal levels of E2 are not defined. A plateau in plasma E2 for more than 3 days suggests poor response to gonadotropins. Conversely excessive response can be gauged by an exponential rise in E2, which helps decide coasting or cancellation. An E2 window of 1,000–1,500 pg/ml is optimal once follicles reach 15 mm [27]. The risk of hyperstimulation is significant with levels more than 3,000 pg/ml [28]. Therefore, E2 monitoring is relevant and should be a part to define optimal response, even though some studies suggest that ultrasound monitoring is sufficient to make decisions during stimulation [29].

Further in antagonist protocols the pattern of E2 during stimulation differs from that in agonist protocols. Plasma E2 levels are higher before the addition of GnRH antagonist, and after the addition of antagonist to control the LH surge, the E2 levels may rise moderately, remain the same or even decline [30]. But these variations in E2 levels do not compromise the cycle outcome. Unlike agonist cycle E2 levels are of little help in adjusting gonadotropin doses after the antagonist has been added to cycle.