MECHANISM

FSH was the first hormone directly linked to ovarian aging (3). It is secreted by the anterior pituitary and promotes the progression of antral follicles into dominant follicles. Feedback from estrogen, inhibin, and activin influence hypothalamic GnRH pulsatility, which determines pituitary FSH expression. Elevated FSH levels can be seen with dwindling reserve, where a greater FSH stimulus is required to drive folliculogenesis, but elevated levels also can be found in normal ovarian reserve if measured at the time of the LH surge. Low FSH levels are seen prior to puberty or with hypogonadotropic hypogonadism. In addition to medical conditions that shift pituitary FSH expression, exogenous hormones and their modulators (clomiphene, letrozole, etc.), cimetidine, phenothiazines, and other medications can also shift levels.

TESTING

Many non-FSH substrates can induce an FSH-like effect. Without describing in detail the spectrum of FSH assays that bypass this challenge, for which an excellent review is available (4), in the clinical setting FSH is typically measured by immunoassay. The sample is usually acquired by phlebotomy (24-hour urine collections are rarely used) on menstrual cycle day three for ovulatory patients, with day one being the first full day of flow.

Testing on cycle days two, four, or five is not unreasonable, but if a normal result would prompt retesting, a day three measurement or a different assay is preferred. Multiple cutoffs are used, with FSH levels of >16.7, >11.4, and <10 mIU/mL reflecting high, moderately high, and normal levels based on the World Health Organization (WHO) Second International Standard (5).

Because ovarian reserve is on a continuum, any cutoff selected should relate to goals of balancing positive and negative predictive values, and this is an issue that applies to other measures of ovarian reserve as well. In amenorrheic patients, a random sample is preferred to testing after hormonally induced menses. In the setting of amenorrhea, a concurrent progesterone level (<2 ng/mL) is a reasonable control to ensure that one is in the follicular phase.

LIMITATIONS

Interpersonal and intercycle variation can be meaningful in patients at risk for moderately elevated FSH, which is why it has been called “Fluctuating Severely Hormone.” The problems with FSH’s sensitivity in part stem from it being a late marker of dwindling ovarian function, as summarized in Stages of Reproductive Aging Workshop + 10 conclusions (6). This limited predictive value is reflected in the NHANES III data, which showed 75% of women aged 40 to 44 years having normal levels at less than 10 mIU/mL, even though ovarian function is typically the rate limiting step at this age, and half of women aged 45 to 49 years had levels less than 11 mIU/mL (7). Sensitivity for FSH is often worse than specificity, with findings ranging from 11-86% and 45-100%, respectively (8). With anti-muellerian hormone (AMH) and antral follicle count (AFC) demonstrating better predictive value for ovarian response than FSH, these are more likely to be the tests of choice (9). Accordingly, relative to emerging alternatives, FSH testing increasingly is seen as less valuable than it used to be for procreative testing and more useful for evaluating perimenopausal status, hypergonadotropic and hypogonadotropic hypogonadism, and central precocious puberty.

MECHANISM

As with FSH, estradiol levels vacillate over the course of a menstrual cycle, peaking in both the late follicular and mid luteal phases. As ovarian reserve declines, the follicular phase shortens because of decreasing feedback inhibition by follicles recruited during the previous cycle. (This is why the first clinical sign of decreasing ovarian reserve is shortening menstrual cycle length.) With the follicular phase starting earlier, estradiol levels start rising closer to menses (and the classic day three FSH peak actually can occur prior to menses). As a result, an elevated day three estradiol level could reflect diminishing ovarian reserve.

Elevated estradiol (>60-80 pg/mL) may also lead to an artificially normal FSH, where higher estradiol levels lead to feedback suppression of FSH. Conversely, estradiol levels <20 pg/mL on day three depending on the circumstances can be consistent with normal ovarian function, hypogonadotropic hypogonadism, or ovarian failure.

TESTING

Estradiol is also typically measured by immunoassay after phlebotomy. The sample is usually drawn at the same time as FSH levels or randomly when assessing amenorrhea. Estrone, the primary postmenopausal estrogen, and estriol, the primary pregnancy estrogen, are not typically tested when evaluating ovarian function. Also, because oral estrogens are typically metabolized into many byproducts (with varying activity), serum estradiol levels often won’t reflect exogenous exposure. (However, transdermal estrogen administration can be monitored through serum levels.) Medical conditions, glucocorticoids, sex steroids, clomiphene, letrozole, GnRH agonists and antagonists, and other medications can alter estradiol levels, just as they could shift FSH levels.

LIMITATIONS

For many conditions, an estradiol level is a reasonable proxy for ovarian inactivity. However, for assessing decreasing ovarian reserve, estradiol is neither a sensitive nor specific assay (9). Accordingly, when used for measuring ovarian reserve, estradiol has its greatest value as an internal control to ensure that one is testing at the expected portion of the menstrual cycle (Figure 2).

Figure 2.

Ovarian, hormonal, and endometrial changes over the menstrual cycle. Adapted from Hall, et al., Hypothalamic gonadotropin-releasing hormone secretion and follicle-stimulating hormone dynamics during the luteal follicular transition (10).

MECHANISM

The clomiphene citrate challenge test combines measurement of FSH and estradiol levels prior to clomiphene exposure and FSH levels after clomiphene exposure. Clomiphene is a selective estrogen receptor modulator (SERM) that inhibits negative feedback inhibition by estradiol on the hypothalamus. Normally, increased estrogen levels decrease GnRH pulsatility, resulting in lower FSH levels through negative feedback. By using clomiphene to block feedback inhibition by estradiol, there is an increase in FSH, which enhances follicular recruitment, and which is why clomiphene can be used for ovulation induction and superovulation.

TESTING

FSH and estradiol levels are assessed through immunoassay, as previously described. The CCCT is performed by having an FSH level drawn on the third day of the menstrual cycle, taking 100 mg of clomiphene orally cycle days five to nine, and then repeating the FSH level on cycle day number ten (11, 12). An estradiol level is also frequently drawn on the third day and sometimes on the tenth day as well.

LIMITATIONS

When assessing ovarian reserve for fertility, FSH is a limited measure of ovarian response and a poor predictor of pregnancy and estradiol is predictive of neither (9). When combining the two through the CCCT, it is difficult to assess the degree of benefit through receiver operator curves (9). If benefit is unclear, cost-effectiveness is even less so. Accordingly, other measures of ovarian reserve are increasingly used instead of the CCCT, although this assay is still more commonly used than other provocative tests, such as the exogenous FSH ovarian reserve test (EFORT) and the GnRH agonist stimulation test (GAST). The CCCT has particularly suboptimal value in anovulatory patients. The reason is that the CCCT is primarily used to help discriminate normal ovarian reserve from poor reserve in patients with potentially borderline function. However, the typical anovulatory patient tends to have robust reserve (PCOS, hypogonadotropic hypogonadism) or poor reserve (primary ovarian insufficiency), so relative to alternative assays, a test designed to elicit subtleties is typically less important in this population.

MECHANISM

Follicular recruitment is in constant flux during the reproductive years, with less than 0.1% of oogonia present at birth ever making it to ovulation. Fluid surrounding numerous oocytes not selected to be the dominant follicle can be seen sonographically prior to regression. The more follicles visualized within the ovary, the greater the probable ovarian reserve, and AFC has been shown to correlate closely with the primordial follicular pool on histologic analysis. (13, 14). Though it remains for debate as to how much a dwindling follicular pool reflects oocyte quality as well as quantity, women with infertility are more likely to have lower antral follicle counts than those without infertility (15). Similarly, women with low antral follicle counts are much more likely to have cancellation for under response with IVF than those with normal counts (16). However, though low quantity in younger women may reflect fewer oocytes with which blastocysts can form, it does not clearly seem associated with higher rates of aneuploidy or miscarriage (17).

TESTING

Antral follicle count can be measured at any time during the menstrual cycle, as well as when a woman is on hormonal contraceptives or is pregnant. Classically, a woman’s AFC is the total number of ovarian follicles measuring between two and nine millimeters, though many studies count follicles up to and including ten millimeters in size (Figure 3).

Figure 3.

Ovarian sonographic imaging of women in their mid-30’s. Figure 3A is from a woman with premature ovarian failure and there are no visualized antral follicles (the sonographically anechoic regions measuring approximately two to nine millimeters within the ovary). Figure 3B is from a woman with tubal factor infertility, and for whom seeing a few follicles within a single plane of the ovary would be normal. Figure 3C is from a woman with polycystic ovarian syndrome. Though her ovary is arguably more multicystic than polycystic (which would typically have follicles concentrated on the periphery of the ovary), she met the criteria for PCOS and her ovary is clearly distinct from those shown in 3A and 3B. Of note, all three ultimately conceived with their own oocytes, so it should be remembered that the absence of visualized antral follicles makes conception far less probable, but not impossible.

Multiple cutoffs are used for what constitutes normal and poor ovarian reserve. Given that antral follicle count varies among cycles, it is reasonable to view the AFC as a continuum, with four total antral follicles reflecting limited reserve, but five antral follicles not being entirely reassuring. Additionally, what constitutes normal is age dependent, where ten total antral follicles may be common for women in their 30’s, but not their teens. Though many measures have been used to define polycystic ovarian morphology, the most accepted standard is that used in the Rotterdam criteria of, “12 or more follicles in each ovary measuring 2 to 9 mm in diameter, and/or increased ovarian volume (>10 ml).” This cutoff was chosen, as it was associated with 75% sensitivity and 98% specificity for distinguishing polycystic ovarian morphology (PCOM) from normal ovaries (18). Another frequently used definition comes from Adams, who considered an ovary polycystic if there were ≥ 10 follicles measuring <9 mm (19). Of note, in the development of guidelines for the WHO on PCOS, sonography was deemed preferable to AMH levels from a pragmatic standpoint. (20)

LIMITATIONS

There is debate as to how much moving outside of the early follicular phase or hormonal modulation such as pregnancy and oral contraceptives will shift the measurement of antral follicle count. Both central and paracrine effects can occur and these are more likely to be meaningful in patients with suboptimal ovarian reserve. However, patients with reassuring ovarian reserve are unlikely to move into a non-reassuring category through these conditions if the ultrasound resolution allows for early antral follicle visualization and measurement.

Patient dependent and observer dependent limitations should also be considered. Patients with elevated BMI (particularly with increased vaginal adiposity) and/or scarring of the pelvis may be more likely to have ovaries with limited resolution for assessment, which could potentially underestimate ovarian reserve. Similarly, large cysts or endometriomas could exert a temporary paracrine effect underestimating reserve. Patients with previous ovarian surgery could also have inclusion cysts appearing similar to antral follicles, but these won’t develop with stimulation or have oocytes at follicular aspiration for IVF. For observer dependent limitations, it should be noted that in some multi-center studies where anti-Müllerian hormone (AMH) is found superior to antral follicle count, one can find most institutions having AMH and AFC equally predictive, but one site has an observer where there is a meaningful difference. This has led some to conclude AMH superior to AFC, but failure to properly train observers prior to research is a limitation to study design and may not necessarily reflect true diminished value in utilizing AFC for assessing ovarian reserve. ASRM 2022 Practice Committee Guidelines note, “When performed in an experienced center, AFC is a reasonable alternative to AMH” (29).

MECHANISM

AMH is a homodimeric glycopeptide that in reproductive aged women is predominantly granulosa cell derived. The role of systemic AMH is not clear, but at the level of the ovary, it is believed to downregulate FSH mediated folliculogenesis. AMH expression is highest in secondary, preantral, and small antral follicles up until approximately 4 mm in size, and it stops being expressed by granulosa cells when the follicle measures in the 4 to 8 mm range (Figure 4).

Figure 4.

The interplay of follicular development and hormonal secretion and responsiveness.

AMH seems to have a role in selecting the dominant follicle in addition to generally mediating preantral follicular recruitment. AMH levels start undergoing a log-linear decline approximately fifteen years prior to menopause and drop to very low levels approximately five years before menopause (21).

The AMH level associated with diminished ovarian reserve is assay specific and depends on the desired balance of sensitivity and specificity, but is typically below 1 ng/mL. The threshold for menopause is typically lower than the lower detectable limit for many assays, being slightly below 0.1 ng/mL (22). AMH <0.5 ng/mL seems associated with fewer than three follicles available at retrieval, 0.5-1 ng/mL with reduced response, 1-3.5 ng/mL with normal response, and >3.5 ng/mL with overresponse, reflecting greater risk for ovarian hyperstimulation syndrome (23). Normal AMH values often exceed 2 ng/mL at 30, 1.5 ng/mL at 35, and 1 ng/mL at 40 as a quick reference for expected reserve at a given age.

The role of weight loss on AMH levels is open for debate, but should be substratified in women with and without PCOS. (51) Though a lack of association cannot be excluded due to limitations in sample size, there does not seem to be a clear shift in AMH with weight loss for non-PCOS patients. However, for those with elevated AMH from PCOS that has been effectively treated through diet, exercise, and/or bariatric surgery, there is improved fertility, even as AMH lowers to more normal levels. (51) It is unclear which signal transduction pathways drive these lower levels, as one would expect weight loss-associated shifts in adiponectin, leptin, and insulin to actually increase AMH through recognized mechanisms. Lifestyle may have a broader impact beyond weight on ovarian reserve, as both AMH and AFC are statistically lower in women with lower socioeconomic status (56).

TESTING

AMH levels are measured through immunoassay on a sample obtained through phlebotomy. Values obtained have the distinct advantage of being equally valid at any point in the menstrual cycle. Because AMH is expressed primarily before FSH responsiveness occurs, it is believed that AMH remains a valid assay even when ovarian suppression occurs through smoking, oral contraceptives, GnRH agonists, and pregnancy (24). Though these factors can lead to transient ovarian suppression, they are unlikely to change levels so much as to meaningfully underestimate true reserve. The magnitude of effect through these reversible factors seems to be low, with age-specific AMH percentiles decreasing by 11% with oral contraceptives and 17% with pregnancy (25). Additionally, AMH levels drawn on day seven of the pill free interval seem to closely correlate with levels seen after oral contraceptive discontinuation (26).

A popular misconception is that just because it is valid to assess AMH throughout the menstrual cycle and under a variety of inhibitory conditions, this should not be mistaken as meaning that AMH levels are static. Though levels of AMH tend to be steady state in perimenopausal patients, for those with higher ovarian reserve, AMH levels fluctuate significantly. This fluctuation, however, is not to the point where a person with robust ovarian reserve is likely to be categorized as having limited reserve (21).

LIMITATIONS

AMH seems to have fewer limitations than most other assays. In fact, it seems to have the advantage that it not only is useful in predicting ovarian response to gonadotropin stimulation, but may even have limited value in predicting pregnancy rates (27). However, like other ovarian reserve assays, it does not appear particularly valuable in predicting viability once pregnancy has already been established. When there is discordance between AFC and AMH levels (e.g., low AFC but normal AMH or vice versa), ovarian response is often a hybrid of the two findings (above those with diminished reserve but less than that of those with normal reserve) (28).

MECHANISM

Inhibin B is similar to AMH in that it is a glycoprotein secreted by preantral follicles, with levels declining with age. Both inhibin A and B downregulate pituitary FSH secretion. However, Inhibin A levels are not used to predict ovarian reserve because they arise primarily from the dominant follicle rather than an earlier follicular cohort and therefore are less predictive. Inhibin B levels are relatively more useful, but overall remain suboptimally predictive, as they are a late finding for diminished ovarian reserve and typically start falling around four years prior to menopause (21).

TESTING

Inhibin levels are measured by immunoassay after phlebotomy. Inhibin B levels fluctuate over the menstrual cycle, with peaks in the early to mid-follicular phase, as well as during ovulation. Accordingly, inhibin B is typically measured on the third day of the menstrual cycle in ovulatory women. Outside of ovarian reserve testing, in postmenopausal women, where inhibin B levels should be consistently low, a random level is particularly good for following granulosa cell tumors (>89% have elevated inhibin B) and also can be useful for following some epithelial cell ovarian tumors.

LIMITATIONS

In addition to significant variation within the cycle, there is also meaningful variation among cycles. Because of limited sensitivity and specificity, this assay has greater value in those far more likely to have diminished reserve. Some have proposed using inhibin B in combination with other assays, but it is the opinion of the American Society for Reproductive Medicine that “combined ovarian reserve tests models do not consistently improve predictive ability over that of single ovarian reserve tests.” (29).

MECHANISM

Follicles, stroma, and vasculature all contribute to ovarian volume. The percentage that each contributes depends on the individual, her age, underlying gynecologic conditions, and where she is during the menstrual cycle.

TESTING

Typically, ultrasound is used to measure the ovary in all 3 dimensions. These measurements are then applied in the formula for calculating the volume of an ellipse (D1 x D2 x D3 x 0.523). An ovarian volume of >10 cm^3 is considered consistent with PCOS. Although increased ovarian stromal volume distinguishes polycystic ovarian morphology from the multicystic ovary, stromal volume is not routinely measured. Alternative approaches that may improve the effectiveness of ovarian volume include the use of trapezoidal volume (30), 3D ultrasound (31), and color Doppler (32).

LIMITATIONS

Ovarian volume shifts in response to normal physiologic changes (such as the presence of a dominant follicle) and coexisting medical conditions (such as endometriomas). Exogenous hormones can decrease ovarian volume (33), even though ovarian reserve itself has not changed. For these reasons, if evaluating the ovaries by ultrasound, antral follicle count is believed to be a better proxy for ovarian reserve.