The importance of adrenal hypoandrogenism in infertile women with low functional ovarian reserve: a case study of associated adrenal insufficiency

For many years hypoandrogenemia has been recognized as a characteristic feature of primary ovarian insufficiency (POI) [1]. More recently, low testosterone (T) levels have also been reported in association with milder cases of POI, so called occult POI (oPOI), characterized by low (age specific) functional ovarian reserve (LFOR) [2].

Over the last decade various animal models and clinical human experience have provided increasing evidence that T is essential for normal follicle growth and maturation during small growing follicle stages. Insufficient androgen receptor (AR) activity on granulosa cells leads to poorer growth of fewer follicles, and to poor oocyte quality in surviving follicles [3].

Albeit still controversial [4, 5], these observations have led to androgen supplementation in women with hypoandrogenic LFOR [6], and to the suggestion that pregnancy success with in vitro fertilization (IVF) in hypoandrogenic LFOR directly correlates with improvements in patients’ testosterone levels [7].

Ovaries (theca cells) and adrenals (zona reticularis) produce the majority of androgens. Consequently, like hyperandrogenism in association with polycystic ovary syndrome (PCOS) [8], hypoandrogenism can be of ovarian and/or adrenal etiology. Though accurate differentiation is not always possible, it is generally accepted that low dehydroepiandrosterone sulfate (DHEAS), almost exclusively produced by the zona reticularis of adrenals, in association with low testosterone levels, strongly suggests adrenal origin of low androgen levels [9‐11].

We recently discovered that, as reflection of adrenal function, peripheral androgen precursor levels in infertile women with LFOR correlate with morning cortisol [12]. This observation suggests that adrenal and ovarian functions may to a degree be interdependent. Such interdependency is also supported by the common embryonic primordium of adrenals and ovaries [13].

We, therefore, in this study investigated this interdependence of adrenals and ovaries based on the recent recognition that LFOR, independent of cause, is usually characterized by peripheral hypoandrogenemia [2]. Based on the presumed origin of patients’ hypoandrogenemia, we then further assessed adrenal function under the hypothesis that adrenal origin of hypoandrogenemia (zona reticularis) may also raise the specter of adrenal insufficiency (AI) in the other two layers of the adrenal cortex.

As further evidence for the hormonal interrelationship of adrenals and ovaries, we here report four cases of previously unknown AI in hypoandrogenic women with LFOR and one case of known primary AI (Addison’s disease), which was treated with glucocorticoid but not androgen supplementation and, therefore, presented with secondary ovarian insufficiency (SOI) due to AI This case had previously been erroneously diagnosed as primary ovarian insufficiency (POI).

Out of 14 patients identified with likely adrenal hypoandrogenism, 10 were found to demonstrate entirely normal adrenal function. Table 1 summarizes the clinical presentations of four who as part of this study for the first time received a diagnosis of AI. One patient was diagnosed with primary AI, two were diagnosed with secondary AI and a fourth with tertiary AI.

The case of tertiary AI was a long standing systemic lupus erythematosus (SLE) patient on corticosteroid therapy who also suffered from long standing infertility. Though she had failed a prior in vitro fertilization (IVF) cycle, she spontaneously conceived once her androgens were supplemented with dehydroepiandrosterone (DHEA) normalizing her abnormally low TT and excessively high sex hormone-binding globulin (SHBG) levels.

We in addition identified a 41-year-old female with established Addison’s disease (primary AI) on glucocorticoid replacement who was not being supplemented with androgens. She based on a greatly elevated FSH level of 44.0 mIU/mL presented with diagnosis of POI. Her TT, FT, DHEA and DHEAS were all below lower cut off levels of normal range, while her SHBG was above normal range. She, thus, with great likelihood demonstrated adrenal hypoandrogenism. Once supplemented with DHEA (25 mg TID), her androgens normalized and her FSH level declined to 14.8 mIU/mL.

This patient’s presenting diagnosis of POI was, therefore, incorrect; a consequence adrenal hypoandrogenism of the zona reticularis, she really suffered from SOI, which apparently was part of a mixed adrenal cortex insufficiency of the zona fasciculata (leading to Addison’s disease) and the zona reticularis (leading to SOI).

Adrenal insufficiency (AI) is a complex, at times life threatening disease, which can be the result of failure of the adrenal glands (primary AI), can be consequence of failure of the hypothalamic/pituitary axis (secondary AI) or can be iatrogenic (tertiary AI) [16, 18]. We here report on four (out of 14) women with adrenal hypoandrogenism (low T and low DHEAS), who, upon adrenal evaluation, were found to suffer from AI, -one case of primary, two cases of secondary and one case of tertiary AI.

We also discovered among those patient one woman who already had been diagnosed with Addison’s disease (primary AI) and received glucocorticoid supplementation. Her treating physician was, however, unaware that, due zona reticularis insufficiency, she also was severely hypoandrogenic. With definition of AI currently restricted to insufficiency of zona fascilulata (glucocorticoids) and zona glomerulosa (mineralocorticoids), adrenal hypoandrogenism of the zona reticularis is, interestingly, excluded from the diagnosis of AI [16, 18].

In presence of severe hypoandrogenism (all of her androgen values were significantly below the lower cut offs of normal range), testosterone-dependent small growing follicle stages [3], likely, arrested. As a consequence, granulosa cell mass and estradiol production declined, AMH levels dropped and FSH increased due to diminished feed back on the pituitary, leading to what can be easily mistaken for a fairly typical clinical POI phenotype (Fig. 1). Due to its adrenal origin, it, however, really represents a form of SOI.

With 4/14 women diagnosed with a form of AI, we here report an unexpectedly high, and likely exaggerated prevalence of AI in infertile women with hypoandrogenic LFOR of adrenal etiology. Further, larger scale studies are obviously needed to assess the true prevalence. However, even here presented small case series suggests that in such women a careful adrenal evaluation appears indicated.

Here presented data in addition suggest that AI may not only include insufficiencies of zona fasciculate and zona glomerulosa but also also of the zona reticularis, leading to adrenal hypoandrogenism. Therefore, one also has to conclude that, at least in women of reproductive years, assessments of peripheral androgens appear indicated if AI is suspected or already diagnosed.

POI is classically associated with hypoandrogenism [1, 2], usually primarily the consequence of insufficient T production of ovarian theca cells. POI, therefore, represents true ovarian insufficiency/aging, while in cases of SOI, once adrenal hypoandrogenism is appropriately supplemented, ovarian function may normalize, and responses to pharmacological stimulation of ovaries may improve. Figure 1 explains the underlying pathophysiology.

Correct differential diagnoses between POI and SOI are, therefore, of importance since a diagnosis SOI often reflects a much better prognosis. We, in SOI, indeed, have seen relative normalization of abnormally high FSH levels (Patient 2 in Table 1 and above noted patient with previously known Addison’s disease), and in Patient 2 even encountered a spontaneous pregnancy following normalization of T and SHBG levels after DHEA supplementation. Such radical changes for the better in ovarian phenotype will practically never be witnessed in cases of POI.

As in women with polycystic ovarian syndrome (PCOS), relative contributions to hyperandrogenisms of adrenals and ovaries are at times difficult to separate [8], so are in women with LFOR contributions to low T. During steroidogenesis, DHEA is via sulfotransferase converted to DHEAS (its 3β-sulfate), a conversion almost exclusively coded by the SULT2A1 gene. Sult2A1 is, however, highly expressed only in the zona reticularis of the adrenal cortex, and practically absent in ovarian tissue [9‐11]. If DHEAS is also abnormally low, low T in women can, therefore, be assumed to be of adrenal origin.

A few more words about the relevance of AI to ovarian function: Though the underlying pathophysiology for this association has remained unresolved, primary AI has been associated with female infertility for decades. [19]. Here suggested pathophysiology of SOI offers a possible explanation.

In developed countries over 90 % of primary AI is considered autoimmune in etiology, frequently coexisting with other autoimmune endocrinopathies [18, 20]. Interestingly, the only so far histopathologically defined autoimmune condition of ovaries in humans, so-called autoimmune oophoritis, practically exclusively only occurs in association with primary AI. This dependency strongly suggests common immunologic epitopes in adrenals and ovaries as targets of shared autoimmune attacks, possibly representing steroidogenic enzymes [21, 22]. Autoantibodies to 21-hydroxylase are widely considered diagnostic of primary AI, and are accepted as evidence for the autoimmune etiology of AI. These autoantibodies, indeed, often precede diagnosis of the disease [18, 23], and approximately 30 % of asymptomatic individuals with positive 21-hydroxylase antibodies will progress to clinical primary AI within five years [24].

Why is this important within here presented discussion?

Addison’s disease can also occur in absence of 21-hydroxylase antibodies, though such cases are rare except in young children and the elderly [15]. Other than 21-hydroxylase, yet unknown antibodies to steroidogenic enzymes and/or other common epitopes between adrenals and ovaries may, therefore, also play an important role in anti-adrenal and anti-ovarian autoimmunity. Autoantibodies to other steroidogenic enzymes have, indeed, been reported [22], though, currently are not considered diagnostic of primary AI.

Autoimmunity to endocrine glands is a commonly observed phenomenon; many such attacks, indeed, occur in combinations [20]. It, therefore, is difficult to imagine that anti-adrenal autoimmunity can affect only zona fasciculate and zona granulosa but will not affect the zona reticularis. For that reason, it appears somewhat puzzling that an autoimmune attack on the androgen-producing zona reticularis is currently not considered a possibility in AI. It appears more reasonable to assume that all three zones of the adrenal cortex can be subjects of autoimmune attacks, sharing risks and manifesting different combinations of involvement. Patients with glucocorticoid deficiency for that reason have to be monitored for the development of mineralocorticoid deficiencies [18]. Here presented study supports the contention that the zona reticularis should also be considered as a potential target of adrenal autoimmunity in presence of autoimmune attacks against the other two zonae.

In the developed world, in absence of tuberculosis and trauma, AI is, practically universally, considered an autoimmune condition [25]. Low T due to adrenal causes, therefore, with great likelihood has also to be considered autoimmune.

Finally, here discussed findings may also have relevance for the PCOS, increasingly viewed as the opposing extreme of LFOR on a spectrum of ovarian function [20]: While LFOR is hypoandrogenic [1, 2] and associated with low follicle yields, PCOS typically presents with high T, excessive follicle recruitment and, therefore, high FOR. At excessively high T levels, follicles, however, arrest at preantral stages [26]. Adrenal androgen production thus influences ovarian function over a wide range, from SOI to secondary PCOS (Fig. 1).

We, therefore, conclude that, functionally, adrenals and ovaries have increasingly to be viewed as a single endocrine unit. The common developmental primordium (primordial germ cells) of adrenals and gonads is supportive of such a concept [13].

This study further strengthens recently reported evidence that adrenal glands and ovaries closely interact in regulating ovarian function via shared androgen production, which, in turn, affects follicle maturation in ovaries. This study, in addition, however, for the first time also demonstrates mutual diagnostic dependencies: LFOR in women of reproductive age, especially if associated with relative hypoandrogenism of adrenal origin (characterized by low DHEAS), should be seen as an indication for adrenal function evaluation. Concomitantly, a diagnosis of AI in women in reproductive years should immediately be considered an indication for evaluation FOR in the patient.