html xmlns=”http://www.w3.org/1999/xhtml” xmlns:mml=”http://www.w3.org/1998/Math/MathML” xmlns:epub=”http://www.idpf.org/2007/ops”>
Introduction
As the attending physician covering the resident gynecology clinic, it was not unusual for a student or a resident to present the case of a woman complaining of hirsutism. When I would ask the presenter for a differential diagnosis, almost always, one of the first responses would be that she could have an androgen-secreting tumor. While true, statistically, this was probably the least likely etiology.
Hyperandrogenism affects 7% of women [1] and can be both physically and psychologically disturbing. Androgen-secreting tumors, however, are an extremely rare cause of this disorder and tend to receive a disproportionate share of attention in the differential diagnosis. Ovarian tumors account for only 0.2% of cases of hyperandrogenism [2].
Androgen-secreting tumors can also arise in the adrenal gland. Briefly, in terms of the adrenal, approximately 60% of adrenocortical carcinomas present with symptoms of hormone excess. The most common presentation is hypercortisolism, though virilization can occur due to secretion of excess androgens. These tumors are extremely rare, with an incidence of only 0.5 to 2 per one million persons per year [3]. The vast majority of adrenal tumors are benign, hormonally inactive adenomas, often found by accident. Generally small (average <4 cm) and unilateral, they are very common and found in approximately 10% of postmortem exams [4]. The majority (85%) are non-functional [5]. There are reported cases of adenomas inducing virilization, but this is extremely unusual [6]. Generally, the biggest risk of these neoplasms is that they can prompt needless surgical exploration.
Moving back to the ovary, a wide variety of tumors may be hormonally active, and even some non-hormonally active tumors can induce the surrounding ovarian stroma to produce androgens [7]. These will be discussed in more detail below. Despite their rarity, these tumors still must be considered, and in most cases eliminated, in the differential diagnosis. How does an androgen-secreting tumor present? What clues do we have? What evaluation should we perform? Before answering these questions, it must be kept in mind that there is no presentation that is always indicative of tumor. A detailed history, including information about family history and social behaviors, along with a thorough physical exam are the most valuable tools when determining the etiology of hirsutism/androgen excess [8]. This is no different when evaluating cases caused by tumors.
In this chapter, we will outline the symptoms and signs useful in making the diagnosis of an androgen-secreting tumor. The value of diagnostic testing, including simple lab work, imaging, and invasive procedures, will be reviewed. Next, we will review the wide variety of tumors capable of producing androgens. And finally, we will discuss treatment, which in the majority of cases, is straightforward.
History and physical examination
Algorithms often point to the rapidity and severity of symptoms as clues to an underlying tumor. According to Speroff et al. [9], “a woman who develops hirsutism after the age of 25 and demonstrates very rapid progression or masculinization over several months” has an increased likelihood of an androgen-secreting tumor. Given this presentation, a full evaluation for the presence of a tumor should be performed regardless of serum testosterone levels. Even with this presentation, however, the majority of cases will not be caused by a tumor. It should also be considered that the concept of an “acute, rapid course” is a relative one. A patient may have a distorted recollection of the process, not noticing the slowly progressive nature of her disorder until suddenly, a certain threshold is reached. In addition, some tumors produce only a moderate amount of androgens; their presentation may be much more indolent and symptoms may develop over several years [10].
Tumors may present any time during the reproductive or post-reproductive years. Very rarely, an ovarian tumor can present with virilization, with or without isosexual precocity, in a child. In contrast, most functional causes of hyperandrogenism show a peripubertal onset [9]. The severity of virilization may support the presence of a tumor. Androgen-secreting tumors can cause severe virilization and “defeminization,” with signs such as clitoromegaly, balding, voice change, and decreased breast size [11]. Even marked signs of virilization, however, while suggestive of a tumor, are not pathognomonic. In addition, if a tumor produces only moderate amounts of androgens, the virilization may not be as severe.
A thorough and careful pelvic examination is essential in the evaluation of the patient. The majority of functional ovarian tumors are palpable on examination [6,9]. There can be very small (<2 cm) tumors, particularly those that develop in the hilus of the ovary, that are non-palpable; these may require additional diagnostic testing for localization.
Diagnostic testing
Laboratory
Some guidelines use a serum testosterone level greater than 200 ng/dL as an indication of a potential tumor [7]. This number is by no means absolute, as many androgen-producing tumors are associated with testosterone levels below this level [9,12,13] and the majority of women with levels this high do not have a tumor. This value needs to be placed into context with the additional information obtained from the woman (e.g., the rapidity of symptom onset) and not used as a stand-alone criterion. Postmenopausal women secrete lower levels of androgens; utilizing a cut-off value of 100 ng/dL may be more appropriate in these women [14]. Levens et al. [15] reviewed laboratory data on 136 women under investigation for androgen-secreting tumors. They found the optimal cut-off point for peripheral serum testosterone in their sample was ≥130 ng/dL. At this level, they found a sensitivity of 93.8% and a specificity of 77.8%.
Measurement of free testosterone is a more sensitive indicator of hyperandrogenism than is total testosterone [8]. For the purposes of screening for an androgen-secreting tumor, however, a routine total testosterone assay is sufficient [9]. The majority of tumors present with clearly abnormal levels of testosterone; the fine discrimination offered by the free testosterone determination is unnecessary in this circumstance.
Levens et al. [15] found no discriminating value to measuring androstenedione. A dehydroepiandrosterone sulfate (DHEA-S) level of ≥700 µg/dL is generally accepted as a marker for abnormal adrenal function and potential tumor [7,9]. The clinical usefulness of measuring DHEA-S has been questioned, especially in the absence of any symptoms suggestive of Cushing’s syndrome [9]. When testosterone is significantly elevated, imaging of the adrenal glands may be a preferable, and more definitive, evaluation for the presence of an adrenal tumor than measuring DHEA-S.
Radiology
When an androgen-secreting tumor is suspected based on history and physical exam, imaging of the ovaries and the adrenal glands should be performed. For the adrenals, imaging with CT scan is a sensitive method for detecting even small neoplasms. Adrenocortical malignancies tend to be large (>6 cm) with an irregular outline [7,9]. Adenomas tend to be smaller (<4 cm) and well circumscribed. It must be remembered that the majority of small adenomas are not responsible for hormone production and additional evaluation is needed before concluding surgical removal is necessary [6]. Additional discussion is beyond the scope of this chapter.
Transvaginal ultrasound (TVUS) is the method of choice for evaluating the ovaries [5,9] and the majority of tumors are large enough to be easily visualized. The challenge arises when a tumor is so small, it has not yet altered the size of the ovary. Despite this, gray-scale TVUS with a 7.5 MHz probe and a color Doppler flow study can visualize even very small (~0.5 cm) tumors [14,16].
Virilizing tumors are most commonly solid or mostly solid, depending on the tumor type, and calcification is extremely rare [17]. Even very small tumors can have a different echogenicity than the surrounding ovary. In addition, these tumors tend to have low impedance-to-flow values [16]. The majority of some tumor types (e.g., Sertoli–Leydig and Leydig) may be moderately or abundantly vascularized [18,19]. It should be noted that the high diastolic flow on Doppler indicative of a tumor can also be caused by a corpus luteum; this potential confusion is less of an issue in postmenopausal women and in those who are anovulatory due to their high androgen levels [17].
The value of other imaging techniques is unclear. In general, CT scan is not considered as sensitive as TVUS for the detection of small ovarian tumors. MRI is potentially useful [20]. One study found a small ovarian tumor appeared as an area of high intensity on diffusion and T2-weighted imaging [16]. The effectiveness of MRI in the evaluation of small ovarian tumors, however, has not yet been fully demonstrated.
Another imaging technique that has been explored is fluorine-18-deoxyglucose positron emission tomography/computed tomography (FDG-PET/CT). After a short period of fasting, 18F-FDG is administered intravenously and increased uptake may be demonstrated in the affected ovary [21]. One of the problems with this technique is that benign luteal cysts may also show increased uptake. The literature contains studies both supporting, and not supporting, the use of this technique.
Interventional testing
Selective venous catheterization and hormonal sampling (SVCHS) is an invasive procedure that has been advocated for the identification and lateralization of very small ovarian tumors not visualized on imaging studies. The resolution of imaging studies has improved since the 1980s and the cases in which this would be applicable today are extremely limited.
The procedure involves selectively sampling testosterone levels from both adrenal glands and from both ovaries. This presents a number of challenges. The success rate of sampling all four sites is 27–45% [16]. In particular, the right ovarian vein is difficult to access due to anatomic issues [17]. In some women, the left ovarian vein may have reflux and drain into the right ovarian vein, thus obscuring any difference in unilateral ovarian secretion [22]. Rupture of the ovarian vein is a potentially serious complication [23].
If prior imaging of the adrenal glands does not demonstrate a tumor, it has been advocated that the adrenal veins not be sampled [24]. If, however, a small adenoma is found on CT scan, sampling of the adrenal veins may definitively rule out the adrenal gland as the source of increased androgens and prevent unnecessary surgery [18].
Often, it is known that the source of excess androgen is ovarian and the key goal then is to lateralize the hypersecretion [14]. Levens et al. [15] looked at venous catheterization data from 136 cases. They found that unilateral lesions of the ovary were associated with a higher ipsilateral (mean 17653 ± 4670 ng/dL) than contralateral (mean 761 ± 150 ng/dL) ovarian effluent concentration. This result, however, was not statistically significant as the high mean ipsilateral value was driven by a small number of extremely high values. They looked at a subset of women with peripheral serum testosterone levels ≥130 ng/dL and calculated the right:left and left:right ovarian testosterone ratios. (In this group, 18 women had right-sided lesions, 14 women had left-sided lesions, and 7 women had bilateral lesions.) They correctly identified 90% of women with a right-sided lesion versus 86% of women with a left-sided or bilateral lesion. When combined with the criterion to identify left-sided lesions, overall, 66% of women were correctly categorized.
If SVCHS is technically difficult to perform or the results are inconclusive, some have advocated the performance of intraoperative ovarian vein sampling [24–26]. If a tumor is very small, the ovary may appear grossly normal on examination and this technique can assist in lateralization of the tumor.
Cases where SVCHS is needed or appropriate are few. It is an invasive procedure with risk attached and should be performed only if it alters patient management (e.g., preserves fertility) [27]. It should be performed in units with special expertise in this area [28]. In a postmenopausal woman with a negative adrenal scan, it may be more appropriate to surgically explore the patient and perform a bilateral oophorectomy, than subject her to a catheterization study, which will be followed by surgery anyway [14].
Summary
The combination of history of rapid onset of symptoms, marked symptomology, and significantly elevated serum testosterone increases the likelihood of a tumor. Even under these circumstances, most women will not have a tumor. The clinician must also be aware of atypical presentations and clinical judgment based on the totality of information is essential. For purposes of ruling out a tumor, in the majority of cases, a routine total testosterone assay is sufficient. The majority of androgen-secreting tumors are palpable on examination. If imaging is required, TVUS is the best method for ruling out ovarian pathology, and CT scan is preferred for adrenal masses. If a tumor is suspected despite negative imaging, and location/lateralization of the tumor will change management, SVCHS may be performed either preoperatively or intraoperatively.
We now turn to the variety of ovarian tumors potentially responsible for the development of virilization in women.
Embryological origin of ovarian tumors
Embryologically, the ovaries initially appear as a pair of longitudinal ridges formed by thickenings of the coelomic epithelium. The proliferation of the cells of the genital ridge into the underlying mesenchyme forms the primitive sex cords, or indifferent gonads. At the same time, primordial germ cells from the yolk sac migrate to and invade the genital ridges. Cells from the adjacent mesonephros also invade the mesenchyme and become closely associated with the germ cells [29,30]. The variety of tumors that can occur in the ovary are all derived from the different cell types that exist in the undifferentiated gonad: mesenchymal, mesonephric, coelomic epithelial, and germ cells.
Any one of these cells, alone or in combination, can become part of an ovarian tumor. In addition, because ovaries and testes develop in a similar fashion during early gestation, cells that are typically associated with testicular tissue, such as Sertoli cells and Leydig cells, can also occur in ovarian tumors [29,31]. Knowing the origin and function of the various cell types of the ovary makes it easier to understand which tumors are likely to have androgen-producing capabilities.
According to the International Agency for Research on Cancer (IARC) classification system [see 29], ovarian tumors can be divided into five general categories. The largest group, epithelial cell tumors, account for 60–70% of all tumors and 90–95% of all ovarian malignancies [29,32]. The stromal component of these tumors can, on very rare occasions, secrete androgens, but this is the exception and will be covered under “idiopathic” tumors [31]. Ovarian sex cord-stromal tumors account for 5–10% of all ovarian tumors and approximately 7% of malignant ovarian tumors [29,32,33]. Approximately 15000–20000 new ovarian sex cord-stromal tumors are diagnosed every year in the United States [32]. Most hormonally active tumors, including those producing androgens, are found in this category. Even in this category, however, the majority of tumors are hormonally inactive. Approximately 70% of sex cord-stromal tumors are benign, inactive fibromas. Potentially active tumors in this category, as well as some of their characteristics, are summarized in Table 13.1 and described below.
Incidence | Androgen secreting | Other hormones | Age | Laterality | Size (diameter) | Malignant | Comments | |
|---|---|---|---|---|---|---|---|---|
2–3% of all ovarian tumors | ||||||||
Adult | 85–95% of granulosa cell tumors | 10% | 75–85% estrogenic | Median age 50 years; peri- and postmenopause | >95% unilateral | Average 12 cm (range 0–30 cm) | 90% stage I (86–96% survival); >stage I (26–49% survival) | |
Juvenile | 5–15% of granulosa cell tumors | Rare | 80% estrogenic | Average 13 years; 97% below age 30 | 98% unilateral | Average 12 cm (range 2.5–32 cm) | 5% malignant; often fatal | 80% in children result in isosexual pseudo-precocity |
One-third as frequent as GCT | 10% if luteinized; rare other | Commonly estrogenic | Mean 59 years; 84% postmenopausal | 95% unilateral | 5–10 cm (range 0–15 cm) | 3% malignant | ||
Sclerosing stromal | 2–6% of sex cord-stromal | Rare | Rarely estrogenic | Mean 27 years; 80% <30 years | Unilateral | 1.5–20 cm | Benign | |
Sertoli–Leydig | <1% of all ovarian tumors | One-third | Occasionally estrogenic | Mean 25 years (15 years, retiform variant; 35 years well differentiated) | 98% unilateral | Average 13.5 cm (majority 5–15 cm) | 12% malignant | Can present with sudden and severe virilization |
Sertoli | Rare; 4% of Sertoli-stromal tumors | Rare | 30% estrogenic | Average 30 years, but can occur any age | Unilateral | Avg. 9 cm (range 4–12 cm) | Most benign | |
0.1% of all ovarian tumors | ||||||||
23% of steroid cell tumors | 12% | Most estrogenic | 80% postmenopausal (range 26–74 years) | 95% unilateral | 0.5–2 cm | Most benign | ||
20% of steroid cell tumors | 80% | Rarely estrogenic or mixed; DHEA-S; androstenedione | Average 58 years (range 32–78 years) | Unilateral | Average 2.4 cm | Rarely malignant | Androgenic signs may develop over many years | |
Not otherwise specified | 58% of steroid cell tumors | 50% | Rarely DHEA-S, androstenedione, estrogen, cortisol, aldosterone | Average 43 years (range 2–80 years) | Average 8.4 cm (range 1.2–45 cm) | 25–40%; extra-ovarian spread at time of diagnosis 20% | Androgenic signs may develop over many years |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
