Adolescent Polycystic Ovary Syndrome Due to Functional Ovarian Hyperandrogenism Persists Into Adulthood

Abstract

Background:

Menstrual irregularity and above-average testosterone levels in adolescence may presage polycystic ovary syndrome (PCOS) in adulthood but persist in only a minority. Prolonged anovulatory cycles in normal adolescents are associated with increased testosterone levels. Thus, questions have been raised about the accuracy of PCOS diagnosed in adolescents.

Objective:

The purpose of this study was to follow-up hyperandrogenic adolescents with features of PCOS to test the hypothesis that adolescent functional ovarian hyperandrogenism (FOH) persists into adulthood.

Study Subjects:

A series of adults previously reported to have adolescent PCOS, with most documented to have FOH by GnRH agonist or dexamethasone androgen-suppression test criteria, were recalled.

Methods:

Recall occurred >3 years after the initial diagnosis and at the age of >18.0 years. Respondents underwent examination, baseline androgen evaluation, and an oral glucose tolerance test after discontinuing oral contraceptive therapy.

Results:

Of the adolescent hyperandrogenic patients, 68% (15 of 22) were traceable, and 60% of those traced returned for follow-up, including half (n = 8) of the original FOH group. The baseline characteristics of respondents and nonrespondents were not significantly different. Patients with FOH were reevaluated when their mean age was 23.0 years (range, 18.4–29.4 years), gynecologic age was 10.7 years (range, 5.5–18.4 years), and body mass index was 42.3 kg/m² (range, 28.3–52.1 kg/m²; P = .02 vs adolescence). Serum free testosterone was 24 pg/mL (range, 10–38 pg/mL, normal, 3–9 pg/mL; not significant vs adolescence); all were oligomenorrheic. Whereas 3 of 8 had impaired glucose tolerance as adolescents, at follow-up 6 of 8 had developed abnormal glucose tolerance (2 with type 2 diabetes mellitus).

Conclusions:

Adolescents with FOH, which underlies most PCOS, uniformly have persistent hyperandrogenism, and glucose tolerance tends to deteriorate. Testing ovarian androgenic function in hyperandrogenic adolescents may be of prognostic value.

Anovulatory menstrual cycles are normally more frequent in adolescents than in adults and usually mature by the end of the second postmenarcheal year (1, 2). Persistent menstrual irregularity is most often due to polycystic ovary syndrome (PCOS) (3, 4), the most common cause of chronic hyperandrogenic anovulation and infertility in adult women (1, 5).

Testosterone serum levels increase during prolonged adolescent anovulatory cycles (6, 7). Population studies have shown that serum androgen levels of healthy adolescent schoolgirls correlate with those in adulthood (8), and above-average levels in midadolescence are associated with subsequent PCOS features (3, 8–10). However, none of these studies provided data on the prognostic value of elevated androgen levels. Among oligomenorrheic midadolescents with elevated total testosterone levels who were identified in a random population sample of 766 participating healthy Danish schoolgirls, 47 returned for examination (10): of these, 38 girls were otherwise asymptomatic, with testosterone elevation persisting in 43% of them; however, 9 girls had clinical or ultrasound features of PCOS, and total testosterone elevation persisted in most of them (78%). These studies may have underestimated the prevalence of hyperandrogenemia because serum free testosterone levels, the single most sensitive indicator of hyperandrogenism (11), were not reported.

Among girls evaluated for symptomatic anovulatory menstrual irregularity without clinical signs of hyperandrogenism, about half have elevated androgen levels (6, 12). Reevaluation of such patients after an average of 2.5 years suggested that hyperandrogenemia resolved in about half (13, 14). Six-year follow-up of adolescents with symptomatic menstrual irregularity of diverse causes revealed that menstrual abnormalities persisted in 62% of the subjects not using hormonal contraception, and 59% of these fulfilled the Rotterdam criteria for PCOS (4). However, the Rotterdam criteria seem too broad for specificity of a diagnosis of PCOS in adolescents (5, 15).

In view of the frequent resolution of adolescent menstrual pattern abnormalities and hyperandrogenemia, there is uncertainty about whether PCOS diagnosed in adolescents is accurately predictive of adult PCOS. Some have expressed the opinion that the diagnosis of PCOS in adolescents should be considered provisional unless all the classic features of the syndrome are present, including polycystic ovary morphology (16), although appropriate criteria for polycystic ovary morphology in adolescents have not been firmly established (15).

It is known that functional ovarian hyperandrogenism (FOH) underlies the great majority of PCOS (17, 18). FOH is typically characterized by 17-hydroxyprogesterone hyperresponsiveness to gonadotropins, in the absence of evidence of a steroidogenic block, as determined by a GnRH agonist (GnRHag) or human chorionic gonadotropin challenge test. An alternative indication of FOH is an elevated testosterone level after adrenocortical suppression by a dexamethasone androgen-suppression test (DAST). Our data indicate that adolescent and adult patients with PCOS not only have similar clinical findings but they also have similar ovarian androgenic dysfunction (1, 15, 19). Thus, we would predict that adolescent FOH would persist into adulthood. To test this hypothesis, we undertook reevaluation of a group of previously reported hyperandrogenemic adolescents whose source of androgen excess was well characterized (19) to test the hypothesis that adolescent PCOS, particularly that associated with FOH, persists into adulthood.

Subjects and Methods

Subjects

The study subjects were respondents to a return receipt letter sent to 22 previously reported (19) adolescents with hyperandrogenism unexplained by known causes such as nonclassic congenital adrenal hyperplasia. They had initially been referred for symptoms suggestive of PCOS, were documented to have hyperandrogenemia, and then consented to undergo determination of the source of their androgen excess by a GnRHag test, a 5-day DAST, and a cosyntropin (ACTH) test; all tests were conducted over a period of 2 months. The recall letter invited return for an outpatient Clinical Research Center reevaluation at no cost >3.0 years after the initial evaluation when all were older than 18.0 years of age. Of the initial group, 32% (n = 7) proved to be untraceable after further follow-up phone calls. Of those traced, 60% (9 of 15) returned for follow-up and comprised the current study group (“respondents”).

Methods

Respondents all underwent a physical examination, baseline androgen measurements, and an oral glucose tolerance test (OGTT) (in 2 subjects baseline glucose tolerance was instead assessed by 24-hour glucose monitoring) after discontinuing use of any oral contraceptive therapy for at least 2 months. Four also consented to an additional cosyntropin test, 5 to a transabdominal pelvic ultrasound examination, 2 to a human chorionic gonadotropin stimulation test, and 1 to a GnRHag test. Subjects were considered oligoanovulatory if menses were abnormally infrequent for age or gynecologic (postmenarcheal) age (1). Clinical research test protocols, laboratory methods, and normative data have been reported previously (18–22). Unconjugated steroid measurements were performed by RIAs validated against liquid chromatography-tandem mass spectrometry (22), and insulin was measured by a double-antibody RIA (18). OGTTs were performed and interpreted according to American Diabetes Association guidelines (23). Insulin resistance was indexed by homeostasis model assessment (HOMA) (24); normative data were derived from the normoandrogenic volunteer subset spanning the study group's age (18). Summary statistics were compiled using Excel (Microsoft Corp), and statistical comparisons were made by paired or unpaired Student t tests or Fisher exact test, as appropriate, using Prism (GraphPad Software, Inc). These studies were approved by the University of Chicago Institutional Review Board.

Results

Baseline characteristics of adolescent study group

Of the original cohort of 22 hyperandrogenic adolescents, 73% had the PCOS type of FOH according to GnRHag or 5-day DAST criteria. Of the original cohort, 55% also had functional adrenal hyperandrogenism (FAH), defined as dehydroepiandrosterone (DHEA) hyperresponsiveness to ACTH; in 3 cases (14% of the entire cohort, 23% of those with FAH), it was isolated, ie, not associated with FOH. Neither an ovarian nor adrenal source for the hyperandrogenemia could be identified in 3 nonrespondents, 2 of whom were untraceable.

Baseline characteristics of respondents (n = 9) (Table 1) and nonrespondents (n = 13) (Table 2) were not significantly different, although there were nonsignificant tendencies for a lower prevalence of FOH and abnormal glucose tolerance in nonrespondents. The plasma total testosterone of respondents (87 ± 50 ng/dL; 3.0 ± 1.7 nM) and nonrespondents (76 ± 39 ng/dL; 2.6 ± 1.4 nM) and the prevalence of FAH (67% vs 46%) were also not significantly different. Ethnicity of respondents was 5 non-Hispanic black and 4 non-Hispanic white and that of nonrespondents was 3 non-Hispanic black, 7 non-Hispanic white, and 3 other.

Table 1.

Hyperandrogenic Adolescents Followed Into Adulthood

Case	Adolescent Evaluation										Adult Recall Evaluation
	Age, y	Gyn Age, y	Menses	Hirsutism Scorea	BMI, kg/cm2	Free Testost, pg/mLb		OGTT	GnRHag	DAST	Age, y	BMI, kg/cm2	Free Testost, pg/mLb	OGTT
Functional ovarian hyperandrogenism
A	11.3	0c	Oligod	1	24.7		12	IGT	Pos	Pos	23.0	38	38	IGT
B	16.4	0.5	2° Amen	13	39.7		72	IGT	Pos	Pos	22.5	43.5	33	T2DM
C	17.2	3.5	2° Amen	10	29		24	N	Pos	Pos	27.6	31.2	29	T2DM
D	13.6	1.8	2° Amen	10	42.1		44	N	Pos	Pos	22.5	47.4	21	N
E	14.8	2.8	Nc	5	28.3		10	N	Pos	N	18.4	28.3	10	IGT
F	13.8	0	2° Amend	14	48.6		43	IGT	N	Pos	19.3	52.1	22	IGT
G	16.6	5.6	Oligo	12	34		31	Ne	N	Pos	29.4	46.2	23	N
H	17.3	8.8	Oligo	20	30.2		33	Ne	N	Pos	21.0	51.5	17	IGT
Mean	15.2	2.9		10.6	34.6		33.6				23.0	42.3	24.1
SD	2.0	3.1		5.8	8.2		20.0				3.8	9.0	9.0
Functional adrenal hyperandrogenism, isolated
I	16.6	5.6	N	19	23.1	11		N	N	N	27.0	23.9	7	N

Abbreviations: Amen, amenorrhea; Gyn, gynecologic; N, normal; Oligo, oligomenorrhea; Pos, positive for FOH; Testost, testosterone; T2DM, type 2 diabetes mellitus.

^aHirsutism is indicated by a score of >7.

^bFree testosterone normal upper limit is 9 pg/mL; to convert to SI units (nmol/L), multiply by 0.0347.

^cPresented with the chief complaint of acanthosis nigricans.

^dAfter perimenarcheal presentation, oligoamenorrhea developed within 1 year of menarche.

^eAssessed by 24-hour glucose monitoring.

Table 2.

Hyperandrogenic Adolescents Not Followed Into Adulthood (n = 13)

	Age, y	Gyn Age, y	Oligoamenorrhea	Hirsutism Score	BMI, kg/cm2	Free Testost, pg/mL	OGTT Abnormal	GnRHag and DAST Abnormal
Mean	15.1	3.1	7/13	12.0	31.9	25.8	2/10*	8/13 GnRHag Pos; 4 also DAST Pos
SD	2.0	1.6		5.5	9.3	14.7

Abbreviations: Gyn, gynecologic; Pos, positive; Testost, testosterone.

^*Data were not available in 3 untraceable patients.

The initial chief concern of the respondents was hirsutism (n = 4, 1 with co-complaints of acne and obesity), menstrual dysfunction (n = 3, 1 with the co-complaint of obesity), or acanthosis nigricans (n = 2, neither with the co-complaint of obesity). Although menses were not abnormal at presentation in 3 of the patients with FOH, they became abnormal within 1 year of menarche in the 2 who were diagnosed before menarche and later in the other. Menses were also normal in the 1 patient with isolated FAH. Impaired glucose tolerance (IGT) was present in 3 of 8 of those with FOH, 1 of whom previously had transient glucocorticoid-induced diabetes mellitus. Fasting glucose and insulin were normal in all, however (Table 3).

Table 3.

Glucose Tolerance Test Results Before and After Recall

Case	Adolescent Evaluation					Adult Recall Evaluation
	Blood Glucose, mg/dL		Serum Insulin, μU/mL		HOMA	Blood Glucose, mg/dL		Serum Insulin, μU/mL		HOMA
	0 Min	120 Min	0 Min	120 Min		0 Min	120 Min	0 Min	120 Min
Functional ovarian hyperandrogenism
A	83	141	47	710	9.6	86	153	360	3610	76.4
B	77	170	17	87	3.2	101	205	16	100	4.0
C	99	136	45	250	11.0	112	214	28	187	7.7
D	98	120	39	333	9.4	98	105	88	337	21.3
E	86	128	33	325	7.0	86	179	12	470	2.5
F	92	156	25	175	5.7	94	150	40	350	9.3
G	92	Na	30	nab	6.8	98	139	40	200	9.7
H	85	Na	30	nab	6.3	117	149	78	138	22.5
Mean	89	142	33	313	7.4	99c	162	83	674	19.2
SD	7.6	18	10	216	2.5	11	36	115	1193	24.3
Functional adrenal hyperandrogenism, isolated
I	94	109	35	75	8.1	94	95	12	25	2.8
Reference range	78–109	88–139	5–56	17–278	1.1–12.5	78–109	88–139	5–56	17–278	1.1–12.5

Conversion factors: divide glucose by 18.0 for mM, and multiply insulin by 6.0 for pM. Bold designates abnormal values.

^aN, blood glucose 24-hour monitoring normal.

^bna, not available.

^cP = .05 vs initial fasting blood glucose.

The respondent adolescents all had either an ovarian or adrenal source for androgen excess, and the vast majority of respondents (89%) had FOH, which constitutes half of the original cohort with FOH, whereas the other respondent had isolated FAH (Table 1).

Among the respondents with FOH, 6 of 8 had clinical evidence of hyperandrogenism and oligoanovulation, ie, the combination of hirsutism and menstrual dysfunction within 1 year of presentation, although these were not necessarily the presenting complaint, even when present. The other 2 patients with FOH presented for acanthosis nigricans, at which time neither menstrual disturbance or hirsutism was apparent. However, 1 developed oligomenorrhea during the first postmenarcheal year (case A) and the other (case E) presented with moderate inflammatory acne.

Follow-up characteristics of respondent study group as young adults

The former adolescent patients were reevaluated as adults at an average age of 23.4 years (range, 18.4–29.4 years) and after 8.1 years (range, 3.7–12.8 years) elapsed time (Table 1); their average gynecologic age was 11.3 years (range, 5.5–16 years). Problems with menstrual abnormality and hirsutism during adolescence persisted and were under treatment; 1 patient gradually became hirsute as an adult. Most were prescribed estrogen-progestin oral contraceptive pills (OCPs) for these symptoms. Two instead took cyclic progestins because of parental objections to OCP use; 3 used OCPs erratically during adolescence because of their interest in becoming pregnant, and they ultimately sought reproductive endocrinology consultation for infertility.

Among the patients with FOH as adolescents, all had persistent elevation of plasma free testosterone (not significant vs adolescent baseline). Body mass index (BMI) had increased in 7 of 8 (P = .02). Six of 8 had developed abnormal glucose tolerance, with 2 having developed type 2 diabetes mellitus (Table 3). One developed endometrial carcinoma at 22 years of age.

The type of androgenic ovarian dysfunction present in adolescence, typical or atypical FOH, persisted in the 2 patients with FOH who consented to a restudy of ovarian androgenic function as adults. Similarly, the eumenorrheic, hirsute patient with isolated FAH in adolescence again had FAH when restudied as an adult and never developed FOH.

Instructive cases

Case A was referred for severe acanthosis nigricans at 11.3 years of age when BMI was 24.7 kg/m² (90th percentile for bone age of 13.9 years). FOH was diagnosed by DAST before menarche at 11.7 years of age, and typical FOH was documented by a GnRHag test, along with FAH, 2 months later. Ovaries were normal by ultrasound. OGTT showed IGT with a very elevated serum insulin response (Table 3). She developed oligomenorrhea within the first postmenarcheal year and gradually developed hirsutism by 23 years (score 13, normal ≤7). She progressed to obesity; IGT persisted and hyperinsulinemia increased remarkably, and acanthosis nigricans persisted. She married in her late teens and because of infertility ovulation was eventually induced by clomiphene citrate treatment, with subsequent delivery of a healthy baby.

Case B presented for primary amenorrhea at 16.4 years of age. She had a history of developing transient diabetes mellitus while receiving a course of steroid therapy for Guillain-Barré syndrome that then resolved. She had hirsutism, obesity, and mild acanthosis nigricans; laboratory studies showed IGT and typical FOH. Upon follow-up, secondary amenorrhea and hirsutism persisted, obesity worsened, and diabetes mellitus developed (Table 3).

Case C underwent abortion for an unwanted early teenage pregnancy; when she discontinued OCPs to attempt pregnancy at 16 years of age, she became amenorrheic and presented at 17.2 years old. She had mild hirsutism, mild obesity, and acanthosis nigricans; typical FOH and FAH were found, as were polycystic ovaries (largest ovary volume of 13.6 mL). Upon follow-up at 27.6 years, obesity had increased slightly, and she developed diabetes mellitus (Table 3). 17-Hydroxyprogesterone was hyperresponsive to human chorionic gonadotropin, as reported previously (21), consistent with persistence of typical FOH.

Case E was referred for acanthosis nigricans at 14.8 years of age when 2.8 years postmenarcheal; BMI was 28.3 kg/m² (93rd percentile for bone age of 14.8 years). Menses were normal. She had moderate inflammatory acne, but not hirsutism. OGTT was normal, but the insulin response was elevated (Table 3). Menses occurred at generally 28- to 30-day intervals. She was then lost to follow-up after 1 year until recalled at 18.4 years of age, by which time she had become oligomenorrheic and developed IGT.

Case F presented with moderate hirsutism, moderate inflammatory acne, and obesity at 13.8 years of age when premenarcheal with bone age of 16.5 years, suggesting primary amenorrhea; FOH was documented by DAST. Menarche occurred at 14.3 years, after which she developed secondary amenorrhea. Polycystic ovaries were found (largest ovary volume was 20 mL), and IGT was documented by OGTT. She declined OCP therapy and took her cyclically prescribed progestin erratically. She remained oligomenorrheic, hirsute, and obese when reevaluated at 5.5 postmenarcheal years. She subsequently married, and unguided efforts to become pregnant were futile. At 22 years of age, at 8.5 years postmenarche, she developed endometrial carcinoma.

Case I was initially evaluated at 16.6 years for moderately severe hirsutism as her sole symptom. She had menses every 30 to 40 days. She was found to have isolated FAH: dehydroepiandrosterone-sulfate (DHEAS) was 198 μg/dL (normal upper limit is 180 μg/dL [4.45 μM]) and the peak DHEA response to 10 μg/m² ACTH was 1530 ng/dL (normal upper limit is 1470 ng/dL [51.0 nM]). She had no evidence of ovarian dysfunction by GnRHag test, DAST, or ultrasound. At 19 years of age, plasma free testosterone remained elevated (19 pg/mL). Upon recall at 27.0 years, BMI and menses remained normal, and hirsutism was an ongoing problem, but she had classic enlarged polycystic ovaries. Free testosterone normalized (7 pg/mL); GnRHag test and DAST results were still normal. However, FAH persisted, with elevated baseline DHEAS (213 μg/dl) and peak DHEA responses to 1.0 μg ACTH (1250 ng/dL, normal upper limit is 1012). She remained unmarried and had not attempted pregnancy.

Discussion

The main finding from this recall series of patients is that the hyperandrogenic ovarian dysfunction (FOH) of PCOS detected in adolescence uniformly predicted PCOS in adulthood. Although adolescent hirsutism alone or oligoamenorrhea alone is not highly predictive of adult PCOS, in this series the combination of the 2, which was present in 75% of our patients with FOH and meets the standard criteria for PCOS (ie, otherwise unexplained evidence of hyperandrogenic anovulation) (5), was highly predictive of ongoing PCOS.

Symptomatic ovulatory dysfunction often remits, and it is only after 1 to 2 years of persistent menstrual abnormality that the probability of ongoing menstrual disturbance rises to 50% or more (1, 2). Our data are consistent with the evidence suggesting that most of the patients with persistent abnormalities have PCOS (4, 10), particularly if other clinical features such as hirsutism, inflammatory acne, acanthosis nigricans, or obesity are present. Thus, it appears that the patients with remitting cases are those in whom physiologic anovulation has waned (ie, maturation of hypothalamic-pituitary-ovarian function has occurred) and those in whom causes such as eating disorders (4) have been identified and treated satisfactorily.

The diagnosis of PCOS in adolescence is confounded by the fact that the increased androgen levels of prolonged anovulatory cycles do not reliably predict later PCOS (8–10). However, our data indicate that persistent frank hyperandrogenemia with clinical evidence of PCOS persists. We previously used specific testing with the GnRHag test and DAST to demonstrate that such girls have ovarian androgenic dysfunction similar to that of adults (15), which had suggested that most PCOS diagnosed in adolescence will persist. The present follow-up study supports that conclusion by demonstrating that documentation of the PCOS type of ovarian androgenic dysfunction by specific testing is predictive of ongoing PCOS. Thus, these tests seem to be useful in distinguishing the hyperandrogenemia that transiently arises from physiologically prolonged anovulatory cycles (6, 7) from adolescent PCOS. To facilitate this testing, we have subsequently shortened the DAST to an afternoon procedure, after which the GnRHag leuprolide acetate can be administered and post-GnRHag sampling performed the following day (22).

PCOS is recognized to be a syndrome of heterogeneous origin (25). Although most of our patients developed the full clinical and biochemical picture of PCOS in the perimenarcheal period, 1 patient (case I) had a variant phenotype that evolved over time. In adolescence, she presented with isolated FAH. While her hirsutism persisted, her mildly elevated plasma free testosterone gradually became normal by 27 years of age. Meanwhile, DHEAS elevation and DHEA hyperresponsiveness to ACTH persisted, and she developed polycystic ovaries. Her latest findings were consistent with PCOS according to the broad Rotterdam criteria (25).

These data also support the concept that the most widely accepted criterion for the diagnosis of PCOS in adolescents, otherwise unexplained hyperandrogenic oligoanovulation (5, 15), does not necessarily encompass the whole clinical spectrum of adolescent PCOS. Our early data indicated that 13% of hyperandrogenic women with positive responses to GnRHag testing were eumenorrheic (20). In 2 patients in the present series (cases A and E), acanthosis nigricans in overweight nonobese girls, 1 of whom had prediabetes (IGT), was the only clinical feature initially: 1 presented shortly before an early menarche and the other presented with normal menses; both eventually became oligomenorrheic. Neither had a polycystic ovary although both had functionally typical FOH. These cases suggest that signs of insulin resistance may occasionally be the sole clinical feature of PCOS early in its course, even in the preteen prepubertal years (26), and a harbinger of infertility.

Metabolic issues were a major problem for most of our adolescents with FOH. At presentation, 62.5% of this group was obese and the remainder were overweight. The group then continued to gain weight, at an average rate of about 1 kg yearly. Upon recall as young adults, fasting blood glucose had risen (P = .05), and 75% had abnormal glucose tolerance with type 2 diabetes mellitus in 25% of the group. Increasing obesity played a role in the development of diabetes by substantially aggravating insulin resistance in some patients, but inherent defects in β-cell secretion, not assessed in this study, exist in PCOS and are essential to diabetes development (27).

Obesity has a complex relationship with PCOS that is incompletely understood. Obesity is recognized to be an aggravating factor in PCOS: the compensatory hyperinsulinemia of obesity-induced insulin resistance stimulates ovarian androgen production (17, 28, 29). On the other hand, obesity may result from the insulin-resistant hyperandrogenemia of PCOS because insulin stimulates adipogenesis (30). It has also been postulated that obesity predisposes to the developmental phase of adolescent PCOS by causing androgen-mediated premenarcheal LH excess (31). However, it is not clear whether premenarcheal obesity and mild androgen excess are the causes of later PCOS, an early manifestation of PCOS, or the result of simple obesity mimicking PCOS because insulin stimulates adipocyte testosterone production (22, 30).

Three of our patients (cases A, C, and F) attempted pregnancy by their late teenage years. One (case E) developed endometrial carcinoma at 22 years of age after repeatedly refusing regular cyclic progestin therapy in her unguided efforts to become pregnant. This cautionary experience suggests that early diagnosis and guidance counseling about preventing endometrial hyperplasia should be important aspects of the management of adolescent PCOS.

The major limitation of our study is its small size, limited follow-up (60% of traceable patients), and low statistical power. Nevertheless, the patients' ovarian, adrenal, and pancreatic β-cell function were well characterized at baseline, and the follow-up was the longest hitherto reported. The shortest duration of follow-up was 3.7 years and the youngest postmenarcheal age at the time of our reevaluation was 5.5 years, but after recall this patient is known to have had on-going oligomenorrhea and to develop endometrial carcinoma at 8.5 postmenarchal years; thus, all were evaluated >6 years postmenarche, at a stage when ovulatory cycles are fully mature (1). In addition, referral bias undoubtedly contributed to the high prevalence of hirsute, oligomenorrheic, obese patients in our series. However, the very high prevalence of hirsutism and obesity in this series does not diminish the fact that PCOS has become recognized more often in nonhirsute and nonobese girls with menstrual disorders in recent years (15). Our data are consistent with the evidence that clinical features in a hyperandrogenemic girl with menstrual dysfunction indicates a very high risk for PCOS (10). Although most of our patients were referred in their midteens, 2 patients in our series demonstrate that the diagnosis can be made shortly before menarche.

In conclusion, our data indicate that adolescents with FOH, the ovarian androgenic dysfunction that underlies the great majority of PCOS, uniformly had persistent hyperandrogenism, BMI increased in most, and glucose tolerance tended to deteriorate. Because 75% of FOH adolescents in this series had hirsutism and abnormal anovulatory symptoms, this further indicates that, at the least, in the great majority of adolescents recognized as meeting the current PCOS criteria, PCOS will persist into adulthood. Testing for ovarian androgenic function in hyperandrogenic adolescents seems to be of prognostic value.