Evidence for Increased 5-Reductase Activity During Early Childhood in Daughters of Women with Polycystic Ovary Syndrome

Context: Polycystic ovary syndrome (PCOS) is a heritable, complex genetic disease. Animal models suggest that androgen exposure at critical developmental stages contributes to disease patho-genesis. We hypothesized that genetic variation resulting in increased androgen production produces the phenotypic features of PCOS by programming during critical developmental periods. Although we have not found evidence for increased in utero androgen levels in cord blood in the daughters of women with PCOS (PCOS-d), target tissue androgen production may be amplified by increased 5 (cid:2) -reductase activity analogous to findings in adult affected women. It is possible to noninvasively test this hypothesis by examining urinary steroid metabolites. Objective: We performed this study to investigate whether PCOS-d have altered androgen metabolism during early childhood. Design, Twenty-one years old, of medical spectrometry. Twenty-four hour steroid excretion rates and precursor to product ratios sug-gestive of 5 (cid:2) -reductase and 11 (cid:3) -hydroxysteroid dehydrogenase activities were calculated. Results: AgedidnotdifferbutweightforlengthZ-scoreswerehigherinPCOS-dcomparedtocontrolgirls ( P (cid:2) .02). PCOS-d had increased 5 (cid:2) -tetrahydrocortisol:tetrahydrocortisol ratios ( P (cid:2) .04), suggesting increased global 5 (cid:2) -reductase activity. There was no evidence for differences in 11 (cid:3) -hydroxysteroid dehydrogenase activity. Steroid metabolite excretion was not correlated with weight. Conclusions: Our findings suggest that differences in androgen metabolism are present in early childhoodinPCOS-d.Increased5 (cid:2) -reductaseactivitycouldcontributetothedevelopmentofPCOS by amplifying target tissue androgen action. ( J Clin Endocrinol Metab 101: 2069–2075, 2016)

women (3). We have found increased circulating testosterone (T) levels in PCOS-d as young as age 8 years (4). Although a number of susceptibility loci for PCOS have been reproducibly mapped (5,6), these loci account for less than 5% of the approximately 70% heritability of PCOS (7). Epigenetic mechanisms may account for some of this so-called "missing heritability" that has been observed in PCOS and other complex genetic traits (8). Phenocopies of PCOS can be produced in animal models by exposure to androgens during gestation (9,10), neonatally (11), or peripubertally (12). Therefore, the PCOS phenotype may result from the interactions of genetic variation resulting in hyperandrogenemia and the epigenetic actions of these androgens in utero as well as during critical postnatal developmental windows.
The source of androgens could be increased glandular androgen production. However, at birth, most studies (13,14) suggest that cord blood androgen levels in PCOS-d are decreased. It remains possible that target tissue androgen action is amplified by upregulation of 5␣reductase activity resulting in increased production of DHT from T (15). If such increases in 5␣-reductase activity were the result of variation in PCOS susceptibility genes, target tissue androgen action would be enhanced in affected individuals when T was available.
Alterations in intracellular steroidogenic enzyme activity could also increase adrenal androgen production. Increased 5␣-reductase or decreased 11␤-hydroxysteroid dehydrogenase type 1 (11␤HSD1) activity would enhance cortisol metabolism resulting in a compensatory increase in ACTH secretion and stimulation of adrenal steroidogenesis (16,17). Indeed, evidence for increased 5␣-reductase and decreased 11␤HSD1 activity has been found in adult women with PCOS (16 -21). It is possible to noninvasively assess global androgen metabolism by measuring urinary metabolites and precursor: product ratios. We tested the hypothesis that 5␣-reductase activity is increased in PCOS-d, who are at genetically high risk for PCOS, by measuring urinary steroid metabolites.

Study population
Twenty-one PCOS-d and 36 control girls aged 1 to 3 years were studied. We chose this age range to ensure that all subjects were at a similar developmental stage when the mini-puberty of infancy was completed, but the infants were still in diapers for ease of urine collection (22). Target sample sizes of 20 cases and 40 controls were calculated for the primary end point, the ratio of 5␣-reduced tetrahydrocortisol (5␣THF) to its 5␤-reduced metabolite, THF (5␣THF/THF), a measure of 5␣-reductase activity (17,23), assuming 80% power and alpha of 0.05 based on the mean and SD of this ratio in previous studies comparing PCOS to control women (17,23). Subjects were recruited by advertisements in local media and on-line, and by contacting women who have previously participated in our studies of PCOS and control adult women. PCOS-d had a mother who fulfilled National Institutes of Health criteria for PCOS (hyperandrogenism and oligo-anovulation with exclusion of other reproductive disorders (1) as confirmed before recruitment by us or by their personal physician. The mothers of control girls had regular menses every 27-35 days and no signs or symptoms of androgen excess by a validated questionnaire (2). All girls were in good health and not taking any medications known to alter steroid hormone metabolism. The Institutional Review Boards of the Feinberg School of Medicine, Northwestern University, and Ann & Robert H. Lurie Children's Hospital of Chicago approved this study. Written informed consent was obtained from the parent or legal guardian of all subjects.

Data collection
Age, height, weight, medical and reproductive histories were ascertained by validated questionnaires (2) in PCOS and control mothers. Timed urine samples were obtained in PCOS-d and control girls through collection of consecutive wet diapers for a minimum of 10 and maximum of 24 hours. Pediatric growth records for height and weight were available in 21 PCOS-d and 30 control girls. Weight for length Z score was calculated based on the US Centers for Disease Control and Prevention 2000 growth charts (24).

Assays
Urine was extracted from diapers as described previously (25) and urinary steroid metabolites measured by a quantitative gas chromatography-mass spectrometry (GC-MS) selected ionmonitoring method (18). Metabolite concentrations were corrected to reflect 24-hour excretion rates by applying the following formula: [(metabolite concentration (g/L) ϫ urine volume (L))/(24/urine collection time (h))]. Substrate-to-product metabolite ratios were calculated to reflect the in vivo net activity of steroidogenic enzymes. Specifically, the ratio of the glucocorticoid metabolite 5␣THF to its 5␤-reduced metabolite, THF (5␣THF/THF), was the primary ratio used to assess 5␣-reductase activity because of the low concentrations of androgen and mineralocorticoid metabolites in this age group (Table 2) (26). The following ratios were used to assess 11␤-hydroxysteroid dehydrogenase type 1 (11␤-HSD1) activity: (5␣THF ϩ THF)/tetrahydrocortisone (THE), and cortols/cortolones (27). Total cortisol production was quantified by the sum of cortisol metabolites: THF ϩ 5␣THF ϩ cortisol ϩ cortols (26). Total cortisone production was quantified by the sum cortisone metabolites: THE ϩ cortisone ϩ cortolones (26). Total glucocorticoid production was quantified by the sum of total cortisol and total cortisone production (26). Total androgen production was quantified by the sum of androsterone (An) ϩ etiocholanolone (Et) (26).

Statistical analysis
Data were log-transformed when necessary to achieve homogeneity of variance. Differences between groups were assessed by two-tailed unpaired t tests or Mann-Whitney tests, if homogeneity of variance was not achieved. Because weight for length Z score was higher in the PCOS-d, Pearson or Spear-man correlation, depending on the normality of the data, was performed on all endpoints to determine whether there was a significant correlation with weight for length Z score. Because there was no correlation of weight for length Z score with any endpoint, no statistical adjustment for weight was necessary. Statistical analyses were performed using SAS 9.4 (SAS Institute, Inc.). Categorical variables were compared by Fisher's exact test. Differences were considered to be significant at P Յ .05. Data are reported as the untransformed mean Ϯ SD.

Results
The PCOS mothers had increased body mass index (BMI) compared to control mothers but their age did not differ (Table 1). Age and ethnicity of the PCOS-d and control girls were similar by design (Table 1). Weight for length Z score (P ϭ .02, Table  1) was higher in the PCOS-d. The length of urine collection did not differ between the groups (P ϭ .28, Table 1).   Figure 1 depicts the steroidogenic enzyme and urinary steroid metabolites investigated. There were no differences between the groups in 24-hour excretion of any of the cortisol metabolites, although there was a trend toward increased ␣-cortol in PCOS-d (P ϭ .07, Table 2). Excretion of the cortisone metabolites and androgen metabolites was also similar ( Table 2). Ex-cretion of the mineralocorticoid metabolites was similar between the groups, with the exception of 5␣-tetrahydrodeoxycorticosterone, which was higher in the PCOS-d (P ϭ .05, Table 2). Finally, excretion of total cortisol metabolites (211 Ϯ 212 PCOS-d vs 139 Ϯ 99 control girls, g/24 hours, P ϭ .16), total cortisone metabolites (276 Ϯ 264 PCOS-d vs 238 Ϯ 188 control girls, g/24 hours, P ϭ .55), total glucocorticoid metabolites (498 Ϯ 466 PCOS-d vs 377 Ϯ 281 control girls, g/24 hours, P ϭ .30, Figure 2) and total androgen metabolites (4.5 Ϯ 4.0 PCOS-d vs 4.1 Ϯ 4.3 control girls, g/24 hours, P ϭ .53, Figure 2) did not differ between the groups.

Discussion
PCOS-d had evidence for increased global 5␣-reductase activity analogous to findings in adult women with PCOS (16, 17, 19 -21). 5␣-reductase converts testosterone to its more potent 5␣-reduced metabolite, DHT (15). Intracellular conversion of testosterone to DHT is responsible for androgen actions in most target tissues (15). In women, increased 5␣reductase activity has been associated with idiopathic hirsutism (28),   .04, A). There was no difference in An/Et, a second ratio indicating 5␣-reductase activity (P ϭ .59, B). This discrepancy may be related to the very low concentrations of these androgen metabolites in this young cohort. There was no significant difference between the groups in total glucocorticoid metabolites (P ϭ .30, C) or total androgen metabolites (P ϭ .53, D). Data are mean and SEM. androgenic alopecia (29), and acne (30). Increased 5␣reductase activity in specific tissues, such as the skin (31) and ovary (21), has been reported in women with PCOS. If genetic variation caused increased 5␣-reductase activity in androgen target tissues of affected PCOS-d, it would result in the amplification of peripheral androgen action by increasing the intracellular production of DHT from T (32,33). These findings are consistent with our hypothesis that increases in target tissue androgen production play a role in the development of PCOS. Increased 5␣-reductase activity would only be physiologically relevant when T was available. Although persistent increases in androgen secretion do not occur until the time of adrenarche (32), there are earlier developmental phases in which androgen secretion may be increased in girls. The fetal adrenal secretes DHEA, an androgen precursor that may undergo extraglandular metabolism to T (34). Although less is known about fetal ovarian androgen production, CYP17A1 is strongly expressed in fetal theca cells starting as early as 33 weeks, suggesting that intrauterine ovarian androgen production is possible (35). A "mini-puberty" of infancy occurs during the first 3-4 months of life during which time the hypothalamic-pituitary-gonadal axis is active, so it is biologically plausible that ovarian androgen production could be increased during this time (22). Following the first year of life, adrenal and ovarian androgen production remains quiescent until the time of adrenarche and subsequent puberty (32,33). Detailed investigation of gonadal and adrenal steroid physiology during infancy and early childhood has been limited because of ethical and practical challenges of performing invasive studies in this age group (36). Analysis of urinary steroid metabolites provides a valuable means to noninvasively investigate sex steroid production in young children (18).
Androgen exposure in utero (9, 10), neonatally (11), or peripubertally (12) produces phenocopies of PCOS in several animal species (37). During gestation, robust placental aromatase activity likely protects the fetus from exposure to the increased circulating androgen levels in mothers with PCOS (38). Nevertheless, it remains possible that increases in fetal glandular or target tissue androgen production are present in females at risk for PCOS, such as those daughters of affected women who inherited PCOS susceptibility genes. No increase in cord blood androgen levels has been found in PCOS-d in studies using sensitive and specific testosterone assay methods (13,14). However, increases in androgen production earlier in gestation or enhanced target tissue 5␣-reductase activity resulting in local DHT increases could have escaped detection in studies using cord blood (13,14). Alternatively, there could be increases in androgen production postnatally during other developmental windows, such as the "mini-puberty" of infancy (22).
In contrast to adult affected women, only the 5␣THF/ THF ratio was increased in PCOS-d, whereas the An/Et ratio did not differ between the groups. However, the excretion of An and Et was near the limits of detection of the assay, limiting the utility of this ratio as a measure of 5␣reductase activity in children of this age, in whom androgen generation is physiologically very low (32,33). In adult women with PCOS, 5␣-reductase was positively correlated with body weight (17,23). Although body weight was significantly higher in PCOS-d compared to control girls, there was no correlation of body weight with any urinary steroid metabolites suggesting weight differences did not contribute to the differences in 5␣-reductase activity we found in PCOS-d.
Increased hepatic 5␣-reductase activity may also enhance cortisol inactivation and clearance, resulting in a compensatory increase in ACTH secretion to normalize circulating cortisol levels, at the expense of increased adrenal androgen production (16,17). The excretion of glucocorticoid metabolites was increased in adult women with PCOS (16,17,23,39), consistent with ACTH-mediated cortisol production. Further, decreased activity of 11␤-HSD1, which reduces inactive cortisone to cortisol, has been reported in some (23,39) but not all (17) studies of adult affected women. This change may also result lower cortisol levels and in compensatory increases in ACTH release (23,39). In the present study, neither glucocorticoid metabolite excretion nor 11␤-HSD1 activity differed in PCOS-d compared to control girls. These findings suggest that increases in zona fasciculata activity are not important in the early origins of PCOS.
Our study has some potential limitations. First, urinary steroid metabolite ratios can estimate global enzyme activity, but do not reflect intracellular steroidogenic enzyme activity in target tissues (28 -30). Nevertheless, this noninvasive approach is a valuable tool for investigation of alterations in steroid metabolism in infants and young children. Second, the difference we observed in 5␣THF/ THF achieved only modest statistical significance (P ϭ .04, Figure 2). However, this difference was similar in magnitude to differences in 5␣-reductase activity observed in lean adult women with PCOS compared to lean control women (17,23). Further, because genetic factors likely contributed to the alterations in steroidogenesis, these effects would be diluted because the PCOS-d group contained both genetically affected and unaffected girls.
In conclusion, our findings suggest that increased 5␣reductase activity is present in girls at risk for PCOS in early childhood. Hence, enzyme expression in target tissues may be primed toward androgen activation in genet-doi: 10.1210/jc.2015-3926 press.endocrine.org/journal/jcem 2073 ically affected PCOS-d before the onset of increased glandular DHEA secretion during adrenarche (32,40). These findings support our hypothesis that genetic variation leading to hyperandrogenemia, specifically increased target tissue production of DHT, contributes to the development of PCOS through epigenetic actions during critical developmental windows. Prospective cohort studies are needed to determine whether those PCOS-d with early childhood activation of 5␣-reductase activity ultimately develop PCOS.