Precocious pubarche (PP) is defined as pubic hair onset before

Transcription

1 ORIGINAL ARTICLE Endocrine Care Precocious Pubarche: Distinguishing Late-Onset Congenital Adrenal Hyperplasia from Jean-Baptiste Armengaud, Marie-Laure Charkaluk, Christine Trivin, Véronique Tardy, Gérard Bréart, Raja Brauner, and Martin Chalumeau Université Paris Descartes (J.-B.A., R.B., M.C.), Paris, France; Assistance Publique-Hôpitaux de Paris (AP-HP) (J.-B.A., M.-L.C., R.B.), Hôpital Bicêtre, Unité d Endocrinologie Pédiatrique, Le Kremlin-Bicêtre, France; Institut National de la Santé et de la Recherche Médicale U953 (J.-B.A., G.B., M.C.), Paris, France; AP-HP (C.T.), Hôpital Necker Enfants Malades, Service d Explorations Fonctionnelles, Paris, France; Groupement Hospitalier Est (V.T.), Centre de Biologie et Pathologie Est, Bron, France; and AP-HP (M.C.), Hôpital Saint Vincent de Paul, Service de Pédiatrie Générale, Paris, France Context: Because precocious pubarche (PP) reveals late-onset congenital adrenal hyperplasia (LO-CAH) in 5 to 20% of cases, an adrenal stimulation test is recommended in all patients presenting with it. This test is stressful and expensive, and results are normal in more than 80% of cases. Objective: Our objective was to identify clinical and plasma predictors of LO-CAH among patients presenting with PP. Design, Setting, and Patients: We conducted a retrospective cohort study that included all patients seen for PP at our hospital between 1999 and 2006 (n 238). All had undergone an ACTH test. Main Outcome Measure: LO-CAH was defined by a post-acth 17-hydroxyprogesterone (17-OHP) plasma level greater than 10 ng/ml and confirmed by mutational analysis of the CYP21 gene. The association of standard clinical and laboratory indicators with LO-CAH was assessed. Results: Ten (4%) of 238 patients had LO-CAH. Basal 17-OHP, 4-androstenedione, and testosterone plasma levels were significantly higher in these patients. A 2-ng/ml threshold for basal 17-OHP plasma levels offered 100% (95% CI, ) sensitivity for the diagnosis of LO-CAH and 99% (95% CI, ) specificity. Conclusion: We identified three plasma predictors of LO-CAH in patients presenting with PP. A selective strategy based on a 2-ng/ml basal 17-OHP plasma level threshold would have safely avoided 99% of the unnecessary ACTH tests among our patients. (J Clin Endocrinol Metab 94: , 2009) Precocious pubarche (PP) is defined as pubic hair onset before age 8 in girls and age 9 in boys (1 5). In 80 to 95% of cases, PP is related to premature adrenarche (PA), the nonpathological precocious secretion of adrenal androgens (6, 7). In 5 to 20% of cases, however, the cause of PP is late-onset congenital adrenal hyperplasia (LO-CAH) (3 5, 8 16), which is due mainly to nonclassic 21-hydroxylase deficiency, a mild adrenal enzyme defect related to a CYP21 gene mutation that causes excessive androgen secretion (3, 17 19). LO-CAH is a rare but potentially severe ISSN Print X ISSN Online Printed in U.S.A. Copyright 2009 by The Endocrine Society doi: /jc Received February 11, Accepted May 8, First Published Online May 19, 2009 condition. If not promptly diagnosed, it can lead to accelerated bone maturation (4), short final height (20), and in adulthood to severe cystic acne (20, 21), hirsutism (22), and hypofertility (23, 24). Early identification of LO-CAH makes it possible to administer substitution therapy that controls symptoms related to androgen excess (4, 11, 12, 16, 25 27). LO-CAH is currently diagnosed by abnormal hormonal response after an adrenal stimulation test with synthetic ACTH (2, 3, 17), specifically an elevated plasma level of ACTH-stimulated 17-hydroxyprogest- Abbreviations: 4A, 4-Androstenedione; AUC, area under the curve; BMI, body mass index; CI, confidence interval; IQR, interquartile range; LO-CAH, late-onset congenital adrenal hyperplasia; LR, likelihood ratio; NPV, negative predictive value; 17-OHP, 17-hydroxyprogesterone; OR, odds ratio; PA, premature adrenarche; PP, precocious pubarche; PPV, positive predictive value; ROC, receiver-operating characteristic. J Clin Endocrinol Metab, August 2009, 94(8): jcem.endojournals.org 2835

2 2836 Armengaud et al. Predictors of Late-Onset Congenital Adrenal Hyperplasia J Clin Endocrinol Metab, August 2009, 94(8): erone (17-OHP) (28). This test requires day-hospitalization in a specialized unit, as well as repeated blood sampling, which is distressing and expensive. Previous studies and guidelines present widely varying indications for ACTH tests in patients with PP. Given that no clinical feature predicts LO-CAH well and that PP may be its earliest manifestation, many consider that an ACTH test should be performed in all children with PP to verify the absence of LO-CAH (4, 5, 9, 10, 12, 13, 15, 16, 26, 27). The application of this systematic strategy assures 100% sensitivity for diagnosing LO- CAH but produces normal ACTH tests in more than 80% of patients. A posteriori these tests are thus unnecessary. Other authors suggest a selective approach toward ACTH tests for patients with high basal 17-OHP (5, 8, 12, 13, 16, 26, 28, 29) or 4-androstenedione ( 4A) plasma levels (16, 28) or with a dehydroepiandrosterone/ 4A ratio greater than 1 (13). However, the predictive value of these thresholds has never been tested in children with PP (5, 12 14, 16, 28, 29). The identification of a highly sensitive and quite specific selective approach to prescription of ACTH tests in patients with PP would make it possible to avoid a posteriori unnecessary normal tests without missing any patient with LO-CAH. The aim of our study was to identify clinical and plasma predictors of LO-CAH and to test their predictive value in children with PP. Patients and Methods We conducted a retrospective hospital-based cohort study. The study included all patients seen for PP in the pediatric endocrinology unit of a teaching hospital in Paris (France) between 1999 and 2006, unless they met the exclusion criteria described below. They were identified by checking computerized hospital charts for the term precocious pubarche and reviewing a local clinical register. During the study period, the local protocol required the systematic prescription of an ACTH test for all patients seen for PP. PP was defined as pubic hair onset before age 8 in girls and age 9 in boys (1 5). Patients were excluded if they had concomitant clinical signs of central puberty (breast development before age 8 in girls or testicular development before age 9 in boys) (5, 8, 15, 16) or signs of adrenal tumors (rapidly progressive systemic virilization) or hypercorticism (Cushing s syndrome) (5, 10). The patients lost to follow-up before an ACTH test was performed were secondarily excluded. The Institutional Review Committee (Comité de Protection des Personnes Ile de France III) stated that this research was found to conform to generally accepted scientific principles and research ethical standards and to be in conformity with the laws and regulations of France, where the research experiment was performed. Written informed consent of the patients or their parents was not judged necessary for this kind of retrospective study. Patients were considered to have LO-CAH if their stimulated 17- OHP plasma level was equal to or greater than 10 ng/ml; otherwise, the patients were considered to have PA (4, 5, 8 10, 12 14, 16, 26 28). The diagnosis of all patients determined to have LO-CAH was confirmed by mutational analysis of the CYP21 gene (18). We considered as potential predictors all clinical and basal hormonal variables available at the time of the first medical examination in our center with a missing data rate lower than 15%. The potential clinical predictors were sex, age at pubic hair onset as reported by the patient or her/his parents (years), weight (SD), height (SD), body mass index (BMI; percentiles), and Tanner stage for pubic hair as recorded at the first physical examination (30 32). The potential plasma predictors were basal plasma levels of 17-OHP, 4A, testosterone, and 11-deoxycortisol, reported in nanograms per milliliter. The ACTH test was performed as internationally recommended: a basal blood sample at 0800 h immediately followed by im infusion of 0.25 mg of tetracosactide acetate (Synacthen) and another blood sample 60 min after the infusion. The basal blood sample included the baseline measurement of serum 17- OHP. A single laboratory measured 17-OHP, 4A, and testosterone by radioimmunological assays (OHP-CT: Cis Bio, Gif-sur-Yvette, France; 4A RIA: Immunotech, Marseille, France; and Testo-CT2: Cis Bio, Gifsur-Yvette, France). Plasma levels of 11-deoxycortisol were measured with the combined RIA developed at Saint-Louis Hospital (Paris). The same kit was used for each potential predictor throughout the study period. We first described the general characteristics of the study population and compared the distributions of the continuous potential predictors between patients with LO-CAH and PA with the nonparametric Mann- Whitney test. The discriminating power of the potential predictors was evaluated by comparing the areas under receiver-operating characteristic (ROC) curves according to the standard method described by Hanley and McNeil (33). Continuous variables were dichotomized according to either a cutoff value reported in the literature (14, 29) or their distribution among the patients without LO-CAH. Weight, height, and BMI as TABLE 1. Distribution of potential predictors LO-CAH (n 10) adrenarche (n 228) Variables n Mean Median IQR Range n Mean Median IQR Range P e Clinical Age at P2 a Weight b ( 2.0) Height b ( 4.0) BMI c Tanner d Plasma (ng/ml) 17-OHP A Testosterone deoxycortisol a Tanner stage II for public hair; age as reported by the patient or his/her parents. b Standardized value for gender and age. c Percentiles. d Pubic hair Tanner stage at first examination. e Nonparametric Mann-Whitney test.

3 J Clin Endocrinol Metab, August 2009, 94(8): jcem.endojournals.org 2837 A 17-hydroxyprogesterone (ng/ml) ,1 2 ng/ml different ages in boys and girls. The predictive value of each potential predictor was evaluated by calculating its sensitivity, specificity, positive and negative predictive values (PPV, NPV), and the likelihood ratios (LR) for a positive and a negative test (LR, LR ). The 95% confidence intervals(cis) werecalculatedwithabinomialapproximation(34). Ifatable containedanemptycell, acorrectedorwascalculatedbyaddinga0.5value in each cell of the table (35). Because of the very unbalanced distribution of the main plasma predictor, we did not adjust for other variables (for example, by using a logical regression model). All statistical analyses were performed with STATA/SE 9 software (Statacorp, College Station, TX). Results B Delta-4-androtenedione (ng/ml) C Testosterone (ng/ml) 0, ,00 0,90 0,80 0,70 0,60 0,50 0,40 0,30 0,20 0,10 0, ng/ml 0.15 ng/ml Late-onset CAH Late-onset CAH Late-onset CAH FIG. 1. Basal hormonal plasma levels at the time of diagnosis of PP. A, 17-OHP; B, 4A; C, testosterone. recorded at the first medical examination were dichotomized around a threshold set at 2 SD (height and weight) or the 97th percentile (BMI) above the mean (or median) value for age and sex; Tanner stage for pubic hair was dichotomized around stage II to offer the best clinical reproducibility. We studied the relations between LO-CAH and the dichotomized potential predictors with a univariate analysis and calculated the odds ratios (OR). Distributions were compared with Fisher s exact test. We looked for an interaction of sex in the relation between the age at pubic hair onset and the diagnosis of LO-CAH because PP is defined at During the study period, 372 patients were seen for PP, 99 (27%) of them met the same exclusion criterion: signs of central puberty concomitant to pubic hair onset. None met either of the other two exclusion criteria, but 35 (13%) of the remaining 273 patients were lost to follow-up before an ACTH test. The analysis was thus based on 238 patients, 75% of them girls. Overall, median age at pubic hair onset was 6.6 yr (mean, 5.9; range, ). Among girls, median age at pubic hair onset was 6.5 yr [interquartile range (IQR), ], and among boys, 7.6 yr (IQR, ). Finally, 10 (4%; 95% CI, 2 7) patients had LO-CAH (nine nonclassic 21-hydroxylase deficiency and one nonclassic 11 hydroxylase deficit), all confirmed by gene mutation analysis. The remaining 228 patients (96%; 95% CI, 93 98) had PA. No statistically significant difference was observed between the distributions of the following clinical variables in patients with LO-CAH and PA: age at pubic hair onset, weight, height, and pubic hair Tanner stage at the time of the first medical examination in our center (Table 1). A trend toward a statistically significant difference was observed between the distributions of BMI among patients with LO-CAH and with PA (median, 1.03 vs. 0.80; P 0.09). No statistically significant difference was observed in the distribution of 11-deoxycortisol plasma levels between the two groups. Basal plasma levels of 17-OHP, 4A, and testosterone were all significantly higher (P 0.001) in patients with LO-CAH than in those with PA (Table 1 and Fig. 1). The areas under the ROC curves (AUC) for 17OHP, 4A, and testosterone were 0.99 (95% CI, ), 0.90 (95% CI, ), and 0.83 (95% CI, ), respectively. The AUC for 17-OHP was statistically significantly larger than the AUC of all other predictors (P 0.05; Fig. 2). After dichotomization, no statistically significant association was observed between the diagnosis of LO-CAH and the following clinical variables: male sex, age at pubic hair onset no greater than 6 yr, weight more than 2 SD, height more than 2 SD, and Tanner stage for pubic hair above stage II at first examination (Table 2). A trend toward a statistically significant association was observed between a BMI above the 97th percentile and the diagnosis of LO-CAH (OR, 3.7; 95% CI, ; P 0.07). We found no interaction of sex in the relation between the age at pubic hair onset and LO-CAH (P 0.53). A basal 17-OHP plasma level greater than 2 ng/ml offered 100% (95% CI, ) sensitivity for predicting LO-CAH, and 99% (95% CI, ) specificity. None of the other plasma predictors considered alone had a better predictive power than basal 17-OHP plasma level. The discriminating

4 2838 Armengaud et al. Predictors of Late-Onset Congenital Adrenal Hyperplasia J Clin Endocrinol Metab, August 2009, 94(8): Sensitivity power of the three plasma predictors was not improved when the distributions were dichotomized around other thresholds (Table 3). Discussion Specificity 17-OHP : 0.99 [95% CI ] 4A : 0.90 [95% CI ]* We describe the largest cohort of children seen for PP thus far reported because all previously published studies had smaller samples (IQR, 15 55; range, 9 171) (5, 8 12, 14 16, 26, 27). In contrast to some reports that patients with LO-CAH, compared with those with PA, are taller, heavier, or have accelerated bone maturation (5, 8, 16), our study did not identify any clinical predictor of LO-CAH; only BMI tended to be distributed differentially between patients with LO-CAH and with PA. We identified three plasma predictors of LO-CAH in children with PP: high basal 17-OHP, 4A, and testosterone plasma levels. The basal 17-OHP plasma level is often reported to be elevated in patients with LO-CAH and has previously been suggested as a good predictor (8, 12 14, 16, 26, 27). Levine et al. (13) suggested that a basal 17-OHP plasma level above 5 ng/ml combined with a dehydroepiandrosterone/ 4A ratio above 1 was a good predictor of LO-CAH. Ibáñez et al. proposed that a basal 17- OHP plasma level above 2 SD was indicative of LO-CAH in patients with PP (26). None of these studies, however, analyzed the discriminating power of these predictors statistically. Clinicians who must decide which patients should undergo an ACTH test require decision-making aid, and this can be obtained only by precise and statistically tested thresholds. We set a basal 17- OHP plasma level threshold at 2 ng/ml to obtain the best sensitivity and specificity. The relevance of this threshold can also be appreciated by comparison to the reference nomogram published by New et al. (28). In the population of this historical study, our threshold would have had 100% sensitivity and specificity. It is also the same as that proposed and prospectively validated by Azziz et al. (29) in a population of unselected hyperandrogenic women. In their study population, where LO-CAH prevalence was 8%, they showed that a basal 17-OHP plasma level below 2 ng/ml effectively ruled out LO-CAH with 100% sensitivity and 99% specificity; the PPV was 80%, and the NPV was close to 100%. In our cohort, where the prevalence of LO-CAH was 4%, the NPV of a basal 17-OHP plasma level below 2 ng/ml was 100% (95% CI, %). A clinician could have safely ruled out LO-CAH in any PP patient in our cohort with a basal 17- OHP plasma level below 2 ng/ml. Because hormonal measurements are highly assay-specific, we may consider the percentile of our threshold value (94th percentile of the distribution of basal 17-OHP plasma levels in the group of patients with PA) to allow standardized comparison with the results obtained with other assays, asproposedforotherpredictivetoolsin pediatric endocrinology (36). Basal 4A and testosterone plasma levels have previously been proposed as potential predictors of LO-CAH, but their predictive values have never been statistically determined (5, 13, 14, 16). In our cohort, a basal 4A plasma level above 0.95 ng/ml or a basal testosterone plasma level above 0.15 ng/ml was strongly and significantly associated with a LO-CAH diagnosis (Table 3). However, their predictive power was lower than that of 17-OHP. We tried to include all consecutive patients seen for PP at our center during the study period. Given the retrospective design of our study, a classification bias was possible. It is likely, however, that any such bias is weak for all patients were examined by the same physician and stringent exclusion criteria were applied (i.e. exclusion of patients with signs of central puberty). Many of the boys also underwent an LHRH test after the ACTH test to rule out central precocious puberty before an increase in testicular volume. The collection of all blood samples at regular times for basal and stimulated levels and their processing according to standardized protocols in the laboratory limited another source of classification bias (37). Relatively few patients were lost to follow-up (n 35; 13%), and the distributions of gender and age at pubic hair onset did not differ between this group and the patients included in our cohort (P 0.30). Selection bias is thus unlikely. The retrospective design of the study resulted in missing data rates greater than 15% for potentially relevant data (growth velocity, advance in bone age, basal ACTH and dehydroepiandrosterone sulfate plasma levels) and their exclusion from the analyses. Nonetheless, none of these indicators has been suggested as a predictor of LO-CAH in the literature. Many clinicians believe bone age advance is greater in patients with LO-CAH than in those with PA, a belief supported by the analysis of our available data (63% of our patients, including 10 patients with LO-CAH): median standardized bone age, 2.0 vs. 0.6 yr (P 0.001). Of the plasma predictors we did assess, the highest missing data rate (3%) concerned the basal testosterone plasma level. Nonetheless, when we used the maximum bias hypothesis, the association between a basal Testosterone: 0.83 [95% CI ]* BMI : 0.66 [95% CI ]* *: p 0.05 after comparison to AUC of 17-OHP FIG. 2. Evaluation of the discriminating power of the best continuous predictors (areas under ROC curves with 95% CI).

5 J Clin Endocrinol Metab, August 2009, 94(8): jcem.endojournals.org 2839 TABLE 2. Univariate analysis LO-CAH (n 10) adrenarche (n 228) Variables n % n % OR 95% CI P c Clinical Gender Boys Girls Age at P2 a 6 yr yr Weight 2 SD SD Height 2 SD SD BMI 97th %ile th %ile Tanner b stage II stage II Plasma (ng/ml) 17-OHP d A Testosterone a Tanner stage II for public hair; age as reported by the patient or his/her parents. b Pubic hair Tanner stage at first examination. c Fisher s exact test. d A 0.5 value was added to all cells when an empty cell was observed (35). testosterone plasma level above 0.15 ng/ml and a diagnosis of LO- CAH remained strong and statistically significant (OR 6.1; 95% CI, ; P 0.01). Although classification bias for LO-CAH in our study is possible, it is improbable because mutation analysis was used to confirm all LO-CAH diagnoses (18). Moreover, we reduced the risk of interindividual variance in interpretation by using the same kit for each plasma predictor throughout the study period. Some other extremely rare causes of LO-CAH must be discussed. LO-CAH due to 11 -hydroxylase deficiency has been reported to induce PP (38). This diagnosis was genetically confirmed in one patient included in our cohort. Although 11 hydroxylase and 21-hydroxylase deficiencies are different diseases, she would nonetheless have been correctly identified with our selective approach because her basal 17-OHP plasma level was 3.7 ng/ml. Excluding her from the analysis did not significantly modify either the sensitivity (100%; 95% CI, ) or the NPV (100%; 95% CI, ) of our main predictor. Some authors suggest that PP may reveal childhood adrenal tumors (39), which usually induce marked clinical signs of rapidly progressive systemic virilization. Because none of our patients presented these signs at their first medical examination, ultrasound imaging of the adrenals was not performed. Moreover, the median period of clinical follow-up 3.7 yr in our cohort (range, ) allowed us to rule out the presence of adrenal tumors among our patients presenting with PP. A selection bias is possible because our patients were recruited through a highly specialized pediatric endocrinology outpatient department. It is likely, however, to be relatively weak because the prevalence of LO-CAH in our cohort was 4%, lower than any of the previously reported rates, which range from 5 to 43% (median, 8%) (5, 8, 9, 12, 14 16, 26, 27). Moreover, our observed sex ratio (0.33) was consistent with previous reports in smaller cohorts (median, 0.2; range, ) (5, 8, 9, 12, 14 16, 26, 27). We identified three plasma predictors among patients with PP that are strongly and significantly associated with LO- CAH. These results can be integrated into a highly sensitive and specific structured strategy (40) usable in daily practice to limit unnecessary ACTH testing, which remains a stressful and costly procedure. In our cohort, according to our strategy, 99% of ACTH tests could have been safely avoided in patients presenting with PP. Such an evidence-based strategy could represent an alternative to the systematic approach that is currently proposed (9, 15, 16, 26) or to a selective approach based on unvalidated predictors (5, 8, 12, 27). Because these TABLE 3. Discriminating power of plasma predictors significantly associated with LO-CAH at selected thresholds Variables Sensitivity Specificity PPV NPV LR ( ) LR ( ) 17-OHP 2.0 ng/ml a a 2.5 ng/ml a a 3.0 ng/ml a a 4A 0.90 ng/ml ng/ml ng/ml Testosterone 0.10 ng/ml ng/ml ng/ml All parameters are reported with their 95% CI. a A 0.5 value was added to all cells when an empty cell was observed (35).

6 2840 Armengaud et al. Predictors of Late-Onset Congenital Adrenal Hyperplasia J Clin Endocrinol Metab, August 2009, 94(8): findings come from a retrospective cohort from a single hospital, they must be prospectively validated to confirm their relevance before any clinical application (36). Acknowledgments The authors would like to thank Pr. Yves MOREL (Groupement Hospitalier Est, Bron, 69677, France) for his collaboration in confirming the diagnosis of LO-CAH by mutational analysis of the CYP2l gene. Address all correspondence and requests for reprints to: Dr. M. Chalumeau, Department of Pediatrics, Hôpital Saint Vincent de Paul, 82, avenue Denfert-Rochereau, Paris Cedex 14, France. martin. chalumeau@svp.aphp.fr. Disclosure Summary: The authors have nothing to disclose. References 1. Grumbach MM, Styne DM 1998 Puberty: ontogeny, neuroendocrinology, physiology and disorders. In: Wilson DJ, Foster DW, Kronenberg HM, Larsen PR, eds. Williams textbook of endocrinology. 9th ed. Philadelphia: WB Saunders; Ibáñez L, Dimartino-Nardi J, Potau N, Saenger P 2000 adrenarche normal variant or forerunner of adult disease? Endocr Rev 21: New MI 2004 An update of congenital adrenal hyperplasia. Ann NY Acad Sci 1038: New MI 2006 Extensive clinical experience: nonclassical 21-hydroxylase deficiency. J Clin Endocrinol Metab 91: Siegel SF, Finegold DN, Urban MD, McVie R, Lee PA 1992 pubarche: etiological heterogeneity. J Clin Endocrinol Metab 74: Ibáñez L, Potau N, Dunger D, de Zegher F 2000 Precocious pubarche in girls and the development of androgen excess. J Pediatr Endocrinol Metab 13(Suppl 5): Sklar CA, Kaplan SL, Grumbach MM 1980 Evidence for dissociation between adrenarche and gonadarche: studies in patients with idiopathic precocious puberty, gonadal dysgenesis, isolated gonadotropin deficiency, and constitutionally delayed growth and adolescence. J Clin Endocrinol Metab 51: Balducci R, Boscherini B, Mangiantini A, Morellini M, Toscano V 1994 Isolated precocious pubarche: an approach. J Clin Endocrinol Metab 79: del Balzo P, Borrelli P, Cambiaso P, Danielli E, Cappa M 1992 Adrenal steroidogenic defects in children with precocious pubarche. Horm Res 37: Hawkins LA, Chasalow FI, Blethen SL 1992 The role of adrenocorticotropin testing in evaluating girls with premature adrenarche and hirsutism/oligomenorrhea. J Clin Endocrinol Metab 74: Kohn B, Levine LS, Pollack MS, Pang S, Lorenzen F, Levy D, Lerner AJ, Rondanini GF, Dupont B, New MI 1982 Late-onset steroid 21-hydroxylase deficiency: a variant of classical congenital adrenal hyperplasia. J Clin Endocrinol Metab 55: Leite MV, Mendonça BB, Arnhold IJ, Estefan V, Nunes C, Nicolau W, Bloise W 1991 Identification of nonclassical 21-hydroxylase deficiency in girls with precocious pubarche. J Endocrinol Invest 14: Levine LS, Dupont B, Lorenzen F, Pang S, Pollack M, Oberfield S, Kohn B, Lerner A, Cacciari E, Mantero F, Cassio A, Scaroni C, Chiumello G, Rondanini GF, Gargantini L, Giovannelli G, Virdis R, Bartolotta E, Migliori C, Pintor C, Tato L, Barboni F, New MI 1980 Cryptic 21-hydroxylase deficiency in families of patients with classical congenital adrenal hyperplasia. J Clin Endocrinol Metab 51: Moayeri H, Rabanni A 2003 Nonclassic 21 hydroxylase-deficient adrenal hyperplasia in patients with isolated precocious pubarche. Acta Medica Iranica 41: Oberfield SE, Mayes DM, Levine LS 1990 Adrenal steroidogenic function in a black and Hispanic population with precocious pubarche. J Clin Endocrinol Metab 70: Temeck JW, Pang SY, Nelson C, New MI 1987 Genetic defects of steroidogenesis in premature pubarche. J Clin Endocrinol Metab 64: Merke DP, Bornstein SR 2005 Congenital adrenal hyperplasia. Lancet 365: Pinto G, Tardy V, Trivin C, Thalassinos C, Lortat-Jacob S, Nihoul-Fékété C, Morel Y, Brauner R 2003 Follow-up of 68 children with congenital adrenal hyperplasia due to 21-hydroxylase deficiency: relevance of genotype for management. J Clin Endocrinol Metab 88: Speiser PW, Dupont B, Rubinstein P, Piazza A, Kastelan A, New MI 1985 High frequency of nonclassical steroid 21-hydroxylase deficiency. Am J Hum Genet 37: Jääskeläinen J, Voutilainen R 1997 Growth of patients with 21-hydroxylase deficiency: an analysis of the factors influencing adult height. Pediatr Res 41: David M, Sempé M, Blanc M, Nicolino M, Forest MG, Morel Y 1994 [Final height in 69 patients with congenital adrenal hyperplasia due to 21-hydroxylase deficiency]. Arch Pediatr 1: Kuttenn F, Couillin P, Girard F, Billaud L, Vincens M, Boucekkine C, Thalabard JC, Maudelonde T, Spritzer P, Mowszowicz I 1985 Late-onset adrenal hyperplasia in hirsutism. N Engl J Med 313: Deneux C, Tardy V, Dib A, Mornet E, Billaud L, Charron D, Morel Y, Kuttenn F 2001 Phenotype-genotype correlation in 56 women with nonclassical congenital adrenal hyperplasia due to 21-hydroxylase deficiency. J Clin Endocrinol Metab 86: New MI 1996 Treatment-induced hypoandrogenism in childhood and puberty in females with virilizing (21-hydroxylase deficiency) congenital adrenal hyperplasia. J Endocrinol 150 Suppl:S31 S DiMartino-Nardi J, Stoner E, O Connell A, New MI 1986 The effect of treatment of final height in classical congenital adrenal hyperplasia (CAH). Acta Endocrinol Suppl (Copenh) 279: Ibáñez L, Bonnin MR, Zampolli M, Prat N, Alia PJ, Navarro MA 1995 Usefulness of an ACTH test in the diagnosis of nonclassical 21-hydroxylase deficiency among children presenting with premature pubarche. Horm Res 44: Török D, Halász Z, Garami M, Homoki J, Fekete G, Sólyom J 2003 Limited value of serum steroid measurements in identification of mild form of 21- hydroxylase deficiency. Exp Clin Endocrinol Diabetes 111: New MI, Lorenzen F, Lerner AJ, Kohn B, Oberfield SE, Pollack MS, Dupont B, Stoner E, Levy DJ, Pang S, Levine LS 1983 Genotyping steroid 21-hydroxylase deficiency: hormonal reference data. J Clin Endocrinol Metab 57: Azziz R, Hincapie LA, Knochenhauer ES, Dewailly D, Fox L, Boots LR 1999 Screening for 21-hydroxylase-deficient nonclassic adrenal hyperplasia among hyperandrogenic women: a prospective study. Fertil Steril 72: Marshall WA, Tanner JM 1969 Variations in pattern of pubertal changes in girls. Arch Dis Child 44: Marshall WA, Tanner JM 1970 Variations in the pattern of pubertal changes in boys. Arch Dis Child 45: Sempe M, Pedron G, Roy-Pernot M 1979 Auxologie. Méthodes et séquences. Paris: Theraplix 33. Hanley JA, McNeil BJ 1983 A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 148: Rothman KJ, Greenland S 1998 Introduction to regression modeling. In: Wilkins LW, ed. Modern epidemiology. 2nd ed. Philadelphia: Lippincott- Raven Publishers; Deeks JJ, Higgins JPT, Altman DG 2004 Analysing and presenting results. In: Alderson P, Green S, Higgins J, eds. Cochrane reviewers handbook , Sect. 8. Oxford, UK: The Cochrane Collection 36. Chalumeau M, Hadjiathanasiou CG, Ng SM, Cassio A, Mul D, Cisternino M, Partsch CJ, Theodoridis C, Didi M, Cacciari E, Oostdijk W, Borghesi A, Sippell WG, Bréart G, Brauner R 2003 Selecting girls with precocious puberty for brain imaging: validation of European evidence-based diagnosis rule. J Pediatr 143: Tworoger SS, Hankinson SE 2006 Use of biomarkers in epidemiologic studies: minimizing the influence of measurement error in the study design and analysis. Cancer Causes Control 17: Zachmann M, Tassinari D, Prader A 1983 Clinical and biochemical variability of congenital adrenal hyperplasia due to 11 -hydroxylase deficiency. A study of 25 patients. J Clin Endocrinol Metab 56: Grossrubatscher E, Vignati F, Possa M, Lohi P 2001 The natural history of incidentally discovered adrenocortical adenomas: a retrospective evaluation. J Endocrinol Invest 24: Chalumeau M, Chemaitilly W, Trivin C, Adan L, Bréart G, Brauner R 2002 Central precocious puberty in girls: an evidence-based diagnosis tree to predict central nervous system abnormalities. Pediatrics 109:61 67