BEHAVIORAL STUDIES in children with congenital adrenal

Transcription

1 /02/$15.00/0 The Journal of Clinical Endocrinology & Metabolism 87(11): Printed in U.S.A. Copyright 2002 by The Endocrine Society doi: /jc Sex-Typed Toy Play Behavior Correlates with the Degree of Prenatal Androgen Exposure Assessed by CYP21 Genotype in Girls with Congenital Adrenal Hyperplasia ANNA NORDENSTRÖM, ANNA SERVIN, GUNILLA BOHLIN, AGNE LARSSON, AND ANNA WEDELL Department of Pediatrics, Karolinska Institute, Huddinge University Hospital (A.N., A.L.), S Stockholm, Sweden; Department of Psychology, University of Uppsala (A.S., G.B.), S Uppsala, Sweden; and Department of Molecular Medicine, Karolinska Institute, Karolinska Hospital (A.W.), S Stockholm, Sweden Previous studies have shown that girls with congenital adrenal hyperplasia (CAH), a syndrome resulting in overproduction of adrenal androgens from early fetal life, are behaviorally masculinized. We studied play with toys in a structured play situation and correlated the results with disease severity, assessed by CYP21 genotyping, and age at diagnosis. Girls with CAH played more with masculine toys than controls when playing alone. In addition, we could demonstrate a doseresponse relationship between disease severity (i.e. degree of BEHAVIORAL STUDIES in children with congenital adrenal hyperplasia (CAH) are important for several reasons. They provide information that is important for the management and follow-up of patients. In addition, the influence of prenatal and neonatal hormonal factors on sex differences in behavior can be studied in girls with CAH, as these children have been exposed to elevated levels of androgens from early fetal development. In the 1960s, Money and Ehrhardt (1) reported that girls with CAH often preferred boys toys and outdoor sports. Since then, several studies have shown that girls with CAH are behaviorally masculinized (2 4). Girls with CAH reached higher scores than their sisters in some cognitive parameters, such as three-dimensional rotational spatial ability, a pattern similar to that seen in comparisons of normal boys and girls (5, 6). Women with CAH showed a typical male pattern for measures of personality traits (7). In direct observational studies of toy play Berenbaum et al. (8, 9) showed that girls with CAH played more with boys toys and less with girls toys than their unaffected sisters. The results regarding the correlation of disease severity with degree of masculinization of behavior have been somewhat contradictory (2, 3, 8, 10). Berenbaum et al. (11) and our group (Servin, A., A. Nordenström, A. Larsson, and G. Bohlin, submitted for publication) have shown that childhood boy-typical interest was strongly associated with the degree of virilization of the genitalia, an indicator of prenatal androgen exposure. It has been argued that the behavioral changes in girls with CAH are the results of the parental treatment triggered by the virilization of genitalia at birth (13). On the other hand, the persistence of sex-atypical interests, activities, and careers in adolescent girls with CAH suggests that they result from the direct Abbreviation: CAH, Congenital adrenal hyperplasia. fetal androgen exposure) and degree of masculinization of behavior. The presence of a parent did not influence the CAH girls to play in a more masculine fashion. Four CAH girls with late diagnosis are also described. Three of the four girls played exclusively with one of the masculine toys, a constructional toy. Our results support the view that prenatal androgen exposure has a direct organizational effect on the human brain to determine certain aspects of sex-typed behavior. (J Clin Endocrinol Metab 87: , 2002) effects of androgens on the developing brain rather than social responses, because these girls were brought up as females (14). The vast majority of girls with CAH have a typical female gender identity (15). CAH constitutes a family of defects in the synthesis of steroid hormones in the adrenal cortex. In more than 90% of the cases it is caused by a defect in the 21-hydroxylase gene (CYP21) (16, 17). The enzyme deficiency results in impaired synthesis of cortisol and aldosterone. The low cortisol level results in increased production of ACTH by the pituitary, which causes hyperplasia of the adrenal glands and increased synthesis of steroid precursors, resulting in high androgen levels. The androgen excess is present from early embryogenesis and results in varying degrees of virilization of the external genitalia in girls depending on the degree of enzyme deficiency. In severe forms, the virilization may result in uncertainty in gender assignment at birth, and the sex of some girls is initially designated as male. The molecular genetics of 21-hydroxylase deficiency have been studied extensively. More than 95% of the patients are homozygous or compound heterozygotes for any of nine different point mutations or deletion of the CYP21 gene. With very few exceptions there is a good correlation between the CYP21 genotype and disease severity (18 20). Deletions or mutations that completely abolish enzyme activity are referred to as null mutations. Patients who are homozygous for null mutations have the most severe form of the disease, with salt loss in the neonatal period and severe prenatal virilization of external genitalia in girls. The I2 splice mutation is slightly less severe; some homozygous patients are not affected by salt loss. The I172N mutation is associated with varying degrees of virilization of external genitalia, but only about 10% of patients with this genotype show signs of salt loss. The V281L mutation is even milder and is associated 5119

2 5120 J Clin Endocrinol Metab, November 2002, 87(11): Nordenström et al. Toy Play Correlates with CYP21 Genotype in CAH with nonclassical CAH without virilization of external genitalia at birth. Untreated, these patients develop symptoms of androgen excess later in life, such as accelerated growth rate, hirsutism, or infertility. At birth, most children with CAH are diagnosed either because of clinical signs or in neonatal screening programs. The treatment consists of glucocorticoid and mineralocorticoid substitution that decreases/normalizes ACTH levels and thereby androgen production. Corrective surgery on the external genitalia is performed when needed (17). In this study we wanted to investigate further the possible influence of disease severity, i.e. the degree of fetal androgen exposure, on toy play and toy preference in girls with CAH. For this purpose we took advantage of the possibility of determining the degree of 21-hydroxylase deficiency by CYP21 genotyping. There is a good genotype-phenotype correlation (18 20), and genotyping is also a more objective way to measure disease severity compared with other methods, such as Prader score or classification according to salt loss. Furthermore, we have previously shown that at birth the level of one of the hormones preceding the enzyme block, 17-hydroxyprogesterone, is correlated to the CYP21 genotype (21). This indicates that the androgen level, to which the fetus is exposed, during intrauterine life is related to the genotype. We measured sex-typed play behavior in a structured play situation. To assess parental influence, toy play was studied when the child was playing alone as well as when a parent was present. A possible influence of postnatal androgen exposure on toy preference was studied in four children with late diagnosis who had been untreated until 3 6 yr of age. Subjects and Methods Study population All families in Sweden with girls with CAH between 1 and 10 yr of age were contacted during and asked to participate. Five families did not agree to participate. A total of 40 girls with CAH in 35 families were included in the study. CYP21 mutation analysis was performed in 39 of these children. The 4 girls who had been diagnosed prenatally and treated with dexamethasone in utero until term were excluded from this study. Four girls had been missed by the screening and were diagnosed late, at 3 6 yr of age. Healthy girls matched for age 2 months on a case by case basis were used as controls. Mutation analysis CYP21 mutation analysis was carried out using allele-specific PCR from genomic DNA prepared from venous blood samples (22). This detects the 95% of alleles that carry any of the common pseudogenederived mutations. Additional rare alleles were characterized by direct DNA sequencing (23). The genotypes were divided into four groups with respect to the severity of the mutation of the allele with the mildest mutation: null, I2 splice, I172N, and V281L (see Tables 1 and 2). We were able to obtain a sample for CYP21 mutation analysis from all but one child. Toy play In a structured toy play situation, 10 different toys that had previously been defined as masculine, feminine, or neutral for children in the presently employed ages were used (9, 24 27). Feminine toys included a doll with a blanket and feeding bottle, Barbie and Ken dolls, a teapot with four cups, and a female doll s head with brush, comb, and mirror. TABLE 1. CYP21 genotypes of the 31 girls with CAH diagnosed in the neonatal period Genotype group No. Age at study yr (mean) Toy play, seconds, median (range) Masculine Feminine Neutral Prader score (mean) SW/SV Genotypes included in the group Null a ( ) 25 (0 144) 0 (0 65) /0 Null/null 10 12Splice (39 478) 0 (0 275) 0 (0 346) 3.4 9/0 12Splice/null 6 12Splice/12Splice 2 12Splice/null or 1 12Splice/12Splice 1172N (0 420) 16 (0 322) 0 (0 386) 2.0 4/4 b 1172N/null N/12Splice N/1172N 3 V28IL ( ) 60 (0 103) 81 (0 180) 0.7 c V28IL/null 2 V281L/12Splice 1 Control (0 420) 18 (0 420) 43 (0 420) The mean Prader score and classification with respect to salt-wasting (SW) or simple virilizing (SV) form of CAH as well as the number of seconds that the girls in the different genotype groups played with the different types of toys are also listed. a Null mutations were CYP21 deletion, V281L L307insT (two mutations in the same allele), L307insT Q318X, I172N ClusterE6 V281L L307insT Q318X R356W (six point mutations in the same allele), Q318X, and W405X. b For one patient, who was adopted, the classification was not available. c The patients in the V281L genotype group were nonclassical. No. TABLE 2. CYP21 genotypes of the four girls with CAH diagnosed later in childhood Genotype Age at diagnosis (yr) Age at study (yr) Toy play, seconds Masculine Feminine Neutral Toy kept I172N/I172N Doll I172N/null Doll V281L/I172N Ball V281L/I172N Doll Patients with late diagnosis.

3 Nordenström et al. Toy Play Correlates with CYP21 Genotype in CAH J Clin Endocrinol Metab, November 2002, 87(11): Masculine toys were a bus, a garage with four cars, a constructional toy (Lincoln logs), and two fighting figures. Neutral toys were a sketchbook and a deck of cards. The toys were arranged in a standard order in a semicircle on the floor in the homes of the children, with every other toy being masculine and feminine and the neutral toys in between. The child was asked to sit in the middle of the semicircle and was videotaped for 7 min when playing alone and for 7 min when playing with her parent. The play order, alone vs. with a parent, was alternated. The families with a girl with CAH could choose whether the mother or the father would participate (two fathers participated). The control families were matched for these factors. The tapes were then scored for the number of seconds that the child played with the different types of toys. Play was defined as the child touching the toy. The person who scored the tapes was blind to the status of the child on the tape. At the end of the visit the children were given a toy to keep as a present. They were able to choose between a doll (feminine), a car (masculine), and a ball (neutral). Statistical analysis The girls with CAH, regardless of disease severity, and the controls were compared with respect to toy play using the Mann-Whitney U test. The relationship between toy play and genotype was analyzed by means of the Spearman rank order correlation coefficient. The groups were graded according to enzyme activity, with the null genotype group being the lowest and the controls the highest. The girls with late diagnosis were compared with respect to toy play with other girls of the same age, both controls and girls with CAH and early diagnosis, using the Kruskal-Wallis and Mann-Whitney U tests. Intraindividual comparison of toy play with and without a parent was analyzed using Wilcoxon s signed ranks test. Fisher s exact test was used to compare the choice of toy to keep as a present. The choice of doll vs. one of the other toys and the choice of car vs. one of the other toys was tested for the CAH girls with early diagnosis (as a group) compared with the controls. The choice of doll vs. one of the other toys was tested for the CAH girls with late diagnosis compared with girls with CAH and early diagnosis of the same age as well as with girls with the same mutations and early diagnosis regardless of age. SPSS computer program 10.1 (SPSS, Inc., Chicago, IL) was used for all the statistical analyses. The level was set at The study was approved by the ethical committee of the Karolinska Institute (Stockholm, Sweden). Informed consent was obtained from all participants. Results The 31 girls with early diagnosis and start of treatment in the neonatal period were divided into 4 genotype groups (Table 1). Three of the groups were of equal size, and 1 was smaller, comprising only 3 children. The mean Prader scores for the different genotype groups are also listed in Table 1. In addition, there were 4 girls with late diagnosis and treatment (Table 2). Two of them belonged to the I172N genotype group and had started treatment at 3 yr of age. The other 2 belonged to the V281L genotype group and had been treated since 6 yr of age (Table 2). We compared toy play for CAH and control girls when the children were playing alone. As expected, there was a significant difference in toy play for masculine toys between the girls with CAH as a group and the controls (P 0.017). In addition, we found a significant correlation between the degree of disease severity as measured by CYP21 genotypes and the amount of time the CAH girls spent playing with masculine toys (r 0.39; P 0.002; Fig. 1A). The correlation was also significant when the controls were excluded (r 0.41; P 0.024). For play with neutral toys the correlation was significant, but less striking (r 0.27; P 0.036), and the coefficient when the controls were excluded was r 0.34; P (Fig. 1C). The correlation was not significant for play FIG. 1. Amount of time (seconds) that the CAH girls played with masculine toys (A), feminine toys (B), and neutral toys (C) in relation to the CYP21 genotype groups. The box plot shows the median values and the 10th, 25th, 75th, and 90th percentiles. The extreme values are denoted with an asterisk, and the outliers are denoted with a circle.

4 5122 J Clin Endocrinol Metab, November 2002, 87(11): Nordenström et al. Toy Play Correlates with CYP21 Genotype in CAH with the feminine toys (Fig. 1B; r 0.20; P with the controls included and r 0.12; P 0.53 with the controls excluded from the calculation). In conclusion, the milder the disease, i.e. the higher the enzyme activity and therefore the lower the androgen level, the less time was spent with the masculine toys, whereas neutral toys were increasingly preferred. As shown in Fig. 1, the girls with CAH in all of the genotype groups played more with the masculine than with the feminine or neutral toys. The control girls also played more with the masculine toys than with the feminine ones (Fig. 1 and Table 1). Figure 2 shows the difference in the amount of time that the girls with early diagnosis in the different genotype groups spent playing with the different types of toys alone compared with when a parent was present. Positive values thus represent time spent with the toys when the child was alone, whereas negative values indicate toys that were chosen more often when the parent was present. For the control girls the influence of the parents was negligible, the median difference with or without a parent was 10 sec for masculine toys. As a group, the CAH girls played somewhat less with the masculine toys when the parent was present (the median difference was 39 sec), but this difference was not significant (P 0.12). When considering the different genotype groups separately, the girls seemed to be increasingly diverted from masculine toys by the parents with decreasing disease severity. The results of the study of choice of toy to keep as a present are shown in Fig. 3. The girls in the null group chose a car more often than a ball, whereas the girls with I2 splice chose a car or a ball equally often. Some of the girls with I172N chose a doll. None of the three girls with V281L and two of the control girls chose a car. The observed difference between girls with CAH and controls in choice of car and choice of doll was significant (P 0.001). There were four girls in this study with late diagnosis (Table 2). The two girls diagnosed at 3 yr of age with the I172N mutation played more with the masculine toys than the girls with the same mutation but early diagnosis and treatment, as shown in Fig. 1A and Tables 1 and 2. They both played exclusively with the constructional toy. In the mildest group, mutation V281L, the two girls with late diagnosis (at age 6 yr) showed different toy play patterns. One girl, studied at 8 yr of age, played in a way similar to that of the girls with the same mutation treated early on. The other girl, studied at age 10 yr, played with the constructional toy the whole time when she played alone. To substantiate these observations, the four girls with a late diagnosis were compared with girls of the same age, 8 10 yr, both controls (n 12) and girls with CAH diagnosed in the neonatal period, including the severe forms (n 8). Play with the constructional toy was analyzed separately. The girls with late diagnosis played significantly more with the constructional toy (median, 420 sec; range, sec) than control girls of the same age (median, 13 sec; range, sec; P 0.01). There was a statistical trend that the girls with late diagnosis played more with the constructional toy than the other girls with CAH of the same age (median, 420 sec; range, sec vs. median, 218 sec; range, sec; P 0.1). In addition, we compared the choice of toy to keep as a present between the different groups of girls of this age. Three of the girls with late diagnosis chose a doll, and the fourth chose a ball. The controls were more likely to choose a ball (car, n 1; ball, n 10; doll, n 1). The CAH girls with early diagnosis chose a car and a ball equally often (car, n 4; ball, n 4). The difference in choice of doll between the patients with late and early diagnosis was significant (P 0.018). The choice of doll as a toy to keep as a present was also compared for the patients with the I172N or V281L mutations regardless of age or late (n 4) vs. early diagnosis (n 12). The girls with late diagnosis chose a doll more often (P 0.052). FIG. 2. The mean difference in the amount of time that the girls in the different genotype groups played with the different types of toys when they played alone compared with when they played with a parent (bars represent time playing alone minus time playing with parent). Positive values thus represent preference for toys when the child was alone, whereas negative values indicate toys that were chosen more often when the parent was present. FIG. 3. Choice of toy to keep as a present in the different genotype groups.

5 Nordenström et al. Toy Play Correlates with CYP21 Genotype in CAH J Clin Endocrinol Metab, November 2002, 87(11): Discussion We found that girls with CAH played more with masculine toys than controls, which is in line with the findings of previous studies (9, 3, 28). In addition, this study is the first to correlate behavior to CYP21 genotype, which is known to reflect the degree of disease severity and thus the degree of fetal androgen exposure. We found a dose-response relationship between disease severity (i.e. degree of fetal androgen exposure) and degree of masculinization of toy play and preference. This finding supports a biological basis for the differences in play behavior between CAH girls and unaffected girls. It has been argued that the masculinization of play behavior that has been seen in girls with CAH can be attributed to parental influence due to expectations of a more masculine behavior in these girls triggered by the virilization of external genitalia at birth (13). Our results do not support this view. When a parent was present, there was no difference in toy play for the girls with the most severe form of CAH, and the girls with less severe forms of CAH played less, rather than more, with masculine toys. The possible effects of prenatal vs. postnatal androgen exposure on behavior have been discussed. Hormones are considered to affect behavior in different ways (29), namely by having an early organizational effect that takes place during certain critical periods of development and/or a later activational effect, for instance during puberty. Maze learning in rodents, for example, is dependent on organizational influences of androgens (29). In this study the results in girls with CAH diagnosed and treated at an early age favor a prenatal, organizational effect of androgens during the development of the central nervous system. On the other hand, even though there are few observations, our results for the girls with CAH diagnosed late indicate that postnatal androgens may have effects as well. The girls with late diagnosis all had less severe forms of CAH and can be assumed to have been exposed to lower levels of androgens in utero. They were diagnosed at 3 or 6 yr of age and therefore had an overproduction of androgens during the first years of life. Three of these girls played with the Lincoln logs the whole time when they played alone. This result was significantly different from the result for the controls in the same age group. In addition, there was a statistical trend that the girls with late diagnosis played more with the constructional toy compared with CAH girls of the same age but with early diagnosis (P 0.1). These results are intriguing because they contrast with the results in the choice of toy to keep as a present. It was striking that 3 of the CAH girls with late diagnosis preferred the doll, and only 1 chose the ball to keep, while among the girls with the same mutations, I172N or V281L, and early diagnosis 2 of 12 chose a doll. This raises questions concerning the possibility that prenatal and postnatal androgen exposure may affect different aspects of cognitive development. In a study by Berenbaum et al. (11) a much less marked effect of postnatal than prenatal androgen exposure in masculinizing play behavior was shown. This is in agreement with the results of our studies. However, in their study the effect on spatial ability or play with constructional toys was not studied specifically. The possibility of a postnatal effect of androgens on central nervous system development has implications for treatment during the first years of life. It has been argued that the glucocorticoid dose should be kept low during the first 1 2 yr to minimize the negative effects on growth, as no effect of androgens on growth or skeletal maturation has been observed during this period (30). However, if elevated androgen levels have an effect on the developing brain, perhaps this regimen should only be used in boys with CAH. We recognize that additional factors, such as upbringing and cultural influences, play vital roles in such complex human characteristics as behavior. We believe, however, that girls with CAH will benefit from an increased understanding and acceptance of their preferences as well as from the acknowledgment that parental expectations, if anything, tend to counteract them. In conclusion, we have found evidence supporting the idea that prenatal androgen exposure has a direct organizational effect on the human brain so as to determine certain aspects of sex-typed behavior. In addition, our data raise questions concerning possible postnatal effects of androgens. Further studies are needed on this subject to clarify the possible sensitive periods and levels of androgens mediating these effects. Acknowledgments We gratefully acknowledge Sheri Berenbaum for her valuable contribution in planning this study. We are grateful to the children and their families for their cooperation. Received September 21, Accepted July 30, Address all correspondence and requests for reprints to: Dr. Anna Nordenström, Department of Pediatrics, Karolinska Institute, Huddinge University Hospital, S Stockholm, Sweden. anna.nordenstrom@klinvet.ki.se. This work was supported by the Swedish Medical Research Council, Grants 4792 and 12198, the Novo Nordisk Foundation, the Märta and Gunnar Philipson Foundation, the Samariten Foundation, and the Frimurare Barnhuset Foundation. References 1. Ehrhardt AA, Epstein R, Money J 1968 Fetal androgens and female gender identity in the early-treated adrenogenital syndrome. Bull Johns Hopkins Hosp 118: Slijper FME 1984 Androgens and gender role behavior in girls with congenital adrenal hyperplasia (CAH). Prog Brain Res 61: Dittmann RW, Kappes MH, Kappes ME, Borger D, Stegner H, Willig RH, Wallis H 1990 Congenital adrenal hyperplasia. I. Gender-related behavior and attitudes in female patients and sisters. Psychoneuroendocrinology 15: Berenbaum SA 1990 Congenital adrenal hyperplasia: intellectual and psychosexual functioning. In: Holmes CS, ed. Psychoneuroendocrinology: brain, behavior, and hormonal interactions. New York: Springer-Verlag; Resnick S, Berenbaum SA, Gottesman II, Bouchard TJ 1986 Early hormonal influences on cognitive functioning in congenital adrenal hyperplasia. Dev Psychol 22: Hampson E, Rovet JF, Altmann D 1998 Spatial reasoning in children with congenital adrenal hyperplasia due to 21-hydroxylase deficiency. Dev Neuropsychol 14: Helleday J, Edman G, Ritzén E, Siwers B 1993 Personality characteristics and platelet MAO activity in women with congenital adrenal hyperplasia (CAH). Psychoneuroendocrinology 18: Berenbaum SA, Hines M 1992 Early androgens are related to childhood sex-typed toy preferences. Psychol Sci 3: Berenbaum SA, Snyder E 1995 Early hormonal influences on childhood sextyped activity and playmate preferences: implications for the development of sexual orientation. Dev Psychol 31: Leveroni CL, Berenbaum SA 1998 Early androgens effects on interest in infants: evidence from children with congenital adrenal hyperplasia. Dev Neuropsychol 14:

6 5124 J Clin Endocrinol Metab, November 2002, 87(11): Nordenström et al. Toy Play Correlates with CYP21 Genotype in CAH 11. Berenbaum S, Duck S, Bry K 2000 Behavioral effects of prenatal versus postnatal androgen excess in children with 21-hydroxylase-deficient congenital adrenal hyperplasia. J Clin Endocrinol Metab 85: Deleted in proof. 13. Quadagno D, Briscoe R, Quadagno J 1977 Effects of perinatal gonadal hormones on selected nonsexual behavioral patterns: a critical assessment of the nonhuman and human literature. Psychol Bull 84: Berenbaum SA 1999 Effects of early androgens on sex-typed activities and interests in adolescents with congenital adrenal hyperplasia. Horm Behav 35: Zucker KJ, Bradley SJ, Oliver G, Blake J, Flemming S, Hood J 1996 Psychosexual development of women with congenital adrenal hyperplasia. Horm Behav 30: White P, Speiser P 2000 Congenital adrenal hyperplasia due to 21-hydroxylase deficiency. Endocr Rev 21: New MI 1998 Diagnosis and management of congenital adrenal hyperplasia. Annu Rev Med 49: Speiser PW, Dupont J, Zhu D, Serrat J, Buegeleisen M, Tusie-Luna MT, Lesser M, New MI, White PC 1992 Disease expression and molecular genotype in congenital adrenal hyperplasia due to 21-hydroxylase deficiency. J Clin Invest 90: Wedell A, Thilén A, Ritzén EM, Stengler B, Luthman H 1994 Mutational spectrum of the steroid 21-hydroxylase gene in Sweden: implications for genetic diagnosis and association with disease manifestation. J Clin Endocrinol Metab 78: Jääskeläinen J, Levo A, Voutilainen R, Partanen J 1997 Population-wide evaluation of disease manifestation in relation to molecular genotype in steroid 21-hydroxylase (CYP21) deficiency: good correlation in a well defined population. J Clin Endocrinol Metab 82: Nordenström A, Thilén A, Hagenfeldt L, Larsson A, Wedell A 1999 Genotyping is a valuable diagnostic complement to neonatal screening for congenital adrenal hyperplasia due to steroid 21-hydroxylase deficiency. J Clin Endocrinol Metab 84: Wedell A, Luthman H 1993 Steroid 21-hydroxylase deficiency: two additional mutations in salt-wasting disease and rapid screening of disease-causing mutations. Hum Mol Genet 2: Wedell A, Ritzén ME, Haglund-Stengler B, Luthman H 1992 Steroid 21- hydroxylase deficiency: three additional mutated alleles and establishment of phenotype-genotype relationships of common mutations. Proc Natl Acad Sci USA 89: Servin A, Bohlin G, Berlin L 1999 Sex differences in 1-, 3-, and 5-year-olds toy-choice in a structured play-session. Scand J Psychol 40: Alexander GM, Hines M 1994 Gender labels and play styles: their relative contribution to children s selection of playmates. Child Dev 65: Fisher-Thompson D 1993 Adult toy purchases for children: factors affecting sex-typed toy selection. J Appl Psychol 14: Connor JM, Serbin LA 1977 Behaviorally based masculine- and feminineactivity-preference scales for preschoolers: correlates with other classroom behaviors and cognitive tests. Child Dev 48: Ehrhardt A, Baker S 1977 Males and females with congenital adrenal hyperplasia. A family study of intelligence and gender-related behavior. In: Lee PA, Plotnick PO, Kowarski AA, Migeon CJ, eds. Congenital adrenal hyperplasia. Baltimore: University Park Press; Beatty W 1992 Gonadal hormones and sex differences in nonreproductive behaviors. In: Gerall A, Moltz H, Ward I, eds. Handbook of behavioral neurobiology. New York: Plenum; Thilén A, Woods K, Perry L, Savage M, Wedell A, Ritzén E 1995 Early growth is not increased in untreated moderately severe 21-hydroxylase deficiency. Acta Paediatr 84: