| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Royal Manchester Children’s Hospital (C.M.H., J.A.J., D.A.P., P.E.C.), Manchester M27 4HA, Regional Molecular Genetics Laboratory (M.C.), St. Mary’s Hospital, Manchester M13 0JH, and Department of Mathematics (P.F.), University of Manchester, Manchester M13 9PL, United Kingdom; and Department of Psychiatry (H.F.L.M.-B., C.D.), Columbia University, New York, New York 10027
Address all correspondence and requests for reprints to: Catherine M. Hall, Endocrine Department, Royal Manchester Children’s Hospital, Manchester M27 4HA, United Kingdom. E-mail: catherine.hall{at}cmmc.nhs.uk.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
To assess the effect of chronic illness, we compared behavioral masculinity in 24 CAH girls and 25 diabetic girls aged 3–12 yr from Manchester using two sensitive questionnaires, and an overall masculinity score M (high = masculine) was derived.
To assess the contributions of pre- and postnatal androgens, the CAH subjects were categorized into genotype groups (G) according to the reported severity of loss of CYP21 function: G1 (n = 10, null mutations), G2 (n = 9, intron 2G), G3 (n = 3, I172N), and G4 (n = 2, unknown loss of function). In CAH girls, relationships between G, Prader degree of genital masculinization at birth, bone age advance, and M were assessed.
CAH girls were less feminine and more masculine than diabetic girls (P < 0.001), who were not significantly different from U.S. controls. Among the CAH girls, those in G1 and 2 were more genitally masculinized than those in G3 and 4 (P < 0.009) and had higher M (P < 0.025). M was negatively correlated with advanced bone age (r = -0.5; P = 0.02).
CAH girls, but not diabetic girls, demonstrated behavioral masculinization. Both physical and behavioral masculinization were related to each other and to genotype, indicating that behavioral masculinization is a consequence of prenatal androgen exposure.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CAH is autosomally recessively inherited, and the phenotype of each patient reflects the less severe allele. There is fairly good correlation between genotype and severity of salt wasting (3, 4, 5, 6, 7). Deletions, large gene-to-pseudogene conversions, and point mutations resulting in complete absence of the enzyme are referred to as null mutations, and female homozygotes for null mutations are most severely affected with respect to masculinization of the external genitalia and salt wasting. The intron 2G splice mutation (655A/C
G) results in a shift in the translational reading frame. Almost all of the mRNA is aberrantly spliced, but in cultured cells a small amount of normally spliced mRNA is detected. Although it is not known what proportion of mRNA is normally spliced in the adrenal gland of subjects with the mutation, most (but not all) subjects who are homozygous or hemizygous for this mutation have salt wasting CAH (0–<5% in vitro enzyme activity). The I172N mutation (999TA) is associated with approximately 1% in vitro enzyme activity and is usually associated with simple virilizing CAH. Other point mutations generate greater enzyme activity and are associated with nonclassical (late-onset) CAH (1). However, the same genotype, even within families, may be associated with clinically heterogeneous phenotypes (8, 9).
Many studies have reported behavioral masculinization in girls with CAH (i.e. boy-stereotypical behavior); young CAH girls tend to play with boys’ toys (e.g. cars) and exhibit low interest in dolls and, later in life, little interest in child rearing (10, 11, 12, 13). CAH girls may be tomboyish with regard to their dress and lack of interest in make-up and jewelry (14). They demonstrate more aggressive behavior than their peers (15) and male-typical behavior in social relations (16). Male-typical cognitive traits including superior spatial abilities (17, 18, 19) and increased frequency of left handedness (20) have been reported in girls with CAH.
Two recent play studies in CAH girls provide evidence of a relationship between prenatal androgens and behavioral masculinization. Berenbaum and colleagues (21) demonstrated that boy-typical play in girls with CAH was significantly associated with inferred prenatal androgen excess (graded by degree of genital virilization and severity of neonatal salt loss) but not with postnatal androgen excess [assessed by bone age advance, increased growth velocity, and 17-hydroxyprogesterone (17OHP) concentrations]. Subsequently, Nordenstrom and colleagues (22) demonstrated that girls with CAH played with masculine toys more than controls, and that there was a significant correlation between the degree of disease severity, as assessed by CYP21 genotype, and the amount of time spent playing with masculine toys, those with null mutations spending longest.
However, several authors have argued that in humans sex assignment at birth influences parental attitudes toward the infant and that these social factors are paramount in determining the gender role behavior of the infant and that hormones play only a minor role (23, 24). Furthermore, Slijper and colleagues (25) reported that behavior was masculinized in both girls with diabetes mellitus and girls with CAH and concluded that chronic illness strongly influences the development of masculinized behavior in girls.
In this study, our objectives were to investigate whether girls with CAH demonstrate behavioral masculinization while controlling for the influence of chronic illness by comparing a group of CAH girls with a group of age-matched diabetic girls. Then, in the CAH girls, we assessed the relative influences of genotype and pre- and postnatal androgen exposure on behavioral and physical masculinization.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Local Ethical Committee approval was granted for the study, and written, informed consent was obtained from all participants. The study group was recruited from the cohort of girls aged 3–12 yr who attend the CAH clinic at the Royal Manchester Children’s Hospital. A control group was recruited from a cohort of the same age range, attending the diabetic clinic at the same hospital. Although the symptoms and prognoses of CAH and diabetes differ, diabetic girls were selected to act as controls because they share very similar life experiences that accompany their respective chronic illnesses: regular outpatient attendances, daily medication, decompensation during intercurrent illnesses, potentially life-threatening crises, and adverse impact on family life, schooling, and social life. The subjects studied were 24 girls with CAH (23 salt losers) [7.8 ± 3.5 (mean ± SD) yr] and 25 girls with diabetes mellitus (9.7 ± 2 yr). Two of the CAH girls were siblings.
In the CAH girls, the relationships between the following variables were assessed: genotype, the degree of genital masculinization at birth, biochemical indices of salt wasting and androgen concentrations at birth and in later life, and bone age (with advanced bone age as a measure of postnatal androgen exposure) and behavioral masculinization.
CAH: CYP21 genotyping
CYP21 mutation detection, based on allele-specific PCR, was performed on genomic DNA from venous blood samples, which were routinely collected at the time of diagnosis. In Manchester, screening for the following mutations, associated with zero enzyme activity, is performed: CYP21 deletion, large pseudogene conversion, and 2108CT (R356W). These are termed null mutations. In addition, seven common point mutations, associated with reduced enzyme activity, are screened for. These include 655A/CG (intron 2G, 0–<5% in vitro activity), 999TA (I172N, 
1% in vitro activity). These 10 mutations account for approximately 90% of 21-hydroxylase carriers in the Manchester population (estimated after analysis of over 400 index cases in-house). Additional rarer alleles were characterized by direct DNA sequencing. The nucleotide positions are as described in the Human Gene Mutation Base (Cardiff, UK), and the PCR methodology is adapted from that of Wedell and Luthman (26).
The genotypes were categorized according to the reported severity of loss of enzyme function: G1, n = 10 (four Asian, six White) with null mutations; G2, n = 9 (two Asian, seven White) homozygous or heterozygous for the intron 2G splice mutation (655A/CG); G3, n = 3 (all White) compound heterozygotes for the I172N mutation (999TA); and G4, n = 2 (both White) compound heterozygotes with either an unknown mutation or a mutation not previously reported with unknown effect on enzyme activity.
Prader stage of virilization
The patient case notes were reviewed independently and blindly (without knowledge of the patient’s other data) by two observers (C.M.H. and J.J.), and a Prader score for genital virilization was assigned on the basis of annotated clinical examination findings, operation notes, and clinical photos. This was possible in 21 subjects.
Auxological data
Annual bone age assessments were extracted from the notes.
Biochemical data
Serum concentrations of testosterone, renin, 17OHP, Na, and K were measured at diagnosis and at regular intervals thereafter. For the purposes of this study, biochemical data at diagnosis in the neonatal period were considered to reflect the prenatal hormonal milieu. Postnatal biochemical data (mean values) for morning 17OHP, testosterone, renin, Na, and K in the first year, the year before the study, and the intervening years were recorded.
Hormonal replacement therapy
The mean daily doses of fludrocortisone (µg/m2) and hydrocortisone (mg/m2) were calculated for the first year of life and in the year preceding the study.
Behavior
Gender-related behavior was assessed with two questionnaires, which were administered to the parents in the family homes by the endocrine specialist nurse.
The Child Game Participation Questionnaire (CGPQ), containing 69 items on play, was devised by Bates and Bentler (27), modified by Meyer-Bahlburg (28), and validated as an instrument for the assessment of gender-related variations in both boys and girls with (physical) intersex conditions and (behavioral) gender identity disorders.
The Child Behavior and Attitude Questionnaire (CBAQ), Male Form (27), modified by Meyer-Bahlburg and colleagues (29), was originally developed for the differentiation of boys with a gender identity disorder. Subsequently, Meyer-Bahlburg and colleagues (29) developed an analogous version for girls, and both versions underwent extensive analysis, rescaling and norming on the basis of community samples.
As both questionnaires had been developed in the U.S., we made some changes in individual items (n = 8) to adjust them to British usage. For instance, "Indian wrestling" was replaced with "arm wrestling," and "play store" with "play shop."
The rationale for using two different questionnaires with five gender-sensitive scales incorporating excellent psychometric properties was to minimize the likelihood of a chance finding, especially as we were using a single informant (the mother).
Statistical analysis
Two-group ANOVA (t test) was used to compare the questionnaire results of the CAH and diabetic groups. The mean age in the diabetic sample was 1.9 yr older than the CAH sample (9.7 vs. 7.8 yr; P = 0.028). Therefore the t tests were repeated as analyses of covariance controlling for age. The CAH sample included six Asians and 18 Caucasians, whereas the diabetic sample was all Caucasian. There was no significant difference between the Asian and Caucasian CAH girls in terms of means or dispersions of gender scores so the two subgroups were combined for comparisons with the diabetic control group.
Community normative data are available for American samples of 6- to 10-yr-old children, and Meyer-Bahlburg and colleagues (30) have shown that both questionnaires also differentiate strongly between boys and girls among 3- to 5- and 11-to 12-yr-olds. No English comparison data are available on community samples of boys and girls so that we could not directly test whether diabetic girls differ from community girls in gender-related behavior. As a compromise approach, given that ethnic differences appear to matter little in heterogeneous U.S. community samples, we compared the 6- to 10-yr-old CAH and diabetic girls from the Manchester sample to the 6- to 10-yr-old antenatally untreated CAH girls and control girls from a study evaluating prenatal dexamethasone exposure using U.S.-based norm scores (30). The community norms are age corrected, and therefore no adjustment for age was required during analysis. There was no significant age difference between the CAH and diabetes girls in this age range.
Two gender-related scales are derived from the CBAQ and three from the CGPQ. To simplify the analysis of behavioral masculinization in the CAH girls, the five scales were combined to derive one even more robust measure of overall gender role behavior. This overall masculinity score M was derived by principal component analysis, with the first principal component being the linear combination of the five variables (scales), which has the largest variance. Multiple comparisons with the Bonferroni post hoc test were used to compare the values of M between different genotype groups.
Pearson correlation and linear regression analysis were used to assess the relationships between variables. In separate analyses, Prader score, 17OHP at birth, fludrocortisone and hydrocortisone doses in the first year of life, and masculinity score M were entered as the dependent variables and genotype group G as the independent variable. In analysis of covariance, M was entered as the dependent variable, G as the factor, and Prader score as the covariate. Nonparametric tests (Kruskal Wallis and Mann Whitney) were used to compare mean values of variables between genotype groups. The statistical package SPSS (version 10.1) was used, with statistical significance assigned to an 
-value of 0.05.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CBAQ. CAH girls scored lower in the femininity scale than diabetic girls but higher on the cross-gender scale for girls (P < 0.001 for both).
CGPQ. CAH girls scored higher on the gender scale and masculinity subscale than diabetic girls (P < 0.001 for both) but lower in the femininity/preschool scale (P = 0.006).
Multiple comparisons of the subgroups of 6- to 10-yr-olds from Manchester (seven CAH girls, 14 diabetic girls) with the 6- to 10-yr-old American subgroup [previously studied by Meyer Bahlburg et al. (30); 38 CAH girls, 22 non-CAH control girls] revealed that the Manchester CAH girls were the most masculine/least feminine, followed by the U.S. CAH girls. Both Manchester and American CAH girls were significantly more masculine and less feminine than Manchester diabetic and American control girls on all scales. The Manchester diabetic girls were comparable to the U.S. control girls, except for one significant difference, on the CBAQ cross gender scale, where they scored lower (P = 0.042) (Table 1
).
|
Relationship between G and Prader stage of genital virilization. The girls in G1 were significantly more virilized [mean (SD) Prader score = 3.1 (0.3)] than the girls in G3 [1.7 (0.6)] and G4 [1.5 (0.7)]; P = 0.002 and 0.003, respectively. The girls in G2 were significantly more virilized [3 (0.6)] than those in G3 and G4; P = 0.006 and 0.008, respectively. G contributed 55% to the variance of the Prader score (Prader score = 3.8 - 0.6 x G; r2 = 0.55; P = 0.001).
Relationship between G and biochemical parameters.
Of the biochemical parameters measured at birth (Na, K, renin, 17OHP, and testosterone), only 17OHP differed by genotype group (Fig. 1). Mean (SD) serum 17OHP at birth was significantly higher in G1 [1233 (798) nmol/liter] than in G3 [295 (270) nmol/liter]; P = 0.049. G accounted for 21% of the variance in 17OHP at birth (17OHP= 1590 - 381 x G; r2 = 0.21; P = 0.028). None of the biochemical parameters measured over subsequent years during steroid treatment was significantly related to G.
|
At the time of the study, there were no significant differences in fludrocortisone or hydrocortisone according to genotype group.
Relationship between G and overall masculinity score M.
The mean (SD) masculinity scores M (high = masculine) in G1–4 were -0.23 (1.8), 1.6 (0.6), -1.5 (0.6), and -3.5 (0.22), respectively (Fig. 2). The mean M in G2 was significantly higher than in G1 (P = 0.043), G3 (P = 0.012), and G4 (P = 0.001). M in G1 was higher than in G4 (P = 0.021). G contributed 23% to the variance in M (M = 1.7 - 0.9 x G; r2 = 0.23; P = 0.03). There was no significant difference in the overall masculinity score between the Asian [M = 0.03 (1.7); n = 6] and White [M = 0.9 (1.6); n = 13] subjects in G1 and G2 (P = 0.32).
|
). Prader stage contributed 44% to the variance in M (M = -4.6 + 1.7 x Prader stage; r2 = 0.44; P = 0.001). In analysis of covariance with M as the dependent variable, G as a factor, and Prader stage as a covariate, M was predicted by G (P < 0.001) alone.
|
Relationship between M and hormonal replacement therapy. M was positively correlated with hydrocortisone dose in the first year of life (r2 = 0.25; P = 0.022). There were no significant relationships between M and fludrocortisone dose in the first year of life or with hydrocortisone or fludrocortisone doses at the time of the study.
Relationship between M and bone age. Advanced bone age at the time of the study was used as a measure of postnatal androgen exposure, and there was a negative correlation between masculinity score and advanced bone age (r = -0.5; P = 0.02; n = 21).
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CAH girls were significantly more masculinized than diabetic girls, who were not significantly different from the U.S. community norms and did not demonstrate masculinized behavior. Our results do not support the hypothesis that chronic illness leads to masculinized behavior (25) but demonstrate that masculinized behavior is a feature of CAH. The conflicting results from Slijper (25) are likely to be due to the fact that her CAH sample included an unusually high percentage of the simple virilizing type (46% of 24), and the latter have been found in other studies to show much less masculinized behavior than girls and women with the salt-losing variety. In fact, Slijper’s data show that it is the simple virilizers who are comparable in gender-related behavior to her diabetic girls whereas the salt-losing CAH girls are much more masculinized.
In the CAH girls, the key observations were that both Prader degree of genital virilization at birth and behavioral masculinization were predicted by genotype and that there was a substantial, significant correlation between the severity of genital virilization and the degree of behavioral masculinization. These data support the hypothesis that prenatal androgen exposure is the major contributor to both physical and behavioral masculinization, and the severity of each is determined by genotype. There was greater variability of M in G1 than in the other groups, and the correlation between M and Prader stage was stronger than between M and G. Perhaps this might be explained by inter-subject variation in androgen sensitivity. If so, girls with severe mutations would be exposed to higher concentrations of androgens than those with milder mutations and more likely to become physically and behaviorally masculinized, but their degree of masculinization would be modulated by their androgen sensitivity.
Although our study does not directly assess parental sex typing, it seems unlikely to us that masculinized behavior in CAH girls is largely due to differential parenting behavior by those parents influenced by their daughters’ genital virilization, as some investigators have hypothesized (23, 24). None of the available data support this hypothesis, regardless of assessment method, e.g. parental self-report (Refs. 31 , 32 , and Resnick, S. M., unpublished doctoral dissertation, 1982), retrospective daughter reports about their parents (33), expressed parental wishes for more femininity in their CAH girls (34), or observation of parents’ sex-typing behavior when they are playing with their CAH daughters (Pasterski, V. L., unpublished doctoral dissertation, 2002). Furthermore, numerous studies have replicated the findings from parent reports, with child reports and behavior observation (reviewed in Ref. 35). Thus, parental bias in the perception of their daughters’ behavior is not likely to be a major explanation of CAH girls’ behavioral masculinization.
The relationship between genotype and genital virilization is consistent with the results of other studies (5, 6, 7). Our study corroborates those of Berenbaum and colleagues (21), who demonstrated a positive correlation between masculinized toy play and genital virilization in girls with CAH and those of Nordenstrom and colleagues (22), who correlated masculinized behavior, as assessed by toy play, with genotype in girls with CAH.
The relationship between genotype and 17OHP concentrations at birth is consistent with previous studies (36, 37, 38). However, we did not observe statistically significant relationships between other biochemical parameters measured at birth and genotype as others have reported (3, 4, 5, 6, 7). Neither did we observe a relationship between serum sodium concentrations at diagnosis and masculinized behavior as described by Berenbaum and colleagues (21). This may be explained by the fact that the subjects were transferred to the care of our team in Manchester from district hospitals, resulting in variability of the timing of initial samples in relation to birth and also some missing data.
Hormonal replacement therapy in the first year of life correlated with genotype in the manner one would expect; those children with the most severe disease required higher doses of hydrocortisone to suppress ACTH in the first year of life and higher doses of fludrocortisone to compensate for renal salt loss. The significant positive correlation between hydrocortisone requirement in the first year of life and masculinity score is consistent with a correlation between clinical disease severity and the degree of masculinized behavior.
A negative relationship between behavioral masculinization and bone age as a marker of postnatal androgen exposure was observed, which is consistent with the observations of Berenbaum and colleagues (21). This might be explained by the fact that those children with more severe disease were treated more aggressively with hydrocortisone from birth, which suppressed postnatal androgen concentrations and retarded bone age. Conversely, children with milder disease would have been treated less aggressively and exposed to higher postnatal androgen concentrations, which promoted skeletal maturation but not behavioral masculinization.
This study has some methodological limitations. One is that the information on the girls’ gender-related behavior is provided only by their mothers, and completing information from the children themselves and/or data from behavior observation would be desirable. However, many studies using a variety of methods and informants have replicated the finding that gender-related behavior in CAH girls is masculinized. It would also be desirable to assess parental socialization practices to test to what extent, if any, parental sex-typing contributes to the masculinization of gender-related behavior in CAH girls, although the existing data make it unlikely.
We conclude that behavioral masculinization in females with CAH is most likely largely as a result of exposure of the fetal brain to high levels of androgens. Furthermore, the degree of behavioral masculinization is correlated with the degree of genital masculinization and determined primarily by genotype, rather than by postnatal androgen exposure, steroid replacement, biochemical status, or ethnicity.
|
Footnotes |
|---|
Received April 28, 2003.
Accepted October 1, 2003.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
-hydroxyprogesterone radioimmunoassay: its application for rapid screening for congenital adrenal hyperplasia. J Clin Endocrinol Metab 45:1003–1008This article has been cited by other articles:
 |
|
 |
|
  C. Ciumas, A. L. Hirschberg, and I. Savic High Fetal Testosterone and Sexually Dimorphic Cerebral Networks in Females Cereb Cortex, May 1, 2009; 19(5): 1167 - 1174. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. L. Claahsen-van der Grinten, K. Noordam, G. F. Borm, and B. J. Otten Absence of Increased Height Velocity in the First Year of Life in Untreated Children with Simple Virilizing Congenital Adrenal Hyperplasia J. Clin. Endocrinol. Metab., April 1, 2006; 91(4): 1205 - 1209. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D.H. Abbott, D.K. Barnett, C.M. Bruns, and D.A. Dumesic Androgen excess fetal programming of female reproduction: a developmental aetiology for polycystic ovary syndrome? Hum. Reprod. Update, July 1, 2005; 11(4): 357 - 374. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. G. Forest Recent advances in the diagnosis and management of congenital adrenal hyperplasia due to 21-hydroxylase deficiency Hum. Reprod. Update, November 1, 2004; 10(6): 469 - 485. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
