Archives of Sexual Behavior

, Volume 37, Issue 1, pp 85–99

Sexual Orientation in Women with Classical or Non-classical Congenital Adrenal Hyperplasia as a Function of Degree of Prenatal Androgen Excess

  • Heino F. L. Meyer-Bahlburg
  • Curtis Dolezal
  • Susan W. Baker
  • Maria I. New
Original Paper: Minot Special Issue

DOI: 10.1007/s10508-007-9265-1

Cite this article as:
Meyer-Bahlburg, H.F.L., Dolezal, C., Baker, S.W. et al. Arch Sex Behav (2008) 37: 85. doi:10.1007/s10508-007-9265-1

Abstract

46,XX individuals with classical congenital adrenal hyperplasia (CAH) due to deficiency of the enzyme, 21-hydroxylase, show variable degrees of masculinization of body and behavior due to excess adrenal androgen production. Increased bisexuality and homosexuality have also been reported. This article provides a review of existing reports of the latter and presents a new study aimed at replicating the previous findings with detailed assessments of sexual orientation on relatively large samples, and at extending the investigation to the mildest form, non-classical (NC) CAH. Also, this is the first study to relate sexual orientation to the specific molecular genotypes of CAH. In the present study, 40 salt-wasters (SW), 21 SV (simple-virilizing), 82 NC, and 24 non-CAH control women (sisters and female cousins of CAH women) were blindly administered the Sexual Behavior Assessment Schedule (SEBAS-A, 1983 ed.; H. F. L. Meyer-Bahlburg & A. A. Ehrhardt, Privately printed). Most women were heterosexual, but the rates of bisexual and homosexual orientation were increased above controls not only in women with classical CAH, but also in NC women, and correlated with the degree of prenatal androgenization. Classifying women by molecular genotypes did not further increase the correlation. Diverse aspects of sexual orientation were highly intercorrelated, and principal components analysis yielded one general factor. Bisexual/homosexual orientation was (modestly) correlated with global measures of masculinization of non-sexual behavior and predicted independently by the degree of both prenatal androgenization and masculinization of childhood behavior. We conclude that the findings support a sexual-differentiation perspective involving prenatal androgens on the development of sexual orientation.

Keywords

Congenital adrenal hyperplasia Sexual orientation Homosexuality Androgen effects Disorders of sex development Hermaphroditism 

Introduction

Sexual orientation is a trait with very large differences between men and women. For instance, Hines (2004, p. 11) reported an effect size d = 6.0, one of the largest for any gender-related behavior or trait (although the distribution of sexual orientation scores in terms of the widely used Kinsey scale raises some doubts about the appropriateness of the usual statistical measure of effect size, which assumes a Gaussian distribution). The demonstration of familiality and heritability of homosexuality (e.g., Pillard & Bailey, 1998) has led to numerous attempts to provide genetic explanations. Given the focus of evolutionary theory on reproduction and survival of the offspring, a sexual orientation of women to men and of men to women is eminently plausible. It is much more difficult to come up with a compelling evolutionary raison d’être for homosexuality and bisexuality. A number of non-endocrine explanatory hypotheses have been formulated. In the framework of evolutionary theory, overdominance (Miller, 2000), kin altruism (Pillard & Bailey, 1998), and sexually antagonistic selection (Hamer & Copeland, 1994) have been suggested as potential mechanisms explaining the gene polymorphism that is presumed to underlie homosexuality, and mathematical models have recently been formulated that should facilitate their empirical testing (Gavrilets & Rice, 2006). However, the variability of homosexual behavior across vertebrate species has not led to a consensus on an explanation in terms of evolutionary theory (Sommer & Vasey, 2006), and the demonstration of learning mechanisms in the acquisition of sexual preferences in animal models (Pfaus, Kippin, & Coria-Avila, 2003) has further complicated the issue. Identification of specific genes has not yet led to consistent success (Hamer, Hu, Magnuson, Hu, & Pattatucci, 1993; Mustanski et al., 2005; Rice, Anderson, Risch, & Ebers, 1999), but new findings on extreme skewing of X-inactivation by DNA-methylation in mothers of gay men have added additional genetic possibilities (Bocklandt, Horvath, Vilain, & Hamer, 2006).

Bearman and Brückner (2002) provided a strong methodological critique of much of the existing genetic literature and presented new data from the AddHealth project that (indirectly) supported a social-influence hypothesis explaining same-sex attractions in adolescents. Other non-genetic explanations include the progressive immunization hypothesis, which is derived from the well-replicated association of homosexuality in males with the number of older brothers in the sibship and assumes that successive pregnancies with male fetuses leads in some mothers to the development of male-specific antigens (Blanchard, 2004; Blanchard & Lippa, 2007). The developmental instability theory explains homosexuality as a perturbance of the complex processes of prenatal brain development by exogenous influences and was originally stimulated by findings of increased non-right handedness among homosexuals of both sexes (Lalumière, Blanchard, & Zucker, 2000), but attempts at finding an association of homosexuality with fluctuating asymmetry as a broader index of developmental instability have been unsuccessful (Rahman, 2005).

The most commonly offered theory places sexual orientation in the context of the sexual differentiation of brain and behavior in general, with a focus on the role of pre- and perinatal sex hormones in this process (e.g., Ellis & Ames, 1987; Rahman & Wilson, 2003). This approach was presumably prompted by the association of human homosexuality with gender-atypical (non-sexual) behavior and goes back to the mid-19th century, when scientific embryology began focusing on the development of the sex-dimorphic reproductive tract and its disorders. Such research yielded medical explanations of somatic hermaphroditism, and analogous medical concepts were applied to the explanation of homosexuality (Ulrichs, 1862, 1868, as cited in Hirschfeld, 1906; see also Kennedy, 1988, Ch. 5). In this context, homosexuality was often categorized as an inversion (of gender roles). With the rapidly advancing techniques of measurement and synthesis of sex hormones in the second half of the 20th century, behavioral-endocrinology research in non-primate mammals demonstrated the profound “organizational” influence of sex hormones during early developmental periods on later mating behavior and sexual orientation, with perinatal androgens and estrogens (derived by aromatization of androgens within brain cells) supporting the development of masculine behavior, and estrogens supporting the defeminization of behavior, followed by “activating” effects of sex-specific hormones from puberty on (Arnold, 2002; Wallen & Baum, 2002). The analogous processes in primates may be limited to androgen effects, but the determinants of sexual orientation in primates are not yet clear (Baum, 2006; Wallen & Baum, 2002). Early attempts to identify sex hormone abnormalities in human homosexuality were unsuccessful in men (Meyer-Bahlburg, 1977) and only partially successful in women (Meyer-Bahlburg, 1979). In the absence of systemic hormone abnormalities and of any signs of somatic intersexuality in human homosexuality, some investigators have suggested that causal endocrine abnormalities might be limited to the central nervous system (“CNS-limited pseudohermaphroditism,” Dörner, 1976), which is compatible with the recently growing evidence of tissue specificity of hormone production and/or metabolism and of hormone receptors. During the last two decades, advances in genetics have broadened the focus of research on sexual differentiation to include the many genes involved in the sexual differentiation of the gonads (Fleming & Vilain, 2005) and, possibly, of the brain (Arnold, 2002, 2004; Gatewood et al., 2006). Although research on the specific genetic mechanisms involved in the brain is still in its early stages, the recent use of cellular-biology techniques to unravel the chain of mechanisms involved in the hormone-based sexual differentiation of specific sex-dimorphic nuclei of the limbic system and the amygdala in the neonatal (Burks, Wright, & McCarthy, 2007; Todd, Schwarz, & McCarthy, 2005; Todd, Schwarz, Mong, & McCarthy, 2007) and pubertal (Zehr, Todd, Schulz, McCarthy, & Sisk, 2006) periods of development is likely to contribute in a major way to the identification of genes with specific functions in these processes.

Recent quite large-scale behavioral data on humans continue to support the “inversion” perspective regarding homosexuality (e.g., Lippa, 2005), and the attempts to find a hormonal cause continue. Considerable efforts have been made to identify somatic markers of prenatal sex hormone effects, such as shifts in the second to fourth finger length ratio (2D:4D) in homosexuals of both sexes (Manning, 2002); the findings are suggestive, but far from uniform (Manning, Churchill, & Peters, 2007; Rahman, 2005). Another such marker may be the reduction of spontaneous otoacoustic emissions in lesbians (McFadden, 2002), which is awaiting replication by independent teams. Clearly, more direct evidence of prenatal sex hormone effects would be desirable. However, experimental variations of the prenatal sex-hormone milieu solely for behavioral research purposes cannot be ethically justified. Money’s team at Johns Hopkins introduced as an alternative the behavioral study of syndromes of intersexuality, which represent naturally occurring extreme variations of the sex-hormone milieu (Ehrhardt, Evers, & Money, 1968). Classical (prenatal-onset) congenital adrenal hyperplasia (CAH) in 46,XX individuals is the most prevalent of the classical intersex syndromes and by far the most thoroughly investigated in terms of endocrinology and psychology. About 90% of CAH patients suffer from the deficiency of the enzyme, 21-hydroxylase (Grumbach, Hughes, & Conte, 2003). As one of several endocrine consequences, 46,XX fetuses with CAH are exposed to unusually high levels of androgens during fetal development, which variably masculinize the genitalia and presumably also the brain and later behavior.

If sexual orientation is sexually differentiated in a similar fashion, 46,XX women with classical CAH should show an increase in bisexuality and homosexuality. Table 1 presents 18 studies on sexual orientation in 46,XX women with classical CAH that were published between 1968 and 2007. The data suggest several conclusions. Most women with classical CAH are heterosexual, but where control groups or population data are used for comparisons, bisexual and homosexual orientation appear increased in CAH women (although four studies did not find such increases: Kühnle, Bullinger, & Schwartz, 1995; Lev-Ran, 1974; Müller, Bidlingmaier, Förster, & Knorr, 1982; Slijper et al., 1992). In 11 studies that separately assessed both erotic/romantic imagery and the gender of actual sex partners, increased bisexuality and homosexuality were seen more commonly in imagery (Ehrhardt, 1979; Ehrhardt et al., 1968; Gastaud et al., 2007; Guth, Witchel, Witchel, & Lee, 2006; Horn, 1997; May, Boyle, & Grant, 1996; Money, Schwartz, & Lewis, 1984; Stikkelbroeck et al., 2003; Zucker et al., 1996), although the differences did not always reach statistical significance, and in two studies (Lev-Ran, 1974; Müller et al., 1982), no CAH woman was positive for bisexuality or homosexuality on either variable.
Table 1

Sexual orientation in women with CAH: Published data

Reference

Country

Sample

Descriptiona

N

Age (yrs)

Imagery

Actual partners

Imagery & partners

Range

M

SD

% Hetb

% Bi/Hoc

No datad

% Hetb

% Bi/Hoc

No datad

% Hetb

% Bi/ Hoc

No datad

Ehrhardt et al. (1968)

USA

CAH: late-treated

23

19–55

32.3

 

39.1

43.5

17.4

52.2

17.4

30.4

   

Lev-Ran (1974)

USSR

CAH: late-treated

18

13–43

26

 

77.7

0.0

23.3

38.8

0.0

62.2

   

Ehrhardt (1979)

USA

CAH: early-treated

13

11–24

16.8

  

15.4e

  

7.7

    

Müller et al. (1982)

Germany

CAH: 2 SW, (rest SV?)

14

18–30

21f

 

64.3

0.0

35.7

78.6

0.0

21.4

   

Money et al. (1984)

USA

CAH:

30

17–26

21g

 

40.0

37.0

23.0

40.0

13.3

46.7

40.0

36.7

23.3

Controls: 15 AIS, 12 MRKS

27

16–?

     

66.7

3.7

29.6

92.6

7.4

0.0

Mulaikal, Migeon, and Rock (1987)

USA

CAH: 40 SW, 40 SV

80

18–69

33.0

    

57.5

5.0

37.5

   

Dittmann, Kappes, and Kappes (1992)

Germany

CAH: 12 SW, 20 SV

34

11–41

         

20.0h/

 

2 no info

           

26.5i

 

Subgroup

9

21–41

      

22.2

  

55.6

 

Controls: Sisters

14

11–31

         

0.0h/

 
            

0.0i

 

Slijper et al. (1992)

Netherlands

CAH:

10

16–33

20.8

    

20.0

0.0

80.0

100.0

0.0

0.0

Kühnle et al. (1995)

Germany

CAH: 20 SW, 17 SV, 8 NC

45

 

27.0

6.6

    

4.4k

    

Controls: Hospital staff & families

46

 

26.3

5.4

    

2.2k

    

May et al. (1996)

UK

CAH:

19

18–37

26.1

5.6

 

26.5e,l

  

10.5

    

Controls: Diabetics

17

18–34

27.0

5.3

 

0.0m

       

Zucker et al. (1996)n

Canada

CAH: 19 SW, 12 SV

31

 

24.4

6.7

66.7

26.7

6.7

80.0

3.3

16.7

   

Controls: Sisters, fem. cousins

15

 

25.6

5.5

100.0

0.0

0.0

100.0

0.0

0.0

   

Horn (1997)

Germany

CAH: 17 SW, 12 SV

29

17–36

23.3

5.1

 

27.6

  

3

  

27.6

 

Stikkelbroeck et al. (2003)

Netherlands

CAH: All SW

8o

18–29

22.8

  

50.0

  

0.0

    

Hines, Brook, and Conway (2004)

UK

CAH: 14 SW, 2 no info

16

 

23.6

6.7

    

31p

    

Controls: Sisters, fem. cousins

15

 

22.7

3.4

    

0.0q

    

Morgan, Murphy, Lacey, and Conway (2005)

UK

CAH:

18

18–36

         

22.2r

 

Johannsen, Ripa, Mortensen, and Main (2006)

Denmark

CAH: 21 SW, 6 SV, 5 NC, 1 no info

33

17–51

30.3

     

12.1s

    

Controls: Civil Registry

33

17–51

      

0.0s

    

Guth et al. (2006)

USA

CAH:

5

23–34

27

 

80.0

20.0

0.0

80.0

0.0

20.0

   

Gastaud et al. (2007)

France

CAHt: 4 P-I, 6 P-II, 11 P-III, 11 P-IV, 3 P-V

35

18–43

29.5

  

20.0

  

5.8

    

Controls: Healthy hospital-staff families

69

19–45

30.0

  

5.7

       

a“CAH” refers to classical CAH (SW, SV, excluding NC) unless otherwise specified

b“% Het” percentage of women with exclusively heterosexual imagery/experience

c“%Bi/Ho” percentage of women with bisexual or exclusively homosexual imagery/experience

d“No data” combines “no sexual imagery/partner experience” and “missing information”

e“Including the women with actual same-sex partner experience

fMedian

gAccording to Zucker and Bradley (1995), p. 143

h“Had or wishes to have long-term/steady relationship with F partner”

iAt least one positive item on a 10-item HOM scale indicating homosexual fantasies, wishes, or experiences

k“Being lesbian and living with a female partner”

l“Strong sense of sexual appreciation of females”

m“Sexual interest in other women”

nLifetime data

oSexual orientation is based on only 6 of the 8 women

pFor the past 12 months

q“None rated themselves as bisexual or homosexual; all indicated their behavior had been exclusively or mainly heterosexual”

rSelf-categorization as homosexual or bisexual; partner experience was not specified

s“Present relationships”

tPrader stage of genital development at birth

The methodology of this research is, however, largely unsatisfactory. Most studies employed relatively small samples, many also lacked a control group, and two (Johannsen et al., 2006; Kühnle et al., 1995) even included non-classical (NC) women in their CAH sample and did not analyze them separately, although NC women become clinically symptomatic only years after birth. Many studies limited their data to the gender of actual sex partners or provided only quite limited detail on imagery, in both cases often without employing systematic assessment methods. In addition, studies differed in sample composition regarding the percentages of mild and severe cases of CAH women and their age distribution. Thus, it is difficult to derive solid estimates of population prevalence figures of bisexual and homosexual orientation in CAH women from these data.

Even if the majority of the findings listed support an association of classical CAH with bisexual or homosexual orientation in women, a causative interpretation of these findings in terms of androgen effects on sexual orientation is not as compelling as would be findings from randomized control trials of androgen treatment. Human studies of this kind only use “quasi-experimental designs” with, in the best case, “patched-up controls” (Kazdin, 2002). One could strengthen the case for the role of androgens by demonstrating a dose–response relationship between the degree of prenatal androgen exposure and the degree of later sexual orientation. Ideally, we would measure prenatal hormone levels repeatedly over the course of fetal development and derive from such measurements an index of the degree of prenatal androgen exposure. However, the health risks of even one-time cross-sectional determinations of androgen levels in the amniotic fluid, for instance, can be justified only if there are compelling medical indications for the procedures involved. A relatively crude alternative is the demonstration of dose–response relations on the group level using the clinical-endocrine or molecular-genetics classification of CAH subtypes that differ in severity, i.e., degree of 21-hydroxylase deficiency and, thereby, degree of androgen excess. Within classical CAH, commonly two major subtypes are distinguished, the more severe salt-wasting (SW) variant and the simple virilizing (SV) variant, and several studies have shown that bisexuality and homosexuality are increased more in the SW than the SV variant (Dittmann et al., 1992; Horn, 1997; Mulaikal et al., 1987; Zucker et al., 1996) or in CAH women with higher Prader stages of genital masculinization at birth, which are also (moderately) correlated with CAH severity (Gastaud et al., 2007); the earlier finding by our team that CAH-SW women with gender dysphoria are gynecophilic fits in with the other data (Meyer-Bahlburg et al., 1996).

Our current study had several goals: (1) To replicate the published findings on sexual orientation in a relatively large sample of adult women with classical CAH using a systematic assessment of multiple aspects of sexual orientation, and to establish at which age the women reach the respective romantic/erotic milestones; (2) to extend the dose–response approach to the mildest form of CAH, the non-classical (NC) variant, which becomes clinically symptomatic (in somatic terms) only after birth, in childhood or adolescence; (3) to examine to what extent sexual orientation and global measures of gender behavior other than sexual orientation are correlated; (4) to test whether the prediction of the behavioral phenotype from the endocrine phenotype can be enhanced by the molecular-genetics classification; (5) to answer the question how commonly CAH women see themselves as men in their romantic/erotic imagery; and (6) to perform a methodological study of the interrelationship of the sexual orientation variables.

Method

Participants

The current study is part of a comprehensive long-term follow-up project of women with CAH. Selection and recruitment of participants were described in detail elsewhere (Meyer-Bahlburg, Dolezal, Baker, Ehrhardt, & New, 2006) and will be only briefly summarized here. During an initial pilot phase of this project, a small number of women with CAH was recruited from two pediatric endocrine clinics in New York City. Subsequent data collection for the main study was limited to the senior endocrine author’s (M.I.N.) clinic, but data from both study phases were combined for the final analysis. Eligible were all adult women with CAH due to 21-hydroxylase deficiency for whom the molecular genetics of the 21-hydroxylase gene had been determined and who spoke English. Geographically, the participating women were spread over the entire United States and other continents. Transportation reimbursement was provided for women within the continental US.

The total analysis sample for CAH women in this report included 40 SW women, 21 SV women, and 82 NC women. Almost all CAH women were on glucocorticoid replacement treatment at the time of the study. A total of 24 non-CAH control women (labeled COS) consisted of sisters and female cousins of participating CAH women. Ages ranged from 18–61 years (subgroup means, 28.8–34.7 years). Not included in this count are two SW women: one patient because he had changed to living as a man and was therefore not administered questionnaires and interviews designed for women, the other because of cognitive limitations which interfered with the standard administration of assessment instruments; the latter was also seriously considering gender change to male at the time of the examination.

In addition to the control group of sisters and female cousins, we included for selected comparisons and illustrations two control groups (labeled COD) from our preceding project on the long-term behavioral after effects of prenatal diethylstilbestrol (DES) exposure (Meyer-Bahlburg et al., 1995; Pillard et al., 1993), which had used the same assessment instrument for sexual behavior and sexual orientation: 67 DES-unexposed female controls and 60 DES-unexposed male controls, of comparable age ranges and subgroup means.

Study procedures were approved by the appropriate institutional review boards, and all participants gave written informed consent.

Measures and Procedure

All women underwent an 8–10 h protocol (often spread out over several days) of standard self-report questionnaires, psychometric tests, physical examinations, and systematic interviews. Sexual orientation was assessed as part of the Sexual Behavior Assessment Schedule (SEBAS-A, 1983 ed.; H. F. L. Meyer-Bahlburg & A. A. Ehrhardt, Privately printed), a comprehensive sexual-history interview schedule that covers psychosexual milestones, sexual orientation, sexual activity level, and sexual dysfunctions. Its administration takes approximately 1 h. The SEBAS-A was placed late in the overall protocol in order to facilitate rapport development between interviewer and interviewee and, thereby, increase disclosure of sensitive information, and the SEBAS-A instructions to the interviewee emphasized the importance of accuracy to enhance the participants’ motivation. All SEBAS-A interviews (as well as most other study interviews) with women were conducted by female interviewers in order to facilitate self-disclosure. Interviewers were clinical psychologists who were specifically trained for sexual research interviewing. Procedures were introduced to keep the interviewers from identifying the group membership of the study participants along with instructions for the women against the disclosure of their medical histories to the interviewers. All interviews were audiotaped to permit monitoring of interviewer performance. Excellent interrater reliability of the SEBAS-A has been demonstrated (Meyer-Bahlburg et al., 1995).

SEBAS-A variables pertinent to sexual orientation covered masturbation fantasies, masturbation erotica, romantic/erotic fantasies during sexual relations with a partner, romantic/sexual daydreams, romantic/sexual nightdreams, sexual attractions, “Total Imagery,” actual sex partners (“Actual Partners”), and overall sexual responsiveness (“Overall Kinsey”). The first six variables addressed “current” sexual orientation, with “current” defined as the 12 months prior to interview, and each was preceded by a question concerning its frequency (e.g., “How often did you have romantic or sexual nightdreams during the past 12 months?”). The remaining three aspects were rated separately for the past 12 months and for lifelong (“Lifetime”) patterns (thus, yielding six variables), with lifelong defined as “since puberty” (for “Total Imagery” and overall sexual responsiveness), or as “since becoming sexually active, excluding prepubertal sexual activities” (for sexual relations); both definitions of lifelong included the past 12 months.

For each sexual-orientation variable, interviewers’ ratings used the Kinsey Rating Scale (Kinsey, Pomeroy, Martin, & Gebhard, 1953) with the following formulations: 0 = entirely heterosexual; 1 = largely heterosexual but incidentally homosexual; 2 = largely heterosexual but also distinctly homosexual; 3 = equally heterosexual and homosexual; 4 = largely homosexual but also distinctly heterosexual; 5 = largely homosexual but incidentally heterosexual; and 6 = entirely homosexual. Since the Kinsey team had not defined “distinct,” “a distinct” homosexual history (Kinsey score 2 or “K2”) was rated when the woman had experiences such as homosexual dreams or fantasies over a period of at least 1-year recurring with some regularity (not less than “about once a month”). Whenever a subscale was rated “K2,” the corresponding global score could not be rated less than “K2.”

The variables on actual sex partners were based on detailed structured interview sections concerning diverse romantic and sexual activities, separately for male and female partners. The definition of sexual relations as used here for actual sex partners required genital contact including but not limited to penile–vaginal intercourse; it did not require orgasm. Total Imagery was a global rating encompassing the preceding six variables on imagery and attractions and taking into consideration the frequencies of the respective experiences as reported by the interviewee. Overall sexual responsiveness was a global rating based on Total Imagery and Actual Partners.

Data Analysis

The study groups were compared on all sexual-behavior and sexual-orientation variables by overall standard parametric procedures (ANOVA), and then, for exploratory purposes, pairwise by independent-samples t-test. If, for a given comparison, any of the demographic variables (age; ethnicity; and mean parental education as an index of socioeconomic status) significantly differed between the study groups and was correlated with the outcome variable in the control group, the potentially confounding influence of the demographic variable was controlled for by including it in a regression analysis. If study groups differed significantly in variability of the outcome variables on Levene’s test for the equality of variances, weighted least square (WLS) regressions were performed for analyses requiring demographic controls; otherwise, t-tests with equal variances not assumed. In addition, we also followed the wide-spread practice of dichotomizing the Kinsey-format scales for statistical tests so as to contrast the number of K0-1 women (exclusively or almost exclusively heterosexual) with the number of K2-6 women. Four-group and two-group comparisons of these dichotomized variables were performed by way of Fisher’s Exact test and its extension to multiple groups. In view of the modest sample sizes of women with the rare syndrome of CAH, both conventionally (p < .05) and marginally (p < .10) significant results were listed. Effect sizes (Cohen’s d) of differences between CAH groups and controls were calculated using the variance of the controls, because of the commonly increased variance of CAH samples on outcome variables. All statistical analyses were conducted using SPSS for Windows Release 13.01 (December 12, 2004) or StatXact, 4.0.1 (2000).

Results

Psychosexual Milestones

Table 2 shows findings on selected items from the SEBAS-A section, Psychosexual Milestones. Substantial minorities of women in all groups experienced same-sex crushes, while fewer women experienced same-sex love and same-sex genital sex. CAH women were increased above control women in all three categories, and SW women were highest (the differences reached significance in several pair comparisons [data not shown]). The NC group was higher than the COS group on all 3 variables (significantly so for genital sex and marginally significantly for love [data not shown]). The ages at first occurrence of these states or events appeared to be relatively late. However, as Table 3 shows, this does not indicate a general delay of psychosexual milestones (heterosexual and homosexual combined) in CAH. Rather, among those women with a history of both heterosexual and homosexual experiences, the first experience tended to be heterosexual. For instance, the number of such women who first had a heterosexual crush as compared to those who first had a homosexual crush breaks down as follows: COS 1:0; NC 8:1; SV 6:0; and SW 9:4. The same holds true of the female COD group (DES-unexposed controls) with 5:3. (For the other milestone categories, women with both heterosexual and homosexual experiences were too few for statistical analysis.)
Table 2

Homosexual milestones

 

COS (n = 22)

NC (n = 81)

SV (n = 21)

SW (n = 39)

pa

Number (%) of participants with the experience

Crush

3 (14%)

13 (16%)

6 (29%)

15 (38%)

.033

Love (yes or maybe versus no)

0 (0%)

4 (5%)

1 (5%)

7 (18%)

.048

Genital sex

0 (0%)

9 (11%)

1 (5%)

6 (15%)

n.s.

Mean age (SD) in years

First crush

15.5 (–b)

15.0 (4.9)

21.8 (7.6)

13.6 (5.6)

(.057)

First love

– (–)

18.8 (2.1)

18.5 (–b)

19.1 (3.8)

n.s.

First genital sex

– (–)

23.2 (4.7)

13.5 (–b)

19.8 (7.5)

n.s.

aUnordered Row by Column Table Test (Stat Xact) for the top three variables; ANOVA (NC, SV, SW) for age at first crush; t-test (NC, SW) for the remaining two variables

bNo SD (n = 1)

Table 3

Psychosexual milestones (homo- and heterosexual combined): Mean age (SD) in years

 

COS

NC

SV

SW

p (ANOVA)

First crush

10.6 (2.3)

9.4 (2.9)

10.3 (3.0)

10.8 (2.9)

(.093)

Menarche

13.4 (1.1)

13.1 (2.1)

14.1 (2.4)

14.1 (2.9)

n.s.

First masturbation

13.9 (6.0)

13.5 (4.4)

14.2 (4.3)

15.4 (5.8)

n.s.

First genital sexa

17.6 (2.7)

17.6 (4.3)

17.3 (3.8)

18.3 (3.8)

n.s

First orgasmb

18.4 (4.5)

17.3 (5.2)

18.9 (4.2)

18.3 (6.5)

n.s.

First love

19.4 (5.3)

18.0 (3.6)

17.9 (3.6)

20.2 (4.3)

(.094)

aWith a partner, any mode

bWith or without a partner, any mode

Sexual Orientation: Lifetime

As can be seen on Table 4, four-group comparisons on the three lifetime Kinsey variables are highly significant. The data show a rather consistent progression of Kinsey score means (and also SDs), except for the SV group on Actual Partners and Overall Kinsey, and of K-2-6% for all three variables with increasing degree of androgenization in terms of 21-OHD severity, although of the 18 respective two-group comparisons of Kinsey scores only 12 were conventionally and 1 marginally significant, and of the 18 two-group comparisons of the dichotomized Kinsey scales only 7 were conventionally and 3 marginally significant (data not shown). Figure 1 shows the Overall Kinsey scores for lifetime sexual orientation, which combines both imagery and actual partner experience (using reverse scoring for the male COD group to facilitate comparisons). Effect sizes d for the Kinsey-score differences between the COS and the CAH subgroups were 1.0 for NC, .9 for SV, 3.0 for SW, and 2.3 for women with classical CAH (SV and SW combined). The effect size for the difference between the male and female DES-unexposed control groups (COD-F and COD-M) from our preceding DES study was 6.7 with the combined SD used as reference (8.1 with the women’s SD used as reference).
Table 4

Lifetime Kinsey-scale scores by subgroup: Means (SDs) and distributions

 

COS

NC

SV

SW

p

M (SD)

n

M (SD)

n

M (SD)

n

M (SD)

n

Total Imagery

0.2 (0.5)

21

0.7 (1.1)

79

1.1 (1.8)

21

1.8 (2.2)

38

<.001a

   K0–1

 

20

 

60

 

14

 

20

 

   K2–6

 

1

 

19

 

7

 

18

 

   No imagery

 

1

 

1

 

0

 

1

 

   Total

 

22

 

80

 

21

 

39

 

   K2–6/(K0-1 + K2-6)

 

5%

 

24%

 

33%

 

47%

.003b

Actual Partners

0.0 (0.0)

18

0.2 (0.6)

77

0.1 (0.4)

20

1.1 (2.3)

28

.001a

   K0-1

 

18

 

74

 

19

 

22

 

   K2-6

 

0

 

3

 

1

 

6

 

   No partners

 

4

 

4

 

1

 

11

 

   Total

 

22

 

81

 

21

 

39

 

   K2-6/(K0-1 + K2-6)

 

0%

 

4%

 

5%

 

21%

.017b

Overall Kinsey

0.2 (0.5)

22

0.7 (0.9)

80

0.6 (1.0)

21

1.7 (2.1)

39

<.001a

   K0-1

 

21

 

60

 

15

 

23

 

   K2-6

 

1

 

20

 

6

 

16

 

   No sexual responsiveness

 

0

 

0

 

0

 

0

 

   Total

 

22

 

80

 

21

 

39

 

   K2-6/(K0-1 + K2-6)

 

5%

 

25%

 

29%

 

41%

.014b

aANOVA

bUnordered Row by Column Table Test (StatXact) for K2-6 versus KO-1 across 4 subgroups

Fig. 1

Overall sexual responsiveness, lifetime, as a function of CAH severity. To facilitate comparisons of the CAH effects with the usual sex difference, DES-unexposed female controls (COD-F) from our preceding DES project have been added on the left and DES-unexposed male controls (COD-M; reverse-scored) on the right. Both individual values (graphed with jitter-function software) and means and SDs are shown

The K2-6% for women with classical CAH (SV and SW combined) was 42% for Total Imagery, 15% for Actual Partners, and 37% for the Overall Kinsey rating. The NC group was significantly higher than the COS group on all three Kinsey scores (p ≤ .003, ≤ .006, and ≤ .002, respectively). The table also shows that the Kinsey scores for lifetime Total Imagery were higher than for lifetime Actual Partners. According to Table 5, the number of women with any actual same-sex partner experience was relatively small, and the number of those with considerable experience in terms of same-sex partner numbers or same-sex occasions was even smaller. In interpreting these data, one has to take into consideration that women with classical CAH, and especially those with the SW variant, had significantly lower lifetime actual-partner numbers and lower total lifetime sex occasions (heterosexual and homosexual combined in both variables) than the other groups (data not shown).
Table 5

Homosexual partner experience, lifetime: Mean (SD) frequency category and frequency distributions

 

COS (n = 22)

NC (n = 80)

SV (n = 21)

SW (n = 38)

p (ANOVA)

Same-sex sex partners

0.0 (0.0)

0.2 (0.45)

0.1 (0.2)

0.5 (1.5)

.037

None

22

71

20

32

 

1

 

6

1

1

 

2–3

 

3

 

2

 

4–6

   

1

 

7–10

     

11–19

   

1

 

20–29

     

30+

   

1

 

% Any

0%

11%

5%

16%

 

Same-sex sex occasions

0.0 (0.0)

0.4 (1.3)

0.4 (1.7)

1.1 (2.6)

(.086)

None

22

71

20

32

 

1

 

3

   

2–3

 

3

   

4–6

 

1

 

1

 

7–10

     

11–19

     

20–29

   

1

 

30–49

     

50+

 

2

1

4

 

Sexual Orientation: Current (Past 12 Months)

The current data on sexual orientation also show Kinsey scores for all three CAH subgroups in an apparent dose–response fashion (Table 6). The gradual increase in Kinsey scores from non-CAH controls to the most severe SW variant applied to all variables, and was highly significant for all categories except for fantasies during partner sex. Again, the NC group had higher Kinsey scores than the COS group on all 9 variables (5 differences were conventionally significant and 1 marginally so), the SW group was significantly higher than the two other CAH variants (versus SV with conventional significance on 1 variable and marginal significance on 2; versus NC with conventional significance on 8 variables), but the difference between SV and NC women reached conventional and marginal significance on only 1 variable each.
Table 6

Mean Kinsey scores (SD) for the past 12 months by subgroup

 

COS

NC

SV

SW

p (ANOVA)

Masturbation erotica

0.8 (1.5)

0.9 (1.3)

3.0 (2.4)

3.4 (2.3)

.004

Masturbation fantasies

0.3 (0.8)

0.9 (1.3)

1.7 (1.9)

2.5 (2.4)

.000

Fantasies during partner sex

0.3 (0.7)

0.8 (1.3)

1.0 (2.0)

2.4 (3.0)

(.069)

Daydreams

0.0 (0.0)

0.6 (1.3)

1.0 (1.8)

2.4 (2.8)

.000

Nightdreams

0.1 (0.3)

0.5 (1.1)

1.2 (2.1)

3.2 (2.9)

.000

Attractions

0.1 (0.3)

0.3 (0.9)

0.9 (1.3)

1.6 (2.2)

.000

Total Imagery

0.2 (0.5)

0.7 (1.1)

1.1 (1.6)

1.9 (2.3)

.000

Actual Partners

0.0 (0.0)

0.1 (0.7)

0.05 (0.2)

1.4 (2.6)

.000

Overall Kinsey score

0.2 (0.5)

0.5 (0.8)

0.9 (1.3)

1.8 (2.3)

.000

Note that the n’s vary somewhat between items because not all women have experience with all of them

Sexual Orientation and Non-sexual Gender-related Behavior

The global lifetime and 12 months Kinsey scores were correlated with selected global variables of gender-related behavior (not including sexual orientation), as listed in Meyer-Bahlburg et al. (2006). The expected correlations were significant and in the predicted direction, but of modest size, and stronger for childhood measures than adulthood measures. For instance, the Overall Kinsey score for lifetime correlated r = −.40 with the Gender scale of the Recalled Childhood Gender Questionnaire-Revised (RCGQ-R), but the Overall Kinsey score for the past 12 months correlated only r = −.27 with the (adult) Hobby Preferences Scale (the high end on both gender scales is feminine).

Cross-gender Imagery

At the end of the imagery section, the participant was asked about the frequency with which they saw themselves as “a person of the opposite sex” in their erotic imagery, separately for the past 12 months and lifetime (since puberty, excluding the past 12 months). As Table 7 shows, all groups except SV included some women with such experience. The percent of women with such experience and the frequency of having that experience, especially for lifetime, was significantly increased in the SW subgroup above all other groups.
Table 7

Seeing self as man during romantic/erotic imagery, by subgroup

 

COS

NC

SV

SW

p (ANOVA or Exact Test)

Mean frequency (categorized), lifetime

0.1 (0.3)

0.1 (0.5)

0.0 (0.0)

0.4 (0.8)

.024

Mean frequency (categorized), past 12 months

0.1 (0.4)

0.1 (0.4)

0.0 (0.0)

0.3 (0.8)

(.074)

% Women, lifetime

10%

6%

0%

21%

.030

% Women, past 12 months

5%

6%

0%

16%

n.s.

Predicting Sexual Orientation from Prenatal Androgenization and Gender Variables

Using hierarchical stepwise regression, we tried to predict the Overall Kinsey Score for the past 12 months successively from the degree of prenatal androgenization in terms of the four CAH-severity groups, childhood gender-related behavior (the Gender scale of the RCGQ-R), adult gender-related behavior (the sum score of the Hobby Preferences Scale), and Seeing Self as Man during Romantic/Erotic Imagery, Frequency: Lifetime (Excluding the Past 12 Months). Prenatal androgenization and RCGQ-R Gender contributed significantly to the variance accounted for and were significant in the final model, but Hobby Preferences and Seeing Self as Man was not (Table 8).
Table 8

Prediction of overall Kinsey score (past 12 months) from degree of prenatal androgenization and gender variables, using stepwise hierarchical regression

 

R2 change

p

Degree of prenatal androgenizationa,b

.130

.000

RCGQ-R Gender scaleb

.062

.002

Hobby Preferences scaleb

.015

.115

Seeing Self as Man During Romantic/Erotic

  

Imagery, Lifelong (excluding past 12 months)c

.014

.135

Final model

 

β

p

Degree of prenatal androgenizationa,b

−.219

.028

RCGQ-R Gender scaleb

−.386

.001

Hobby Preferences scaleb

.167

.144

Seeing Self as Man During Romantic/Erotic Imagery, lifelong (excluding the past 12 months)c

.118

.135

aDegree of prenatal androgenization coded SW = 1, SV = 2, NC = 3, COS = 4

bHigh score = feminine

cHigh score = masculine

Role of the Molecular Genotype

Finally, we wanted to test whether the CAH subtypes as defined by endocrine criteria or their classification by molecular genotype is more closely associated with sexual orientation. We employed the widely used severity classification of molecular genotypes of the 21-OH gene (White & Speiser, 2000), which is based on the in-vitro synthesis of 21-OH expressed as % of synthesis by the intact genotype. It yields four groups, listed here in ascending severity and with the major molecular genotypes: (1) Genotypes V281L or P30L (20–50% or 30–60% 21-OH; all patients are NC); (2) I172N (1+% 21-OH; 10% of patients are SW); (3) I2 splice (0+% 21-OH; most patients are SW); (4) “Null”/gene deleted (0% 21-OH; all patients are SW). In compound heterozygotes the classification is based on the allele with the milder mutation. We had severity-classifiable genotypes for 28 SW women, 17 SV women, and 79 NC women. Using stepwise hierarchical regression, we either entered the molecular genotype classification first, and the clinical endocrine classification second, or vice versa. As Table 9 shows, neither procedure indicated that the variable entered second added significant predictive variance to sexual orientation outcome. Thus, the two classifications did not differ in their association with the Kinsey score. In this particular subsample, the endocrine and genetic classifications correlated Pearson r = .90 with each other, i.e., there was little room for divergence between molecular genotype and endocrine phenotype.
Table 9

Prediction of overall Kinsey score (past 12 months) from endocrinea and geneticb classification of CAH, using stepwise hierarchical regression

 

R2 change

p

Genetic entered first, clinical second

Genetic

.100

.000

Clinical

.007

.356

Clinical entered first, genetic second

Clinical

.102

.000

Genetic

.004

.453

aEndocrine classification coded SW = 1, SV = 2, NC = 3

bGenetic classification coded by molecular genotypes: V281L or P30L = 1; I172N = 2, I2 splice = 3, Gene deleted = 4

Intercorrelations among Sexual Orientation Variables

For methodological purposes, we constructed an intercorrelation matrix of all 12-month Kinsey variables. All of the Kinsey variables were quite highly intercorrelated, with Pearson r ranging from .50 to .92 (median = .78). A principal components analysis of all these variables (except for Masturbation Erotica which had too few women with the relevant experience and excluding the two global variables, Total imagery and Overall Kinsey score) yielded one general factor accounting for 77.2% of the variance. The factor loadings for the individual variables were: Daydreams .96, Attractions .93, Masturbation Fantasies .92, Fantasies During Partner Sex .88, Night Dreams .81, and Actual Partners .76. On the basis of the principal components analysis, we constructed a 6-item unit-weighted scale of sexual orientation. This scale was correlated with the interviewer rating of overall sexual responsiveness during the past 12 months (at the end of the sexual-orientation section) with Pearson r = .94 (p < .001).

Discussion

Our data clearly showed increased sexual orientation towards females (i.e., bisexuality and homosexuality) in women with classical CAH (SV and SW combined) compared to non-CAH controls. Note that our data constitute underestimates: The two 46,XX individuals with SW-CAH who were excluded from analysis for technical reasons were both living with female partners at the time of the evaluation and had a history of sexual responsiveness to females. Thus, our study corroborates earlier findings on sexual orientation in CAH. We also clearly replicate earlier demonstrations of increased bisexual/homosexual orientation in SW women compared to SV women, and earlier reports indicating that these shifts are more strongly accentuated in romantic/erotic imagery than actual sex-partner experiences.

More surprising is the finding of increased bisexual/homosexual orientation in NC women above controls in the diverse variables evaluated here. This finding is in line with our earlier report on mild, but significant shifts of the same NC women towards masculinized gender-related behavior other than sexual orientation (Meyer-Bahlburg et al., 2006). In the endocrine literature, NC is usually described as a syndrome characterized by onset of clinical (somatic, physiological) symptoms of androgen excess after birth, in childhood or later. Since there is consensus that the masculinization of gender-related behavior in classical CAH is due to the effect of prenatal androgens on the developing brain, these behavioral shifts in NC women were not expected. The finding raises the question whether the mild androgen excess that is likely to be present in NC fetuses from the first trimester on, but is insufficient to noticeably affect the sexual differentiation of the genitalia, is nevertheless sufficient to slightly affect the sexual differentiation of the brain. Alternatively, the data suggest an unexpected postnatal effect of mild but persistent androgen excess on brain and gender-related behaviors. Our study does not provide data that would help us to argue in favor of one or the other explanation.

In conjunction with the other CAH subgroups and the control women, our NC data further strengthen the notion of a dose–response relationships of androgens with sexual orientation, at least on the subgroup level, given that the other hormone abnormalities seen in the CAH syndrome (e.g., deficiency of cortisol and aldosterone, excess of ACTH and 17-hydroxyprogesterone) are not known to be associated with masculinization of gender-related behavior in animals or humans (although specific studies of this kind are yet to be conducted). As we had found analogous relationships with CAH severity for non-sexual gender-related behavior (Meyer-Bahlburg et al., 2006), we also could confirm significant, but modest-sized correlations of sexual orientation scores with non-sexual gender-related behaviors, which had been shown by others (e.g., Hines et al., 2004). Such findings are also in line with an understanding of sexual orientation in the context of sexual differentiation. Given the many differences in sexual orientation and associated variables between men and women, the question arises, whether the increased sexual orientation towards females associated with CAH severity in 46,XX individuals is a model of the role of androgens in sexual-orientation development in males rather than lesbian women. Early developmental research on non-sexual sex-dimorphic behavior in animals sought to explain sex differences in behavior (e.g., Beatty, 1979), and experimental manipulations of pre- and perinatal androgen levels served to show to what extent one could create male-typical behavior in females so treated. Later, such findings were also used to explain interindividual variations of gender-related behavior among females (e.g., Hines, 2004), and there is some supportive evidence for this approach in humans. That this may apply also to sexual orientation in at least a subgroup of women is suggested by the fact that earlier research has repeatedly shown that about one third of homosexual women have (modestly) increased levels of androgens (Meyer-Bahlburg, 1979).

One of the major limitations of the interpretation of findings in classical CAH in relation to animal studies is the hormone treatment that females with classical CAH typically receive throughout postnatal life. The classical animal study of sexual differentiation of brain and behavior exposes the female fetus to androgen treatment during the known early hormone-sensitive period of sexual differentiation of the brain and then again around the time of puberty/young adulthood. That combination tends to maximize the development of cross-gender sexual behaviors, especially when ovariectomy after puberty preceded the treatment with testosterone propionate, as shown by Dörner (1976, p. 164) in the rat. By contrast, human 46,XX individuals with CAH are initially exposed to excess, endogenous androgens from their adrenals during the presumed hormone-sensitive prenatal period of sexual differentiation of the brain, and in the severe variants at even higher levels than normal males and more chronically so (Carson et al., 1982; Pang, Levine, Chow, Faiman, & New, 1979; Pang et al., 1980, 1985; Wudy, Dörr, Solleder, Djalali, & Homoki, 1999); however, from birth on, the excess androgen levels are suppressed by glucocorticoid replacement therapy (sometimes to levels even lower than is normal for healthy females), the ovaries are left in place, and female puberty is induced by an endogenous, largely normal female hormonal milieu. Unfortunately, there are no behavioral studies of non-human primates that mimic the androgen history of 46,XX humans with CAH, so that we cannot be sure if we should expect to see much more bisexuality and homosexuality in 46,XX individuals without postnatal androgen suppression. The early study by Ehrhardt et al. (1968) on late-treated women with CAH who did not have glucocorticoid treatment available before later adolescence or adulthood does not show massive increases in bisexuality or homosexuality compared to our data, but most of those early studied women must have had milder (non-SW) variants of CAH implying relatively less prenatal androgen excess, or they would not have survived. Also the fact that the effect size for sexual orientation in SW as the most severe CAH variant does not reach even half the effect size for the sex difference despite chronically high prenatal androgen levels may be related to the other male-atypical features of 46,XX CAH development and/or reflect the absence of Y-based genes in the 46,XX brain, which are suspected of participating in the sexual differentiation of the brain in other vertebrates (Arnold, 2002; Gatewood et al., 2006).

Another potentially confounding factor is the fact that insufficient glucocorticoid replacement or treatment interruptions lead to virilization, i.e., somatic symptoms of androgen excess, and, if occurring early enough during childhood development, also to stunting of growth, while overtreatment brings about variable degrees of obesity, all of which may reduce attractivity to men and romantic approaches by men, and to related body-image concerns of the CAH women. This raises the question whether associated inhibition of romantic practice in adolescence, perhaps also supported by less-stereotypic feminine leisure time activities and, at least for some CAH women, relative isolation in the peer group, might increase the chance of bisexual/homosexual development rather than exclusively a direct effect of androgens on brain circuits that regulate sexual behavior. If so, this might be an example of the interaction of social and biological factors in the development of bisexuality/homosexuality, for which Bearman and Brückner (2002) argued.

In our data set, the clinical-endocrine classification of CAH was highly correlated with the classification based on molecular genetics, and knowledge of one did not add to the predictive power of the other in terms of gender outcome. Thus, there is no suggestion that the molecular-genetic defect influences sexual-orientation outcome through any other physiological route than the hormonal abnormalities caused by the degree of enzyme deficiency.

It is noteworthy that our data suggest an influence of CAH severity on cross-gender identity in sexual situations. Our data here again represent an underestimate because of the exclusion of two SW patients from the current data analysis, one having undergone gender change to male, the other considering it (see above). Perhaps, gender-atypicality in non-sexual behavior along with later emerging atypical sexual orientation facilitate in some CAH women an identification with selected aspects of the male role, which may subsequently broaden and thereby lead to late overall gender identity change, as it has been documented to occur in some 46,XX individuals with CAH (Meyer-Bahlburg et al., 1996).

Our data show that current sexual orientation is not only predicted from the degree of prenatal androgen exposure as indicated by the CAH-severity classification, but in addition also from the degree of masculinization of gender-related behavior during childhood. The latter variable could reflect variable brain responsiveness to prenatal androgens as well as postnatal psychosocial influences, provided retrospective reporting bias can be ruled out. This study does not provide an opportunity to decide between these options.

The association of CAH severity with outcome appears in all aspects of sexual orientation assessed in our protocol and plausibly explains the strong general factor that emerged from the principal-component analysis. This finding, along with the high correlation of the resulting summary scale with the global interviewer rating of overall sexual responsiveness, suggests the feasibility of a simplified self-report version of a multi-item sexual orientation assessment.

The current data set on sexual orientation suffers from the same overall limitations as the previous one on gender development (Meyer-Bahlburg et al., 2006), namely, small sample size, questionable representativeness, and cross-sectional design. However, our findings on a dose–response relationship of androgens and sexual orientation appear even stronger than for non-sexual gender-related behavior and make a persuasive case for the extension of this association to women with NC CAH. Overall, our findings support a sexual-differentiation perspective involving prenatal androgens on the development of sexual orientation.

Acknowledgments

The project described was supported in part by USPHS Grants HD-38409, RR06020 (GCRC), and by Grant Number U54 RR01-9484 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH). Its contents are solely the responsibilities of the authors and do not necessarily represent the official views of NCRR or NIH. We thank the study women for their participation in this research. Ms. Rhoda Gruen served as interviewer trainer. Ms. Patricia Connolly assisted in word processing.

Copyright information

© Springer Science+Business Media, LLC 2007

Authors and Affiliations

  • Heino F. L. Meyer-Bahlburg
    • 1
  • Curtis Dolezal
    • 1
  • Susan W. Baker
    • 2
  • Maria I. New
    • 2
  1. 1.New York State Psychiatric Institute & Department of PsychiatryColumbia UniversityNew YorkUSA
  2. 2.Department of PediatricsMount Sinai School of MedicineNew YorkUSA

Personalised recommendations