Abstract
The authors discuss previous research showing differences in physiologic/genital versus subjective/cognitive sexual arousal in women, as well as differences in the areas of the brain activated in women versus men [1,2]. In addition, a significant proportion of women do not report distress although they meet other criteria for sexual dysfunction [3]. All of this information suggests sexual dysfunction may manifest as many different symptoms in various women. As such, it seems reasonable to hypothesize that female sexual dysfunction (FSD) may be related to various causative factors.
The authors point out one previous small similar study that demonstrated a genetic contribution to FSD [4]. Two other studies demonstrated a modest genetic effect on orgasm frequency [5,6]. One important difference in the current study is the focus on lifetime sexual function [7]. In addition, this study used a multivariate twin model to examine the covariation of FSD dimensions between genetic and environmental components.
Methods
Participants were drawn from the TwinsUK registry, which consisted of unselected volunteers registered since 1993. These twins were comparable in disease prevalence, and demographic and lifestyle characteristics to age-matched British women who were representative of the general population in range of sexual lifestyles and behaviors. Zygosity was determined by a questionnaire or DNA genotyping. Data were collected in 2008 and 2009 using the Female Sexual Function Index-Lifelong (FSFI-LL) questionnaire, which was sent to 3175 individuals with a 50% response rate. Respondents were excluded for homosexuality or for omitting more than five of the 19 items in the FSFI-LL or more than two items in the Female Sexual Distress Scale. Any other missing data were imputed by item-specific means of the nonmissing values, and were separated into four age groups (18–30, 31–45, 46–55 and 56–85 years). In total, 1489 respondents (41.8%) were included in the analysis, with 244 monozygotic pairs (MZ), 189 dizygotic pairs (DZ) and 623 women whose co-twins did not participate.
The FSFI-LL was used to assess lifelong sexual functioning. It assesses six dimensions of women's sexual functioning since they became sexually active including two items measuring desire, four items for arousal, four items for lubrication, three items for orgasm, three items for satisfaction and three items for pain with responses indicated on a Likert-type scale. Lower scores suggest greater sexual problems, and higher scores suggest fewer. A cut-off score of 26.55 is used to determine dysfunction from healthy function [8].
Data were entered and analyzed with Stata (version 10, StataCorp LP, TX, USA). Mean differences in age and FSFI-LL scores between MZ and DZ twins were compared with unpaired two-tailed t-tests. All phenotypes except desire and arousal were either log- or square root-transformed, and standard age corrections were applied. Univariate genetic modeling was performed, which involved breaking the phenotypic variance of FSFI-LL scores into additive genetic, dominant genetic, common environmental and unique environmental components. The goodness of fit of the genetic models was evaluated, and Akaike's Information Criterion, a parsimony fit index, was used to choose the best-fitting models.
Multivariate modeling, to examine covariation of FSD dimensions into genetic and environmental factors, involved cross-twin cross-trait correlations to obtain an initial impression, and use of three multivariate statistical models to formally quantify results. Initial correlations suggested including desire, arousal, lubrication and orgasm in the formal model analysis, while not including satisfaction or pain, due to lower phenotypic intercorrelations. The three statistical models used were Cholesky decomposition, which shows the correlations between the four independent genetic and environmental factors and breaks down the variance of the phenotypes into separate additive genetic and nonshared environmental effects: independent pathway, which examines whether the covariance between desire, arousal, lubrication and orgasm can be explained by a single shared genetic and shared environmental factor (meaning all phenotypes share common etiological factors); and common pathway which would suggest a latent phenotype underlies FSD. Again, likelihood χ2 tests and the Akaike's Information Criterion were used to determine goodness of fit.
Results
The mean age of participants was 56.3 years, and there were no significant differences between MZ and DZ twins in age and FSFI-LL scores, suggesting the twin groups were well matched. In addition, no significant differences in sexual functioning were found between women who had provided full data compared with women whose missing values were imputed.
MZ twin correlations were greater than twice the DZ correlations, suggesting genetic effects in every FSD dimension. MZ twin correlations also substantially deviated from unity, suggesting nonshared environmental influences also contribute. The best model for all dimensions showed only additive genetic and nonshared environmental influences. Additive genetic effects accounted for 22% (sexual satisfaction) to 39% (overall FSD) of variance. Nonshared environmental factors accounted for 61% (overall FSD) to 78% (sexual satisfaction) of variance.
In the multivariate model-fitting analysis, the Akaike's Information Criterion index of parsimony showed the Cholesky model to fit best, which was then used to calculate both genetic and environmental factor correlations between FSD dimensions (additive genetic correlation [rA]) [7]. It showed genetic correlation of arousal with desire (rA = 0.86), as well as with lubrication (rA = 0.9), and orgasm (rA = 0.92). Desire showed smaller correlations with lubrication (rA = 0.6) and orgasm (rA = 0.59). Genetic differences were not significantly different, since 95% CIs overlapped. Bivariate heritabilities ranged from 35 to 44%, implying approximately one-third of covariance between the dimensions was due to additive genetic factors and the remaining covariance due to nonshared environmental factors. The influence of common shared environmental factors was negligible.
The model produced two relatively weak genetic factors (A1 and A2). A1 was shared by all phenotypes and represented 7% (lubrication) to 33% (desire) of the variation. A2 accounted for 7% (arousal), 10% (lubrication) and 17% (orgasm) of the variation. Four nonshared environmental factors (E) were identified, and they weighed more heavily on the variation, accounting for up to 65% for factor E1 on desire. Three of the four factors influenced all four dimensions to some degree, but they seemed somewhat dimension specific (E1 for desire at 65%, E2 for arousal at 55% and E3 for lubrication at 56%).
Discussion
These results showed that all dimensions of FSD are somewhat attributable to genetics (about one-third of the variance), and to a greater degree, to nonshared environmental factors. The data analysis did not suggest a relationship between common environmental factors and FSD. The largest correlations overall between FSD dimensions were between desire and arousal, arousal and lubrication, arousal and orgasm, and lubrication and orgasm. This analysis implies these dimensions were influenced by both similar genetics and similar nonshared environmental factors (such as similar difficult events). Correlations between desire and lubrication, and desire and orgasm were lower.
Overall this analysis suggests etiological heterogeneity to FSD. Most importantly, it indicates a difference between FSD dimensions that are more psychological, such as desire, in contrast to those that are both psychological and physiological, such as arousal, and those that are mostly physiological, such as lubrication and orgasm. This leads to the conclusion that an individual woman's sexual dysfunction may be caused to some degree by genetics, but is influenced more by unique biological, psychological or other environmental factors.
Conclusion
This paper clearly demonstrates that FSD is heterogeneous, caused by a genetic substrate for FSD that is influenced significantly by environmental factors. Future research can be targeted toward correlating specific etiologic factors with individual FSD dimensions, then diagnostic categories can be further refined, and research can be targeted toward developing treatments for separate causative factors based on presenting symptoms.
Future perspective
This work elaborates on previous genetics studies regarding potential etiology of FSD and pathophysiology in both humans and animals. A review by Clayton shows FSD to be multifactorial, with hormonal, neurobiological, and psychosocial contributions [9]. Pfaus also demonstrated the complicated potential contributions of neurobiology in his discussion of certain neurotransmitters that contribute to sexual excitation, as opposed to others that contribute to sexual inhibition [10]. Additionally, a trial of bupropion (physiologic intervention) for the treatment of hypoactive sexual desire disorder suggested that intrapsychic and interpersonal barriers (environmental factors) to improvement must also be addressed [11], indicating that all etiologic factors must be managed for successful outcomes.
Executive summary
Two genetic factors influence female sexual dysfunction (FSD; ~30% of variance):
– One genetic loci impacts all four dimensions;
– The other loci impacts only three dimensions.
Environmental factors account for the remainder of the variance (70%):
– Different dimensions of FSD are affected by separate nonshared environmental factors.
Specific genetic factors are associated with specific dimensions of sexual function.
Environmental factors modify expression of the genotype to equal the phenotype (nature and nurture).
The interaction between genetic and environmental factors explains the heterogeneity of FSD.
FSD should be classified into several distinct categories. In particular, desire should be distinct from arousal/lubrication.
Future research can focus on narrowing diagnostic categories within the scope of FSD, and targeting treatment toward separate causative factors.
Footnotes