Twin Studies and Quantitative Genetics in Premature Ejaculation Research

Fig. 10.1

A simplified univariate ACE model. A is the additive genetic component of the trait, and is set at 1.0 for MZ twins, and 0.5 for DZ twins. C is the shared environmental component, and is set to 1.0 for both MZ and DZ twins. E is the non-shared environmental component and is by definition unique for each twin individual; hence no covariance between twins is measured for the E component. Latent variables are illustrated with circles (A, C and E); the observed variable is illustrated with a rectangle (Phenotype). Causal paths are illustrated with single-headed arrows and covariance paths with double-headed arrows

Few studies have investigated PE using twin or family data; to my knowledge, PE has only been investigated in two twin populations. The first of these studies was conducted using data from a population-based sample of nearly 4,000 twins and siblings of twins in Finland [5, 7, 8]. This research group found a significant A component explaining about 28 % of the variance in PE [5]. No significant effects of shared environment were found, and thus the C component could be omitted without significant reduction of model fit. In other words, most of the variance in PE (72 %) was explained by E effects. PE was estimated using a composite score based on self-report questionnaire items inquiring about, for example, ejaculation latency time, frequency of occurrence of anteportal ejaculation, and subjective perception of ejaculatory control. Recently, a research group in Hungary [9] investigated PE using twin data, however the authors of this study noted that their estimates of concordance rates for PE between MZ and DZ twins were not trustworthy due to significant misfit of the statistical model (likely due to small sample size). PE was estimated using a dichotomous query (yes/no) and was defined as “ejaculation before introitus or as occurring when a lack of ejaculatory control interferes with sexual and emotional wellbeing in one or both partners” (p. 147). Although a valid model could not be established in the study, the concordance rates between MZ and DZ pairs suggested a heritable component.

10.1.1 Multivariate Models

Using bi- or multivariate models, one can calculate genetic and environmental correlations between two or more traits in addition to the heritability estimates that are calculated with a standard, univariate ACE model. A genetic correlation is an estimate of the similarity of the underlying A effects of two (or more) traits. In other words, if a genetic correlation is very high, it can be assumed that the same genes contribute to the A component in both traits. To provide an example, I computed heritability estimates as well as genetic and environmental correlations using two phenotypic variables commonly used as indicators of PE in the literature: subjective perception of ejaculatory control, and self-reported ELT. Both variables were responded to on a five-point Likert scale. The variable measuring ejaculatory control (How often have you felt that you could decide when to ejaculate?) had the following response options: (1) never or rarely; (2) less than 50 % of the time; (3) about 50 % of the time; (4) more than 50 % of the time; (5) almost always or always. The variable measuring ELT (On average, during vaginal or anal intercourse, how much time elapses between when you first enter your partner with your penis and when you first ejaculate?) had these response options: (1) less than one minute; (2) 1–5 min; (3) 5–10 min; (4) more than 10 min; (5) I usually do not ejaculate. The questionnaire is more thoroughly elaborated elsewhere [5, 7, 8]. The analysis was conducted using responses from 225 MZ and 213 DZ pairs (876 individuals in total), and its results can be seen in Fig. 10.2. A bivariate Mx script for ordinal data was fitted to the data set.

Fig. 10.2

A bivariate AE model illustrating genetic (A) and non-shared environmental (E) effects on ejaculatory control and ejaculation latency time, as well as genetic (r _g) and non-shared environmental (r _e) correlations. The shared environmental component C has been dropped due to insignificant impact on either phenotype. A and E effects are indicated with single-headed arrows, whereas genetic and non-shared environmental correlations are indicated with double-headed arrows

In the parameter estimation procedure, the shared environmental component C was estimated to 1.16 × 10⁻¹¹ for ejaculatory control and 1.25 × 10⁻¹¹ for ELT (i.e., virtually zero). In other words, C had virtually no influence on either phenotype, thus C could be dropped without reducing model fit, leaving us with a more parsimonious AE model. As seen in Fig. 10.2, about 8 % of the variance in ejaculatory control, and 25 % of the variance in ELT, were under genetic control (the remaining 92 and 75 % of the total phenotypic variance, respectively, accounted for by non-shared environment). The genetic correlation between the A components, however, was quite high at r _g = 0.88, indicating that largely the same genes influence both ejaculatory control and ELT. The non-shared environmental correlation, on the other hand, was rather modest (r _e = 0.22) indicating that quite different environmental factors contribute to ejaculatory control and ELT. In short, both ejaculatory control and ELT are under quite modest genetic influence. However, the minority share of the phenotypic variance in both traits that is under genetic control is likely affected by largely the same genes.

Only gold members can continue reading. Log In or Register to continue