- Top of page
- CONFLICT OF INTEREST
Premature ejaculation (PE) is widely acknowledged as the most commonly reported sexual dysfunction in men . Whereas significant improvements to the understanding of the biological aetiology of PE have been achieved in the last two decades, much is still inconclusive regarding the aetiology of the condition, and treatment options are at early stages with somewhat random success .
Although a hereditary aetiology to PE was already suggested in the 1940s , definite evidence for a genetic component in PE could not be established until a recent series of twin studies found that around 30% of the total phenotypic variance in PE is attributable to heritable factors [2,4–6]. Since then, molecular genetic studies have attempted to identify polymorphic regions in predominantly serotonergic [7–10] and dopaminergic [11,12] genes that could explain some of the genetic variation in ejaculatory function. Most of these studies have focused on a polymorphism in the promoter region of the serotonin transporter gene (5-HTTLPR), with equivocal results. Janssen et al.  reported no difference in terms of allele frequencies between a group of PE patients and controls, but found a significant difference in intravaginal ejaculation latency time (ELT) within the group of PE patients, so that carriers of the ‘short’ allele had significantly shorter intravaginal ELTs. A Turkish study , on the other hand, reported significantly shorter intravaginal ELTs for ‘long’ allele carriers. However, it should be noted that the allele frequencies in the population of the latter study differed significantly from what would be expected under Hardy–Weinberg equilibrium, indicating that the results should be interpreted with some caution . Additionally, Safarinejad  reported that carriers of the ‘long’ allele of this polymorphism responded significantly better to sertraline (a selective serotonin reuptake inhibitor) treatment than controls. Further evidence for the involvement of serotonergic genes was provided in a study on Han Chinese subjects, which revealed an association between two different polymorphisms of the 5-HT2C serotonin receptor, with carriers of the –759T and –697C genotypes having elevated odds of PE . A subsequent study conducted by Jern et al. , also focusing on serotonin receptor single nucleotide polymorphisms (SNPs), failed to replicate the findings of Luo et al. . However, Jern et al.  found associations between self-reported ejaculation latency time and two SNPs linked to the 5-HTR1B receptor gene (one of these associations did not remain significant after correction for multiple testing), with the rs11568817 locus having the most pronounced association. At least two studies have, to our knowledge, focused on the dopamine transporter gene, also generating equivocal results. In the first of these, Santtila et al.  reported that individuals who were homozygous for the 10-repeat allele of a variable tandem repeat polymorphism in the dopamine transporter gene DAT1 had significantly elevated scores on a composite variable measuring different aspects of PE, whereas Safarinejad  found in his study that the 9-repeat allele was more prevalent among PE patients.
Serotonergic and dopaminergic neurotransmission aside, there has been evidence for several decades for the involvement of the neuropeptide oxytocin (OXT) in the ejaculatory process. Oxytocin is released at orgasm, which is shown by elevated levels of plasma OXT at the time of ejaculation , and OXT has also been shown to increase contractility in the ejaculatory tissues, including the epididymis, vas deferens and prostate [16–20]. However, it has been proposed that OXT has a facilitatory role on tissue contraction through steroidogenesis, rather than a direct contractile effect . Furthermore, the closely related neuropeptide arginine vasopressin (AVP) has very similar contractile effects on ejaculatory tissues to those of OXT , and there is evidence that the contractile effects of OXT on mammalian tissue are mediated via arginine vasopressin 1A receptors (AVPR1A) and not via OXT receptors (OXTR) . The OXT also increases sperm numbers in both man  and various mammals [21,22].
Clément et al.  showed that intracerebroventricular administration of an OXT antagonist in the rat had a dose-dependent inhibitory effect on sexual responses (i.e. both ejaculatory and erectile) induced by a dopamine agonist, implying that OXT may also act centrally on the ejaculatory response acting as a mediator on dopaminergic neurotransmission. When delivered intrathecally at the sixth lumbar segment, the OXT antagonist impaired ejaculation – but not erection – providing a very interesting complement to the evidence for the presence of an ‘ejaculation generator’ (a group of lumbar spinothalamic neurons that, when severed, will completely disrupt ejaculation but has no other apparent effects on sexual behaviour in rats) in the spinal cord . In humans, a recent case study found that anorgasmia in an elderly male could be successfully treated using nasally administered OXT . The OXT neurons are also related to serotonergic effects on ejaculation, as selective serotonin reuptake inhibitor-induced delayed ejaculation appears to be regulated by a gradual desensitization of serotonergic 5-HT1A receptors on OXT neurons .
To date, no study has investigated the effects of genetic polymorphisms in OXTR and AVPR genes on ejaculatory function. Based on the evidence for OXT and AVP involvement in the ejaculatory process, we hypothesized that SNPs in these regions would have a significant effect on ejaculatory function. As sexual desire  and frequency of sexual activities  have been found to affect ejaculatory function, we decided to control for the effects of these along with the effects of age. To test our hypothesis, the effects of six AVPR1A, one AVPR1B and twelve OXT receptor gene (OXTR) SNPs.
- Top of page
- CONFLICT OF INTEREST
The present study involved 1517 male twins and non-twin brothers of twins (mean age 26.43 years, sd 4.87, range 18–45). The participants were a subset from the second data collection of the Genetics of Sex and Aggression sample carried out in 2006, which is a population-based sample of young adult Finnish twins and their siblings (see  for a more detailed description of the sample). Saliva samples for DNA analyses were available from all participants. Each SNP had valid data from at least 1491 individuals (i.e. for each of the 19 SNPs, there were valid genotype data for between 1491 and 1517 individuals). A total of 1274 individuals had valid genotyping data and valid information for all phenotypic instruments. All participants provided informed consent. The research plan was approved by the Ethics Committee of the Abo Akademi University.
Four different self-reported indicators were used to measure ejaculatory function: Ejaculation Latency Time, Number of Thrusts, Anteportal Ejaculation and Feeling of Control. Descriptive statistics regarding the items measuring ejaculatory function are presented in Table 1. Participants were asked to consider their responses to these questionnaire items over a period of the last 2 years. Participants who had not engaged in partnered sexual activity during this time were excluded from further statistical analyses. The indicator variables were subjected to factor analyses to determine their reliability (see [5,11] for a more detailed description of this procedure). Factor analyses conducted on these indicator variables  have shown good factorability (Kaiser–Meyer–Olkin measure of sampling adequacy 0.62, Bartlett's test of sphericity chi-squared (6) 413.07, P < 0.001), with all items loading clearly on one factor containing 42.76% of the variance. A composite score measuring ejaculatory function was subsequently computed by summing the raw variables (the variable measuring anteportal ejaculation was reflected before this operation because it had a different direction compared with the other variables). As a composite score can be expected to have superior reliability to any individual indicator variable, we first tested the effects of all SNPs against the composite score. Next, we tested any SNPs with a significant association with the composite score against individual PE indicator variables to elucidate what aspect of ejaculatory function was affected by the SNP.
Table 1. Descriptive statistics for the individual variables measuring ejaculatory function
|Variable/Description||Mean (sd)*||Options and frequencies (n, valid %)|
|Ejaculation Latency Time||3.18 (0.85)||1) <1 min (31, 2.0)|
|On average, during intercourse, how much time elapses between when you first enter your partner (vaginally or anally) with your penis and when you first ejaculate?||2) 1–5 min (297, 19.3)|
|3) 5–10 min (513, 33.4)|
|4) >10 min (451, 29.4)|
|5) I usually do not ejaculate (20, 2.0)|
|Missing percentage (213, 13.9)|
|Number of Thrusts||3.80 (0.53)||1) No thrusts at all (7, 0.5)|
|How many thrusts have you typically been able to perform before ejaculation?||2) 1–5 thrusts (19, 1.2)|
|3) 6–10 thrusts (53, 3.5)|
|4) >10 thrusts (1206, 78.6)|
|5) I usually do not ejaculate (22, 1.4)|
|Missing percentage (228, 14.9)|
|Anteportal Ejaculation||1.19 (0.50)||1) Never or very rarely (1126, 73.4)|
|In what proportion of sexual experiences do you involuntarily ejaculate before intercourse has started?||2) Less than 50% of the time (142, 9.3)|
|3) Around 50% of the time (31, 2.0)|
|4) More often than 50% of the time (6, 0.4)|
|5) Almost always or always (5, 0.3)|
|Missing percentage (225, 14.7)|
|Ejaculatory Control||2.83 (1.18)||1) Never or very rarely (212, 13.8)|
|How often have you felt that you could decide when to ejaculate?||2) Less than 50% of the time (320, 20.8)|
|3) Around 50% of the time (317, 20.7)|
|4) More often than 50% of the time (349, 22.7)|
|5) Almost always or always (117, 7.6)|
|Missing percentage (220, 14.3)|
To control for sexual desire, six items from the Sexual Desire Inventory  were used. This scale has been shown to have good internal consistency (Cronbach's α= 0.77; ). To create a composite score measuring sexual desire, the six items of the scale were summed and divided by their number.
We used three items from the modified Desired and Actual Sexual Activity Scale , a modification of the Derogatis Sexual Functioning Inventory , to investigate frequency of partnered sexual activities. Participants were asked to indicate how frequently they engaged in oral, vaginal and anal sex using a nine-point scale (0 = not at all, 1 = less than once per month, 2 = once or twice a month, 3 = once per week, 4 = two or three times per week, 5 = four to six times per week, 6 = once a day, 7 = two or three times per day, 8 = four or more times per day).
For DNA extraction and genotyping saliva samples were collected using OrageneTM DNA self-collection kits that were posted to the participants and returned by them by mail. The participants were instructed to follow the manufacturer's instructions in collecting the samples (DNA Genotek, Inc., Kanata, ON, Canada) and to deposit approximately 2 mL saliva into the collection cup. When an adequate sample was collected, the cap was placed on the cup and closed firmly. The collection cup is designed so that a stabilizing solution from the cap is released when closed. This solution mixes with the saliva and stabilizes the saliva sample for long-term storage at room temperature or in low-temperature freezers. Genotyping of SNPs was conducted by KBioscience in the UK (http://www.kbioscience.co.uk) using the KASPar chemistry, a competitive allele-specific PCR SNP genotyping system performed with FRET quencher cassette oligos (for details regarding the laboratory and its equipment, please visit http://www.kbioscience.co.uk/). The SNPs analysed in the present study are presented in Table 2 and Fig. 1.
Table 2. Descriptive statistics for the single nucleotide polymorphisms (SNPs) in the oxytocin and arginine vasopressin 1A/1B receptor genes
|RS number||SNP||Minor allele frequency in sample population (%)||Common homozygotes, n (%)||Heterozygotes, n (%)||Rare homozygotes, n (%)|
|Oxytocin receptor SNPs|| || || || || |
| rs237897||A/G||A: 49||387 (25.8)||752 (50.2)||360 (24.0)|
| rs2268493||C/T||C: 40||552 (36.5)||723 (47.8)||236 (15.6)|
| rs75775||G/T||T: 24||867 (57.8)||553 (36.9)||80 (5.3)|
| rs1042778||G/T||T: 40||545 (36.0)||736 (48.6)||232 (15.3)|
| rs11720238||G/T||T: 12||1161 (77.1)||317 (21.0)||28 (1.9)|
| rs2254298||A/G||A: 7||1311 (86.9)||190 (12.6)||8 (0.5)|
| rs237887||A/G||G: 41||528 (35.1)||717 (47.6)||260 (17.3)|
| rs4686302||C/T||T: 14||1113 (73.7)||368 (24.4)||29 (1.9)|
| rs53576||A/G||A: 41||527 (35.3)||698 (46.8)||266 (17.8)|
| rs7632287||A/G||A: 29||758 (50.1)||627 (41.4)||129 (8.5)|
| rs4564970||C/G||C: 4||1391 (92.5)||113 (7.5)||–|
| rs1488467||C/G||C: 3||1416 (93.4)||100 (6.6)||–|
|Arginine vasopressin receptor SNPs|| || || || || |
| rs3021529||A/G||A: 10||1239 (81.7)||259 (17.1)||19 (1.3)|
| rs1587097||C/T||T: 6||1331 (88.0)||157 (11.6)||7 (0.5)|
| rs10877970†||C/T||C: 14||1130 (74.9)||337 (22.3)||42 (2.8)|
| rs11174811||A/C||A: 10||1227 (81.5)||261 (17.3)||17 (1.1)|
| rs3759292‡||A/G||G: 0.1||1512 (99.7)||4 (0.3)||–|
| rs1042615||A/G||A: 42||520 (34.5)||721 (47.8)||267 (17.7)|
| rs35369693*||C/G||C: 5||1352 (89.7)||153 (10.1)||3 (0.2)|
Figure 1. Schematic representations of the oxytocin (A) and arginine vasopressin 1A (B) receptor genes, with the location of the 18 analyzed single nucleotide polymorphisms in these genes (see Table 1).
Download figure to PowerPoint
No SNPs deviated significantly from what would be expected under Hardy–Weinberg equilibrium, with the exception of rs10877970 (chi-squared value 7.34, P < 0.005). Therefore, this SNP was excluded from any further analyses. The total allelic frequency of the A allele of the rs3759292 SNP was extremely low (0.1%; four heterozygotic carriers) in the study population so this SNP was also excluded from subsequent analyses.
To compute descriptive statistics and frequencies, SPSS 17.0 for Windows was used (SPSS Inc., 2008; SPSS Inc., Chicago, IL, USA). The effect of the allelic variants was calculated with a linear mixed-effects model using the ‘lme4’ package (http://lme4.r-forge.r-project.org/) in R 2.10.1 (http://www.R-project.org). This model appropriately controls for between-subjects dependence, which was necessary because the sample in the present study consisted of twins and siblings. In the model, polymorphism effects were inserted as factors. As sexual desire and frequency of sexual activities have been shown to be associated with PE , and because the association between age and PE is somewhat unclear (although age effects appear to be marginal at best ), we decided to control for these by including them as covariates in the polymorphism analyses. A Markov Chain Monte Carlo method was used to resample (10 000 times) the posterior distribution of the parameter intervals for the model parameters  to receive 95% Bayesian credible intervals (BCI) for significance testing of the parameters. The BCI are, in this context, to be interpreted as confidence intervals that can be adjusted to a P value that accounts for multiple testing.
To correct for multiple testing, a significance threshold required to keep Type I error rate at 5% was calculated using a linkage disequilibrium correlation measure  for correlated SNPs. As SNPs situated close to one another are often in linkage disequilibrium and therefore correlated, a Bonferroni correction for multiple testing (which simply divides the value of α by the number of SNPs in the model) is commonly regarded as too stringent [33,34]. Using Nyholt's approach , a correlation matrix of SNPs is used to calculate a corrected significance threshold using the equation proposed by Li and Ji . This procedure resulted in an effective number of independent SNPs of 14 and a corrected P value of 0.004 (i.e. a P value lower than 0.004 for the main effect of any individual SNP is considered significant when taking multiple testing into consideration).
Of the 17 SNPs, seven had rare homozygotic genotypes (i.e. with fewer than 30 individuals being homozygous for the less common allele). These were rs11720238, rs2254298 and rs4686302 (OXTR SNPs); rs3021529, rs1587097 and rs11174811 (AVPR1A SNPs); and rs35369693 (AVPR1B SNP). These SNPs were tested for significant mean differences (of the variable measuring PE) between allele groups. At an α level of 0.05 (i.e. not correcting for multiple tests) no significant phenotypic means were found between any allele groups except between the C:C and C:G allele groups of the rs35369693 AVPR1B SNP. However, the C:C group for this SNP consisted of only three individuals, and this effect was nullified when correction for multiple testing was applied. Hence, we decided to combine rare homozygotes with the heterozygote for these seven SNPs to improve statistical power. In addition, two SNPs (rs4564970 and rs1488467) had such low frequencies of the less common allele that no individuals were homozygous carriers of it in the study population. In the case of these two variables, only the heterozygote could be tested against the common homozygote.
Missing values were imputed using the expectation maximization procedure of SPSS 17.0 (SPSS Inc.) if the participant had responded to at least one questionnaire item. Variables were imputed using available data from a number of questionnaires concerning sexuality and aggression.
- Top of page
- CONFLICT OF INTEREST
None of the AVPR polymorphisms had a significant association with the PE composite score when controlling for multiple tests (Table 3). Among the OXTR SNPs, a significant heterozygote effect was found for rs75775, so that carriers of the G:T genotype had significantly elevated levels of PE-related problems compared with the G:G carriers (estimated marginal means MG:G= 14.43 and MG:T= 14.81; PE composite variable range 4.00–19.68). It should also be noted that the T:T genotype was relatively rare (80 individuals compared with 867 carriers of the other homozygote), and as such a significant effect between the homozygotes could possibly be detected in a larger sample.
Table 3. Main effects of oxytocin and vasopressin receptor gene single nucleotide polymorphisms (SNPs) on premature ejaculation composite score
|RS number||Genotype||Main effect of SNP|
|Estimate|| se || t ||95% BCI||99.6343% BCI|
|Oxytocin receptor SNPs|| || || || || || |
| rs237897||A : A vs G : A||0.007||0.108||0.065||−0.190 to 0.228||−0.284 to 0.340|
|A : A vs G : G||0.110||0.123||0.889||−0.124 to 0.353||−0.242 to 0.461|
| rs2268493||C : C vs C : T||−0.072||0.126||0.574||−0.319 to 0.171||−0.442 to 0.290|
|C : C vs T : T||−0.161||0.131||1.236||−0.423 to 0.077||−0.552 to 0.196|
| rs75775 || G : G vs G : T || 0.329 || 0.105 || 3.130 || 0.124 to 0.536 || 0.028 to 0.629 |
|G : G vs T : T||0.128||0.226||0.570||−0.292 to 0.573||−0.509 to 0.821|
| rs1042778||G : G vs T : G||0.055||0.095||0.575||−0.136 to 0.239||−0.215 to 0.319|
|G : G vs T : T||−0.018||0.132||0.138||−0.282 to 0.241||−0.414 to 0.350|
| rs11720238||G : G vs:G : T/T : T||−0.058||0.103||0.563||−0.261 to 0.142||−0.366 to 0.234|
| rs2254298||G : G vs A : G/A : A||0.241||0.128||1.876||−0.262 to 0.139||−0.0124 to 0.619|
| rs237887||A : A vs G : A||−0.125||0.096||1.298||−0.307 to 0.064||−0.400 to 0.169|
|A : A vs G : G||−0.057||0.127||0.444||−0.316 to 0.183||−0.428 to 0.297|
| rs4686302||C : C vs C : T/T : T||−0.044||0.098||0.451||−0.227 to 0.145||−0.317 to 0.235|
| rs53576||A : A vs G : A||0.066||0.121||0.545||−0.163 to 0.305||−0.296 to 0.421|
|A : A vs G : G||0.147||0.126||1.167||−0.094 to 0.386||−0.226 to 0.515|
| rs7632287||A : A vs G : A||0.058||0.163||0.355||−0.163 to 0.305||−0.441 to 0.509|
|A : A vs G : G||−0.022||0.161||0.139||−0.332 to 0.294||−0.490 to 0.462|
| rs4564970||C : G vs G : G||0.070||0.164||0.426||−0.258 to 0.367||−0.419 to 0.517|
| rs1488467||C : G vs G : G||−0.092||0.174||0.532||−0.445 to 0.226||−0.608 to 0.388|
|Arginine vasopressin 1A/1B receptor SNPs|| || || || || || |
| rs3021529||G : G vs A : G/A : A||−0.034||0.111||0.306||−0.246 to 0.184||−0.338 to 0.298|
| rs1587097||C : C vs T : C/T : T||−0.014||0.133||0.106||−0.280 to 0.238||−0.399 to 0.365|
| rs11174811||C : C vs A : C/A : A||−0.098||0.111||0.880||−0.229 to 0.130||−0.408 to 0.228|
| rs1042615||A : A vs A : G||−0.042||0.121||0.351||−0.281 to 0.181||−0.421 to 0.305|
|A : A vs G : G||−0.039||0.127||0.305||−0.284 to 0.210||−0.411 to 0.333|
| rs35369693||G : G vs C : G/C : C||−0.267||0.143||1.871||−0.532 to 0.022||−0.677 to 0.154|
Next, we tested the OXTR SNP rs75775 against the individual PE indicators. As in the case of the PE composite score, a significant heterozygote effect was detected (i.e. that the G:T heterozygote differed significantly from both homozygotes), so that G:T genotype carriers had longer ejaculation latencies (estimate: 0.094; se= 0.047, t= 2.01; BCI = 0.002–0.186; note that BCIs should be interpreted so that the estimate is considered significant for a given P value – here, P < 0.004 – if the interval does not contain zero) and better perceived ejaculatory control (estimate: 0.197; se= 0.065, t= 3.05; BCI = 0.074–0.323). No significant effects were detected for the indicator variables measuring anteportal ejaculation and number of thrusts.
- Top of page
- CONFLICT OF INTEREST
The findings of the present study revealed a significant association between the rs75775 SNP in the OXTR gene, so that the heterozygote had an elevated risk of PE in comparison to both homozygotes. This effect was significant when controlling for 19 simultaneously conducted tests and several covariates that have been associated with PE in the literature .
Although both OXT and AVP have a clearly documented role in ejaculatory function (e.g. ), the polymorphisms in the AVPR and OXTR genes that were analysed in the present study (and arguably cover the respective genes rather well) appear in this study to have rather weak associations with phenotypic indicators of PE. It may be that OXT and AVP are involved in the physiological mechanisms of ejaculation rather than in triggering the ejaculatory reflex itself; in other words, the effects of OXT and AVP are prominent once the ejaculatory reflex has begun, but their effects on the physiological process that starts ejaculation may be limited. This interpretation would be in line with previously documented effects of OXT and AVP in ejaculatory functioning [19,23]. Also, five out of seven AVPR SNPs had genotype distributions in which one of the homozygotes was extremely rare (<1.5%), so failure to detect significant association may be because of sample size. It is known that diseases and observable differences in phenotypic traits may be the result of both common and rare variants; it has been argued that rare genetic variants may be rare explicitly because they have a more profound phenotypic impact . Therefore, AVPR gene-related polymorphisms with rare variants may be suitable target genotypes for studies of the genetic aetiology of PE, but such studies would probably require very large sample sizes; or a case–control design rather than the population-based design used in the present study.
The objectives of genetic association studies of ejaculatory function are not limited to basic charting of the aetiology of ejaculatory dysfunction – information gained from such studies could also provide some direction for the development of new pharmacological treatment alternatives, including pharmacogenetic interventions (i.e. interventions tailored to the patient's genotype to maximize treatment efficacy or reduce side-effects). The findings of the present study concur with the conclusions of Andersson and Abdel-Hamid . They noted that oxytocinergic drugs are unlikely targets for successful pharmaceutical treatments. Although OXT receptors are involved in ejaculatory control, they argue, the exact modus of their involvement is unclear, and it would probably be difficult to develop an oxytocinergic drug with specific ejaculation-delaying effects, and the administration of any such drug would present further problems. Our results, although preliminary, suggest that OXTR and AVPR genes have limited effects on ejaculatory function, and as there are no OXTR-specific or AVPR-specific drugs with any demonstrated efficacy on ejaculatory function for humans, polymorphisms in these genes are not likely to be suitable targets for pharmacogenetic intervention studies.
In the present study, we used a measure of ejaculatory function that included parameters that can be considered both objective (i.e. ELT, frequency of occurrence of anteportal ejaculation, number of thrusts) and subjective (i.e. ejaculatory control) indicators of ejaculatory function. We also used retrospective self-report data in the present study (e.g. instead of a stopwatch, in the case of ELT). However, it has been shown that self-reported ejaculatory latencies are interchangeable with stopwatch measured ELTs, and that patient-reported outcomes in general are largely reliable . Furthermore, it would have been extremely difficult to assess ejaculatory latency by stopwatch in a sample exceeding 1000 men. It should perhaps also be noted that current proposals for a suitable definition and diagnostic criteria require assessment of both objective and subjective indicators of PE [38,39].
This study was conducted on a population-based sample. In a clinical sample of PE patients, risk alleles for PE are expected to occur with increased prevalence, so effects of rare genotypes that could not be detected in our population-based sample could be documented in a clinical sample. Considering that an intravaginal ELT of <1 min, which is indicative of lifelong PE, is quite rare (around 1% of the population ), it is conceivable that genotypes with a strong main effect on PE would be relatively rare in the general population.
In conclusion, the rs75775 SNP in the OXTR is associated with PE, so that G:T genotype carriers are at elevated risk compared with both homozygotes. Several SNPs in the AVPR1A gene had rare genotypes in our sample, hence insufficient statistical power could mask potential effects of these. Replication of this study is warranted.