Osteocalcin is not a strong determinant of serum testosterone and sperm count in men from infertile couples
Verena Schwetz, Division of Endocrinology and Metabolism, Department of Internal Medicine, Medical University of Graz, Auenbruggerplatz 15, 8036 Graz, Austria. E-mail: email@example.com
Osteocalcin (OC) – released by osteoblasts and known as a marker of bone turnover – has been suggested to influence male fertility in murine models by enhancing testosterone production and sperm count. Results from clinical studies are scarce, however. The aim of this cross-sectional study was to investigate the proposed association of OC, undercarboxylated osteocalcin (ucOC) or carboxylated osteocalcin (cOC) with testosterone and sperm count in a cohort of 159 young male adults from infertile couples. Semen analysis was performed. Testosterone, free testosterone, LH, OC and ucOC were measured in serum samples after an overnight fast. cOC and OC correlated weakly but significantly with testosterone (OC: r = 0.165, p = 0.040, cOC: r = 0.193, p = 0.017), but not after adjusting for age and body mass index (BMI) or waist–hip ratio (WHR). %ucOC (ucOC levels expressed as percentage of total OC) correlated inversely with LH (r = −0.184, p = 0.023) and remained significant after the same adjustment. No significant correlations were observed between OC, cOC, ucOC, %ucOC and sperm count, semen volume and number of vital spermatozoa. In binary logistic regression analyses, none of the parameters of OC were predictors of oligozoospermia after adjusting for age and BMI or WHR. The weak association between %ucOC and LH has marginal clinical importance because of the lack of associations of parameters of OC with testosterone and sperm count. The current data thus cannot support the notion that OC is associated with male fertility in young men from infertile couples.
About 7% of all men are affected by fertility problems (Krausz, 2011), while male causes of infertility are found in approximately half of involuntarily childless couples (Krausz, 2011). Fertility is affected by numerous genetic and environmental factors (Krausz, 2011). Recently, the bone-secreted hormone osteocalcin (OC) has been discovered to influence testosterone production by the testis and to regulate male fertility.
Osteocalcin is produced by osteoblasts and is known as a marker of bone turnover. It is released into the systemic circulation both in its undercarboxylated (ucOC) and carboxylated form (cOC; Ferron et al., 2010). Recently, ucOC was shown to enhance insulin sensitivity in muscle, liver and adipose tissue (Lee et al., 2007) and insulin secretion in the beta cells (Ferron et al., 2010). Bone might thus function as an endocrine organ. Clinical studies have yielded controversial results – among them, many authors have described associations of OC, ucOC, but also cOC with parameters of glucose metabolism (summarized in Schwetz et al., 2012) and mortality (Lerchbaum et al., 2013).
This newly discovered endocrine action of bone and the regulation of bone remodelling by the gonads have suggested a feedback loop between the gonads and the skeleton. As summarized by Clarke, androgens have beneficial effects on bone (Clarke & Khosla, 2009). In an intervention study, the elimination of both testosterone and oestradiol diminished serum OC, while both sex hormones enhanced OC levels (Falahati-Nini et al., 2000). The other part of the feedback loop suggests that OC has an influence on testosterone production, i.e. that bone influences reproductive functions (Ducy et al., 2000; Karsenty, 2006). Indeed, lately the testes have been understood as a target organ of OC (Oury et al., 2011).
In murine models, OC induced testosterone production in Leydig cells and thus reduced germ cell apoptosis, but failed to have an influence on oestrogen or testosterone production in the ovaries (Oury et al., 2011). In the ovaries, the relevant G protein-coupled receptor (GPRC6A) is not expressed (Pi et al., 2005). In OC knockout mice, litter size was significantly smaller, testes size and weight were decreased, as were the weights of epididymides, seminal vesicles, spermatozoa and levels of testosterone; levels of luteinizing hormone (LH) were increased as compared with Esp knockout mice. However, the percentage of motile, abnormally shaped and dead spermatozoa did not differ between OC knockout and wild-type mice. OC production was only found in osteoblasts, not however, in the testis itself (Oury et al., 2011).
Clinical studies found significant correlations of total OC with testosterone in boys aged 11–14 years (Kirmani et al., 2011) and weak correlations of total OC and cOC with total testosterone in 204 18- to 20-year-old Finnish men (Valimaki et al., 2004). Another revealed that ucOC and ucOC/total OC ratio were positively associated with serum free testosterone levels in men with type 2 diabetes (Kanazawa et al., 2013). In the same study, ucOC and ucOC/total OC were also negatively associated with LH (Kanazawa et al., 2013). The authors suggest that a reduction in ucOC would lead to low testosterone levels, which would in turn raise LH by feedback loop of the gonads (Kanazawa et al., 2013). Hannemann and colleagues have also recently revealed a positive association between OC and total testosterone in the general population and in male patients with bone disorders (Hannemann et al., 2013). Other studies found contradictory results concerning the association of testosterone with bone turnover markers. No effect of testosterone stimulation on OC production or even a decrease was shown in one study (Clarke & Khosla, 2009). In 40 healthy men and 80 men with osteoporosis, no association of OC with testosterone was revealed (Legrand et al., 2001). To date, no data on semen quality are available.
Therefore, the aim of our study was to investigate whether the reported association of OC with testosterone exists in young adult men from infertile couples – an age group that has not been investigated yet – and whether OC, ucOC, or cOC serve as a predictor of sperm count.
Materials and methods
Male adults referred to semen analysis were recruited from the outpatient clinic of the Department of Gynecology and Obstetrics at the Medical University of Graz between 2010 and 2012. These men were either part of an infertile couple or simply in desire of having children and therefore came for semen analysis. None of the patients had diabetes or a history of bone disease or fractures. None of the men included took medications known to affect bone, mineral, energy metabolism or spermatogenesis including vitamin D supplementation. Eleven patients on bone-active medication or on medication known to effect fertility were excluded from analysis. The study protocol was approved by the ethics committee of the Medical University of Graz. Written informed consent was obtained from each patient.
Standard anthropometric data, i.e. weight, height, waist circumference, hip circumference and blood pressure, were obtained from each participant. Basal blood samples were drawn in the morning between 8 and 9 a.m. after an overnight fast. Testosterone, free testosterone and LH were measured as samples were collected. Serum samples were then stored at −70 °C and later batch analysed for total OC and ucOC. FAI was calculated as (total testosterone: SHBG) × 100.
Semen analysis was carried out using standard procedures as recommended by the World Health Organization (WHO, 2010). Criteria for sample collection are collection of a complete sample in a private room near the laboratory by masturbation into a container after a minimum of 2 days and a maximum of 7 days of abstinence and avoidance of large changes in temperature (WHO, 2010). The time between collection of the semen sample and the microbiological investigation did not exceed 3 h. Oligozoospermia was defined as a sperm count of <15 mio/mL according to the WHO criteria 2010 (WHO, 2010).
Total OC was measured in serum by electrochemiluminescence immunoassay (Roche, Mannheim, Germany) detecting all forms of OC with a similar affinity. cOC was also measured in serum and was separated from ucOC by adsorption on hydroxyapatite (Price & Williamson, 1981; Price et al., 1981; Pietschmann et al., 1988). ucOC levels are expressed as percentage of the total OC (%ucOC) measured before incubation with hydroxyapatite (Gundberg et al., 1998; Motyl et al., 2010). To validate the method, we performed measurements of the same samples by means of three different methods – the hydroxyapatite-binding method following two distinct protocols as well as a commercially available enzyme immunoassay (EIA; Takara Bio Europe/SAS, Saint-Germain-en-Laye, France). We observed a Spearman correlation coefficient of 0.95 between one hydroxyapatite-binding method and the EIA (p < 0.001). cOC was calculated by subtracting ucOC from total OC. SHBG (Roche, Basel, Switzerland) and total testosterone (Siemens, Erlangen, Germany) were measured by luminescence immunoassay, LH was measured by enzyme immunoassay (Radim diagnostics, Italy).
Data are presented as median with interquartile range. Kolmogorov–Smirnov test and descriptive statistics were used to evaluate distribution of data. Parametric tests were performed for normally distributed data. For non-normally distributed data, non-parametric tests were applied or continuous parameters were logarithmically transformed for parametric tests. Spearman correlation analyses were performed for non-normally distributed data, for normally distributed data, Pearson correlation was applied. Stepwise linear regression analyses were performed to evaluate whether parameters of OC predict testosterone or LH. Binary logistic regression analyses were used to calculate whether oligozoospermia (yes/no) is predicted by parameters of OC. Adjustments for age and BMI/WHR were carried out. A p-value of <0.05 was considered statistically significant. All analyses were performed using spss version 18.0 (SPSS Inc., Chicago, IL, USA).
In our study cohort (baseline characteristics are shown in Table 1), 48 of 159 (30.2%) patients had oligozoospermia, 111 of 159 (69.8%) had normozoospermia. cOC and OC but not ucOC or %ucOC correlated with body mass index (BMI; OC: r = −0.209, p = 0.008, cOC: r = −0.206, p = 0.010) and waist–hip ratio (WHR; OC: r = −0.218, p = 0.006, cOC: r = −0.221, p = 005). No correlations were observed between parameters of OC and blood pressure, HbA1c and fasting glucose. cOC and OC but not ucOC correlated weakly but significantly with testosterone (OC: r = 0.165, p = 0.040, cOC: r = 0.193, p = 0.017). %ucOC correlated inversely with testosterone (r = −0.172, p = 0.034), supporting the positive correlation of the carboxylated fraction of OC with testosterone. In the correlations with free testosterone, LH and free androgen index (FAI), only %ucOC, but not OC, cOC or ucOC, correlated inversely with LH (r = −0.184, p = 0.023). Free testosterone did not correlate with OC (r = −0.016, p = 0.846), cOC (r = −0.011, p = 0.893), %ucOC (r = −0.109, p = 0.178) or ucOC (r = −0.054, p = 505). After adjusting for age and BMI or WHR in stepwise linear regression analyses, however, using testosterone as dependent variables, only age, but neither OC, ucOC, %ucOC nor cOC were independent predictors of testosterone. OC, ucOC and cOC did not predict LH after adjustment for age and BMI/WHR. Only %ucOC retained statistical significance as a weak predictor of LH (β = −0.186, p = 0.022, WHR/BMI and age were not significant). BMI and WHR as well as age were considered relevant confounders as they correlated with testosterone and free testosterone.
Table 1. Characteristics of study participants (n = 159)
|Systolic blood pressure (mmHg)||135||122–145|
|Diastolic blood pressure (mmHg)||88||80–94|
|Parameters of semen analysis|
|Sperm count (mio/mL)||34.4||12.1–74.9|
|Semen volume (mL)||3.0||2.0–4.2|
|Vital spermatozoa (%)||33||22–48|
|Free testosterone (pg/mL)||10.11||7.77–12.38|
|Total OC (ng/mL)||21.1||17.1–27.1|
|25-OH-vitamin D (ng/mL)||26.6||17.1–39.6|
|Fasting glucose (mg/dL)||88||80–96|
Age correlated with semen volume (r = −0.254, p = 0.001), vital spermatozoa (r = −0.213, p = 0.007), testosterone (r = −0.201, p = 0.012), free testosterone (r = −0.307, p < 0.001), FAI (r = −0.307, p < 0.001), FSH (r = 0.206, p = 0.011), BMI (r = 0.168, p = 0.035) and WHR (r = 0.345, p < 0.001), but not with sperm count (r = −0.018, p = 0.826) or LH. However, no correlations were observed between any of the fractions of or total OC and sperm count, semen volume and number of vital spermatozoa.
In binary logistic regression analyses (crude model in brackets, for corrected model see Table 2) with oligozoospermia/normozoospermia as dependent variable and OC (β = −1.001, p = 0.455), ucOC (β = −0.623, p = 0.510), %ucOC (β = −0.607, p = 0.703) and cOC (β = −0.627, p = 0.649) as independent variable, none of the parameters of OC were predictors of sufficient or insufficient sperm count after adjusting for age and BMI or WHR.
Table 2. Results of binary logistic regression analyses with oligozoospermia (yes/no) as dependent variable and parameters of OC (i.e. OC, cOC, ucOC, %ucOC) as independent variables
We observed significant correlations of cOC and OC with testosterone, and of %ucOC with LH – however, only the association between %ucOC and LH remained significant after adjustment. This association was weak, however, and its clinical significance seems marginal as no associations with testosterone and sperm count were found. None of the parameters of OC were predictors of oligozoospermia. The current data thus cannot strengthen the assumed association of OC, testosterone and human male fertility as observed in murine models (Oury et al., 2011).
Our results might thus seem in contrast with the studies by Kirmani et al. (2011), Kanazawa et al. (2013) and Hannemann et al. (2013). The first study (Kirmani et al., 2011) found a significant correlation of total OC with testosterone in boys aged 11–14 years, the second (Kanazawa et al., 2013) found a significant correlation between ucOC and ucOC/total OC ratio with serum free testosterone levels in men with type 2 diabetes. Of note, in the first study no such correlation was found for boys younger than 11 or older than 14 years. Therefore, this finding might originate from the fact that boys aged 11–14 years are in a phase of rapid skeletal growth and in puberty. This observation might thus not support the postulated influence of OC on testosterone and fertility in human adults. Differences in our results to the study by Kanazawa et al. (2013) might arise from an age difference of about 25 years between the study groups and the presence of diabetes in these men. As for the association of total testosterone and OC in the study by Hannemann, a possible stimulatory effect of total testosterone on OC secretion cannot be excluded as the underlying mechanism of the observed associations, as also noted by the authors (Clarke & Khosla, 2009; Hannemann et al., 2013). Before adjustment we find a weak association of testosterone and total OC as well as cOC which disappears after adjustment. It can be concluded that either this is an effect of our small sample size or that OC is only a weak determinant of testosterone levels in humans.
Of note, parameters of semen analysis were not available in any of the publications cited. Also, to the best of our knowledge, no other clinical study has investigated the suggested association of OC with sperm count or any other parameter of semen analysis in humans. Previous studies (Kindblom et al., 2009; Pittas et al., 2009) have only shown associations of OC with glucose metabolism – our associations of cOC and OC with BMI and WHR are consistent with these data.
The basis of discrepancy in terms of male fertility between this study and the one by Kanazawa (Kanazawa et al., 2013) is subject to speculation, but might be based on the different age groups and on diverging pre-existing conditions (non-diabetic vs. diabetic) in the study participants. The differing results reflect the controversies that have also arisen as to the role of OC in glucose metabolism, emphasized by recent publications that failed to show an association of OC with the development of type 2 diabetes in middle-aged males (Hwang et al., 2012) and suggest only a minor and age-specific association of OC and glucose metabolism in non-diabetic women (Lu et al., 2012).
It may well be that in contrast to murine models, OC has a limited, less pronounced effect on testosterone levels in humans. We presume that both the study by Kanazawa as well as our own study have a sample size (69 and 159) too small to satisfyingly elucidate this question. However, there are data in support of our assumptions. It is known that antiresorptive bone-active medication suppresses bone turnover by inhibiting osteoclasts and as such lower serum OC levels. This would suggest that osteoporotic men – often hypogonadic and thus suffering from secondary osteoporosis – on antiresorptive treatment would have even lower levels of testosterone than before and the consecutive clinical signs. No such observations were, however, made in any of the clinical trials (MacLean et al., 2008; Watts & Diab, 2010). However, decreases in sperm count would presumably go undetected as in none of these studies semen analyses were performed.
It remains questionable whether the two studies (Kirmani et al., 2011; Kanazawa et al., 2013) that have shown human correlations between OC and testosterone have included apt subjects to actually answer these questions. In the first (Kirmani et al., 2011), adolescents were investigated, in the second, 60-year-old diabetics were examined (Kanazawa et al., 2013). Our study is the first to include young adult men from infertile couples who were expected to have normal bone turnover. This is necessary to minimize the effect of confounders.
There are limitations to our study. They include the cross-sectional character of our analysis. Furthermore, no other bone turnover markers other than OC were measured, which might impact the quality of our study. We did not evaluate alcohol consumption which could affect male fertility (Sadeu et al., 2010; Krausz, 2011). In addition, we were able to recruit a number of men, but the sample size might be too small to conclude further consequences. However, our sample size exceeds the sample sizes in most previous studies on OC and male fertility. Another limitation might be the high prevalence of oligozoospermia in our sample which might have confounded a possible association with OC. Besides, there is a large biological variability in semen quality (Castilla et al., 2006). We tried to overcome this problem by repeating semen analysis in men with oligozoospermia. We then applied the mean of semen volume, sperm count and vital sperms for further analysis.
Despite these limitations, our study is the first to investigate the association of OC with parameters of semen analysis in young adult men from infertile couples. We find that OC cannot be considered a strong determinant of testosterone and semen quality in our cohort of men. In addition to measuring total OC, we also measured ucOC and calculated cOC to include possible carboxylation-specific effects of OC. Further studies with larger sample sizes are needed to clarify the discussed findings on the subject of OC and male fertility.
We greatly acknowledge Univ. Prof. Dr. Uwe Lang and Ass. Prof. Dr. Johann Auner for their support.
VS, RG, MG, NH, NS, OT, MB and EL performed the research; VS, EL, TP and BOP analysed the data; EL, MG and BOP designed the study; and VS wrote the manuscript.
The authors have nothing to disclose.
This work was supported by BioPersMed (COMET K-project 825329) which is funded by the Federal Ministry of Transport, Innovation and Technology (BMVIT) and the Federal Ministry of Economics and Labour/the Federal Ministry of Economy, Family and Youth (BMWA/BMWFJ) and the Styrian Business Promotion Agency (SFG).