Discussion
This study confirms previously observed positive associations between circulating 25(OH) D and total and free testosterone levels before and after adjustment for a variety of potential confounders. The shape of the dose–response curves indicate that associations between 25(OH)D and testosterone are strongest in lower ranges of 25(OH)D. Unlike for vitamin D, we did not observe any seasonal variation of testosterone concentrations.
This is the fourth study to show positive associations between vitamin D and testosterone. Both VDR and vitamin D metabolizing enzymes were shown to be expressed in human Leydig cells, where testosterone in the male testes is synthesized, suggesting that vitamin D might affect the production of male reproductive hormones. This hypothesis is supported by the observation that VDR knockout mice are characterized by hypergonadotropic hypogonadism. It cannot be excluded that the here observed associations between 25(OH)D and testosterone reflect nonbiological correlations because of similar distributions of confounding factors or because of common biological precursors such as cholesterol. However, positive associations between 25(OH)D and testosterone persisted after adjusting for a variety of factors that are associated with both 25(OH)D levels and testosterone levels such as age, BMI and physical activity. Furthermore, adjustment for total cholesterol did not change the observed associations between 25(OH)D and total or free testosterone.
Vitamin D status as reflected by 25(OH)D levels is mainly determined by sun exposure (ultraviolet light conversion of 7-dehydrocholesterol in the skin), while dietary intake generally plays a more minor role in vitamin D supply. Although our findings of a positive association between 25(OH) vitamin D and testosterone confirm the previous cross-sectional findings among 2299 men routinely referred for angiography in the German LURIC study, we did not observe any seasonal variation of total testosterone, which contrasts this study's findings of parallel seasonal variation of 25(OH)D and testosterone. It is noteworthy that the two studies differ substantially with respect to vitamin D status. Based on 25(OH)D levels, in the German study, 63% of participants had a deficient vitamin D status (25 (OH) D <50 nmol/l), whereas in our sample, only 25% of participants were vitamin D deficient. The poor vitamin D supply in the German sample may be partly explainable by the geographical latitude that is further away from equatorial latitudes than US latitudes. In addition, while in the US, milk and other foods are commonly fortified with vitamin D, no fortification is undertaken in Germany, so that the dietary vitamin D intake is substantially lower than that in the US. Finally, the German sample of men who were referred for coronary angiography may have had exceptionally low 25(OH)D levels because coronary artery disease has been linked to a poor vitamin D status. In a study relating vitamin D levels to sperm quality in humans, the effect of vitamin D on sperm motility was most pronounced in men with vitamin D deficiency. In a recent multicentre cross-sectional study among 3369 community-dwelling European men, of whom 37% were in the deficiency range, 25(OH)D was positively associated with total and free testosterone in age- and study centre-adjusted models, but not after multivariate adjustment for health and lifestyle factors. In the same study, no seasonal variation of testosterone was observed. Seasonal variation of testosterone has been investigated in numerous other cross-sectional and longitudinal studies, yielding inconsistent results. In line with our finding, no seasonal variation of testosterone has been observed in four studies.
Our investigation of the dose–response relationship suggests that associations between 25(OH)D and testosterone are stronger at the lower end of vitamin D concentrations. However, owing to the low proportion of vitamin D–deficient men in our study, we were not able to investigate the associations at the very low end of 25(OH)D levels (only 10 participants had 25(OH)D levels <25 nmol/l).
The observed statistical interaction by vasectomy requires confirmation by further studies. A biological explanation why vitamin D may be associated with higher testosterone levels only in men who had not had a vasectomy is lacking so far, and it cannot be excluded that this finding was because of chance. Vasectomy is not strongly associated with changes in hormonal profiles. In our study testosterone, SHBG and vitamin D levels did not differ substantially by vasectomy status, but men who had had a vasectomy were younger and more likely to reside in sunny areas than men who had not had a vasectomy. Thus, it is possible that the observed effect modification was because of differences in the distribution of confounding factors by vasectomy status.
Strengths of our study include the ability to adjust for a variety of potential confounders. Although residual confounding cannot be excluded, our findings suggest an association between 25(OH)D and testosterone independent of major determinants of 25(OH)D/testosterone such as BMI, smoking and physical activity. Our study also has several limitations. Owing to the cross-sectional design, no conclusions with respect to causality or directionality of the vitamin D–testosterone association can be drawn. Another drawback of our study is that total testosterone was measured by an immunoassay, which has lower accuracy and sensitivity than methods based on mass spectrometry, which is considered the gold standard. Owing to the strong variation by analysis batch and the low intraperson consistency of free testosterone measurements, the associations observed with free testosterone should be interpreted with caution. Furthermore, our findings are only generalizable to a limited extent, because this is a sample of middle aged mainly Caucasian men with a rather homogenous socio-economic status.
In conclusion, the present study confirms previous findings of a positive association between vitamin D status as reflected by 25(OH) vitamin D levels and testosterone levels, although in our study, this finding was not supported by parallel seasonal variation patterns of 25(OH) vitamin D and testosterone. Whether these cross-sectional findings reflect a causal relationship deserves further scientific examination.