Similarly, the inverse association between ED and FT or BT appeared to be independent of hypertriglyceridemia and hyperglycemia, whereas in groups with elevated BP, low HDL-C or central obesity, it might be underpowered to detect the relatively small associations observed in other groups. Concerning the changes of testosterone levels across age, FT or BT rather than TT appeared to be more correlated with age. We explained that TT did not change much with age probably because SHBG went up with age. It has been reported that serum testosterone levels gradually fall with advancing age, whereas SHBG levels increase with age and present a rapid increase in the old, consistent with the current study. Along these lines, although FT and BT were decreased among the aged men, the unbound testosterone may have a high probability to combine with SHBG because the elevation of SHBG level showed an enhanced rate when these men got older. Thus the net result of these changes of unbound testosterone and SHBG is to present that TT did not change much with age. Moreover, TT levels in many aged men were very closed to those found in young healthy subjects, which might indirectly support this result from present study. Additionally, we observed that the increase of SHBG across ED status paralleled with its increase with age, thus the positive association of SHBG with ED might be a CPI-613 reflection of age. However, we observed that the association of TT with ED and its association with age showed a reverse direction, thus the positive association between TT and ED was probably due to the increase in SHBG. So that testosterone levels are strongly related to SHBG concentrations. In addition, genetic variants in the SHBG locus have been associated with a substantial variation in testosterone concentrations, and the SHBG polymorphism could affect testosterone binding to SHBG. Although the association between TT and ED in the current study was independent of age, the possibility of SHBG interaction could not be ruled out. As shown in our study, both TT and SHBG were gradually increased across the ED status, although the increase in TT was relatively small in absolute terms. Moreover, it is critical to highlight that although the positive association between TT and ED remained statistically significant after adjusting the putative confounders, the magnitude of this association was modest. Recently, the EMAS demonstrated that there was a testosterone threshold for the relationship between TT and ED. TT was associated with worse sexual functioning at concentrations of 8 nmol/l or less, whereas the relationship came to a plateau at TT levels over 8 nmol/l. This finding was consistent with the evidence from the animal trials that androgen requirement for sexual behavior was less than the amount normally present Moreover, evidences from a meta-analysis also suggested that such a testosterone threshold on sexual function might exist in men. Nevertheless the EMAS group also suggested that the relationship between testosterone and function.