Total testosterone is the most valuable indicator of metabolic syndrome among various testosterone values in middle-aged

Endocrine Journal 2015, 62 (2), 123-132 Original Total testosterone is the most valuable indicator of metabolic syndrome among various testosterone values in middle-aged Japanese men Makito Tanabe 1), Yuko Akehi 1), Takashi Nomiyama 1), Junji Murakami 2) and Toshihiko Yanase 1) 1) Department of Endocrinology and Diabetes Mellitus, Faculty of Medicine, Fukuoka University, Fukuoka 814-0180, Japan 2) Department of Preventive Medicine, Iizuka Hospital, Iizuka 820-8505, Japan Abstract. Endogenous testosterone is known to be protective against metabolic syndrome (MetS) in men. While various markers of testosterone status including serum total testosterone (TT), free testosterone (measured using analogue ligand RIA [aft]), calculated FT (cft), calculated bioavailable testosterone (cbt), and sex-hormone binding globulin (SHBG) are recognized, it is unclear which of these markers are the most appropriate ones for the detection of MetS. We measured various testosterone values and metabolic markers in 249 healthy Japanese males (mean age 52.7 ± 7.4 yr) and analyzed which testosterone value is most associated with various metabolic parameters, including MetS as diagnosed according to the International Diabetes Federation (IDF, 2009 version) or with the Japanese criteria. Age had no effect on the TT level but significantly decreased aft, cft, and cbt levels and significantly increased the SHBG level. All testosterone values and SHBG showed weak inverse relationships with the metabolic markers BMI, waist circumference, insulin, HOMA-R, and HOMA-β, with the strongest relationship being to TT. TT and SHBG were significantly lower in men with MetS than in men without MetS. All testosterone values gradually decreased as the number of MetS components increased. Multivariate analysis revealed that the TT median value of <4.0 ng/ml was the only significant marker for the detection of MetS. These results were essentially the same regardless of whether the diagnosis of MetS was based on the IDF or the Japanese criteria. In conclusion, among various testosterone values, TT is the most reliable indicator of MetS in middleaged Japanese men. Key words: Testosterone, Late onset hypogonadism, Metabolic syndrome ENDOGENOUS androgen, such as testosterone, has been shown to have a protective effect against obesity and metabolic syndrome (MetS) in men, as evidenced by many experimental and epidemiological investigations [1]. A recent meta-analysis reported that low blood testosterone levels are a risk factor of mortality and death by cardiovascular disease [2]. Blood testosterone level is known to decrease gradually with aging. A low testosterone level sometimes causes various clinical symptoms, called late onset hypogonadism (LOH) and its diagnostic guideline has been proposed in Japan [3]. MetS is one of the clinical characteristics of LOH [3]. Total testosterone (TT) level in serum is composed Submitted Jul. 10, 2014; Accepted Oct. 1, 2014 as EJ14-0313 Released online in J-STAGE as advance publication Oct. 23, 2014 Correspondence to: Toshihiko Yanase, M.D., Department of Endocrinology and Diabetes Mellitus, School of Medicine, Fukuoka University, 7-45-1 Nanakuma, Jonan-ku, Fukuoka 814-0180, Japan. E-mail: tyanase@fukuoka-u.ac.jp The Japan Endocrine Society of the following three fractions: i) sex hormone binding globulin (SHBG) (35 75%), ii) albumin (25 65%) and iii) free testosterone (FT). Only 0.5 3% of TT exists as FT, which is bioactive [4]. The albuminbound testosterone fraction can be easily dissociated from albumin, so this fraction is also considered to be bioactive. Hence, the sum of albumin-bound testosterone and FT are defined as bioavailable testosterone (bt). Calculated bioavailable T (cbt) and calculated FT (cft) have been proposed [5] and these values are available by a simple calculation at the site of the Free & Bioavailable Testosterone calculator (http://www. issam.ch/freetesto.htm). The direct measurement of FT by the analog ligand RIA (aft) generally shows good correlation with FT as measured by equilibrium dialysis or cbt [5], but is not usually thought to be a reliable index of FT. The aft has been reported to represent a variable fraction (20 60%) of FT as measured by equilibrium dialysis [5]. However, the measurement

124 Tanabe et al. of FT by equilibrium dialysis is a very time-consuming method and is not realistic in a daily clinical situation. The absolute value of aft is known to be very low, around 10 20% of the value of cft. Considering the above circumstances, cbt is thought to be the most reliable and realistic marker to reflect the true testosterone bioactivity. It has not been fully understood which testosterone marker is the most reliable indicator for MetS. Interestingly, SHBG might also be a candidate indicator for the detection of MetS, as low serum SHBG levels have been reported to be associated with the occurrence of MetS [6, 7] and type 2 diabetes mellitus (DM) [6, 8]. One possible mechanism for the reduction of SHBG in MetS and type 2 DM is hyperinsulinemia caused by insulin resistance, which can cause impaired SHBG synthesis in the liver, as evidenced by an in vitro experiment in cultured liver cells [9]. On the other hand, a trial of weight loss intervention using a very low calorie diet and behavior modifications revealed that the treatment group showed increases in the levels of SHBG compared with the no intervention group [10]. Thus, the causal association between low SHBG and MetS might be bidirectional. The relationship between MetS/type2 DM and testosterone or SHBG has recently been reviewed [1, 11, 12]. Low TT is partially mediated by low SHBG, but low testosterone bioactivity itself was also reported to correlate with MetS and type 2 DM by a bidirectional mechanism. Lowered testosterone bioactivity has been reported to increase lipoprotein lipase activity and TG catabolism in blood, causing TG uptake into adipose tissue. Conversely, fat accumulation in the body is considered to suppress gonadotropin secretion, mediated by inflammatory cytokines from adipose tissue or by leptin resistance in the hypothalamus. With this hypothesis, a vicious cycle of low testosterone and fat accumulation has been proposed. In this respect, the use of relatively asymptomatic healthy subjects is desirable to clarify the true relationship between various testosterone values and MetS. Early detection of MetS in males using a simple method is very important because the absolute risk of cardiovascular disease in patients with MetS is much higher in males than females in Japan [13]. Testosterone is a promising candidate marker for such detection of male MetS, but it is uneconomical and unrealistic to measure multiple values of testosterone or SHBG because the Japanese insurance system has not approved their simultaneous measurements. In this respect, the present study was undertaken to clarify which testosterone value is the most important marker for MetS in middle-aged Japanese men. Materials and Methods Subjects The study protocol was approved by the Institutional Review Boards at Fukuoka University Hospital and Iizuka Hospital. Written informed consent was obtained from all subjects for participation in the study. The first 304 subjects who visited the Department of Preventive Medicine at Iizuka Hospital for a health check-up were recruited. Of these, 53 subjects, who were taking drugs either for one or both diabetes mellitus or hyperlipidemia as revealed from a questionnaire, were excluded, because these subjects would have made the data values unreliable. Subjects who were taking antihypertensive drugs and considered to be hypertensive were not excluded, even though their blood pressure levels on the examination day were normal. Among the remaining 251 individuals, two additional subjects who lacked waist circumference data were also excluded. Therefore, 249 asymptomatic and healthy men (aged 52.7 ± 7.4 years, mean ± SD) were analyzed. Fasting blood samples were collected in the morning and measured for lipids, including total cholesterol (TC), triglycerides (TG), high density lipoprotein-cholesterol (HDL-C), and markers for glucose metabolism, including fasting blood glucose (FBG), insulin, glycohemoglobin (HbA1c) and homeostasis model assessment of insulin resistance (HOMA-R) and beta cell function (HOMA-β). Insulin sensitivity was determined as HOMA-R, which was calculated as [fasting immunoreactive insulin (F-IRI) (μu/ml) FBG (mmol/l)]/22.5 [14]. HOMA-β, as an index of insulin secretion capability, was calculated as [F-IRI (μu/ml) 20]/[FPG (mmol/l) 3.5] [14]. In addition, the anthropometric data, including height, body weight, body mass index (BMI), waist circumference and blood pressure were measured. Blood samples were also used to measure various testosterone markers, including TT, FT measured by analog ligand RIA (aft), cft, cbt and SHBG. The method for calculating cft and cbt is available at International Society for the Study of the Aging Male (ISSAM) (http://www. issam.ch/freetesto.htm). The diagnosis of MetS in men was undertaken

Testosterone and metabolic syndrome 125 according to one international (International Diabetes Federation [IDF], 2009 version) [15] and one domestic (Japanese [16]) criterion. The IDF criterion requires three or more of the following conditions: i) elevated BP of either one or both a systolic or diastolic BP of 130/85 mmhg, ii) an elevated FBG level of 100 mg/ dl, iii) an elevated serum TG level of 150 mg/dl, iv) a decreased HDL-C level of 40 mg/dl and v) abdominal obesity of a waist circumference of 90 cm. The Japanese criterion requires the presence of abdominal obesity of a waist circumference of 85 cm and two or more of the following components: i) elevated BP of either one or both a systolic or diastolic BP of 130/85 mmhg, ii) an elevated FBG level of 110 mg/dl, iii) lipid abnormality of either one or both of a serum TG level 150 mg/dl or an HDL-C level 40 mg/dl. Measurements of various testosterones and SHBG Serum concentrations of TT, aft and SHBG were measured with the collaboration of SRL Co. Ltd. (Tokyo, Japan). TT was measured by electro-chemiluminescence immunoassay (ECLIA) (ECL TESTO II, Roche, Mannheim, Germany). The measurement of aft was done by RIA (Coat-A-Count: Diagnostic Products Corp., Los Angeles, CA, USA) [17]. SHBG was measured by Immunoradiometric assay (IRMA) using the IRMA-Count SHBG kit (Siemens Japan KK, Tokyo, Japan). The detection limits of TT, aft and SHBG stated by the manufacturers were 0.025 ng/ml, 0.15 pg/ml and 0.04 nmol/l, respectively. The intraassay coefficient of variation of TT, aft and SHBG was <10%, <15% and 2.8 5.3%, respectively. The values of cbt and cft were calculated by the Free & Bioavailable Testosterone calculator at (http://www.issam.ch/free testo.htm) based on the values of TT and SHBG. Statistical analysis All data are expressed as means ± standard deviation. Comparative analyses of continuous variables between two groups were performed, as appropriate, using the unpaired t-test. Binary logistic regression analyses were executed to determine predictors for MetS and odds ratio with a 95% confidence interval and were calculated as both univariable and multivariable analyses. When there was collinearity between two variables, either variable was excluded from the multiple logistic regression model. All statistical analyses were performed using IBM SPSS version 18.0. Values of P <0.05 were considered statistically significant. Results The clinical characteristics of 249 subjects and the comparison of various values between the MetS group (N = 61) and the non-mets group (N = 188) based on the IDF criterion are summarized in Table 1. BMI, waist circumference, TG, FBG and HOMA-R were significantly higher in the MetS group than in the non- Table 1 Clinical characteristics of 249 subjects and comparison of various values between MetS (IDF criterion) group (N = 61) and non-mets (N = 188) group Total (N = 249) With MetS (N = 61) Without MetS (N = 188) Age (years) 52.7 ± 7.4 52.6 ± 6.7 52.7 ± 7.7 0.884 BMI (kg/m 2 ) 23.9 ± 2.9 26.3 ± 3.0 23.1 ± 2.4 <0.001 Waist C (cm) 86.7 ± 8.0 94.2 ± 7.4 84.3 ± 6.6 <0.001 TT (ng/ml) 4.92 ± 1.61 4.12 ± 1.41 5.17 ± 1.59 <0.001 aft (pg/ml) 10.42 ± 3.49 9.69 ± 3.32 10.66 ± 3.53 0.059 cft (pg/ml) 80.9 ± 24.2 72.8 ± 21.2 83.5 ± 24.6 0.003 cbt (ng/ml) 1.88 ± 0.56 1.72 ± 0.49 1.93 ± 0.57 0.007 SHBG (nmol/l) 47.8 ± 16.6 41.4 ± 16.0 50.0 ± 16.3 <0.001 TG (mg/dl) 126.4 ± 66.3 179.5 ± 81.1 109.2 ± 50.0 <0.001 HDL-C (mg/dl) 54.7 ± 11.5 49.3 ± 9.9 56.5 ± 11.4 <0.001 FBG (mg/dl) 100.3 ± 12.9 111.8 ± 17.6 96.6 ± 8.1 <0.001 HOMA-R 1.66 ± 1.20 2.84 ± 1.66 1.27 ± 0.65 <0.001 Data are expressed as mean±sd. were determined by unpaired t test. Values were statistically significant at P <0.05. BMI, body mass index; TT, total testosterone; aft, free testosterone measured by analogue RIA; cft, calculated free testosterone; cbt, calculated bioavailable testosterone; SHBG, sex hormone binding globulin; TG, triglycerides; HDL-C, high density lipoprotein-cholesterol; FBG, fasting blood glucose; HOMA-R, homeostasis model assessment-resistance; Waist C, waist circumference.

126 Tanabe et al. Table 2 Correlation coefficient values between testosterone types Correlation coefficient values (R) aft cft cbt SHBG Age TT 0.531 0.730 0.710 0.638-0.098 aft 0.725 0.740-0.082-0.427 cft 0.986 0.019-0.403 cbt - 0.047-0.430 SHBG 0.351 Correlation coefficients were calculated from linear regression analysis as determined using the Pearson product moment correlation coefficient method (parametric). P<0.001. P <0.05 was considered to be statistically significant. TT, total testosterone; aft, free testosterone measured by analogue RIA; cft, calculated free testosterone; cbt, calculated bioavailable testosterone; SHBG, sex hormone binding globulin. Table 3 Association between various testosterone values and metabolic markers Correlation coefficient values (R) TT aft cft cbt SHBG BMI - 0.321-0.233-0.180-0.180-0.262 Waist C - 0.372-0.227-0.225-0.227-0.279 TG - 0.298 0.010-0.127-0.111-0.284 HDL-C 0.077 0.119-0.019-0.001-0.101 LDL-C 0.043 0.020 0.057 0.075-0.006 FBG - 0.166-0.182-0.137-0.133-0.079 Insulin - 0.276-0.146-0.194-0.178-0.206 HbA1C - 0.100-0.158-0.107-0.125 0.014 HOMA-R - 0.265-0.170-0.197-0.185-0.179 HOMA-β - 0.220-0.069-0.141-0.126-0.190 Correlation coefficient was determined using Pearson product moment correlation coefficient method (parametric). P <0.05 was considered to be statistically significant. P <0.05, P <0.01, P <0.001. TT, total testosterone; aft, free testosterone measured by analogue RIA; cft, calculated free testosterone; cbt, calculated bioavailable testosterone; SHBG, sex hormone binding globulin. MetS group. HDL-C and all testosterone values except aft were significantly lower in the MetS group than in the non-mets group (Table 1). The association between age and each of the various testosterone values and the mutual correlations of the testosterone values and SHBG are shown in Table 2. The TT level did not show any significant change with age. However, there was a significant inverse relationship with age and aft, cft and cbt. Also, SHBG showed a significant positive correlation with age. In addition, testosterone values significantly correlated with the other testosterone values to various degrees. The cft showed the strongest positive correlation with cbt (R = 0.986). Although the actual absolute value of aft was almost 13 % of that of cft, aft did show a positive correlation with cft (R = 0.725) and cbt (R = 0.740). The level of TT showed a significant positive correlation with that of SHBG (R = 0.638). All of the testosterone values and SHBG showed weak inverse correlations with BMI, waist circumference, insulin and HOMA-R (Table 3). Among them, TT showed the strongest coefficient value to all the components associated with obesity, insulin resistance and insulin secretion, namely BMI, waist circumference, TG, insulin, HOMA-R and HOMA-β with the exception of FBG. These results suggest that even in normal, healthy subjects, all endogenous testosterone values, with TT being the most predominant, are associated with fat and glucose metabolism including both insulin sensitivity and insulin secretion. All testosterone and SHBG values gradually decreased significantly as the number of MetS components based on the IDF criterion increased (Table 4). Statistical analysis revealed that this phenomenon was also most statisti-

Testosterone and metabolic syndrome 127 Table 4 Relationships between changes of various testosterone values to increases in the number of MetS (IDF criterion) components Age (years) TT (ng/ml) aft (pg/ml) cft (pg/ml) cbt (ng/ml) SHBG (nmol/l) None (N = 188) Numbers of MetS components 3 4 (N = 41) (N = 15) 5 (N = 5) (ANOVA) 52.7 ± 7.7 52.3 ± 6.9 53.2 ± 6.2 53.0 ± 7.4 0.976 5.17 ± 1.59 4.36 ± 1.48 3.78 ± 1.30 10.66 ± 3.53 10.45 ± 3.20 8.83 ± 3.08 83.5 ± 24.6 75.7 ± 20.1 70.0 ± 24.9 1.93 ± 0.57 1.79 ± 0.47 1.64 ± 0.56 50.0 ± 16.3 42.4 ± 16.2 39.0 ± 16.7 3.19 ± 0.31 <0.001 6.00 ± 1.97 # 0.007 57.9 ± 11.5 0.008 1.34 ± 0.33 0.015 40.3 ± 14.4 0.005 Data are expressed as means±sd. P <0.05, P <0.01, P <0.001 vs. none group, # P <0.01 vs. three components as determined by ad hoc Fisher s LSD method after ANOVA. The subject number that satisfied no component was 188. The subject number that satisfied three, four, and five components was 41, 15, and 5, respectively. Components of MetS are: i) either one or both of systolic or diastolic BP of 130/85 mmhg, ii) an elevated fasting blood glucose (FBG) level defined as 100 mg/dl, iii) serum triglyceride level 150 mg/dl, iv) serum HDL-cholesterol level <40 mg/dl, and v) abdominal obesity defined by a waist circumference of 90cm. TT, total testosterone; aft, free testosterone measured by analogue RIA; cft, calculated free testosterone; cbt, calculated bioavailable testosterone; SHBG, sex hormone binding globulin. The statistical significance of the changes was analyzed by Fisher s analysis of variance. P <0.05 was considered to be significant. Table 5 Logistic regression analysis for the detection of MetS (IDF criterion) Univariate analysis Multivariate analysis Variables Odds Ratios (95% CI) Odds Ratios (95% CI) Age (years) 0.997 (0.959 1.037) 0.883 0.998 (0.949 1.049) 0.935 TT <4 ng/ml 4.856 (2.630 8.967) <0.001 3.763 (1.669 8.485) 0.001 aft <10 pg/ml 1.679 (0.932 3.002) 0.084 0.890 (0.389 2.038) 0.783 cft <80 pg/ml 2.185 (1.192 4.007) 0.012 1.660 (0.611 4.511) 0.321 cbt <1.7 ng/ml 1.993 (1.111 3.575) 0.021 0.785 (0.275 2.242) 0.651 SHBG <47 nmol/l 2.895 (1.556 5.385) <0.001 1.938 (0.947 3.966) 0.070 χ 2 value was 2.446 (P = 0.964) as determined by the Hosmer Lemeshow test. Univariate and multivariate analyses were executed. The cut-off value in each variable was the median value. P <0.05 was considered to be statistically significant. In multivariate analysis, TT was the only variable to detect MetS in both analyses. cally significant in TT (P <0.001). Using binary logistic regression analysis, we examined testosterone markers that were statistically associated with the occurrence of MetS as defined by the IDF criterion (Table 5). As no cut-off values of testosterone levels for the detection of MetS are available, median values of the serum levels of TT <4 ng/ml, aft <10 pg/ml, cft <80 pg/ml, cbt <1.7 ng/ml and SHBG <47 nmol/l were tentatively examined as variables in logistic regression analyses. The usefulness of these values to diagnose MetS was examined by univariate or multivariate analysis. While univariate analysis revealed that the values of TT, cft, cbt and SHBG are statistically significant as markers for MetS defined by the IDF criteria, multivariate analysis revealed that TT was the only significant marker for MetS (Table 5). We also performed the same analyses using the Japanese MetS criterion. With this definition, the patient characteristics are summarized in Table 6. Among the 249 subjects, 58 were diagnosed as having MetS while

128 Tanabe et al. Table 6 Clinical characteristics of 249 subjects and comparison of various values between the MetS (Japanese criterion) group (N = 58) and the non-mets (N = 191) group Total (N = 249) With MetS (N = 58) Without MetS (N = 191) Age (years) 52.7 ± 7.4 53.8 ± 7.4 52.3 ± 7.4 0.173 BMI (kg/m 2 ) 23.9 ± 2.9 26.2 ± 2.7 23.2 ± 2.6 <0.001 Waist C (cm) 86.7 ± 8.0 93.5 ± 6.4 84.7 ± 7.3 <0.001 TT (ng/ml) 4.92 ± 1.61 4.33 ± 1.44 5.09 ± 1.62 0.002 aft (pg/ml) 10.42 ± 3.49 9.52 ± 3.34 10.70 ± 3.51 0.024 cft (pg/ml) 80.9 ± 24.2 76.4 ± 25.9 82.3 ± 23.6 0.105 cbt (ng/ml) 1.88 ± 0.56 1.79 ± 0.61 1.91 ± 0.54 0.157 SHBG (nmol/l) 47.8 ± 16.6 42.7 ± 16.2 49.4 ± 16.4 0.007 TG (mg/dl) 126.4 ± 66.3 176.7 ± 83.7 111.2 ± 51.2 <0.001 HDL-C (mg/dl) 54.7 ± 11.5 48.7 ± 10.1 56.5 ± 11.2 <0.001 FBG (mg/dl) 100.3 ± 12.9 111.2 ± 19.1 97.0 ± 7.9 <0.001 HOMA-R 1.66 ± 1.20 2.68 ± 1.74 1.35 ± 0.74 <0.001 Data are expressed as means±sd. were determined by the unpaired t test. Values were statistically significant at P <0.05. BMI, body mass index; TT, total testosterone; aft, free testosterone measured by analogue RIA; cft, calculated free testosterone; cbt, calculated bioavailable testosterone; SHBG, sex hormone binding globulin; TG, triglycerides; HDL-C, high density lipoprotein-cholesterol; FBG, fasting blood glucose; HOMA-R, homeostasis model assessment-resistance; Waist C, waist circumference. Table 7 Relationships between changes of various testosterone values to increases in the number of MetS (Japanese criterion) components Numbers of MetS components Waist <85cm Waist only Waist + 1 Waist + 2 Waist + 3 (ANOVA) (N = 109) (N = 26) (N = 56) (N = 48) (N = 10) Age 51.6 ± 7.1 51.9 ± 6.9 54.0 ± 8.0 54.1 ± 7.6 52.6 ± 6.4 0.178 (years) TT 5.47 ± 1.67 4.88 ± 1.48 4.46 ± 1.40 4.48 ± 1.48 3.64 ± 0.99 <0.001 (ng/ml) # aft 11.29 ± 3.63 9.99 ± 3.15 9.87 ± 3.24 9.78 ± 3.38 8.26 ± 2.97 0.007 (pg/ml) cft 86.3 ± 25.1 78.0 ± 20.3 76.6 ± 20.5 78.3 ± 27.8 67.1 ± 10.7 0.024 (pg/ml) cbt 2.01 ± 0.57 1.81 ± 0.48 1.76 ± 0.46 1.83 ± 0.64 1.59 ± 0.36 0.017 (ng/ml) SHBG 51.9 ± 15.4 49.8 ± 16.1 44.4 ± 17.6 43.5 ± 15.6 38.7 ± 19.6 0.003 (nmol/l) Data are expressed as means±sd. P <0.05, P <0.01, P <0.001 vs. none group, # P <0.01 vs. three components as determined by ad hoc Fisher s LSD method after ANOVA. The number of subjects that showed waist circumference <85 cm was 109. The number of subjects who satisfied only the condition of waist circumference 85 cm was 26. In addition to the abdominal obesity, the number of subjects who satisfied one (n = 56), two (n = 48) or three (n = 10) components of MetS, viz., either one or both a systolic or a diastolic BP of 130/85 mmhg, an elevated fasting blood glucose (FBG) level defined as 110 mg/dl and lipid abnormality defined as either one or both of a serum triglyceride level 150 mg/dl or a HDL-cholesterol level 40 mg/ dl, are listed. TT, total testosterone; aft, free testosterone measured by analogue RIA; cft, calculated free testosterone; cbt, calculated bioavailable testosterone; SHBG, sex hormone binding globulin. The statistical significance of the changes was analyzed by Fisher s analysis of variance. P <0.05 was considered significant. 191 were not. Additionally, subjects with MetS diagnosed by the Japanese criterion showed significantly lower concentrations of TT, aft and SHBG than subjects without MetS (Table 6); the P value was lowest in the case of TT. Unexpectedly, the levels of cbt and cft did not differ with the presence or absence of MetS. All of the testosterone values decreased significantly as the number of MetS components as defined by the Japanese criterion increased (Table 7). Binary logistic regression analysis also revealed that while all testosterone variables were significantly associated with the detection of MetS by univariate analysis, TT was the only significant indicator of MetS by multivariate analysis (Table 8). These trends (Tables 7 and 8) by Japanese criteria were essentially the same as observed in the analysis using the IDF criterion (Tables 4 and 5).

Testosterone and metabolic syndrome 129 Table 8 Logistic regression analysis for the detection of MetS (Japanese criterion) Univariate analysis Multivariate analysis Variables Odds Ratios (95% CI) Odds Ratios (95% CI) Age (years) 1.014 (0.975 1.054) 0.484 1.014 (0.965 1.065) 0.590 TT <4 ng/ml 5.084 (2.722 9.495) <0.001 4.050 (1.779 9.220) <0.001 aft <10 pg/ml 1.977 (1.078 3.625) 0.028 0.991 (0.428 2.293) 0.991 cft <80 pg/ml 2.391 (1.281 4.465) 0.006 1.446 (0.516 4.058) 0.483 cbt <1.7 ng/ml 2.278 (1.254 4.137) 0.007 0.820 (0.283 2.377) 0.820 SHBG < 47 nmol/l 2.352 (1.268 4.363) 0.007 1.617 (0.779 3.355) 0.197 χ 2 value was 6.463 (P = 0.596) as determined by the Hosmer Lemeshow test. Univariate and multivariate analyses were executed. P <0.05 was considered statistically significant. In multivariate analysis, TT was the only variable to detect MetS in both analyses. Discussion Testosterone is a crucial determinant of male body composition. Fat accumulation is reported to have an inverse relationship with blood testosterone levels [1, 11, 12, 18]. In another report, reducing circulating testosterone levels in healthy young males by a gonadotropin releasing hormone (GnRH) analogue resulted in an increased rate of somatic fat and a decrease in energy consumption in the stable state [19]. This phenomenon has also been observed in patients with prostate cancer, treated with GnRH agonist. These patients showed fat accumulation, insulin resistance, and a high prevalence of diabetes [20]. In vitro, testosterone, via androgen receptor (AR) signaling, stimulates the differentiation of stem cells into myocytes, but inhibits the differentiation into adipocytes [21]. Also in mature adipocytes, testosterone stimulates lipolysis induced by catecholamines [22]. In vivo, testosterone inhibits lipoprotein lipase activity in human abdominal fat tissue and reduces the uptake of triglycerides into adipocytes [23]. In male AR knockout mice, subcutaneous and visceral fat accumulation and obesity were observed, but not in female AR knockout mice [24]. These reports suggest that the intrinsic testosterone-ar system increases the energy consumption and reduces fat accumulation, and the impairment of this system is possibly related to obesity and MetS. In the present study of 249 healthy middle-aged men, we determined the serum levels of various testosterone values, including TT, aft, cft, cbt and SHBG and investigated the relationship between these values and various metabolic markers related to MetS. All the various testosterone and SHBG values showed a weak, but significant, inverse association with BMI, waist circumference and glucose metabolism. The presence of MetS was significantly associated with lowered TT, cft, cbt and SHBG (IDF criterion) or with lowered TT, aft and SHBG values (Japanese criterion). For both criteria, TT showed the smallest (Tables 1 and 6). In addition, it is worthy of note that TT showed the strongest inverse correlation with many of the metabolic markers that are related to obesity and insulin resistance (Table 3). Importantly, while all serum values of aft, cft and cbt significantly decreased with age, TT did not change with age. Furthermore, multivariate analysis revealed that lowered TT, tentatively defined as <4 ng/ml, was the only detection marker for MetS. It is controversial whether serum TT levels decrease with aging or not [25 29]. As an example, a slight decrease of TT with aging was observed in the Framingham Heart Study [25]. However, recent large scale investigations targeting healthy Chinese men [26] and healthy Japanese men [27-29] have repeatedly concluded that the serum levels of TT were unchanged by aging. According to one report, the differences that have been observed may be owing to racial differences but not to a difference in the assay method [29]. Specifically, even though they examined the TT levels of 498 Japanese men using the same method (liquid chromatography tandem mass spectrometry) in the same laboratory as performed in the Framingham Heart Study [25], the TT levels of Japanese men remained unchanged by aging [29]. These facts indicate that serum TT levels in Japanese men are age-independently stable and are

130 Tanabe et al. an indicator of MetS, at least in middle-age. In previous reports, differences have been observed in the relationship between blood testosterone levels and MetS, although the various types of testosterone were not investigated, as was done in our study. One report described that in 1226 men with low levels of TT and SHBG, but not cft, were significantly associated with MetS [30]. Meta-analysis of 55 observational studies revealed that MetS in men was associated with TT and FT, and that SHBG was associated with MetS in both men and women [31]. In a meta-analysis of longitudinal studies, TT levels at baseline were lower in a MetS subgroup compared with non-mets [32]. In prospective studies, one report revealed that the basal cft level was an independent predictor for the occurrence of MetS [26], but this was not observed in another metaanalysis study [32]. Collectively, although the analysis method for FT was not always clear in the above reports, especially in the meta-analysis studies, most reports seem to support that TT, compared with FT, may have a closer relationship with MetS. The bioavailable testosterone fraction, namely bt or cbt, is generally considered to be the most reliable bioactive testosterone marker. However, our results clearly indicated that among various testosterone markers including cbt, TT is still the most reliable testosterone value for the detection of MetS. An inverse strong association of TT and various metabolic factors has also been reported in 1150 Japanese men aged 30 years, although other testosterone values have not been examined [28]. In the present study, SHBG showed a significant inverse relationship with waist circumference, BMI, blood insulin levels, HOMA-R and HOMA-β. Additionally, a significantly lower SHBG level was observed in subjects with MetS compared with those without MetS, according to both MetS criteria. These results suggest that a relatively low SHBG level may be a risk factor for MetS or for impaired glucose metabolism. A decrease in SHBG level can result in a decrease in the largest of the TT fraction, the SHBG-bound T, and thus result in a decrease in the TT level. This may partly account for the reason that TT is the strongest detection marker for MetS. While the in multivariate analyses were much lower in SHBG compared with those of aft, cft and cbt, the value of SHBG was not statistically significant. Thus, bioactive fractions of testosterone are suggested to be also important for the pathogenesis of MetS. The measurement of serum SHBG is not a realistic prospect as it is not covered by the health insurance system in Japan. In this respect, the use of TT as a diagnostic tool of MetS, which reflects both low testosterone and low SHBG, is covered by Japanese health insurance and would, therefore, be a reasonable option. In summary, various testosterone values were examined in healthy male subjects and all values were significantly associated with markers of lipid and glucose metabolism, suggesting that endogenous testosterone may be protective against MetS in healthy subjects. Among all the various testosterone values and SHBG, TT was the only age-independent marker and also the strongest marker for the detection of MetS in Japanese males. Acknowledgments This research was partially supported by a Grantin-Aid for Scientific Research (B) Japan Society for the Promotion of Science (JSPS) (ID: 23390248). We greatly thank Mrs. Kusamoto K, Ochi M, Onimaru T and Makita R in the Department of Preventive Medicine of Iizuka Hospital for their great help in this study. Conflicts of Interest The authors declare no conflict of interest relevant to this manuscript. References 1. Zitzmann M (2009) Testosterone deficiency, insulin resistance and the metabolic syndrome. Nat Rev Endocrinol 5: 673-681. 2. Araujo AB, Dixon JM, Suarez EA, Murad MH, Guey LT, et al. (2011) Endogeneous testosterone and mortality in men: a systematic review and meta-analysis. J Clin Endocrinol Metab 96: 3007-3019. 3. Namiki M, Akaza H, Shimazui T, Ito N, Iwamoto T, et al. (2008) Clinical practice manual for late-onset hypogonadism syndrome. Int J Urol 15: 378-388. 4. Dunn JF, Nisula BC, Rodbard D (1981) Transport of steroid hormones: binding of 21 endogeneous steroids to both testosterone-binding globulin and corticosteroid-binding in human placenta. J Clin Endocrinol Metab 53: 58-68. 5. Vermeulen A, Verdonck L, Kaufman JM (1999) A criti-

Testosterone and metabolic syndrome 131 cal evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab 84: 3666 3672. 6. Laaksonen DE, Niskanen L, Punnonen K, Nyyssonen K, Tuomainen TP, et al. (2004) Testosterone and sexhormone-binding globulin predict the metabolic syndrome and diabetes in middle-aged men. Diabetes Care 27: 1036-1041. 7. Kupelian V, Page ST, Araujo AB, Travison TG, Bremner WJ, et al. (2006) Low sex hormone-binding globulin, total testosterone, and symptomatic androgen deficiency are associated with development of the metabolic syndrome in nonobese men. J Clin Endocrinol Metab 91: 843-850. 8. Ding EL, Song Y, Manson JE, Hunter DJ, Lee CC, et al. (2009) Sex hormone-binding globulin and risk of type 2 diabetes in women and men. N Engl J Med 361: 1152-1163. 9. Kalme T, Koistinen H, Loukovaara M, Koistinen R, Leinonen P (2003) Comparative studies on the regulation of insulin-like growth factor-binding protein-1 (IGFBP-1) and sex hormone-binding globulin (SHBG) production by insulin and insulin-like growth factors in human hepatoma cells. J Steroid Biochem Mol Biol 86: 197-200. 10. Kaukua J, Pekkarinen T, Sane T, Mustajoki P (2003) Sex hormone and sexual function in obese men losing weight. Obes Res 11: 689-694. 11. Grossmann M (2011) Low testosterone in men with type 2 diabetes: significance and treatment. J Clin Endocrinol Metab 96: 2341-2353. 12. Rao PM, Kelly DM, Jones TH (2013) Testosterone and insulin resistance in the metabolic syndrome and T2DM in men. Nat Rev Endocrinol 9: 479-493. 13. Ninomiya T, Kubo M, Doi Y, Yonemoto K, Tanizaki Y, et al. (2007) Impact of metabolic syndrome on the development of cardiovascular disease in a general Japanese population: the Hisayama study. Stroke 38:2063-2069. 14. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, et al. (1985) Homeostasis model assessment:insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 28: 412-419. 15. Alberti K.G.M.M, Eckel RH, Grundy SM, Zimmet PZ, Cleeman JI, et al. (2009) Harmonizing the metabolic syndrome: A Joint interim statement of the International Diabetes Federation task force on epidemiology and prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity. Circulation 120: 1640-1645. 16. The Examination Committee of Criteria for Metabolic syndrome (2005) The definition and criteria of metabolic syndrome. Nihon Naika Gakkai Zasshi (J Jpn Soc Int Med) 94: 794 809. (In Japanese) 17. Yanase T, Haji M, Natori S, Ogo A, Nawata H, et al. (1988) Fundamental and clinical studies on a solidphase radioimmunoassay kit for the quantitative measurement of free testosterone in serum. Clinical Endocrinol (Tokyo) 69: 357-362. (In Japanese) 18. Yanase T, Fan W, Kyoya K, Min L, Takayanagi R, et al. (2008) Androgens and metabolic syndrome: lessons from androgen receptor knockout (ARKO) mice. J Steroid Biochem Mol Biol 109: 254-257. 19. Mauras N, Hayes V, Weich S, Rini A, Helgeson K, et al. (1998) Testosterone deficiency in young men: marked alterations in whole body protein kinetics, strength, and adiposity. J Clin Endocrinol Metab 83: 1886-1892. 20. Keating NL, O Malley AJ, Smith MR (2006) Diabetes and cardiovascular disease during androgen deprivation therapy for prostate cancer. J Clin Oncol 24: 4448-4456. 21. Singh R, Artaza JN, Taylor WE, Gonzalez-Cadavid NF, Bhasin S (2003) Androgens stimulate myogenic differentiation and inhibit adipogenesis in C3H 10T1/2 pluripotent cells through an androgen receptor-mediated pathway. Endocrinology 144: 5081-5088. 22. Xu XF, De Pergola G, Bjorntorp P (1991) Testosterone increases lipolysis and the number of β-adrenoreceptors in male rat adipocytes. Endocrinology 128: 379-382. 23. Marin P, Oden B, Bjorntorp P (1995) Assimilation and mobilization of triglycerides in subcutaneous abdominal and femoral adipose tissue in vivo in men: effects of androgens. J Clin Endocrinol Metab 80: 239-243. 24. Fan W, Yanase T, Nomura M, Okabe T, Goto K, et al. (2005) Androgen receptor null male mice develop lateonset obesity caused by decreased energy expenditure and lipolytic activity but show normal insulin sensitivity with high adiponectin secretion. Diabetes 54: 1000-1008. 25. Haring R, Xanthakis V, Coviello A, Sullivan L, Bhasin S, et al. (2012) Clinical correlates of sex steroids and gonadotropins in men over the late adulthood: the Framingham Heart Study. Int J Androl 35: 775-782. 26. Li JY, Li XY, Li M, Zhang GK, Ma FL, et al. (2005) Decline of serum levels of free testosterone in aging healthy Chinese men. Aging Male 8: 203-206. 27. Iwamoto T, Yanase T, Koh E, Horie H, Baba K, et al. (2004) Reference ranges of total serum and free testosterone in Japanese male adults. Nihon Hinyokika Gakkai Zasshi 95: 751-760. (In Japanese) 28. Tsujimura A, Miyagawa Y, Takezawa K, Okuda H, Fukuhara S, et al. (2013) Is low testosterone concentration a risk factor for metabolic syndrome in healthy middle-aged men? Urology 82:814-819. 29. Taya M, Koh E, Izumi K, Iijima M, Maeda Y, et al. (2014) Comparison of testosterone fractions between Framingham Heart Study participants and Japanese participants. Int J Urol 21: 689-695.

132 Tanabe et al. 30. Li C, Ford ES, Li B, Giles WH, Liu S (2010) Association of testosterone and sex hormone-binding globulin with metabolic syndrome and insulin resistance in men. Diabetes Care 33: 1618-1624. 31. Brand JS, van der Tweel I, Grobbee DE, Emmelot-Vonk MH, van der Schouw YT (2011) Testosterone, sex hormone-binding globulin and the metabolic syndrome: a systemic review and meta-analysis of observational studies. Int J Epidemiol 40: 189-207. 32. Corona G, Monami M, Rastrelli G, Aversa A, Tishova Y, et al. (2011) Testosterone and metabolic syndrome: a meta-analysis study. J Sex Med 8: 272-283.