Decline of serum levels of free testosterone in aging healthy Chinese men

Abstract

Objective. To investigate the age-related change of serum androgen levels in healthy men and to define a cut-off value of serum testosterone for the diagnosis of androgen deficiency in the aging male.

Method. 1080 healthy men aged 20 to ⩾70 years old were enrolled in Beijing, Shanghai, Xian and Chongqing. Luteinizing hormone (LH), follicle-stimulating hormone (FSH), total testosterone (T), calculated free testosterone (cFT), sex hormone binding globulin (SHBG), 17beta-oestradiol (E2), the T/LH ratio, and T/SHBG as a free testosterone index (FTI) were all determined.

Results. Serum total T did not significantly decline, but the cFT, T/LH and FTI progressively decreased with aging. To determine androgen deficiency, the 10th percentile value of men <40 years was defined as the lower cut-off value for cFT, T/LH or FTI, which were 0.3 nmol/L, 2.8 nmol/IU, and 0.4 nmol/IU respectively. With the median value of cFT of men aged between 20 and 49 years as the criterion, the level of cFT was lower in 2.82% of men from 40 to 49 years, in 19.53% from 50 to 59 years, in 22.57% from 60 to 69 years, and in 33.19% of men ⩾70 years. Taking the above value of cFT as the cut-off point, the prevalence of androgen deficiency in men 40–49 years was 13.0%, 31.8% in men 50–59 years, 30.1% in men 60 to 69 years, and 46.7% in men >70 years.

Conclusions. (i). While serum total T values do not decline with aging, the levels of cFT gradually decline with aging; (ii) when using the value of cFT of the 10th percentile of men aged 20 to 39 years as the cut-off point, the prevalence of androgen deficiency was <15% before the age of 50 years, and about 30% thereafter, approaching 45% after the age of 70 years; and (iii) in this study the values of T/LH paralleled those of cFT closely; therefore, T/LH could serve as a surrogate for cFT.

Keywords:

• Calculated free testosterone
• T/LH
• aging men

Introduction

In 1940s, Werner was the first to describe a climacteric-like syndrome in men, with symptoms of sexual dysfunction, loss of libido, palpitations, flushing, episodic sweating, loss of energy, depression, and mood swings [1]. Over the past two decades a large number of studies have documented that serum total testosterone decline with aging – particularly free, non-bound testosterone and bio-available testosterone – while the serum level of sex hormone binding globulin (SHBG) increase with aging [2-6]. Almost all commentaries dismiss the concept of a male menopause or andropause, but rather describe partial androgen deficiency in the aging male (PADAM) [5-7] or late onset hypogonadism (LOH) [8]. It still remains a point of debate how to define the criteria for testosterone deficiency in elderly men [8-10]. Age-appropriate reference values are lacking; therefore, it has been proposed that the same criteria as for the general male population be used [11]. We conducted a large-scale study of 1080 healthy adult men, aged 20 years and above, in four major cities in China (Beijing, Shanghai, Xian and Chongqing), between December 2002 and December 2003. Depending on this database, the cut-off point for definition of androgen deficiency was set to be the 10th percentile value of men aged 20 to 39 years. We did not observe a decline of serum total testosterone with aging, but in conformity with studies in other parts of the world, there was a progressive decrease of free testosterone levels [2-6].

Subjects and methods

Subjects

Two hundred and seventy subjects were enrolled from each of the following four cities–Beijing, Shanghai, Xian and Chongqing–in China. These subjects were divided in nine subgroups, each comprising 30 subjects, based on age, the categories being: 20–29, 30–39, 40–49, 50–54, 55–59, 60–64, 65–69, 70–74, and ⩾70 years old. Written informed consent was obtained from each participant. Men had to be aged ⩾20 years old to participate in this study. Subjects could only be included in this study if there was: (i) no acute systemic disease; (ii) normal functions of heart, liver, and kidney; (iii) no diabetes or other endocrine disease; (iv) no drug addiction; and (v) if they were living independently (not institutionalized).

Anthropometric and hormone measurements

The history of diseases and medication was taken for each subject, and a physical examination performed, with body mass index (BMI, kg/m2), and waist–hip ratio (WHR, cm/cm) being measured. A peripheral blood sample was drawn between 07:30 h to 08:30 h on the morning following the examination, and serum was kept in −80°C until hormone determination.

Luteinizing hormone (LH), follicle-stimulating hormone (FSH), total testosterone (T), and 17beta-oestradiol (E2) were determined by Automated Chemiluminescence System (ACS: 180®, Bayer) with intra-assay CVs of 4.8%, 3.0%, 6.5%, and 8.3%, respectively, and inter-assay CVs of 6.6%, 4.6%, 7.0%, and 8.6%, respectively. SHBG measurement was performed by ILISA (IBL, Hamburg). The intra-assay CV was 5.5%, and the inter-assay CV was 9.0%. The free testosterone (cFT) was calculated using the formula developed by Vermeulen et al. [12], which has been considered to be a reliable method [9]. We further calculated the ratio of T (nmol/L) to LH (IU/L), and the free testosterone index (FTI) to be the ratio of T (nmol/L) to SHBG (nmol/L).

Statistical analysis

The serum androgen values were not normally distributed; therefore percentiles were used for statistical analysis. The 10th percentile value of androgen at age less than 40 years was arbitrarily set to be the cut-off point of androgen deficiency.

As there was no significant difference between hormone measurements in men aged 40 to 49 years and those from the younger group (20–39 years), they were grouped together. Men over 50 years old were regrouped by 10-year intervals. The Kruskall-Wallis test was used to compare differences in hormone levels between age groups, and ANOVA was used to determine prevalence of androgen deficiency, which we defined in this study as being values lying below the 10th percentile of the values obtained from participating males under 39 years.

Results

The serum hormone levels, expressed as median values for each age subgroup (in decades), are presented in Table I. The serum level of total T did not change significantly with aging. However, the values for cFT, T/LH ratio (TSI), and FTI significantly decreased with aging, whereas this trend was weak for the ratio of T/E2. By contrast, LH, FSH, SHBG and E2 serum levels increased with aging. Spearman's analysis (Table II) showed a negative correlation between cFT, T/LH or FTI with age, and with serum levels of LH or FSH.

Table I.  Age-related change of serum sex hormones expressed by median (n).

	20 Age (yrs)	40 Age (yrs)	50 Age (yrs)	60 Age (yrs)	70 Age (yrs)	K_W Age (yrs)	p
LH (IU/L)	3.10 (175)	3.85 (92)	4.00 (218)	4.50 (185)	6.60 (192)	156.04	0.0000
FSH (IU/L)	3.30 (171)	5.30 (89)	6.80 (218)	8.30 (179)	13.30 (178)	319.91	0.0000
T (nmol/L)	19.83 (180)	21.58 (95)	20.20 (222)	20.96 (189)	20.77 (192)	4.78	0.3103
CFT (nmol/L)	0.44 (199)	0.41 (95)	0.37 (222)	0.35 (188)	0.30 (192)	140.52	0.0000
E2 (pmol/L)	90.42 (172)	112.91 (86)	110.17 (210)	115.84 (176)	114.38 (186)	30.61	0.0000
SHBG (nmol/L)	32.15 (162)	43.30 (77)	44.20 (197)	47.90 (169)	54.40 (152)	137.29	0.0000
TSI (T/LH)	6.55	6.02	4.62	4.61	3.28	96.74	0.0000
FTI (T/SHBG)	0.64	0.55	0.44	0.43	0.37	234.20	0.0000
T/E2	0.21	0.18	0.19	0.18	0.18	10.00	0.0404

Table II.  Spearman's analysis for the relation between androgen and aging.

	Age	BMI (kg/m^2)	WHR	LH (IU/L)	FSH (IU/L)
T (nmol/L)
r =	0.0496	−0.1516	−0.0553	0.0163	0.0508
P =	0.1716	0.0001	0.1681	0.6484	0.1644
n =	793	621	622	779	752
CFT (nmol/L)
r =	−0.4348	−0.0435	−0.1504	−0.2704	−0.3785
P =	< 0.0001	0.3187	0.0005	< 0.0001	< 0.0001
n =	648	528	530	675	660
SHBG (nmol/L)
r =	0.4239	−0.1623	0.0844	0.3230	0.3782
P =	< 0.0001	0.0002	0.0520	< 0.0001	< 0.0001
n =	640	529	531	676	661
TSI
r =	−0.3407	0.0017	0.0126	−0.8106	−0.4963
P =	< 0.0001	0.9667	0.7575	< 0.0001	< 0.0001
n =	748	607	608	779	748
FTI
r =	−0.5250	0.0603	−0.3432	−0.3462	−0.4579

The 10th and 90th percentile values of T, cFT, TSI, and FTI are shown in Table III. The 10th percentile value of the results of men aged 20 to 39 years were grouped together and were subsequently considered to be the lower limit of normal values for men below middle age. On the basis of this calculation, the lower cut-off value for cFT to define androgen deficiency was 0.3 nmol/L, 2.8 nmol/IU for TSI, and 0.4 nmol/IU for FTI. For the purpose of this study, the percentage of men (in each age category) with levels above and below this cut-off point of cFT, TSI, and FTI are presented in Table IV. The cFT serum level started to decrease in a small number of men in their fifth decade, and became more profound with aging. The prevalence of androgen deficiency is presented in Table V. Using cFT as a criterion, 13% of adult men showed androgen deficiency between 40 to 49 years, approximately 30% between the ages of 50 to 69 years, and about 45% in those men over 70 years.

Table III.  The 10th and 90th percentiles for t, cFT, TSI, and FTI.

	T (nmol/L)	CFT (nmol/L)	TSI	FTI
	10	10	10	10
Age (yrs)	90	90	90	90
20∼	11.485	0.297	2.804	0.416
	27.725	0.608	12.429	0.976
40∼	8.779	0.286	1.795	0.359
	32.930	0.587	11.299	0.940
50∼	11.434	0.225	1.495	0.296
	29.529	0.493	10.261	0.724
60∼	13.880	0.215	1.317	0.290
	31.334	0.482	9.738	0.623
70∼	12.353	0.190	1.182	0.253
	30.987	0.404	6.704	0.503

Table IV.  The rate of decline for cFT, TSI, and FTI, expressed as median.

Age (yrs)	cFT	TSI	FTI
20∼	100	100	100
40∼	97.18 (−2.82)	91.91 (−8.09)	85.96 (−14.04)
50∼	80.47 (−19.53)	70.55 (−29.45)	68.64 (−31.36)
60∼	77.43 (−22.57)	70.51 (−29.49)	66.08 (−33.92)
70∼	66.81 (−33.19)	50.20 (−49.80)	56.89 (43.11)

Table V.  The relationship between partial androgen deficiency and aging.

	40∼	50∼	60∼	70∼	X^2	p
CFT (nmol/L)	13.04%	31.76%	30.14%	46.67%	57.042	0.0000
TSI	14.46%	23.35%	22.84%	42.20%	54.162	0.0000
FTI	17.39%	40.54%	47.26%	71.11%	129.548	0.0000

Discussion

An age-related decline of serum total testosterone, and particularly serum free- or bio-available testosterone, has been reported by a number of cross-sectional and longitudinal studies [2-6]. Our data fail to illustrate a decline in serum total testosterone with male aging, but do reflect previous findings regarding the decline of serum cFT in healthy men, with a progressive increase of serum SHBG levels with aging [13]. FT or bio-available T is the proportion of circulating T not bound to SHBG or albumin, respectively, and is regarded as the physiologically active fraction of total testosterone [12]. The etiology of the decline of serum T with aging is partly the result of testicular failure, and partly a result of diminished stimulation of the Leydig cell by decreased output of LH [6]. We did not find serum total T to decline with aging, but serum LH levels rose significantly with aging. This is most likely an expression to a certain extent of a compensated testicular failure with aging [14], which sometimes fails [14], but probably not in the sample of men we studied, as evidenced by their stable serum total T levels with aging. The maintenance of relatively stable total T levels with aging is probably largely explained by the simultaneous increase of SHBG serum level. Serum SHBG is suppressed by circulating insulin, a feature of obesity or diabetes type II, which is rather common in aged men [2],[15], but this did not affect our finding of an increase of serum SHBG with aging.

Serum T falls with aging and, as a result of the feedback mechanism, serum LH usually – but not always – rises with aging; we reasoned that the T/LH ratio in elderly men might provide insight in their gonadal status. Our data show that the decline in the value of T/LH with aging parallels closely the value of cFT. Using the T/LH ratio, there was approximately 7% underestimation in two age groups (50–59 years and 60–69 years). It merits further study as to whether the T/LH ratio is an acceptable surrogate for the cFT. In China, SHBG is not routinely measured in clinical practice, although LH is. The introduction of T/LH would be more convenient and cost effective. The FTI is the calculated value of T/SHBG, and is an indirect measure of free testosterone [3]. In our data, the FTI differed more from cFT than T/LH. With FTI there were overestimations of 3% at 40–49, 10% at 50–59, 17% at 60–69, and 25% at ⩾70 years. Therefore, in the context of our study, T/LH was a better surrogate marker of cFT, and more useful than FTI in clinical practice.

The present study shows a progressive increase in serum levels of SHBG, LH, FSH, and E2 in relation to aging. The increase of SHBG is likely to result from the decrease of androgen and growth hormone system (GH and IGF-1) with aging [16]. The elevations of LH and FSH are an expression of the decline in testicular function [6]. Circulating levels of E2 are mainly derived from the conversion of androgens to estrogens in peripheral tissue by aromatase [17]. As described earlier [17], we found a positive relationship between serum levels of E2 and T (r = 0.1509, p < 0.001; data not shown). The circulating levels of E2 in aging men may be within normal range or slightly elevated with aging [15],[17]. The normal or mildly elevated concentrations of E2 probably suggest an increase in activity of the enzyme aromatase [17], which results in a decline of the T/E2 ratio, but the change is not large.