Eur J Endocrinol
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

DOI: 10.1530/EJE-07-0873
European Journal of Endocrinology, Vol 159, Issue 4, 459-468
Copyright © 2008 by European Society of Endocrinology
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
EJE-07-0873v1
159/4/459    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Lapauw, B
Right arrow Articles by Kaufman, J M
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Lapauw, B
Right arrow Articles by Kaufman, J M
CLINICAL STUDY

The decline of serum testosterone levels in community-dwelling men over 70 years of age: descriptive data and predictors of longitudinal changes

B Lapauw1,2,*, S Goemaere2,*, H Zmierczak2, I Van Pottelbergh1,{dagger}, A Mahmoud1, Y Taes1, D De Bacquer3, S Vansteelandt4 and J M Kaufman1,2

1 Department of Endocrinology2 , Unit for Osteoporosis and Metabolic Bone Diseases3 Department of Public Health, Ghent University Hospital, De Pintelaan 185, 6K12 I.E., B-9000 Ghent, Belgium4 Department of Applied Mathematics and Computer Science, Ghent University, Krijgslaan 281, 9000 Ghent, Belgium

(Correspondence should be addressed to B Lapauw; Email: bruno.lapauw{at}ugent.be)

* B Lapauw and S Goemaere contributed equally to this work Back

{dagger} I Van Pottelbergh is now at Department of Endocrinology, OLV-Hospital Aalst, Aalst, Belgium Back

Objective: This study was designed to assess longitudinal changes in serum testosterone levels, explore relationships with aging, genetic-, health-, and lifestyle-related factors, and investigate predictors of changes in healthy elderly men.

Design: Population-based, longitudinal, 4-year observational study in 221 community-dwelling men aged 71–86 years at baseline.

Methods: Hormone levels assessed by immunoassay, anthropometry, questionnaires on general health, and genetic polymorphisms. Predictors of changes in testosterone levels explored using linear mixed-effects modeling for longitudinal analyses.

Results: Total testosterone (TT), free testosterone, and bioavailable testosterone (BioT) levels decreased with aging, decreases in BioT being most marked. No changes in sex hormone-binding globulin (SHBG) or estradiol (E2), while LH and FSH levels increased during follow-up. Subjects who gained weight displayed a greater decline in TT levels, mainly due to decreasing SHBG levels. However, baseline body composition was not predictive of subsequent changes in testosterone levels. Baseline E2 (P=0.023 to 0.004), LH (P=0.046 to 0.005), and FSH (P<0.002) levels were independently positively associated with a faster decline in testosterone fractions, although only FSH remained significant when adjusting for baseline testosterone (P=0.041–0.035). Carriers of a ‘TA’ haplotype of the estrogen receptor {alpha} gene (ER{alpha}) PvuII and XbaI polymorphisms displayed a slower decline of TT and BioT (P=0.041–0.007).

Conclusions: In elderly men with already low serum testosterone levels, a further decline was observed, independent of baseline age. The identification of FSH levels as a predictor of this decline appears to reflect the testicular mechanisms of aging-related changes in testosterone production, whereas associations with E2 and ER{alpha} polymorphisms are suggestive of estrogen-related processes, possibly related to changes in the neuroendocrine regulation of testosterone production.


   Introduction
Top
 Introduction
Subjects and methods
 Results
 Discussion
 Declaration of interest
 References
 
It is now widely accepted that normal chronological aging in men is associated with a progressive decline in androgen levels. More than 20% of healthy men over 60 years of age present with serum levels below the reference range for young men. In contrast to women, who experience a rather precipitous and profound reduction of endogenous estradiol (E2) production during menopausal transition, the decline of testosterone production in aging men is much more gradual and individually highly variable. This decline results from both testicular changes and altered neuroendocrine regulation of luteinizing hormone (LH) secretion (1). Between the third and eighth decade of life, a 20–30% decrease in population mean serum testosterone levels can be observed. As serum levels of sex hormone-binding globulin (SHBG) increase with age, the decrease in serum-free testosterone (FT) and non-SHBG-bound testosterone levels, often referred to as bioavailable testosterone (BioT), is of even greater magnitude reaching up to 50% over the same age range (2, 3, 4, 5).

Understanding the physiology of this decline is important because testosterone has actions in many tissues and low serum testosterone levels have been associated with a variety of adverse conditions, which comprise increased risk of diabetes mellitus and impaired glucose tolerance, reduced bone and muscle mass, increased (abdominal) fat mass, impaired sexual function, decreased quality of life, and even increased mortality risk (1, 6, 7, 8). These features resemble the symptoms observed in young hypogonadal men, and in accordance with this view, treatment of hypogonadism in both young and older men with low serum testosterone may result in an improvement of some of these conditions (1, 9). However, a causal relationship between these clinical changes and the age-related decrease in testosterone levels remains a matter of debate. Indeed, many of these conditions are non-specific, multifactorially determined, and often associated with normal aging per se (1). Up to now, the risk–benefit ratio for testosterone supplementation in elderly men has not been established (10).

While chronological aging per se seems to be associated with declining testosterone production, co-morbid chronic illnesses, and certain medications as well as the aging-related changes in body composition may accentuate and contribute to these changes (1, 4, 11, 12). Furthermore, lifestyle-related factors such as smoking may also influence testosterone levels (2, 7, 11). In addition, the possibility of genetic polymorphisms influencing this decline cannot be excluded (13). Most information about changes in testosterone levels with aging results from cross-sectional studies (2, 3, 5, 7, 14, 15), while information from longitudinal studies remains more limited (4, 16, 17, 18, 19). Moreover, it should be noted that most studies report on testosterone levels of middle-aged men and relatively few studies have focused specifically on men older than 70 years, whereas many of the adverse conditions potentially associated with the age-related decline of testosterone levels occur relatively late in life.

In this population-based longitudinal study, we followed a homogenous group of 221 community-dwelling elderly men, aged 71–86 years at baseline, for a period of 4 years. We report on the distribution and changes of gonadal steroid and gonadotropin concentrations. Using mixed-effects modeling, predictors of longitudinal changes in androgen levels were explored.


   Subjects and methods
Top
 Introduction
 Subjects and methods
Results
 Discussion
 Declaration of interest
 References
 
Study subjects

Subjects were recruited from the population registry of the semi-rural community of Merelbeke (Belgium). A sample of 352 community-dwelling men, aged 71–86 years, agreed to participate (initial participation rate was 47.1%). This longitudinal, population-based study was specifically designed to investigate the process of aging, focusing on hormonal changes, bone metabolism, and body composition in elderly men at yearly intervals over a period of 4 years. The study was approved by the ethics review board of the Ghent University Hospital (Belgium). All the participants gave written informed consent for participation in this study and completed questionnaires pertaining to general health (Geriatric Depression Scale, Short Form-36 and Rapid Disability Rating Scale-2 (20)), dietary habits, and physical activity. Baseline characteristics, exclusion and inclusion criteria have been described extensively in previous publications (20, 21, 22). Following exclusions because of past or current history of disorders or treatments known to affect androgen status, 221 eligible subjects were included in the longitudinal analyses, for a total of 912 observations. Five serial measurements were available in 125 subjects, for the remaining subjects two to four data points were available. Reasons for loss to follow-up were institutionalization, withdrawal of consent, death, and occurrence of diseases or treatments known to affect androgen levels. All exclusions were censored before hormonal analysis. Longitudinal changes in BioT over 4 years in 214 men of this longitudinal cohort have previously been shown in Fig. 1 of the review article under reference (1).


Figure 1
View larger version (48K):
[in this window]
[in a new window]

  		Figure 1 Evolution of mean TT and BioT levels over time (profile plots with local regression smoother and 95% confidence intervals are depicted).

 
Hormonal assays

Each consecutive year, venous blood was obtained between 0800 and 1000 h after overnight fasting and serum was stored at –80 °C until batch analysis. Blood collection was completed over a period of 2 months during summer, except for the first year when blood collection was performed from March till June. Commercial immunoassays were used to determine serum levels of total testosterone (TT), follicle stimulating hormone (FSH), and LH (Medgenix, Fleurus, Belgium), E2 (Clinical Assay, DiaSorin s.r.l., Saluggia, Italy, according to a modified protocol that doubles the serum amount (22)), SHBG (Orion Diagnostica, Espoo, Finland), insulin (Pharmacia & Upjohn Diagnostics AB), insulin-like growth factor-1 (IGF-1), IGF-binding protein 3 (IGF-BP3; Diagnostic Laboratory Systems Inc., Webster, TX, USA), and leptin (Linco Research Inc., St Louis, MO, USA). Testosterone and SHBG levels were determined at all visits, whereas E2, LH, and FSH were assessed at the first and last study visits. All samples from the same subject were assayed in a single assay run using duplicate measurements. The intra- and interassay coefficients of variation (CV) were 4.8 and 5.1% for testosterone and 3.6 and 6.6% for SHBG respectively. Intra- and inter-CV for all other measurements were below 10% and 15% respectively. Serum non-SHBG-bound testosterone (BioT), FT, non-SHBG-bound E2 (‘bioavailable E2’; BioE2), and free E2 (FE2) were calculated from serum TT, total E2, SHBG, and albumin concentrations using a previously validated equation derived from mass action law (23, 24).

Determination of gene polymorphisms

Genomic DNA was extracted from EDTA-treated blood using a commercial kit (Qiagen Midi Kit, Qiagen Inc).

Androgen receptor gene (AR) CAG repeat length  As previously described (21), PCR to amplify exon 1 of the AR gene with primers 5'-AGCCTGTTGAACTCTTCTGAGC-3' (sense) and 5'-CTGCCTTACACAACTCCTTGGC-3' (antisense). After ethanol precipitation, the amplified fragment was directly sequenced on a ABI Prism 310 sequencer (ABI Prism, Perkin–Elmer Applied Biosystems, Foster City, CA, USA), using BigDye Terminator Cycle Sequencing Reaction Kit (ABI Prism, Perkin–Elmer Applied Biosystems). Fragment length size was determined running GeneScan-400HD Analysis Software (ABI Prism, Perkin–Elmer Applied Biosystems).

CYP19 gene (TTTA)n repeat length  As previously described (22), (TTTA)n repeat length was assessed by fragment analysis of the PCR products using primers published by Haiman et al. (25). The forward primers were 5'-labeled with a fluorescent dye for automated fragment analysis on a ABI Prism 310 sequencer (ABI Prism, Perkin–Elmer Applied Biosystems), using BigDye Terminator Cycle Sequencing Reaction Kit (ABI Prism, Perkin–Elmer Applied Biosystems). To confirm repeat length, homozygotic representatives of the different observed allele lengths (7, 8, 9, 10, 11, 12, 13) were sequenced. The allelic variant consisting of (TTTA)7 repeats was the most frequently observed one in this population of healthy elderly men. Subjects were classified in three genotype groups according to the presence of the shortest allele length.

Estrogen receptor {alpha} gene (ER{alpha})  Both XbaI and PvuII ER{alpha} single nucleotide polymorphisms (SNPs) were determined by a single PCR fragment in accordance with a study by Yamada et al. (26). The amplified products were then digested with the XbaI and PvuII restriction enzymes (New England Biolabs Inc., Beverly, MA, USA). Haplogenotypes were constructed using Phase software (27). Subjects were then defined by the presence of a ‘TA’ haplotype.

Body composition

At each visit study subjects had their body weight measured to the nearest 0.1 kg on a calibrated balance scale in light indoor clothing without shoes. Height was measured to the nearest 0.1 cm after removal of shoes. Body mass index (BMI) was calculated as the weight (kg) divided by the height squared (m2). Bioelectrical impedance analysis (Bodystat 1500, Bodystat Ltd, Isle of Man, UK) was used to estimate fat and lean mass percentage. The CV was 1.3 and 0.5% for fat mass and lean mass percentage, respectively, as calculated from duplicate measurements in 15 study subjects.

Statistical analysis

Cross-sectional analyses  Continuous variables were described in terms of mean±S.D. if their distribution was normal according to the Kolmogorov–Smirnov test, and in terms of median, first and third quartiles otherwise. Bivariate partial correlations, adjusting for age and BMI or fat mass percentage, were used to show associations between hormonal and various clinical parameters. SPSS 12.0 software package (SPSS Inc., Chicago, IL, USA) was used for the descriptive analyses and bivariate correlations; a P value <0.05 was considered to indicate statistical significance; all P values were two-tailed.

Longitudinal analyses  Linear mixed-effects models with random intercepts and autoregressive residual correlation structure were used for longitudinal analysis. Using likelihood ratio tests, no evidence of random time slopes was found. Parameters for baseline predictors were estimated via restricted maximum likelihood estimation and model-based confidence intervals (CI) and P values are reported. No adjustments for baseline testosterone levels were performed when it was anticipated that these levels could be affected by the considered exposure (e.g. genetic polymorphism, smoking, etc.) (28). All the analyses were performed in the software package SAS 9.1.3, Service Pack 4 (SAS Institute, Inc., Cary, NC, USA).


   Results
Top
 Introduction
 Subjects and methods
 Results
Discussion
 Declaration of interest
 References
 
Baseline clinical and hormonal characteristics

Baseline clinical characteristics of the study population are described in Table 1 (n=218; three subjects excluded from baseline data because of temporary use of medication). Mean age at baseline was 74 years (first and third quartiles: 73–78 years); 40 out of 218 (18.3%) were current smokers, 32 (20.1%) consumed more than two alcoholic beverages per day, 36 (16.5%) were obese (BMI≥30 kg/m2), 80 (36.7%) reported the use of at least three prescription medications, and 71 (32.6%) reported to be physically active for >2 h per week. After 4 years of follow-up, 20 out of 152 (13.2%) were current smokers, 16 (10.5%) consumed more than two alcoholic beverages per day, 23 (15.1%) were obese, and 78 (51.3%) reported the use of at least three prescription medications. More men than at baseline, i.e. 95 (62.5%), reported to be physically active for >2 h per week.


View this table:
[in this window]
[in a new window]

  		Table 1 Clinical characteristics of the study population (n=218) at baseline.

 
Median length of the AR CAG repeat polymorphism was 21 (first to third quartiles: 20–24). Sixty (28.8%) subjects had seven (TTTA) repeats of the CYP19 on both alleles, 97 (46.6%) had only one allele with seven (TTTA) repeats, and in 51 (24.5%) subjects the number of (TTTA) repeats exceeded seven on both alleles. Regarding the ER{alpha} polymorphisms, 72 (33.0%) subjects were homozygous for the ‘TA’ haplotype, 105 (48.5%) had only one allele with the ‘TA’ haplotype, and 41 (18.8%) had no ‘TA’ haplotype.

Age at baseline was negatively associated with FT, BioT ({rho}=–0.20 to –0.22; P<0.01), and positively with SHBG, LH, and FSH levels ({rho}=0.13–0.27; P<0.05). In the age range of our study participants (71–86 years), mean cross-sectional changes of –0.66% (95% CI –1.77; 0.44%), 1.15% (95% CI –2.11; –0.41%), and 1.76% (95% CI –2.75; –0.77%) per year were observed for TT, FT, and BioT respectively. After controlling for age, BMI was negatively associated with TT, FT, BioT, and SHBG ({rho}=–0.14 to –0.30; P<0.05), and positively with FE2 and BioE2 levels ({rho}=0.15; P<0.05). Both absolute (kg) and relative (%) fat mass displayed a negative association with all three testosterone fractions ({rho}=–0.23 to –0.33; P<0.001). Absolute fat mass was negatively associated with SHBG levels ({rho}=–0.24; P<0.001). After controlling for absolute fat mass, insulin levels were negatively associated with SHBG levels ({rho}=–0.15; P=0.016) and positively with BioE2 levels ({rho}=0.15; P=0.025). Controlling for total body weight, both IGF-1 and IGF-BP3 were negatively associated with SHBG levels ({rho}=–0.21 to –0.29; P<0.01), and IGF-BP3 displayed a positive association with FT and BioT levels ({rho}=0.14–0.16; P<0.05). As reported previously, serum leptin levels were negatively correlated with all three testosterone fractions after adjusting for absolute fat mass (29).

No significant differences in sex steroid or SHBG levels between current smokers, past smokers, or subjects who never smoked were found (all P>0.129). Among current or past smokers, the number of pack year smoked was inversely correlated with both FT ({rho}=–0.20; P=0.011) and BioT ({rho}=–0.20; P=0.009) after controlling for age and BMI. As reported previously, no associations between hormone or SHBG levels and AR CAG repeat length (21) or CYP19 TTTA repeat length (22) were observed in this elderly population. Neither for the ER{alpha} XbaI and PvuII SNPs nor for the combined haplogenotypes were differences in hormone levels detected (all P>0.32; data not shown).

Evolution of hormonal parameters over time

Table 2 presents descriptive statistics for number of participants, age, and BMI at all visits, as well as for testosterone and SHBG levels. During the 4-year follow-up, 69 subjects left the study. Dropout analysis using logistic regression showed that dropout was associated with lower FT levels and a lower score of functional capacity and mental status at their last visit, lower physical activity scores at baseline and the ‘TA,X’ haplogenotype of the ER{alpha} (data not shown).


View this table:
[in this window]
[in a new window]

  		Table 2 Evolution of age, body mass index (BMI), number of participants, and hormonal levels over time.

 
The evolution of TT and BioT levels is displayed in Fig. 1. Mean TT levels declined from 444±148 to 414±157 ng/dl over the 4-year period. The greatest decrease was observed between visits 1 and 2, which was reflected in SHBG levels with generally higher levels at visit 1 compared with visit 2. The evolution of FT and BioT showed a more gradual decline (mean levels at visit 1 were 7.4±2.1 and 173±51 ng/dl vs 6.9±2.2 and 153±50 ng/dl at visit 5 for FT and BioT respectively). Longitudinal data analysis, adjusting for baseline age and BMI, revealed a yearly mean decrease of –9.5 ng/dl (95% CI –12.7; –6.2 ng/dl), –0.16 ng/dl (95% CI –0.21; –0.10 ng/dl), and –6.1 ng/dl (95% CI –7.3; –4.9 ng/dl) for TT, FT, and BioT respectively. This corresponds to yearly mean procentual decreases of –1.26% (95% CI –2.58; –0.01), –1.33% (95% CI –2.61; –0.01), and –2.43% (95% CI –3.78; –1.08) for TT, FT, and BioT respectively. No significant overall changes in SHBG levels were detected (data not shown).

Percentages of participants with testosterone levels below the reference range for young men (320 ng/dl for TT, 6.5 ng/dl for FT, and 140 ng/dl for BioT (1)) are displayed for all visits in Fig. 2. At baseline, 19.3, 32.1, and 25.7% of participants presented with levels below this reference range for TT, FT, and BioT respectively. At visit 5, this rose to 30.3, 43.4, and 38.8% respectively.


Figure 2
View larger version (16K):
[in this window]
[in a new window]

  		Figure 2 Percentage of participants with testosterone levels below reference range for young men (320 ng/dl for TT, 6.5 ng/dl for FT, and 140 ng/dl for BioT) at each visit.

 
Over the 4-year follow-up period, a modest decrease in BMI was noted (mean {Delta}=–0.10±1.37 kg/m2). After adjusting for baseline age, subjects presenting with an increase in BMI (>0 kg/m2: n=70 or 42.2%) presented a greater decline in TT levels compared with those with a decrease in BMI (<0 kg/m2: n=96 or 57.8%): –58.4 vs –17.9 ng/dl respectively (P=0.019). No differences in changes of FT or BioT levels were noted between these groups (P>0.16). This reflected the mean decrease in SHBG levels in subjects presenting with an increase in BMI, compared with increasing SHBG levels in subjects with a decrease in BMI: –1.7 vs +2.8 nmol/l respectively (P=0.015). Additional adjustment for baseline T or SHBG levels did not alter these findings (data not shown).

Baseline predictors of changes in T and SHBG levels

Lifestyle- and health-related factors, body composition, insulin, and IGF-1  Age at baseline was not associated with the evolution of T or SHBG levels over time (all P>0.50; data not shown), neither were indicators of quality of life, psychological and functional well-being, medical history, or number of medications (either as a continuous variable or as more or less than three prescribed medications), physical activity, or frequency of alcoholic consumptions at baseline (all P>0.20; data not shown). Adjusting for baseline age and BMI, current or former smoking at baseline was associated with a faster yearly decline of FT levels with 0.13 ng/dl (P=0.037) compared with subjects who never smoked; similar trends for TT (P=0.12) and BioT (P=0.09) did not reach statistical significance (Table 3). These observations held true when additionally adjusting for baseline T levels (data not shown). Among smokers, the number of pack year smoked was not associated with changes in testosterone levels (P>0.50).


View this table:
[in this window]
[in a new window]

  		Table 3 Baseline predictors of subsequent changes in T levels, using linear mixed-effects modeling for longitudinal data analysis.

 
BMI at baseline was not associated with changes in testosterone or SHBG levels, neither was absolute nor relative fat or lean mass. There was no difference in the rate of change of testosterone levels between subjects who were obese (BMI >30 kg/m2) and/or overweight (BMI >25 kg/m2) at baseline compared with those who were not. Baseline glucose, insulin, leptin, IGF-1, or IGF-BP3 levels were not associated with changes in testosterone or SHBG levels (all P>0.33; data not shown).

Sex steroids and gonadotropins  After controlling for age and BMI, baseline E2, LH, and FSH levels were associated with a faster decline in testosterone levels over time. However, for E2 and LH this association was no longer significant after adjusting for the predictive effects of baseline testosterone levels, whereas a significant trend was still observed for higher baseline FSH levels being associated with a faster decline in T levels (of 0.40, 0.007, and 0.15 ng/dl per IU/l increase in FSH levels for TT, FT, and BioT respectively; Table 3). The evolution of BioT levels associated with median, upper, and lower quartiles of baseline FSH levels is illustrated in Fig. 3. No associations with the changes in SHBG levels were found (data not shown).


Figure 3
View larger version (19K):
[in this window]
[in a new window]

  		Figure 3 Mean evolution of BioT levels over time between median, upper, and lower quartiles according to baseline FSH levels.

 
Genetic polymorphisms  Neither the CAG repeat length of the AR nor the TTTA repeat length of the CYP19 gene was associated with subsequent changes in TT, FT, BioT, or SHBG levels (data not shown). The polymorphisms of the ER{alpha} were significantly associated with the evolution of testosterone levels over time. Specifically, after controlling for baseline age and BMI, a slower yearly average decline of TT levels was observed with 19.0 ng/dl (P=0.009) for carriers of the (‘TA,TA’) haplogenotype and with 15.0 ng/dl (P=0.024) for carriers of the (‘TA,X’) haplogenotype as compared with those carrying the (‘X,X’) haplogenotype. Similarly, a slower yearly average decline of BioT levels was observed with 7.6 ng/dl (P=0.007) for carriers of the (‘TA,TA’) haplogenotype and 5.2 ng/dl (P=0.041) for carriers of the (‘TA,X’) haplogenotype when compared with those carrying the (‘X,X’) haplogenotype. For FT, a significantly slower decline was found for carriers of the (‘TA,TA’) haplogenotype (of 0.26 ng/dl per year; P=0.026) when compared with those carrying the (‘X,X’) haplogenotype (Table 3). This indicates a slower decline in testosterone levels in carriers of a ‘TA’ allele compared with subjects without a ‘TA’ allele. None of the genetic polymorphisms were associated with changes in SHBG levels (data not shown). Adjusting for the predictive effects of baseline testosterone or E2 levels did not alter these observations (data not shown).


   Discussion
Top
 Introduction
 Subjects and methods
 Results
 Discussion
Declaration of interest
 References
 
In this study we present 4-year longitudinal data on testosterone levels in community-dwelling elderly men and discuss predictors of the observed changes. We demonstrated that even in elderly subjects with already decreased testosterone levels, a further decline in testosterone levels can still be observed. Furthermore, in accordance with previous reports (4, 5, 6, 7, 18) the decrease is least apparent for TT and most accentuated for calculated BioT levels, due to the relatively stable SHBG levels. Consistent with the results of the Massachusetts Male Aging Study (MMAS), the observed longitudinal changes in testosterone levels were of greater magnitude than the baseline cross-sectional trends (4). In addition, in these elderly men the magnitude of these longitudinal changes was comparable to three previous longitudinal studies covering a broader age range (4, 16, 19), and slightly higher than the results obtained by the Baltimore Longitudinal Study on Aging (BLSA) and by Zmuda et al. (17, 18).

In our cross-sectional analyses, age was negatively associated with FT and BioT and positively with SHBG, LH, and FSH levels. These results are consistent with those reported in other studies (7, 15). However, within the considered older age group, age at baseline was not predictive of subsequent changes in testosterone or SHBG levels, indicating a decrease in testosterone production over time which is independent of age. This is in agreement with the results of the Multiple Risk Factor Intervention Trial (MRFIT study) (17) and with the previously discussed observation of a similar magnitude of decline among different populations and age ranges, but at variance with the results of the BLSA, where age at baseline was negatively associated with TT levels over time (18). Further, health status at baseline was predictive of longitudinal changes in neither testosterone nor SHBG levels. However, the inclusion criteria we applied yielded a population of healthy elderly men, and it cannot therefore be concluded that co-morbidity is not an important determinant of testosterone levels (1).

Consistent with previous findings (3, 7, 12, 15), BMI at baseline was negatively associated with TT, FT, BioT, and SHBG levels. Moreover, those who gained weight over the 4-year period presented a greater decline in TT, explained by a decrease in SHBG levels, compared with those with a decrease in BMI. These results suggest that changes in BMI over time modulate age-related changes in TT and SHBG, but in our study population neither FT nor BioT levels were affected. This is in agreement with the findings of the BLSA and MMAS in elderly men (18, 30) and the results of the CARDIA Male Hormone Study in a younger population (31).

However, in this study no baseline body composition parameters (BMI, fat, or lean mass as assessed by bioelectrical impedance) were predictive of subsequent changes in testosterone or SHBG levels. This is consistent with the results of the MRFIT study (17), but in contrast with the results from the larger BLSA and MMAS cohorts, where obesity at baseline predicted a greater decline over time in TT and SHBG, and even in FT levels in the MMAS, (12, 18). It remains unclear whether body composition influences sex steroid hormone concentrations over time or whether, conversely, sex steroid hormone concentrations alter body composition; most likely this relationship is bidirectional.

Regarding the influence of genetic polymorphisms considered in this study, we observed an association of ER{alpha} polymorphisms with changes in testosterone levels, indicating a slower decline in testosterone levels in carriers of a ‘TA’ allele compared with non-carriers. At variance with the longitudinal results from Krithivas et al. (13), no effect of AR CAG repeat length on the evolution in testosterone levels was found. However, it should be noted that their study population was clearly younger with a time interval of 8 years between visits, which could result in a greater absolute decline in testosterone levels.

In this study, baseline E2, LH, and FSH levels were predictive of a subsequent decline in testosterone levels. However, when controlling for baseline testosterone levels, only FSH levels remained significantly associated with a decline in testosterone levels. This observation could be explained by the fact that both in the case of primary hypogonadism and of aging in healthy men, the rise in FSH levels is steeper than that of LH levels (4, 16, 32), and by the greater ultradian variability of LH levels. Moreover, we have already reported on the negative association between testicular volume and FSH levels in this elderly population, with FSH levels being an independent determinant of total testicular volume (33). As a consequence, in elderly patients presenting with relatively high FSH levels despite having low-normal serum testosterone levels, biochemical follow-up in the next years may be considered.

As a whole, these findings point toward a substantial testicular factor in the observed decline in testosterone production, whereas the association with E2 levels and the ER{alpha} polymorphisms is suggestive of estrogen-related processes and might be related to changes in the neuroendocrine regulation of testosterone production in elderly men, which are characterized by relative deficiency of LH secretion (1). Indeed, E2 levels did not change in our study population, despite the overall decline in testosterone levels, which suggests increased aromatase activity (34), and circulating E2 is known to be a major player in LH negative feedback regulation in men (35). Nevertheless, a comparative study in young and older men of the effects of aromatase inhibition has failed to disclose a greater restraining action of E2 on LH and testosterone levels in the elderly men when compared with young men (36).

Notwithstanding the observation of a clear age-related decline in serum testosterone levels, this does not necessarily mean that the same holds true at a tissue level. Bélanger et al. have already suggested that the observed decline in androgen levels could be partially compensated by an increase in steroid-converting enzymes in peripheral tissues, especially the 5{alpha}-reductase activity resulting in tissue dihydrotestosterone levels unaffected by aging (14). On the other hand, there is also evidence of decreased androgen tissue concentrations (1) and of reduced androgen sensitivity, with reports of decreased numbers of AR in various tissues in elderly men (37, 38). However, up to now no practical and clinically useful biological marker of androgen activity at tissue level is available. Therefore, one has to further rely on serum testosterone levels as an indirect and less than optimal parameter of androgen activity at tissue level.

This is the first study to report on yearly based, longitudinal changes in testosterone levels in a fairly homogenous population of community-dwelling men over 70 years of age at baseline, with 20% of subjects over 80 years. Notably, this age group is often underrepresented in the available literature but is steadily increasing in the general population. The strength of this study is the yearly time interval for longitudinal assessment of serum testosterone levels and it is a population, rather than clinic or complaint-based, study. Moreover, most men were in relatively good state of health, suggesting that the studied longitudinal changes largely reflect effects related to aging per se; adjusting for illnesses and medication produced little, if any, change in the observed overall effects of aging. Mixed-effects modeling analysis for longitudinal data treats every measurement as a separate outcome, but accounts for the interdependence of repeated measurements from the same individual. Finally, all samples from the same subject were analyzed in a single assay run, eliminating the possibility of within-subject variation due to interassay variability.

A limitation of our study is that it is based on single time-point hormonal measurements for each visit. Nevertheless, it has been shown previously that single time-point estimates are a valid approach for population studies (39). Further, bioelectrical impedance analysis has been shown to have a suboptimal accuracy and is limited by various factors for assessing body composition, which could possibly mask associations of baseline body composition with subsequent changes in testosterone levels. Another possible concern is the clearly higher TT and SHBG levels at visit 1 compared with the subsequent visits. This could point toward a selection bias, with dropout between visit 1 and 2 of subjects with higher testosterone levels. However, dropout analysis showed that dropout was associated with lower testosterone levels at the previous visit. While mixed-effects model analyses naturally allow for dropout to be associated with the outcome history (40), and thus with lower testosterone levels at the previous visit, we could not account for the possibility that dropout was associated with higher/lower testosterone levels at the current (or future) visits. Another hypothesis could be a seasonal effect on testosterone levels, since blood sampling at visit 1 was mainly performed from March till June, whereas for the subsequent visits this was done from July till August. However, from the study by Brambilla et al. it appeared that seasonal variation is unlikely to be an important source of variation in serum sex steroid levels in men (41). Finally, the present study was performed in a well-characterized and specific subset of the population consisting of generally healthy men of high age with a relatively short follow-up period, and the results should therefore not be extrapolated to the general population. In the general population, factors related to co-morbidity are likely to accentuate the observed age-related hormonal changes.

In conclusion, a decline in serum testosterone levels, especially in the non-SHBG-bound and FT fractions, was shown to occur in generally healthy community-dwelling men aged 71–86 years with already decreased testosterone levels compared with healthy young men. Moreover, this decline appeared to be independent of age at baseline, indicating a rather constant decline in testosterone production with aging. Further, subjects who gained weight during follow-up displayed a greater decline in TT and SHBG levels, whereas baseline body composition was not predictive of subsequent changes in testosterone levels. The predictive value of FSH levels appears to confirm the involvement of testicular changes as an important factor for the decline in testosterone production and might warrant future follow-up in case of low-normal testosterone levels. In addition, the predictive effects of E2 and the ER{alpha} polymorphisms suggest a contribution of estrogen-related processes in the decline of serum testosterone levels, which might relate to the known involvement of changes in the neuroendocrine regulation of testosterone production in elderly men.


   Declaration of interest
Top
 Introduction
 Subjects and methods
 Results
 Discussion
 Declaration of interest
References
 
The authors declare that there is no conflict of interest that would prejudice the impartiality of this scientific work.


   Funding
 
This work was supported by the Flemish Fund for Scientific Research (FWO Vlaanderen) grants G.0331.02 and G.0404.00.


   Acknowledgements
 
The authors are indebted to K Toye, M Daems, R De Muynck, H Myny, and I Bocquaert for their excellent technical assistance. Determination of AR gene CAG repeat length was performed in the laboratory of C Sultan and S Lumbroso, Department of Reproductive Endocrinology, Lapeyronie Hospital, Montpellier, France. Finally, the authors would like to thank the Municipal Council and population of Merelbeke for their support and participation in the study.


   References
Top
 Introduction
 Subjects and methods
 Results
 Discussion
 Declaration of interest
 References
 

    1. Kaufman JM & Vermeulen A. The decline of androgen levels in elderly men and its clinical and therapeutic implications. Endocrine Reviews 2005; 26:833–876.[Abstract/Free Full Text]

    2. Vermeulen A, Kaufman JM & Giagulli VA. Influence of some biological indexes on sex hormone-binding globulin and androgen levels in aging or obese males. Journal of Clinical Endocrinology and Metabolism 1996; 81:1821–1826.[Abstract]

    3. Ferrini RL & Barrett-Connor E. Sex hormones and age: a cross-sectional study of testosterone and estradiol and their bioavailable fractions in community-dwelling men. American Journal of Epidemiology 1998; 147:750–754.[Abstract/Free Full Text]

    4. Feldman HA, Longcope C, Derby CA, Johannes CB, Araujo AB, Coviello AD, Bremner WJ & McKinlay JB. Age trends in the level of serum testosterone and other hormones in middle-aged men: longitudinal results from the Massachusetts Male Aging Study. Journal of Clinical Endocrinology and Metabolism 2002; 87:589–598.[Abstract/Free Full Text]

    5. Yeap BB, Almeida OP, Hyde Z, Norman PE, Chubb SAP, Jamrozik K & Flicker L. In men older than 70 years, total testosterone remains stable while free testosterone declines with age. The Health in Men Study. European Journal of Endocrinology 2007; 156:585–594.[Abstract/Free Full Text]

    6. Barrett-Connor E, von Muhlen DG & Kritz-Silverstein D. Bioavailable testosterone and depressed mood in older men: the Rancho Bernardo Study. Journal of Clinical Endocrinology and Metabolism 1999; 84:573–577.[Abstract/Free Full Text]

    7. Muller M, den Tonkelaar I, Thijssen JH, Grobbee DE & van der Schouw YT. Endogenous sex hormones in men aged 40–80 years. European Journal of Endocrinology 2003; 149:583–589.[Abstract]

    8. Khaw KT, Dowsett M, Folkerd E, Bingham S, Wareham N, Luben R, Welch A & Day N. Endogenous testosterone and mortality due to all causes, cardiovascular disease, and cancer in men: European Prospective Investigation Into Cancer in Norfolk (EPIC-Norfolk) Prospective Population Study. Circulation 2007; 116:2694–2701.[Abstract/Free Full Text]

    9. Wang C, Swerdloff RS, Iranmanesh A, Dobs A, Snyder PJ, Cunningham GR, Matsumoto AM, Weber T, Berman N & the Testosterone Gel Study Group. Transdermal testosterone gel improves sexual function, mood, muscle strength, and body composition parameters in hypogonadal men. Journal of Clinical Endocrinology and Metabolism 2000; 85:2839–2853.[Abstract/Free Full Text]

    10. Snyder PJ. Hypogonadism in elderly men – what to do until the evidence comes. New England Journal of Medicine 2004; 350:440–442.[Free Full Text]

    11. Mohr BA, Guay AT, O'Donnell AB & McKinlay JB. Normal, bound and nonbound testosterone levels in normally ageing men: results from the Massachusetts Male Ageing Study. Clinical Endocrinology 2005; 62:64–73.[CrossRef][Medline]

    12. Derby CA, Zilber S, Brambilla D, Morales KH & McKinlay JB. Body mass index, waist circumference and waist to hip ratio and change in sex steroid hormones: the Massachusetts Male Ageing Study. Clinical Endocrinology 2006; 65:125–131.[CrossRef][Medline]

    13. Krithivas K, Yurgalevitch SM, Mohr BA, Wilcox CJ, Batter SJ, Brown M, Longcope C, McKinlay JB & Kantoff PW. Evidence that the CAG repeat in the androgen receptor gene is associated with the age-related decline in serum androgen levels in men. Journal of Endocrinology 1999; 162:137–142.[Abstract]

    14. Belanger A, Candas B, Dupont A, Cusan L, Diamond P, Gomez JL & Labrie F. Changes in serum concentrations of conjugated and unconjugated steroids in 40- to 80-year-old men. Journal of Clinical Endocrinology and Metabolism 1994; 79:1086–1090.[Abstract]

    15. Orwoll ES, Lambert LC, Marshall LM, Phipps K, Blank J, Barrett-Connor E, Cauley J, Ensrud K, Cummings S & for the Osteoporotic Fractures in Men Study Group. Testosterone and estradiol among older men. Journal of Clinical Endocrinology and Metabolism 2006; 91:1336–1344.[Abstract/Free Full Text]

    16. Morley JE, Kaiser FE, Perry HM III, Patrick P, Morley PM, Stauber PM, Vellas B, Baumgartner RN & Garry PJ. Longitudinal changes in testosterone, luteinizing hormone, and follicle-stimulating hormone in healthy older men. Metabolism 1997; 46:410–413.[CrossRef][Web of Science][Medline]

    17. Zmuda JM, Cauley JA, Kriska A, Glynn NW, Gutai JP & Kuller LH. Longitudinal relation between endogenous testosterone and cardiovascular disease risk factors in middle-aged men: a 13-year follow-up of former multiple risk factor intervention trial participants. American Journal of Epidemiology 1997; 146:609–617.[Abstract/Free Full Text]

    18. Harman SM, Metter EJ, Tobin JD, Pearson J & Blackman MR. Longitudinal effects of aging on serum total and free testosterone levels in healthy men. Journal of Clinical Endocrinology and Metabolism 2001; 86:724–731.[Abstract/Free Full Text]

    19. Liu PY, Beilin J, Meier C, Nguyen TV, Center JR, Leedman PJ, Seibel MJ, Eisman JA & Handelsman DJ. Age-related changes in serum testosterone and sex hormone binding globulin in Australian men: longitudinal analyses of two geographically separate regional cohorts. Journal of Clinical Endocrinology and Metabolism 2007; 92:3599–3603.[Abstract/Free Full Text]

    20. T'Sjoen GG, De Vos S, Goemaere S, Van Pottelbergh I, Dierick M, Van Heeringen C & Kaufman JM. Sex steroid level, androgen receptor polymorphism, and depressive symptoms in healthy elderly men. Journal of the American Geriatrics Society 2005; 53:636–642.[CrossRef][Web of Science][Medline]

    21. Van Pottelbergh I, Lumbroso S, Goemaere S, Sultan C & Kaufman JM. Lack of influence of the androgen receptor gene CAG-repeat polymorphism on sex steroid status and bone metabolism in elderly men. Clinical Endocrinology 2001; 55:659–666.[CrossRef][Medline]

    22. Van Pottelbergh I, Goemaere S & Kaufman JM. Bioavailable estradiol and an aromatase gene polymorphism are determinants of bone mineral density changes in men over 70 years of age. Journal of Clinical Endocrinology and Metabolism 2003; 88:3075–3081.[Abstract/Free Full Text]

    23. Vermeulen A, Verdonck L & Kaufman JM. A critical evaluation of simple methods for the estimation of free testosterone in serum. Journal of Clinical Endocrinology and Metabolism 1999; 84:3666–3672.[Abstract/Free Full Text]

    24. Szulc P, Claustrat B, Munoz F, Marchand F & Delmas PD. Assessment of the role of 17beta-oestradiol in bone metabolism in men: does the assay technique matter? The MINOS study. Clinical Endocrinology 2004; 61:447–457.[CrossRef][Medline]

    25. Haiman CA, Hankinson SE, Spiegelman D, De V I, Colditz GA, Willett WC, Speizer FE & Hunter DJ. A tetranucleotide repeat polymorphism in CYP19 and breast cancer risk. International Journal of Cancer 2000; 87:204–210.[CrossRef][Web of Science][Medline]

    26. Yamada Y, Ando F, Niino N, Ohta S & Shimokata H. Association of polymorphisms of the estrogen receptor a gene with bone mineral density of the femoral neck in elderly Japanese women. Journal of Molecular Medicine 2002; 80:452–460.[CrossRef][Web of Science][Medline]

    27. Stephens M, Smith NJ & Donnelly P. A new statistical method for haplotype reconstruction from population data. American Journal of Human Genetics 2001; 68:978–989.[CrossRef][Web of Science][Medline]

    28. Glymour MM, Weuve J, Berkman LF, Kawachi I & Robins JM. When is baseline adjustment useful in analyses of change? An example with education and cognitive change.. American Journal of Epidemiology 2005; 162:267–278.[Abstract/Free Full Text]

    29. Van den Saffele JK, Goemaere S, De Bacquer D & Kaufman JM. Serum leptin levels in healthy ageing men: are decreased serum testosterone and increased adiposity in elderly men the consequence of leptin deficiency? Clinical Endocrinology 1999; 51:81–88.[CrossRef][Medline]

    30. Mohr BA, Bhasin S, Link CL, O'Donnell AB & McKinlay JB. The effect of changes in adiposity on testosterone levels in older men: longitudinal results from the Massachusetts Male Aging Study. European Journal of Endocrinology 2006; 155:443–452.[Abstract/Free Full Text]

    31. Gapstur SM, Kopp P, Gann PH, Chiu BCH, Colangelo LA & Liu K. Changes in BMI modulate age-associated changes in sex hormone binding globulin and total testosterone, but not bioavailable testosterone in young adult men: the CARDIA Male Hormone Study. International Journal of Obesity 2006; 31:685–691.[Web of Science]

    32. Christ-Crain M, Meier C, Huber PR, Zimmerli L & Mueller B. Value of gonadotropin-releasing hormone testing in the differential diagnosis of androgen deficiency in elderly men. Journal of Clinical Endocrinology and Metabolism 2005; 90:1280–1286.[Abstract/Free Full Text]

    33. Mahmoud AM, Goemaere S, El-Garem Y, Van Pottelbergh I, Comhaire FH & Kaufman JM. Testicular volume in relation to hormonal indices of gonadal function in community-dwelling elderly men. Journal of Clinical Endocrinology and Metabolism 2003; 88:179–184.[Abstract/Free Full Text]

    34. Vermeulen A, Kaufman JM, Goemaere S & Van Pottelbergh I. Estradiol in elderly men. Aging Male 2002; 5:98–102.[Medline]

    35. Raven G, de Jong FH, Kaufman JM & de Ronde W. In men peripheral estradiol levels directly reflect the action of estrogens at hypothalamo-pituitary level to inhibit gonadotropin secretion. Journal of Clinical Endocrinology and Metabolism 2006; 91:3324–3328.[Abstract/Free Full Text]

    36. T'Sjoen GG, Giagulli VA, Delva H, Crabbe P, De Bacquer D & Kaufman JM. Comparative assessment in young and elderly men of the gonadotropin response to aromatase inhibition. Journal of Clinical Endocrinology and Metabolism 2005; 90:5717–5722.[Abstract/Free Full Text]

    37. Roth GS & Hess GD. Changes in the mechanisms of hormone and neurotransmitter action during aging: current status of the role of receptor and post-receptor alterations. A review. Mechanisms of Ageing and Development 1982; 20:175–194.[CrossRef][Web of Science][Medline]

    38. Roehrborn CG, Lange JL, George FW & Wilson JD. Changes in amount and intracellular distribution of androgen receptor in human foreskin as a function of age. Journal of Clinical Investigation 1987; 79:44–47.[Web of Science][Medline]

    39. Vermeulen A & Verdonck G. Representativeness of a single point plasma testosterone level for the long term hormonal milieu in men. Journal of Clinical Endocrinology and Metabolism 1992; 74:939–942.[Abstract]

    40. Molenberghs G & Kenward M. The direct likelihood method. In Missing Data in Clinical Studies, edn 1, ch 7, pp 77–92. Eds G Molenberghs & M Kenward, New York: John Wiley & Sons, 2007.

    41. Brambilla DJ, O'Donnell AB, Matsumoto AM & McKinlay JB. Lack of seasonal variation in serum sex hormone levels in middle-aged to older men in the Boston area. Journal of Clinical Endocrinology and Metabolism 2007; 92:4224–4229.[Abstract/Free Full Text]

Received 3 June 2008
Accepted 30 June 2008




This article has been cited by other articles:


Home page
 Eur J EndocrinolHome page
B. B Yeap, S A P. Chubb, Z. Hyde, K. Jamrozik, G. J Hankey, L. Flicker, and P. E Norman
Lower serum testosterone is independently associated with insulin resistance in non-diabetic older men: the Health In Men Study
Eur. J. Endocrinol., October 1, 2009; 161(4): 591 - 598.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
EJE-07-0873v1
159/4/459    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Lapauw, B
Right arrow Articles by Kaufman, J M
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Lapauw, B
Right arrow Articles by Kaufman, J M

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS