Association of age, hormonal, and lifestyle factors with the Leydig cell biomarker INSL3 in aging men from the European Male Aging Study cohort

Abstract Background Aging in men is accompanied by a broad range of symptoms, including sexual dysfunction, cognitive and musculoskeletal decline, obesity, type 2 diabetes, cardiovascular disease and hypertension, organ degeneration/failure, and increasing neoplasia, some of which are associated with declining levels of Leydig cell‐produced testosterone. High natural biological variance, together with multiple factors that can modulate circulating testosterone concentration, may influence its interpretation and clinical implications. Insulin‐like peptide 3 is a biomarker of Leydig cell function that might provide complementary information on testicular health and its downstream outcomes. Objectives To characterize insulin‐like peptide 3 as a biomarker to assess gonadal status in aging men. Methods and materials A large European multicenter (European Male Aging Study) cohort of community‐dwelling men was analyzed to determine how insulin‐like peptide 3 relates to a range of hormonal, anthropometric, and lifestyle parameters. Results and discussion Insulin‐like peptide 3 declines cross‐sectionally and longitudinally within individuals at approximately 15% per decade from age 40 years, unlike testosterone (1.9% per decade), which is partly compensated by increasing pituitary luteinizing hormone production. Importantly, lower insulin‐like peptide 3 in younger men appears to persist with aging. Multiple regression analysis shows that, unlike testosterone, insulin‐like peptide 3 is negatively dependent on luteinizing hormone and sex hormone‐binding globulin and positively dependent on follicle‐stimulating hormone, suggesting a different mechanism of gonadotropic regulation. Circulating insulin‐like peptide 3 is negatively associated with increased body mass index or waist circumference and with smoking, and unlike testosterone, it is not affected by weight loss in obese individuals. Geographic variation in mean insulin‐like peptide 3 within Europe appears to be largely explained by differences in these parameters. The results allowed the establishment of a European‐wide reference range for insulin‐like peptide 3 (95% confidence interval) adjusted for increasing age. Conclusion Insulin‐like peptide 3 is a constitutive biomarker of Leydig cell functional capacity and is a robust, reliably measurable peptide not subject to gonadotropin‐dependent short‐term regulation and within‐individual variation in testosterone.


INTRODUCTION
Aging in adult men is a continuous process, largely correlating with the concentrations of circulating hormones that originate in or are dependent upon the testes. The best known of these hormones is testosterone (T), whose reduced concentrations have been linked to sexual dysfunction, obesity, metabolic syndrome, cardiovascular disease, cognitive decline, and physical frailty. However, there is considerable variability in the concentration of T, its biologically relevant proportion, and the responsiveness of various organ systems. Consequently, only clearly reduced levels of T, together with significantly increased symptoms of hypogonadism, are considered sufficient to constitute a clinical hypoandrogenemia (late-onset hypogonadism [LOH] or more appropriately functional hypogonadism) warranting treatment. 1 LOH, thus defined, represents only 2.1% of the community-dwelling adult male population. 2 How much circulating T is required for optimal health is unclear and must be seen in the context of lifestyle factors as well as morbidity. This is due in part to its bioavailability, because less than 2%, on average, of total T is not protein-bound and thus available for interacting with specific androgen receptors (so-called free testosterone [FT]). This is a parameter that in many conditions better correlates with disease incidence. Secondly, T varies across the day, and from day to day, and its synthesis by the testes is acutely feedback regulated by the hypothalamo-pituitary-gonadal (HPG) axis, which can compensate for declining T production by increasing the output of the gonadotropin luteinizing hormone (LH) as men age.
Although anatomical evidence indicates some loss of mature Leydig cells with age, 3,4 reduced T can also be because of declining hypothalamic or pituitary control, to reduced testicular sensitivity to LH, or to a deficiency of the testes to make T. 5,6 The last of these can be assessed from the ratio of circulating T to LH concentrations, a parameter also referred to as Leydig cell functional capacity, which markedly and progressively declines with age in men, beginning already at about age 30 years; like its component parameters, it is subject to high within-individual variance. A major diagnostic problem is that low T and hypogonadal symptoms do not always match each other. 2 Most men with low T have no symptoms, and those with hypogonadal symptoms often have normal T levels; only monitoring T may therefore not provide the most accurate diagnosis, and the applicability of other parameters, such as insulin-like peptide 3 (INSL3), should be tested to improve diagnostic accuracy.
If we could better understand the multiple interrelationships impacting the age-related decline in circulating T and its relationship to increasing age-dependent morbidity, then appropriate intervention at an earlier age and applied to a larger proportion of men might be able to redress the functional decline with aging and its consequent socioeconomic as well as health burden. A specific biomarker of Leydig cell functional status, in addition to T, that is independent of other (extratesticular) modifying factors could contribute to a more objective assessment of the age-related decline in male gonadal function.
INSL3 is a peptide hormone secreted into the systemic circulation by the steroidogenic Leydig cells of the testes. 7 Importantly, and unlike T, INSL3 appears to be secreted in an acutely constitutive manner, simply reflecting the number and extant differentiation status of Leydig cells; it is independent of acute regulation by gonadotrophic hormones of the HPG axis. 7,8 Whereas the best-known function of INSL3 is the regulation of testicular descent in the fetus, 9 in the adult, it may act as a germ cell survival factor. 10 It is also associated with the maintenance of bone strength; genetic defects for either INSL3 or its cognate receptor RXFP2 in mice and humans are linked to osteopenia and osteoporosis. 11 INSL3 has additionally been suggested to ameliorate liver fibrosis 12 as well as deteriorating kidney function. 13 Following an initial cross-sectional study of community-living men in Australia, 14,15 the present investigation was aimed at developing a more detailed understanding of the relationship between circulating INSL3 and hormonal/lifestyle factors in a larger multicenter European population of community-dwelling aging men. This population study takes advantage of the extensive phenotyping carried out in the EMAS cohort and additionally profits from the longitudinal nature of this study in addition to the cross-sectional data. 16 The results describe the cross-sectional and longitudinal characteristics of INSL3 in the normal aging male population, in comparison with T, and reinforce its role as an independent biomarker of Leydig cell functional capacity. Consequently, most analyses presented here represent the EMAS second phase, and samples from the first phase are only used for longitudinal comparison. Descriptive statistics for the parameters used for the second-phase subjects analyzed here are given in Table 1 and do not differ from those presented previously 16 for the entire cohort. Body weight was measured to the nearest 0.1 kg, and height to the nearest 1 mm. Body mass index (BMI) was calculated as kg/m 2 .

Subjects, anthropometric data
Waist circumference (WC; cm) was measured three times for each subject, and the average was taken.

Questionnaires to evaluate smoking, alcohol, etc
Smoking habits were obtained by postal questionnaires, which recorded whether the participant was a current, past or non-smoker.
For this analysis, current smokers were compared with pooled past smokers and non-smokers. Alcohol consumption was provided in terms of approximately weekly intake, whereby for multivariate analysis, 0-1 alcoholic drinks per week were allocated "low" (0), 2-4 drinks per week "moderate" (1), and >4 drinks per week "high" (2).

Hormone assays
Blood samples were collected before 10.00 AM, and separated sera were stored at -80 • C. Total T was measured by gas chromatographymass spectrometry (MS) 18 in the same laboratory for all samples. The coefficient of variation (COV) was less than 3.5% within and between runs and the detection limit was 0.17 nmol/L. LH, follicle-stimulating hormone (FSH), and sex hormone-binding globulin (SHBG) were all measured by the Modular E170 platform electrochemiluminescence immunoassay (Roche Diagnostics, Mannheim, Germany). The withinand between-assay COVs for LH were 1.88% and 3.01%, respectively (detection limit 0.10 U/L), for FSH were 0.9% and 1.9% (detection limit 0.10 U/L), and for SHBG were 1.70% and 3.18% (detection limit 0.35 nmol/L), respectively. 19 FT was estimated by calculation (cFT) from total T and SHBG using the Vermeulen formula. 20 INSL3 was measured using an established and well-validated, sensitive, and specific time-resolved fluorescent immunoassay. 7 The inter-and intra-COVs were <8% and <3%, respectively, and the limit of detection was 20 pg/ml. In direct comparison with a new liquid chromatography-MS/MS procedure, this assay yielded almost identical values. 21

Statistics
Descriptive statistics for the phase 2 subjects analyzed here are given in

Characteristics of the EMAS cohort
The EMAS cohort at baseline has been described in detail in several previous publications. 17,19 Similarly, the follow-up longitudinal phase of the study at an average of 4.3 years after initial sampling has also been described. 16 shown) did not differ significantly from those of the original cohort, as described previously. 16,22 Only the phase 2 samples were used for the detailed cross-sectional analysis, whereas all paired samples from four centers were used for the longitudinal analysis.

Between-center differences in INSL3
The EMAS cohort comprises men from eight different countries, representing the northern, eastern, southern, and western regions of Europe, in approximately equal numbers. While between-center differences for most parameters were minimal (Table 1), there were significant differences between centers for INSL3 for phase 2 ( Figure 1A).
Similar differences were also seen for the four centers for which data are available for phase 1 (not shown). In phase 2, Leuven registered the highest mean INSL3 (1.21 ng/ml), followed by Manchester (1.08 ng/ml), with Tartu and Lodz registering the lowest mean INSL3 values (0.81 and 0.87 ng/ml, respectively). These differences were significant (p < 0.05).
To adjust for age, the INSL3 concentration (mean and 95% CI) at a given age of 65 years (close to the actual mean age of the cohort; 63 ± 10 years) was calculated by interpolation from the center-specific regressions of INSL3 against age ( Figure 1B). Because mean age was similar between centers, these age-adjusted results are essentially comparable to the simple means ( Figure 1A).
Data were combined for subsequent analyses, with center differences being taken into account when adjusting parameters (see later).

Effect of age on INSL3 and other reproductive hormones at baseline and follow-up
The regression of the phases 1 and 2 data for the four centers (n = 961) for which data from both phases were available shows that the withinindividual change in INSL3 concentration with age is minimal (Figure 2 Figure 3A, a horizontal dashed line is drawn at 0.4 ng/ml (fifth percentile for the whole cohort) to emphasize the large increase in potentially hypogonadal subjects between ages 40 and 85 years. Figure 3B indicates the means, quartiles, and 95% CI for each decade, implying changing reference ranges depending upon the age of the subject.
Comparing cross-sectional data from phase 2 for T, cFT, LH, and the T/LH ratio using simple regression analysis against age, T declines much more gradually (1.9% per decade) than INSL3 ( Figure S1). cFT also declines significantly at 10.8% per decade. In contrast, LH increases markedly with age at an average of 23.6% per decade, with the result that the T/LH ratio consequently declines significantly at 14.3% per decade. SHBG also increased at 15.5% per decade, and FSH increased at 33.4% per decade. All regressions were highly significant ( Figure S1).

Relationships between INSL3, total T, cFT, and other hormonal, anthropometric, and lifestyle parameters
Simple unadjusted correlation analysis ( Table 2) indicated that INSL3 was significantly positively correlated with total T, cFT, the T/LH ratio, and alcohol intake. It was negatively correlated with age, SHBG, LH, FSH, BMI, and WC. The results for T show that T is positively correlated with SHBG, LH, and smoking, although not alcohol ( Table 2). The correlation matrix for cFT is essentially similar to that for T, although it correlates negatively with SHBG. When subsequently adjusted for subject age ± center (not shown), partial correlation analysis shows that correlations are still essentially similar to those in Table 2, although INSL3 is now also significantly negatively correlated with smoking (R = -0.110, p < 0.001, n = 1874), as well as to alcohol consumption (R = 0.067, p = 0.004, n = 1874).   (Table 3; excl.). The size of the standardized beta coefficient reflects the relative importance of those parameters, noting that gonadotropins appear to be more important for T than for INSL3. Modeling of cFT is essentially similar to that for T (not shown).

Effect of weight gain or loss for individual men between phases 1 and 2
For those individuals who indicated substantial changes (weight loss or weight gain) in either BMI (±2 units) or WC (±4 cm) over the 4.3 years (average) between the two phases of the study, this had no significant impact on circulating INSL3 levels ( Figure 4). The generally negative effect on INSL3, irrespective of weight gain or loss, can largely be attributed to the age-linked association alluded to above (Figure 2), whereby INSL3 is reduced on average by 15.1% per decade; this is represented by the dashed line in Figure 4. Similar analyses were carried out for T, cFT, and SHBG ( Figure S2) largely confirmed the results from a previous study, 23 whereby SHBG was significantly affected by weight loss; T showed a similar trend, although because of the reduced numbers than the previous study, this did not attain significance; cFT showed no effect.  Note: Waist circumference was not included in the analysis as being colinear with body mass index (BMI). Std. β is the standardized beta coefficient for the model with its t-value and significance (p-value). "excl." indicates a parameter excluded through lack of statistical significance in the final optimal model. Abbreviations: FSH, follicle-stimulating hormone; LH, luteinizing hormone; SHBG, sex hormone-binding globulin.

Effect of age
reliable index of mature Leydig cell numbers and their differentiation status, reflecting their functional capacity. This view is supported by the similar decline of ca. 14.3% per decade in the T/LH ratio ( Figure   S1D), also regarded as an index of Leydig cell functional capacity, 26,27 and by the comparable decline (44% over 30 years) reported for mature Leydig cell numbers. 3 This is the first study, however, to provide longi-

Effect of the HPG axis
The INSL3 concentration correlates well (

Effect of obesity and lifestyle factors
Of the lifestyle parameters measured, and as shown earlier, 15 Figure S3); the impact is relatively smaller (-9.4%) for the former than T (-21.5%), cFT (-14.5%), and SHBG (-17.4%). For T, this has been linked partly to the ability of adipose tissue to use aromatase to convert androgens to estrogens, 31,32 although direct effects of obesogens or inflammatory cytokines on the testes are not excluded. 32 Partly also, this may be because BMI can influence SHBG levels ( Figure S2), which in turn may regulate hypothalamic feedback and hence gonadotropin production. 33

Geographic variation in INSL3 concentration
Leuven and Manchester had significantly higher mean circulating INSL3 concentrations than the total cohort average, and Lodz and Tartu had significantly lower concentrations. In contrast, the T concentration indicates only small between-center differences ( Table 1) found little age-dependent change within a Danish cohort of younger men (age 19-40 years, 0.9-2.7 ng/ml; age 50-60 years, 0.9-2.5 ng/ml).
We also observed a similar INSL3 reference range for a cohort of 18-year-old Swedish men (0.4-1.9 ng/ml). 28

CONCLUSION
Together, these findings show that in normal subjects, Leydig cell functional capacity as reflected in circulating insulin-like peptide 3 is highly consistent within an individual over long periods and that insulin-like peptide 3 is a powerful long-term index of Leydig cell status. 38