madman
Super Moderator
Again you heard it here first only at Nelson's ExcelMale!
@Nelson Vergel
Was going to drop this bomb a few days ago but you let down the curtains early LOL!
Reference Intervals for Free Testosterone in Adult Men Measured Using a Standardized Equilibrium Dialysis Procedure (2022)
Ravi Jasuja, Ph.D.; Karol M. Pencina, Ph.D.; Daniel J. Spencer, BS; Liming Peng, MS; Fabiola Privat, BS; Waljit Dhillo; MD, Ph.D.; Channa Jayasena, MD, Ph.D.; Frances Hayes, MD; Bu B. Yeap, MB, BS, Ph.D.; Alvin M. Matsumoto, MD; Shalender Bhasin, MB, BS
ABSTRACT
Background: Free testosterone (FT) determination may be helpful in evaluating men suspected of testosterone deficiency, especially in conditions with altered binding-protein concentrations. However, methods for measuring FT by equilibrium dialysis and reference intervals vary among laboratories.
Objective: To determine reference intervals for FT in healthy, nonobese men by age groups as well as in healthy young men, 19 to 39 years, using a standardized equilibrium dialysis procedure.
Methods: We measured FT in 145 healthy, nonobese men, 19 years or older, using a standardized equilibrium dialysis method performed for 16 hours at 37ºC using undiluted serum and dialysis buffer that mimicked the ionic composition of human plasma. FT in the dialysate was measured using a CDC-certified liquid chromatography-tandem mass spectrometry assay.
Results: In healthy nonobese men, the 2.5th, 10th, 50th, 90th, and 97.5th percentile values for FT were 66, 91, 141, 240, and 309 pg/mL, respectively; corresponding values for men 19 to 39 years, were 120, 128, 190, 274 and 368 pg/mL, respectively. FT levels by age groups exhibit the expected age-related decline. FT levels were negatively associated with body mass index, age, and SHBG levels. Percent FT was lower in middle-aged and older men than young men adjusting for SHBG level.
Discussion: Further studies are needed to determine how these reference intervals apply to the diagnosis of androgen deficiency in clinical populations and in men of different races and ethnicities in different geographic regions.
Conclusion: Reference intervals for free FT levels (normative range 66-309 pg/mL [229- 1072 pmol/L] in all men and 120-368 pg/mL [415-1274 pmol/L] in men, 19-to-39 years), measured using a standardized equilibrium dialysis method in healthy nonobese men, provide a rational basis for categorizing FT levels. These intervals require further validation in other populations, in relation to outcomes, and in randomized trials.
INTRODUCTION
Most of the circulating testosterone is bound to sex hormone-binding globulin (SHBG) and human serum albumin (HSA), and a smaller fraction is bound to cortisol-binding globulin and orosomucoid; only less than 4.0% of circulating testosterone is unbound or free (1). Because serum total testosterone concentration represents the sum of unbound and protein-bound testosterone, the circulating levels of sex hormone binding globulin, HSA, and, to a lesser extent, other binding proteins affect the total testosterone concentrations (2-9). Common conditions, such as obesity, type 2 diabetes mellitus, and metabolic syndrome are associated with low SHBG levels (4,5,8,10,11); consequently, many men with these conditions have low total but normal free testosterone levels (12-15). Similarly, the mice expressing a human SHBG transgene exhibit increased circulating SHBG and total testosterone concentrations; but, in spite of markedly elevated SHBG and total testosterone concentrations, their free testosterone concentration is unaffected (16). Accordingly, practice guidelines of many professional societies (11,17,18), including the Endocrine Society (11), but not all societies (19,20) recommend the determination of free testosterone concentrations in men, who are being evaluated for hypogonadism and who have conditions that alter binding protein concentrations, or whose initial total testosterone concentrations are at or near the lower limit of the normal range.
An expert panel of the Endocrine Society (9) reviewed the various methods for determining free testosterone (4,5,9,21-32) and concluded that each method has some inherent limitations but that the equilibrium dialysis method is the reference standard against which all other methods should be compared (33). However, substantial heterogeneity in the procedures used by various laboratories for performing the equilibrium dialysis assay has contributed to variability in the reported free testosterone values. For instance, the equilibrium dialysis can be performed by adding 3H-labeled testosterone to the serum sample and free testosterone derived from the total testosterone concentration and the dialyzed fraction of 3H-testosterone; the procedures that use 3H-testosterone are susceptible to error due to tracer impurities and the potential for tritium exchange with water (9). Furthermore, the determination of free testosterone concentration by equilibrium dialysis is affected greatly by the assay conditions, including the buffer composition, the incubation time, and the temperature. Historically, some laboratories have included a carrier protein such as gelatin or bovine serum albumin in the dialysis buffer which affects the partitioning of testosterone among binding proteins and the measured free testosterone concentration. The equilibrium dialysis procedure can also be performed by direct measurement of testosterone on both sides of the dialysis chamber (24,25). The accuracy and precision of the total testosterone assay also affect the precision and accuracy of the measured free testosterone concentration. Most commercial laboratories do not report the buffer composition and other dialysis conditions which renders it difficult to evaluate their methods; the procedures for equilibrium dialysis have varied even in published reports from academic research laboratories (2,24,32-37). Because of the wide variation in the equilibrium dialysis procedures, the reference ranges are not generalizable across laboratories.
Although epidemiologic studies agree that free testosterone levels decline with advancing age, these reports are largely based on calculated testosterone levels (38-41). No epidemiologic study has reported the distribution of free testosterone levels in a carefully selected prospective sample of healthy men across a wide age range using the equilibrium dialysis method, widely considered the reference method.
In an effort to generate reference intervals for free testosterone concentrations in men that can be used across laboratories, we report here a detailed description of a non-proprietary, standardized equilibrium dialysis procedure that can be replicated in any qualified laboratory using commercially available reagents. Instead of using a tritium-labeled tracer, we used liquid chromatography-tandem mass spectrometry (LC-MS/MS) to directly measure testosterone concentrations on both sides of the dialysis chamber. We employed standardized dialysis conditions at 37ºC for a duration of time shown to be sufficient for reaching equilibrium. Using this validated equilibrium dialysis method, we evaluated the distribution of free testosterone concentrations in a prospectively-collected sample of community-living healthy nonobese men, 19 years or older. The blood samples were collected in the morning before 10 AM after an overnight fast. Total testosterone levels in the serum as well as in the dialysate were measured using a validated LC-MS/MS assay that is certified by the Hormone Standardization Program for Testosterone (HoST) of the US Centers for Disease Control and Prevention (42). We report the distribution of free testosterone concentrations in healthy young men, 19 to 39 years, as well by age groups, generated using this standardized equilibrium dialysis procedure.
METHODS
The study was approved by the Institutional Review Board of the Massachusetts General Brigham Healthcare System. The participants provided written informed consent.
Study Population
The participants were healthy community-dwelling men, 19 years of age or older, with no known disease of the hypothalamus, pituitary, or the testes, and had normal physical examinations including normal testicular volume, blood counts, and chemistries. Those with body mass index >30 kg/m2, any acute or unstable chronic illness, AST or ALT > 1.5 times the upper limit of normal, diabetes mellitus, or serum creatinine >1.5 mg/dL were excluded. We also excluded men using prescription drugs, testosterone, anabolic steroids, or dietary supplements containing any androgen including DHEA and androstenedione. Stable replacement doses of thyroid hormones and anti-hypertensive drugs were allowed.
Study Procedures
Potential participants who responded to advertisements or letters were asked structured questions to verify key eligibility criteria. Those who met the major eligibility criteria were invited for an in-person screening visit during which written informed consent was obtained, and detailed medical history, medication, height and weight, a directed physical examination, urinalysis, and blood collection for blood counts and chemistries were performed. The men, who met all the inclusion criteria and none of the exclusion criteria, were invited back for Visit 2 during which an early morning blood sample was obtained for hormone measurements before 10 AM after an overnight fast. The serum was separated, aliquoted, and stored at -80C and never thawed until the time of the assay.
A standardized equilibrium dialysis procedure coupled with liquid chromatography-tandem mass spectrometry assay to measure free testosterone levels
Free testosterone concentrations in human serum samples were determined using a standardized protocol of equilibrium dialysis coupled with liquid chromatography-tandem mass spectrometry (LC-MS/MS) assay for testosterone. Equilibrium dialysis was performed in 96-well plates (Harvard Apparatus, Holliston, MA) with semi-permeable membranes that allow species less than 10 kDa to pass through (41,43). The dialysis buffer composition simulated the ionic strength conditions of the human plasma: 90 mM NaCl, 3 nM KCl, 1.3 mM KH2PO4, 1.9 mM, 1.1 mM MgSO4·7H2O, 5 mM 0.30 g urea and 23 mM HEPES sodium salt, 30 nM HEPES acid, 8 mM sodium azide, and 1 mL of 0.06% DL lactic acid. 200 µL of dialysis buffer was added to the “buffer side” and 200 µL of undiluted serum was added to the “sample side”.
To standardize the dialysis procedure, in preliminary experiments, we determined the time required to achieve equilibrium in the dialysis and compared the effect of incubation at room temperature versus 37ºC. After the dialysis conditions had been standardized in these preliminary experiments, the dialysis plates were incubated for 24 hours at 37ºC, after which 150 µL aliquots were removed from each side for testosterone measurement using a validated LC-MS/MS assay that is certified by the Center for Disease Control’s Hormone Standardization Program (HoST) (44,45). The dialysis plates were rotated throughout the course of the experiment to disrupt the formation of the Nernst diffusion layer and minimize its influence on the diffusion rate.
Standardized Measurement of Testosterone Using a HoST-Certified Liquid Chromatography-Tandem Mass Spectrometry Assay
Aliquots from each side of the dialysis chambers were transferred to 96 well plates and d3-testosterone was added to each sample as an internal standard. Solid phase extraction was performed using Strata-X 33µ Polymeric Reversed Phase in 96-well plates (Phenomenex 8E-S100-AGB), conditioned with 0.5mL methanol followed by 1mL deionized water prior to loading of samples. The wells were washed with 1mL water followed by 40% methanol. The eluates were evaporated to dryness under N2 gas at 40ºC and the residues were reconstituted with methanol/deionized water (1:1) for LC-MS/MS analysis, as described previously (45).
A Shimadzu UFLC XR system (Shimadzu Scientific Instruments, Pleasanton, CA) consisting of DGU-20A5R solvent degasser, LC-20 AD XR Binary Pump A and B, CTO-20 AC Column Oven, SIL-20 AC XR Autosampler, a Kinetex® 2.6μm C18, 50 ×2.1mm column (Phenomenex, Torrance, CA) was used for chromatographic separations. A linear gradient elution from 40% to 100% methanol in 2 min at a flow rate of 0.6 ml/min was used and the eluant was fed into an AB Sciex 5500 QTRAP® (AB Sciex, Foster City, CA) equipped with a Turbo V ion source; ESI ionization probe in positive ion mode was used for detection. Multiple Reaction Monitoring (MRM) transitions 289/109 and 289/97 for testosterone and 292/109 and 292/97 for the internal standard (D3-testosterone) were used for the quantification of testosterone. The data acquisition and processing were carried out using the Analyst 1.6.3/1.7.0 software (AB Sciex, Foster City, CA). As quality control measures, the National Institute of Standards and Technology (NIST) male and female controls, and house quality control samples were assayed concurrently. The lower limit of quantitation for testosterone was 1 ng/dL, and the assay exhibited a linear range from 1 to 1000 ng/dL. In an independent experiment, the recovery by LC-MS/MS assay was found to be 102 ± 2% in the aliquots of charcoal-stripped serum spiked with known quantities of testosterone. The laboratory is certified by the HoST Certification Program of the CDC, an accuracy-based benchmark that evaluates the assay's performance by sending every three months 40 quality control samples with testosterone concentrations that extend across the entire range of physiologic testosterone range in men, women, and children. The overall mean (95% confidence interval) bias from the HoST quality control pools was 0% (-1.2 to +1.2%). The bias was 0.7% (- 2.3 to +3.6%) in the low range (-7.7 to 24.0 ng/dL); -8 (-2.7 to + 1.0%) in the medium range (24.7 to 318.0 ng/dL); and 0.1 (-1.5 to +1.8%) in the high range (329 to 942 ng/dL).
The intra-assay CVs were <10%, and inter-assay CVs of the equilibrium dialysis method were 11.5 and 14.5% in the two male quality control pools, 14.2 and 14.2% in the two female quality control pools, and 9.8% at the lower limit of quantitation (1 pg/mL).
Serum SHBG levels were measured using a chemiluminescent immunoassay (43,44) with a lower limit of quantitation of 0.33 nmol/L. The linear range is 0.33 nmol/L to 200.0 nmol/L. The intra-assay CV is 4.5 to 4.8% and the inter-assay CV is 5.2 to 5.5% across the male range.
Statistical Methods
The baseline characteristics of study participants are presented across age groups and overall, as means and standard deviations or as percentages. The distributional properties of variables were inspected graphically and quantitatively. 2.5th, 10th, 25th, 50th, 75th, 90th, and 97.5 th percentile values are provided for absolute hormone values (free and total testosterone and SHBG concentrations) and for percent free testosterone.
The associations between hormone levels and age and BMI were analyzed using linear regression models. The magnitude of these relationships was assessed using R-squared metrics and Pearson’s correlation coefficients. The non-linear relationship between percent free testosterone and SHBG was assessed using polynomial regression models adjusted to age. The two-sided type I error was set at 0.05 for all hypotheses testing. Statistical analyses were conducted using SAS 9.4 (SAS Institute, Inc, Cary, NC) and R software version 4.2.0. (R Foundation).
RESULTS
Standardized dialysis conditions in protein-free dialysis buffer provide a robust framework for free testosterone determination.
Steady-state equilibrium dialysis is considered the reference method for the separation of free and protein-bound testosterone fractions (9,25). We determined the effect of temperature on the percent of testosterone dialyzed since some laboratories use dialysis at room temperature because of the convenience of incubating the plates at room temperature. The dialysis performed at room temperature (23oC) resulted in lower free testosterone values in the buffer side than at 37oC (Figure 1A). Accordingly, subsequent assays were performed at 37oC in a temperature-controlled oven.
To determine the time needed to reach steady-state distribution of free testosterone across the dialysis membrane, we conducted a time course experiment in which 6 to 8 replicates of serum samples were dialyzed for 1, 2, 4, 6, 8, 12, 16, 24, 36, and 48 hours at 37oC and testosterone levels on the sample and the buffer side were determined by LC-MS/MS (Figure 1B). We found that equilibrium was achieved between 12 to 16 hours and percent free testosterone in the dialysate did not differ among wells that were incubated for 12, 16, 24, 36, or 48 hours at 37oC. The intra-assay CV was 3.9% at 24 hours.
Study Participants
Among the subjects who were screened on the telephone, 230 underwent in-person screening in the clinical research unit, and 147 met all eligibility criteria (STROBE DIAGRAM, Figure 2). One hundred and forty-five enrolled participants completed the study, in whom total and free testosterone levels were available, and were included in the analyses. The study participants were ethnically and racially diverse and included 12.4% Hispanic or Latinx, 86.2% non-Hispanic, and 1.4% other; 69.0% were White, 9.7% Africa-Americans, 15.2% Asian, and 6.2% others. The descriptive data summarizing the subjects’ characteristics across age groups is presented in Table 1. The mean ± SD age of the study participants was 48.9 ± 19.5 years, and the body mass index was 25.6 ± 2.9 kg/m2.
Distribution of free testosterone, total testosterone, and SHBG levels in the community-dwelling healthy men
As expected, serum-free testosterone levels were negatively associated with age (rho=-0.65) and were lower in middle-aged and older men than in young men (Table 3). Serum-free testosterone levels also were negatively associated with body mass index (rho=- 0.23) and SHBG levels (rho=-0.28) in unadjusted analyses as well as in analyses adjusted for age. SHBG levels were positively associated with age (rho=0.49) and higher in older men than young men (Table 2)
In healthy nonobese men, 19 years or older, the 2.5th, 10th, 25th, 50th, 75th, 90th and 97.5th percentile values for absolute free testosterone were 66, 91, 116, 141, 190, 240, and 309 pg/mL, respectively (Table 3). To convert standard units (pg/mL) to SI units (pmol/L), please divide the concentrations in pg/mL by 0.2885. In the subset of men, 19 to 39 years, the corresponding 2.5 th, 10th, 25th, 50th, 75th, 90th and 97.5th percentile values for absolute free testosterone were 120, 128, 149, 190, 228, 274, and 368 pg/mL. By convention, the 2.5th percentile of the reference sample defines the lower limit of the reference range and the 97.5th percentile value defines the upper limit (46,47). By this convention, the normal range in the reference sample of men 19 to 39 years is 120 to 368 pg/mL (415 to 1274 pmol/L) and, in all men, the range is 120 to 368 pg/mL [229 to 1072 pmol/L], respectively.
The Effect of Age on the Relation Between SHBG and Free Testosterone Concentrations
Not only were the absolute free testosterone concentrations lower in older men than young men, as expected (38,39,45,48), but we found that compared to young men, 19 to 39 years, the proportion of circulating total testosterone that was free also was significantly lower in middle-aged men (40 to 59 years) (mean difference = -0.72%, 95% CI: -1.15, -0.29; p-value<0.001), and in older men, 60 or older (mean difference=-1.42%, 95% CI: -1.78, - 1.06; p-value<0.001) even after adjusting for SHBG concentrations. Interestingly, across the range of SHBG concentrations, young men, 19 to 39 years, exhibited higher percent free testosterone compared to men, 40 years or older, and compared to men 60 years or older (Figure 3). The age-adjusted association between percent-free testosterone and SHBG levels was statistically significant and non-linear (partial R-squared=0.32; p<0.001).
DISCUSSION
The guidelines of some professional societies recommend measuring free testosterone levels in men who are being evaluated for testosterone deficiency especially when alterations in binding protein levels are suspected or when the total testosterone levels are only slightly below or slightly above the diagnostic threshold (11,17,18). Using this standardized equilibrium dialysis method coupled with a HoST-certified LC-MS/MS assay for measuring testosterone in the dialysate, described in detail here to enable its easy replication across laboratories, we report for the first time the distribution of free testosterone levels in a prospectively collected sample of carefully screened healthy men, 19 years or older. The range of free testosterone levels (2.5th to 97.5th percentile values) in healthy nonobese men, 19 years or older, is 66 to 309 pg/mL (229 to 1072 pmol/L), and in a reference sample of nonobese healthy young men, 19 to 39 years, is 120 to 368 pg/mL (415 to 1274 pmol/L). The distribution of free testosterone levels by age group is also reported. These normative data can potentially be applied after appropriate cross-calibration to other laboratories that perform the equilibrium dialysis under similar standardized conditions and use a testosterone assay that is certified by an accuracy-based standardization program such as the Center for Disease Control and Prevention's HoST program. Further studies are needed to determine how these reference intervals apply to the diagnosis of androgen deficiency in clinical populations and in men of different races and ethnicities in different geographic regions.
We found that in addition to the lower absolute free testosterone levels, as expected, the percent free testosterone also were lower in older men than young men; this might be expected as SHBG levels were higher in older men than young men. However, the percent free testosterone was lower in older men than young men at any given SHBG level. Thus, the relation of free testosterone to total testosterone differed between young and older men across a range of SHBG concentrations. The mechanisms of age-related decline in percent free testosterone even after adjusting for SHBG levels are not clear; it is possible that posttranslational modifications of SHBG or other binding proteins and/ or changes in the set point of the hypothalamic-pituitary-testicular feed-forward and feedback homeostatic loops might contribute to the age-related changes in the relation between percent free and total testosterone levels.
There is agreement that free testosterone levels decline with advancing age (10,38,40,48). Yet, there is a remarkable paucity of data on the distribution of free testosterone levels measured using a standardized equilibrium dialysis procedure across a wide age range in healthy men. Very few cohort studies have measured free testosterone levels using equilibrium dialysis (10,35,49). Vermeulen et al (10) and Pirke and Doerr (35) used immunoassays for total testosterone and equilibrium dialysis to measure free testosterone in convenience samples of adult men, and have reported the age-related decline in free testosterone levels. Our study is unique in reporting free testosterone levels using a standardized dialysis procedure coupled with direct measurement of testosterone in the dialysate using an LC-MS/MS assay that has been certified by the CDC's accuracy-based Hormone Standardization Program for Testosterone (HoST) in samples that were prospectively collected in the morning before 10 AM after an overnight fast and stored at - 80C and never thawed. We also report here free testosterone levels by age group. To minimize the confounding influence of disease and obesity, we carefully selected healthy, non-obese men. Unlike some other epidemiologic studies, which included only middle-aged and older individuals (38,39), we included both young and older individuals. The data have internal consistency, as indicated by the expected negative association of testosterone with age and body mass index.
This study also has some limitations. These reference ranges were derived from single morning samples, which discount the pulsatile and diurnal secretory rhythms. Previous analyses show that early morning testosterone levels, obtained in a manner similar to that used by physicians in practice, are associated cross-sectionally with symptoms and clinical outcomes (38-40,48). We report reference ranges in men <40 years of age but also provide the distribution of free testosterone levels by age group. This approach of generating the reference range in healthy young men is analogous to the use of T-scores for bone mineral density. Although the sample included men of various races and ethnicities, the number of nonwhites was not large enough to offer sufficient power to detect meaningful differences in free testosterone levels among racial or ethnic subgroups. The data on geographic and racial differences in total testosterone levels are inconsistent (50,51), and no study has examined racial and ethnic differences in free testosterone levels using equilibrium dialysis in different geographic regions of the world. Additional investigations of multi-ethnic cohorts to evaluate the generalizability of the proposed reference limits to men of other races and ethnicities in different regions of the world are important. Although the sample size was within the IFCC guidelines for analytes with normal distribution, there were relatively small numbers of men within each decade of age and larger sample size would provide more robust estimates of the reference ranges by age decades.
Our normative ranges are similar to those reported by pioneers of this field using the legacy method (2) but the percent free testosterone in our reference sample differs from that reported by another laboratory (10,52); this difference in percent free testosterone from that reported by another research laboratory (<2.8%) could be due to differences in buffer composition or in other assay conditions that are not apparent in the published methods (10,52). The normative ranges of most commercial laboratories have changed substantially in recent years suggesting changes in their procedures over time (53,54); because procedures used by the commercial laboratories and the details of how reference ranges were derived are not published, an evaluation of these procedures was not feasible.
These reference ranges, generated in a reference sample of healthy men, should not be applied to other assays in other laboratories without appropriate cross-calibration of assays. Differences in study populations, time of sample collection, and testosterone assays can contribute to the differences in reference ranges. The adoption of a standardized procedure for measuring free testosterone and cross-calibration of the testosterone assays against an accuracy-based benchmark such as the CDC's HoST program will facilitate the application of these reference ranges across laboratories.
The data here defines reference intervals from a population of healthy nonobese men using a standardized equilibrium dialysis procedure coupled to a HoST-certified LC-MS/MS assay and represent an important first step. Further studies are needed to determine how well these reference limits can be applied to the diagnosis of androgen deficiency in clinical populations and in people of different races and ethnicities in different geographic regions. The association of low free testosterone defined using these criteria with incident outcomes in epidemiologic studies should be studied. Importantly, randomized trials are needed to determine whether testosterone therapy improves outcomes in men, who have free testosterone below these reference limits.
@Nelson Vergel
Was going to drop this bomb a few days ago but you let down the curtains early LOL!
Reference Intervals for Free Testosterone in Adult Men Measured Using a Standardized Equilibrium Dialysis Procedure (2022)
Ravi Jasuja, Ph.D.; Karol M. Pencina, Ph.D.; Daniel J. Spencer, BS; Liming Peng, MS; Fabiola Privat, BS; Waljit Dhillo; MD, Ph.D.; Channa Jayasena, MD, Ph.D.; Frances Hayes, MD; Bu B. Yeap, MB, BS, Ph.D.; Alvin M. Matsumoto, MD; Shalender Bhasin, MB, BS
ABSTRACT
Background: Free testosterone (FT) determination may be helpful in evaluating men suspected of testosterone deficiency, especially in conditions with altered binding-protein concentrations. However, methods for measuring FT by equilibrium dialysis and reference intervals vary among laboratories.
Objective: To determine reference intervals for FT in healthy, nonobese men by age groups as well as in healthy young men, 19 to 39 years, using a standardized equilibrium dialysis procedure.
Methods: We measured FT in 145 healthy, nonobese men, 19 years or older, using a standardized equilibrium dialysis method performed for 16 hours at 37ºC using undiluted serum and dialysis buffer that mimicked the ionic composition of human plasma. FT in the dialysate was measured using a CDC-certified liquid chromatography-tandem mass spectrometry assay.
Results: In healthy nonobese men, the 2.5th, 10th, 50th, 90th, and 97.5th percentile values for FT were 66, 91, 141, 240, and 309 pg/mL, respectively; corresponding values for men 19 to 39 years, were 120, 128, 190, 274 and 368 pg/mL, respectively. FT levels by age groups exhibit the expected age-related decline. FT levels were negatively associated with body mass index, age, and SHBG levels. Percent FT was lower in middle-aged and older men than young men adjusting for SHBG level.
Discussion: Further studies are needed to determine how these reference intervals apply to the diagnosis of androgen deficiency in clinical populations and in men of different races and ethnicities in different geographic regions.
Conclusion: Reference intervals for free FT levels (normative range 66-309 pg/mL [229- 1072 pmol/L] in all men and 120-368 pg/mL [415-1274 pmol/L] in men, 19-to-39 years), measured using a standardized equilibrium dialysis method in healthy nonobese men, provide a rational basis for categorizing FT levels. These intervals require further validation in other populations, in relation to outcomes, and in randomized trials.
INTRODUCTION
Most of the circulating testosterone is bound to sex hormone-binding globulin (SHBG) and human serum albumin (HSA), and a smaller fraction is bound to cortisol-binding globulin and orosomucoid; only less than 4.0% of circulating testosterone is unbound or free (1). Because serum total testosterone concentration represents the sum of unbound and protein-bound testosterone, the circulating levels of sex hormone binding globulin, HSA, and, to a lesser extent, other binding proteins affect the total testosterone concentrations (2-9). Common conditions, such as obesity, type 2 diabetes mellitus, and metabolic syndrome are associated with low SHBG levels (4,5,8,10,11); consequently, many men with these conditions have low total but normal free testosterone levels (12-15). Similarly, the mice expressing a human SHBG transgene exhibit increased circulating SHBG and total testosterone concentrations; but, in spite of markedly elevated SHBG and total testosterone concentrations, their free testosterone concentration is unaffected (16). Accordingly, practice guidelines of many professional societies (11,17,18), including the Endocrine Society (11), but not all societies (19,20) recommend the determination of free testosterone concentrations in men, who are being evaluated for hypogonadism and who have conditions that alter binding protein concentrations, or whose initial total testosterone concentrations are at or near the lower limit of the normal range.
An expert panel of the Endocrine Society (9) reviewed the various methods for determining free testosterone (4,5,9,21-32) and concluded that each method has some inherent limitations but that the equilibrium dialysis method is the reference standard against which all other methods should be compared (33). However, substantial heterogeneity in the procedures used by various laboratories for performing the equilibrium dialysis assay has contributed to variability in the reported free testosterone values. For instance, the equilibrium dialysis can be performed by adding 3H-labeled testosterone to the serum sample and free testosterone derived from the total testosterone concentration and the dialyzed fraction of 3H-testosterone; the procedures that use 3H-testosterone are susceptible to error due to tracer impurities and the potential for tritium exchange with water (9). Furthermore, the determination of free testosterone concentration by equilibrium dialysis is affected greatly by the assay conditions, including the buffer composition, the incubation time, and the temperature. Historically, some laboratories have included a carrier protein such as gelatin or bovine serum albumin in the dialysis buffer which affects the partitioning of testosterone among binding proteins and the measured free testosterone concentration. The equilibrium dialysis procedure can also be performed by direct measurement of testosterone on both sides of the dialysis chamber (24,25). The accuracy and precision of the total testosterone assay also affect the precision and accuracy of the measured free testosterone concentration. Most commercial laboratories do not report the buffer composition and other dialysis conditions which renders it difficult to evaluate their methods; the procedures for equilibrium dialysis have varied even in published reports from academic research laboratories (2,24,32-37). Because of the wide variation in the equilibrium dialysis procedures, the reference ranges are not generalizable across laboratories.
Although epidemiologic studies agree that free testosterone levels decline with advancing age, these reports are largely based on calculated testosterone levels (38-41). No epidemiologic study has reported the distribution of free testosterone levels in a carefully selected prospective sample of healthy men across a wide age range using the equilibrium dialysis method, widely considered the reference method.
In an effort to generate reference intervals for free testosterone concentrations in men that can be used across laboratories, we report here a detailed description of a non-proprietary, standardized equilibrium dialysis procedure that can be replicated in any qualified laboratory using commercially available reagents. Instead of using a tritium-labeled tracer, we used liquid chromatography-tandem mass spectrometry (LC-MS/MS) to directly measure testosterone concentrations on both sides of the dialysis chamber. We employed standardized dialysis conditions at 37ºC for a duration of time shown to be sufficient for reaching equilibrium. Using this validated equilibrium dialysis method, we evaluated the distribution of free testosterone concentrations in a prospectively-collected sample of community-living healthy nonobese men, 19 years or older. The blood samples were collected in the morning before 10 AM after an overnight fast. Total testosterone levels in the serum as well as in the dialysate were measured using a validated LC-MS/MS assay that is certified by the Hormone Standardization Program for Testosterone (HoST) of the US Centers for Disease Control and Prevention (42). We report the distribution of free testosterone concentrations in healthy young men, 19 to 39 years, as well by age groups, generated using this standardized equilibrium dialysis procedure.
METHODS
The study was approved by the Institutional Review Board of the Massachusetts General Brigham Healthcare System. The participants provided written informed consent.
Study Population
The participants were healthy community-dwelling men, 19 years of age or older, with no known disease of the hypothalamus, pituitary, or the testes, and had normal physical examinations including normal testicular volume, blood counts, and chemistries. Those with body mass index >30 kg/m2, any acute or unstable chronic illness, AST or ALT > 1.5 times the upper limit of normal, diabetes mellitus, or serum creatinine >1.5 mg/dL were excluded. We also excluded men using prescription drugs, testosterone, anabolic steroids, or dietary supplements containing any androgen including DHEA and androstenedione. Stable replacement doses of thyroid hormones and anti-hypertensive drugs were allowed.
Study Procedures
Potential participants who responded to advertisements or letters were asked structured questions to verify key eligibility criteria. Those who met the major eligibility criteria were invited for an in-person screening visit during which written informed consent was obtained, and detailed medical history, medication, height and weight, a directed physical examination, urinalysis, and blood collection for blood counts and chemistries were performed. The men, who met all the inclusion criteria and none of the exclusion criteria, were invited back for Visit 2 during which an early morning blood sample was obtained for hormone measurements before 10 AM after an overnight fast. The serum was separated, aliquoted, and stored at -80C and never thawed until the time of the assay.
A standardized equilibrium dialysis procedure coupled with liquid chromatography-tandem mass spectrometry assay to measure free testosterone levels
Free testosterone concentrations in human serum samples were determined using a standardized protocol of equilibrium dialysis coupled with liquid chromatography-tandem mass spectrometry (LC-MS/MS) assay for testosterone. Equilibrium dialysis was performed in 96-well plates (Harvard Apparatus, Holliston, MA) with semi-permeable membranes that allow species less than 10 kDa to pass through (41,43). The dialysis buffer composition simulated the ionic strength conditions of the human plasma: 90 mM NaCl, 3 nM KCl, 1.3 mM KH2PO4, 1.9 mM, 1.1 mM MgSO4·7H2O, 5 mM 0.30 g urea and 23 mM HEPES sodium salt, 30 nM HEPES acid, 8 mM sodium azide, and 1 mL of 0.06% DL lactic acid. 200 µL of dialysis buffer was added to the “buffer side” and 200 µL of undiluted serum was added to the “sample side”.
To standardize the dialysis procedure, in preliminary experiments, we determined the time required to achieve equilibrium in the dialysis and compared the effect of incubation at room temperature versus 37ºC. After the dialysis conditions had been standardized in these preliminary experiments, the dialysis plates were incubated for 24 hours at 37ºC, after which 150 µL aliquots were removed from each side for testosterone measurement using a validated LC-MS/MS assay that is certified by the Center for Disease Control’s Hormone Standardization Program (HoST) (44,45). The dialysis plates were rotated throughout the course of the experiment to disrupt the formation of the Nernst diffusion layer and minimize its influence on the diffusion rate.
Standardized Measurement of Testosterone Using a HoST-Certified Liquid Chromatography-Tandem Mass Spectrometry Assay
Aliquots from each side of the dialysis chambers were transferred to 96 well plates and d3-testosterone was added to each sample as an internal standard. Solid phase extraction was performed using Strata-X 33µ Polymeric Reversed Phase in 96-well plates (Phenomenex 8E-S100-AGB), conditioned with 0.5mL methanol followed by 1mL deionized water prior to loading of samples. The wells were washed with 1mL water followed by 40% methanol. The eluates were evaporated to dryness under N2 gas at 40ºC and the residues were reconstituted with methanol/deionized water (1:1) for LC-MS/MS analysis, as described previously (45).
A Shimadzu UFLC XR system (Shimadzu Scientific Instruments, Pleasanton, CA) consisting of DGU-20A5R solvent degasser, LC-20 AD XR Binary Pump A and B, CTO-20 AC Column Oven, SIL-20 AC XR Autosampler, a Kinetex® 2.6μm C18, 50 ×2.1mm column (Phenomenex, Torrance, CA) was used for chromatographic separations. A linear gradient elution from 40% to 100% methanol in 2 min at a flow rate of 0.6 ml/min was used and the eluant was fed into an AB Sciex 5500 QTRAP® (AB Sciex, Foster City, CA) equipped with a Turbo V ion source; ESI ionization probe in positive ion mode was used for detection. Multiple Reaction Monitoring (MRM) transitions 289/109 and 289/97 for testosterone and 292/109 and 292/97 for the internal standard (D3-testosterone) were used for the quantification of testosterone. The data acquisition and processing were carried out using the Analyst 1.6.3/1.7.0 software (AB Sciex, Foster City, CA). As quality control measures, the National Institute of Standards and Technology (NIST) male and female controls, and house quality control samples were assayed concurrently. The lower limit of quantitation for testosterone was 1 ng/dL, and the assay exhibited a linear range from 1 to 1000 ng/dL. In an independent experiment, the recovery by LC-MS/MS assay was found to be 102 ± 2% in the aliquots of charcoal-stripped serum spiked with known quantities of testosterone. The laboratory is certified by the HoST Certification Program of the CDC, an accuracy-based benchmark that evaluates the assay's performance by sending every three months 40 quality control samples with testosterone concentrations that extend across the entire range of physiologic testosterone range in men, women, and children. The overall mean (95% confidence interval) bias from the HoST quality control pools was 0% (-1.2 to +1.2%). The bias was 0.7% (- 2.3 to +3.6%) in the low range (-7.7 to 24.0 ng/dL); -8 (-2.7 to + 1.0%) in the medium range (24.7 to 318.0 ng/dL); and 0.1 (-1.5 to +1.8%) in the high range (329 to 942 ng/dL).
The intra-assay CVs were <10%, and inter-assay CVs of the equilibrium dialysis method were 11.5 and 14.5% in the two male quality control pools, 14.2 and 14.2% in the two female quality control pools, and 9.8% at the lower limit of quantitation (1 pg/mL).
Serum SHBG levels were measured using a chemiluminescent immunoassay (43,44) with a lower limit of quantitation of 0.33 nmol/L. The linear range is 0.33 nmol/L to 200.0 nmol/L. The intra-assay CV is 4.5 to 4.8% and the inter-assay CV is 5.2 to 5.5% across the male range.
Statistical Methods
The baseline characteristics of study participants are presented across age groups and overall, as means and standard deviations or as percentages. The distributional properties of variables were inspected graphically and quantitatively. 2.5th, 10th, 25th, 50th, 75th, 90th, and 97.5 th percentile values are provided for absolute hormone values (free and total testosterone and SHBG concentrations) and for percent free testosterone.
The associations between hormone levels and age and BMI were analyzed using linear regression models. The magnitude of these relationships was assessed using R-squared metrics and Pearson’s correlation coefficients. The non-linear relationship between percent free testosterone and SHBG was assessed using polynomial regression models adjusted to age. The two-sided type I error was set at 0.05 for all hypotheses testing. Statistical analyses were conducted using SAS 9.4 (SAS Institute, Inc, Cary, NC) and R software version 4.2.0. (R Foundation).
RESULTS
Standardized dialysis conditions in protein-free dialysis buffer provide a robust framework for free testosterone determination.
Steady-state equilibrium dialysis is considered the reference method for the separation of free and protein-bound testosterone fractions (9,25). We determined the effect of temperature on the percent of testosterone dialyzed since some laboratories use dialysis at room temperature because of the convenience of incubating the plates at room temperature. The dialysis performed at room temperature (23oC) resulted in lower free testosterone values in the buffer side than at 37oC (Figure 1A). Accordingly, subsequent assays were performed at 37oC in a temperature-controlled oven.
To determine the time needed to reach steady-state distribution of free testosterone across the dialysis membrane, we conducted a time course experiment in which 6 to 8 replicates of serum samples were dialyzed for 1, 2, 4, 6, 8, 12, 16, 24, 36, and 48 hours at 37oC and testosterone levels on the sample and the buffer side were determined by LC-MS/MS (Figure 1B). We found that equilibrium was achieved between 12 to 16 hours and percent free testosterone in the dialysate did not differ among wells that were incubated for 12, 16, 24, 36, or 48 hours at 37oC. The intra-assay CV was 3.9% at 24 hours.
Study Participants
Among the subjects who were screened on the telephone, 230 underwent in-person screening in the clinical research unit, and 147 met all eligibility criteria (STROBE DIAGRAM, Figure 2). One hundred and forty-five enrolled participants completed the study, in whom total and free testosterone levels were available, and were included in the analyses. The study participants were ethnically and racially diverse and included 12.4% Hispanic or Latinx, 86.2% non-Hispanic, and 1.4% other; 69.0% were White, 9.7% Africa-Americans, 15.2% Asian, and 6.2% others. The descriptive data summarizing the subjects’ characteristics across age groups is presented in Table 1. The mean ± SD age of the study participants was 48.9 ± 19.5 years, and the body mass index was 25.6 ± 2.9 kg/m2.
Distribution of free testosterone, total testosterone, and SHBG levels in the community-dwelling healthy men
As expected, serum-free testosterone levels were negatively associated with age (rho=-0.65) and were lower in middle-aged and older men than in young men (Table 3). Serum-free testosterone levels also were negatively associated with body mass index (rho=- 0.23) and SHBG levels (rho=-0.28) in unadjusted analyses as well as in analyses adjusted for age. SHBG levels were positively associated with age (rho=0.49) and higher in older men than young men (Table 2)
In healthy nonobese men, 19 years or older, the 2.5th, 10th, 25th, 50th, 75th, 90th and 97.5th percentile values for absolute free testosterone were 66, 91, 116, 141, 190, 240, and 309 pg/mL, respectively (Table 3). To convert standard units (pg/mL) to SI units (pmol/L), please divide the concentrations in pg/mL by 0.2885. In the subset of men, 19 to 39 years, the corresponding 2.5 th, 10th, 25th, 50th, 75th, 90th and 97.5th percentile values for absolute free testosterone were 120, 128, 149, 190, 228, 274, and 368 pg/mL. By convention, the 2.5th percentile of the reference sample defines the lower limit of the reference range and the 97.5th percentile value defines the upper limit (46,47). By this convention, the normal range in the reference sample of men 19 to 39 years is 120 to 368 pg/mL (415 to 1274 pmol/L) and, in all men, the range is 120 to 368 pg/mL [229 to 1072 pmol/L], respectively.
The Effect of Age on the Relation Between SHBG and Free Testosterone Concentrations
Not only were the absolute free testosterone concentrations lower in older men than young men, as expected (38,39,45,48), but we found that compared to young men, 19 to 39 years, the proportion of circulating total testosterone that was free also was significantly lower in middle-aged men (40 to 59 years) (mean difference = -0.72%, 95% CI: -1.15, -0.29; p-value<0.001), and in older men, 60 or older (mean difference=-1.42%, 95% CI: -1.78, - 1.06; p-value<0.001) even after adjusting for SHBG concentrations. Interestingly, across the range of SHBG concentrations, young men, 19 to 39 years, exhibited higher percent free testosterone compared to men, 40 years or older, and compared to men 60 years or older (Figure 3). The age-adjusted association between percent-free testosterone and SHBG levels was statistically significant and non-linear (partial R-squared=0.32; p<0.001).
DISCUSSION
The guidelines of some professional societies recommend measuring free testosterone levels in men who are being evaluated for testosterone deficiency especially when alterations in binding protein levels are suspected or when the total testosterone levels are only slightly below or slightly above the diagnostic threshold (11,17,18). Using this standardized equilibrium dialysis method coupled with a HoST-certified LC-MS/MS assay for measuring testosterone in the dialysate, described in detail here to enable its easy replication across laboratories, we report for the first time the distribution of free testosterone levels in a prospectively collected sample of carefully screened healthy men, 19 years or older. The range of free testosterone levels (2.5th to 97.5th percentile values) in healthy nonobese men, 19 years or older, is 66 to 309 pg/mL (229 to 1072 pmol/L), and in a reference sample of nonobese healthy young men, 19 to 39 years, is 120 to 368 pg/mL (415 to 1274 pmol/L). The distribution of free testosterone levels by age group is also reported. These normative data can potentially be applied after appropriate cross-calibration to other laboratories that perform the equilibrium dialysis under similar standardized conditions and use a testosterone assay that is certified by an accuracy-based standardization program such as the Center for Disease Control and Prevention's HoST program. Further studies are needed to determine how these reference intervals apply to the diagnosis of androgen deficiency in clinical populations and in men of different races and ethnicities in different geographic regions.
We found that in addition to the lower absolute free testosterone levels, as expected, the percent free testosterone also were lower in older men than young men; this might be expected as SHBG levels were higher in older men than young men. However, the percent free testosterone was lower in older men than young men at any given SHBG level. Thus, the relation of free testosterone to total testosterone differed between young and older men across a range of SHBG concentrations. The mechanisms of age-related decline in percent free testosterone even after adjusting for SHBG levels are not clear; it is possible that posttranslational modifications of SHBG or other binding proteins and/ or changes in the set point of the hypothalamic-pituitary-testicular feed-forward and feedback homeostatic loops might contribute to the age-related changes in the relation between percent free and total testosterone levels.
There is agreement that free testosterone levels decline with advancing age (10,38,40,48). Yet, there is a remarkable paucity of data on the distribution of free testosterone levels measured using a standardized equilibrium dialysis procedure across a wide age range in healthy men. Very few cohort studies have measured free testosterone levels using equilibrium dialysis (10,35,49). Vermeulen et al (10) and Pirke and Doerr (35) used immunoassays for total testosterone and equilibrium dialysis to measure free testosterone in convenience samples of adult men, and have reported the age-related decline in free testosterone levels. Our study is unique in reporting free testosterone levels using a standardized dialysis procedure coupled with direct measurement of testosterone in the dialysate using an LC-MS/MS assay that has been certified by the CDC's accuracy-based Hormone Standardization Program for Testosterone (HoST) in samples that were prospectively collected in the morning before 10 AM after an overnight fast and stored at - 80C and never thawed. We also report here free testosterone levels by age group. To minimize the confounding influence of disease and obesity, we carefully selected healthy, non-obese men. Unlike some other epidemiologic studies, which included only middle-aged and older individuals (38,39), we included both young and older individuals. The data have internal consistency, as indicated by the expected negative association of testosterone with age and body mass index.
This study also has some limitations. These reference ranges were derived from single morning samples, which discount the pulsatile and diurnal secretory rhythms. Previous analyses show that early morning testosterone levels, obtained in a manner similar to that used by physicians in practice, are associated cross-sectionally with symptoms and clinical outcomes (38-40,48). We report reference ranges in men <40 years of age but also provide the distribution of free testosterone levels by age group. This approach of generating the reference range in healthy young men is analogous to the use of T-scores for bone mineral density. Although the sample included men of various races and ethnicities, the number of nonwhites was not large enough to offer sufficient power to detect meaningful differences in free testosterone levels among racial or ethnic subgroups. The data on geographic and racial differences in total testosterone levels are inconsistent (50,51), and no study has examined racial and ethnic differences in free testosterone levels using equilibrium dialysis in different geographic regions of the world. Additional investigations of multi-ethnic cohorts to evaluate the generalizability of the proposed reference limits to men of other races and ethnicities in different regions of the world are important. Although the sample size was within the IFCC guidelines for analytes with normal distribution, there were relatively small numbers of men within each decade of age and larger sample size would provide more robust estimates of the reference ranges by age decades.
Our normative ranges are similar to those reported by pioneers of this field using the legacy method (2) but the percent free testosterone in our reference sample differs from that reported by another laboratory (10,52); this difference in percent free testosterone from that reported by another research laboratory (<2.8%) could be due to differences in buffer composition or in other assay conditions that are not apparent in the published methods (10,52). The normative ranges of most commercial laboratories have changed substantially in recent years suggesting changes in their procedures over time (53,54); because procedures used by the commercial laboratories and the details of how reference ranges were derived are not published, an evaluation of these procedures was not feasible.
These reference ranges, generated in a reference sample of healthy men, should not be applied to other assays in other laboratories without appropriate cross-calibration of assays. Differences in study populations, time of sample collection, and testosterone assays can contribute to the differences in reference ranges. The adoption of a standardized procedure for measuring free testosterone and cross-calibration of the testosterone assays against an accuracy-based benchmark such as the CDC's HoST program will facilitate the application of these reference ranges across laboratories.
The data here defines reference intervals from a population of healthy nonobese men using a standardized equilibrium dialysis procedure coupled to a HoST-certified LC-MS/MS assay and represent an important first step. Further studies are needed to determine how well these reference limits can be applied to the diagnosis of androgen deficiency in clinical populations and in people of different races and ethnicities in different geographic regions. The association of low free testosterone defined using these criteria with incident outcomes in epidemiologic studies should be studied. Importantly, randomized trials are needed to determine whether testosterone therapy improves outcomes in men, who have free testosterone below these reference limits.
