- Email: [email protected]
ORIGINAL RESEARCH—ENDOCRINOLOGY: Comparison of Free Testosterone Results by Analog Radioimmunoassay and Calculated Free Testosterone in an Ambulatory Clinical Population
1948
ORIGINAL RESEARCH—ENDOCRINOLOGY Comparison of Free Testosterone Results by Analog Radioimmunoassay and Calculated Free Testosterone in an Ambulatory Clinical Population jsm_1473
1948..1953
Sergio A. Moreno, MD, Anita Shyam, MD, and Abraham Morgentaler, MD Men’s Health Boston, Harvard Medical School, Boston, MA, USA DOI: 10.1111/j.1743-6109.2009.01473.x
ABSTRACT
Introduction. The most widely used method for measuring free testosterone (FT) is by analog immunoassay (aFT); however, this assay has been criticized as unreliable based on laboratory studies in small groups of men. Calculated FT (cFT), derived from total testosterone (TT) and sex-hormone binding globulin (SHBG) values has been recommended in its place. There are limited data comparing aFT and cFT in clinical populations. Aim. The purpose of this study was to compare aFT with cFT in a population of ambulatory men in a clinical setting. Methods. Medical records were reviewed for 100 randomly selected men in a urology practice, yielding 140 test results complete for TT, aFT, and SHBG. Calculated FT was determined via an online calculator. Comparisons were made with Pearson rank coefficients. Main Outcome Measures. Pearson rank correlation between aFT and cFT. Results. Mean patient age was 52.3 ⫾ 14.3 years (range 24–80). Mean TT was 443.0 ⫾ 208.3 ng/dL (range 110– 1276). Mean aFT was 1.22 ⫾ 0.54 ng/dL (range 0.24–3.8) and mean cFT 9.4 ⫾ 4.5 ng/dL (range 1.8–27.8). Mean SHBG was 34.2 ⫾ 19.5 nmol/L (range 9–127). A strong correlation was observed for aFT and cFT (r = 0.88, P < 0.0001), particularly at low concentrations. Significant correlations were also noted between aFT and TT (r = 0.73, P < 0.0001), and between cFT and TT (r = 0.82, P < 0.0001). Numerical values for aFT were approximately one-eighth of the values obtained for cFT. Neither aFT nor cFT correlated with SHBG. Conclusions. A strong correlation was observed between aFT and cFT in this clinical population of ambulatory men. Different sets of reference values must be applied for each of these tests. Moreno SA, Shyam A, and Morgentaler A. Comparison of free testosterone results by analog radioimmunoassay and calculated free testosterone in an ambulatory clinical population. J Sex Med 2010;7:1948–1953. Key Words. Testosterone; Hypogonadism; Testosterone Deficiency; Analog Free Testosterone; Calculated Free Testosterone; Testosterone Blood Testing
Introduction
T
here is a growing appreciation of the importance of testosterone deficiency (TD) in men [1–6]. Traditionally, the diagnosis has been made on the basis of serum total testosterone (TT); however, there is a general consensus that serum free testosterone (FT) may more accurately reflect bioavailable T. Bioavailable T consists of the free and albumin-bound fractions of testosterone, and although bioavailable T can also be measured or J Sex Med 2010;7:1948–1953
calculated, much of the focus in recent years has been on FT. Yet, there is considerable uncertainty regarding the reliability and utility of currently available FT assays, particularly the “analog” radioimmunoassay (aFT) available in most laboratories. Equilibrium dialysis (EqD) has been considered the “gold standard” technique for determining FT concentrations [7]; however, this technique is time-consuming, non-automated, labor-intensive, and expensive [8,9]. For this reason, use of EqD © 2009 International Society for Sexual Medicine
1949
Analog and Calculated Free Testosterone has been largely restricted to research laboratories. Only 7 out of 122 laboratories reported using EqD in the September 2004 Ligand Special Survey of the College of American Pathology [10]. A survey of academic and community hospital laboratories in six U.S. states revealed that none of 24 offered EqD, and it was only made available from a national reference laboratory upon special request [11]. The most widely used method for measuring FT in the United States is by aFT, with 70% of U.S. laboratories reporting use of this assay [10]. However, the Endocrine Society and a group of international andrology societies have discouraged use of aFT, asserting it is inaccurate as values obtained by this assay are considerably lower than those obtained via EqD [1,2]. Instead, these groups have recommended use of calculated FT (cFT) for clinical practice, based on its strong correlation with EqD [1,2,12]. Results for cFT are obtained by a complex equation involving concentrations for TT, sex hormone binding globulin (SHBG), and albumin, but can more easily be obtained via an online calculator. The calculation is simplified by the use of a constant value for albumin (4.3 g/dL) in place of measured concentrations, without negatively affecting the correlation of cFT results with those obtained by EqD [9]. The criticism of aFT’s reliability despite its widespread use has created confusion for clinicians, some of whom have argued that aFT provides valuable clinical information regarding testosterone status in the adult man [13]. Moreover, in the landmark study by Vermeulen et al. in which various FT assays were directly compared, the correlation of aFT with EqD approached unity (r = 0.937), and was exceeded by only a minor degree by cFT (r = 0.987) [9]. However, numerical values for aFT were a fraction of those obtained by cFT and EqD, leading to the conclusion that this assay provided inaccurate results. Additional studies performed in small numbers of men in research laboratories have suggested that aFT provided no independent information from TT [14], and that it may not even detect dialyzable T [15]. These studies would appear to support the assertion that aFT may not be clinically useful. On the other hand, a literature search identified only a single publication that directly compared aFT and cFT results, demonstrating a strong correlation (r > 0.80) in 471 Japanese men aged 40–79 years [16].
The goal of this study was to examine the relationship of aFT, cFT, and TT results in a population of ambulatory men seen in a urological clinical practice. Methods
We retrospectively reviewed the medical records of 100 randomly selected men seen in a urology practice specializing in andrology, providing a total of 140 sets of test results that were complete for TT, aFT, and SHBG. The maximum number of samples from any one individual was three. Socio-demographic data and medical history were obtained. Blood samples were obtained during clinic hours of 8:00 am to 5:00 pm. Commercial immunoassays were used for measurement of TT (Tosoh Bioscience, San Francisco, CA, USA) and FT (Diagnostic Products Corp., Los Angeles, CA, USA). SHBG was measured by Immulite/ Chemiluminescence (Diagnostic Products Corp., Los Angeles, CA, USA). Results for cFT were obtained by using the online FT calculator at http://www.issam.ch/freetesto.htm. A standard value of 4.3 g/dL was used for albumin. Tests for TT were performed in-house, while aFT and SHBG were performed by a national reference laboratory (Quest Diagnostics). Statistical analyses were performed with Minitab statistical program version 15 (Minitab Inc., State College, PA, USA). Pearson’s rank test coefficients were used to evaluate relationships between continuous variables. All results are expressed as means ⫾ standard deviation. A P value of <0.05 was considered significant. Results
Mean patient age was 52.3 ⫾ 14.3 years (range 24–80). Erectile dysfunction was reported by 51 patients (51%). The most common additional medical conditions were hypertension (32%), infertility (21%), and dyslipidemia (19%). Fortythree patients (43%) were receiving testosterone therapy when blood tests were obtained. No patient in this study had significant renal or hepatic disease. The characteristics of the study population are presented in Table 1. Mean TT was 443.0 ⫾ 208.3 ng/dL (range 110–1276). Mean aFT was 1.22 ⫾ 0.54 ng/dL (range 0.24–3.8), and mean cFT 9.4 ⫾ 4.5 ng/dL (range 1.8–27.8; multiply by 34.7 to convert to pmol/L). The mean J Sex Med 2010;7:1948–1953
1950 Table 1
Moreno et al. General characteristics of population sample
Characteristics
Value*
Subjects Age (years) Erectile dysfunction Comorbidities Hypertension Infertility Dyslipidemia Diabetes mellitus Coronary artery disease Hormonal profile TT (ng/dL) aFT (ng/dL) cFT (ng/dL) SHBG (nmol/L)
100 52.3 ⫾ 14.3 (24–80) 51 (51) 32 21 19 7 4
(32) (21) (19) (7) (4)
Discussion
There is a great need to have a simple, readily available blood test to identify men with symptomatic TD [17]. Although TT has been used for this purpose, it is now recognized that TT may inaccurately reflect bioavailable serum T in some men
443.0 ⫾ 208.3 (110–1276) 1.22 ⫾ 0.54 (0.24–3.8) 9.4 ⫾ 4.5 (1.8–27.8) 34.2 ⫾ 19.5 (9–127)
*Values expressed as means ⫾ standard deviation (range), frequency (%). TT = total testosterone; aFT = analog free testosterone; cFT = calculated free testosterone; SHBG = sex hormone binding globulin.
SHBG was 34.2 ⫾ 19.5 nmol/L (range 9–127) and the median was 30 nmol/L. Figure 1 shows a strong correlation between aFT and cFT (r = 0.88, P < 0.0001). Examination of the scatter plot reveals a particularly tight correlation at lower values, with greater dispersion at higher concentrations. For aFT values above the median, the correlation with cFT was r = 0.81 (P < 0.0001), and for values below the median, it was r = 0.73 (P < 0.0001). Numerical values for aFT were approximately one-eighth of the values obtained for cFT. Significant correlations were also noted between aFT and TT (r = 0.73, P < 0.0001; Figure 2), and between cFT and TT (r = 0.82, P < 0.0001; Figure 3). No significant correlation was noted between aFT and SHBG (r = -0.13, P = 0.141), or cFT and SHBG (r = -0.12, P = 0.168). However, a significant correlation was observed between TT and SHBG (r = 0.41, P < 0.0001). A sub-analysis was performed by dividing the population into two groups based on SHBG values below or above the median to investigate how these tests correlated in men with relatively high or low SHBG concentrations. Numerically stronger correlations were noted in men in the lower SHBG sub-population. In men with SHBG below the median, the correlation between aFT and cFT was r = 0.90 (P < 0.0001); aFT and TT, r = 0.87 (P < 0.0001); and cFT and TT, r = 0.97 (P < 0.0001). For men with SHBG over the median, the corresponding r values were: aFT and cFT, r = 0.85 (P < 0.0001); aFT and TT, r = 0.76 (P < 0.0001); and cFT and TT, r = 0.85 (P < 0.0001). J Sex Med 2010;7:1948–1953
Figure 1 Analog free testosterone (afT) versus calculated free testosterone (cFT).
Figure 2 Analog free testosterone (aFT) versus total testosterone (TT).
Figure 3 Calculated free testosterone (cFT) versus total testosterone (TT).
Analog and Calculated Free Testosterone because of its tight binding to SHBG, which can vary considerably in concentration from one individual to another. A number of factors may increase SHBG, most notably age, and elevated levels of SHBG may produce normal-appearing TT results even when bioavailable T is depressed. This effect is illustrated by the minor decrease in TT seen with age, in contrast to a substantial decline in FT [18]. This study compared results from the most widely used FT assay, aFT, with results from the FT assay recommended by the Endocrine Society and others, cFT. The recommendation to use cFT clinically is based on studies showing a strong correlation with the “gold standard” technique of EqD, which is impractical for clinical practice and not widely available. The goal of this study was straightforward: How well do results by aFT compare with results from an established, recommended test, i.e. cFT? The primary results of this study revealed a very strong correlation of aFT with cFT (r = 0.88, P < 0.0001). Statistical texts indicate that correlations greater than 0.75 represent “very good to excellent relationships.” [19] Higher or lower values noted by cFT were associated with proportionately higher or lower values by aFT. Visual investigation of the scatterplot comparing aFT and cFT revealed particularly tight correlations at low concentrations, where the diagnosis of TD is made, and slightly greater dispersion at higher concentrations. Weaker but significant correlations were noted between aFT and TT, and cFT and TT. The strong relationship between aFT and cFT appears to be robust, as the study population included men with a wide range of total testosterone values (110–1276 ng/dL), and a broad age range (24–80 years). Moreover, this study included men who were deficient in T at the time of blood testing as well as men with high T values while receiving T therapy. Further investigation revealed that the strong correlation between aFT and cFT held true for men with relatively high or relatively low SHBG, defined as values above or below the median. No significant relationship was noted for either aFT or cFT results with SHBG concentrations, which is reassuring for a clinical test intended to avoid the confounding influence of SHBG on the interpretation of TT results. Numerical results for aFT were approximately one-eighth the values for cFT. This observation has been made previously, and this unfortunate aspect of the aFT assay has resulted in the criticism
1951 that the aFT assay is inaccurate. However, the strong correlation indicates this discrepancy is a calibration issue, one that has already been corrected in clinical practice by the use of a different set of reference values. Because the aFT assay is proprietary (originally Diagnostic Products Corp, now Siemens), it is unknown how numerical values were originally assigned when the test was introduced several decades ago. One may speculate that if the test were introduced today, with FT norms already established by EqD, numerical aFT results would have been calibrated differently to provide values more consistent with those obtained by cFT or EqD. The data in this study were obtained during routine clinical practice, with blood tests for aFT and SHBG performed by a national laboratory that performs many thousands of these tests annually. These results should thus have wide clinical applicability, as similar results would presumably be obtained by a clinician practicing anywhere in the United States. This differs from research studies of aFT that have been performed in small numbers of men under laboratory conditions that may differ from the automated, controlled environment of a large commercial laboratory. The results of this study are consistent with a study of 471 Japanese men that revealed a strong correlation between aFT and cFT with an r value of 0.803 [16]. In that study, the two tests had a 98.7% correspondence for results more than two standard deviations below the mean, representing the standard range categorized as abnormally low [19]. In the current study, there was 100% concordance between aFT and cFT values that fell in this range. Others have also noted a strong and highly significant correlation between aFT and cFT [20]. It is important to note that cFT values in this study were obtained using a standard value of 4.3 g/dL for albumin, which has been shown in healthy men to provide accurate results [9]. It is unknown how the correlation between aFT and cFT would have been influenced by the use of actual albumin values, which may vary considerably in clinical populations. A technical consideration is that blood tests were obtained throughout the day and were not restricted to the early morning, as recommended by some authors [2]. Because this study compared results performed on the same blood samples, it is unlikely that time of day would have influenced the correlation between aFT and cFT. Moreover, substantial blunting of diurnal T variation has been shown in men over 40 years [21–23], which should J Sex Med 2010;7:1948–1953
1952 minimize any impact of time of day on results presented in this study. A number of publications support the clinical relevance of aFT in men. Its use in clinical practice has been found helpful with the diagnosis and monitoring men with TD [13]. Men with characteristic symptoms of TD and aFT results <1.5 ng/dL had largely similar rates of symptomatic response to testosterone therapy regardless of whether TT was severely reduced (<200 ng/dL), more moderately reduced (200–300 ng/dL), or “normal” (>300 ng/dL) [24], suggesting that aFT results below this threshold may have greater clinical utility in the diagnosis of TD than TT, a concept that is consistent with consensus opinions regarding FT in general [1,2]. Other studies have found that men with aFT concentrations less than 1.0 ng/dL have a significantly increased risk of prostate cancer compared with men with milder FT reductions [25], and that aFT concentrations are significantly associated with severity of Peyronie’s disease [26]. These studies argue that aFT appears to have clinical and biological relevance. A number of criticisms of aFT appear in the literature. The Endocrine Society Guidelines assert that aFT results “are affected by alterations in SHBG and are inaccurate.” The single citation for this statement is the classic paper comparing various FT assays by Vermeulen et al. [9], who found a remarkably strong correlation between aFT and EqD of 0.937, albeit with numerically lower aFT values, as noted above. The impact of SHBG concentrations on aFT values noted by Vermeulen in a population of 28 men was weak (r = 0.348). A significant correlation between SHBG and aFT was also noted by Winters et al. [14] in 29 men and by Guay in 76 women [27]. However, no significant impact of SHBG on aFT results was noted by Okamura et al. [20] in a much larger population of 471 men, nor in the current study involving 140 samples. The explanation for these discrepant results is uncertain, but may possibly be influenced by gender and technical issues. In this regard, it is worth noting that aFT results in the study by Okamura et al. and in the present study were obtained from high-volume clinical laboratories. Moreover, the near-unity correlation of 0.97 for aFT and TT noted by Winters et al. [14] has not been reproduced in this or other studies [9,16], and suggests methodological differences in performance of the aFT assay. Despite recommendations against its use, the strong correlation between aFT and cFT suggests J Sex Med 2010;7:1948–1953
Moreno et al. that aFT may serve as a clinically useful test for FT. A lower set of reference values is required for aFT compared with cFT or EqD, and has already been in use by clinical laboratories for many years. Strengths of aFT include automation, speed, convenience, and reasonable cost. Further investigation of the clinical utility of aFT is warranted, particularly in prospective studies of defined groups such as hypogonadal men prior to and following treatment. In addition, it would be valuable to determine whether aFT can reliably identify men with symptoms of testosterone deficiency and predict clinical response to treatment, and to compare such results to prospectively obtained EqD values. Conclusion
In this clinical population of ambulatory men seen in a urological practice, aFT results correlated strongly with cFT results, and were independent of SHBG. Different sets of reference values are required for each test due to differences in numerical values. Corresponding Author: Abraham Morgentaler, MD, One Brookline Place, #624, Brookline, MA 02445, USA. Tel: 617-277-5000; Fax: 617-277-5444; E-mail: [email protected] Conflict of Interest: None.
Statement of Authorship
Category 1 (a) Conception and Design Abraham Morgentaler; Anita Shyam; Sergio A. Moreno (b) Acquisition of Data Abraham Morgentaler; Anita Shyam; Sergio A. Moreno (c) Analysis and Interpretation of Data Abraham Morgentaler; Anita Shyam; Sergio A. Moreno
Category 2 (a) Drafting the Article Abraham Morgentaler; Sergio A. Moreno (b) Revising It for Intellectual Content Abraham Morgentaler; Sergio A. Moreno
Category 3 (a) Final Approval of the Completed Article Abraham Morgentaler; Anita Shyam; Sergio A. Moreno
Analog and Calculated Free Testosterone References 1 Wang C, Nieschlag E, Swerdloff R, Behre HM, Hellstrom WJ, Gooren LJ, Kaufman JM, Legros JJ, Lunenfeld B, Morales A, Morley JE, Schulman C, Thompson IM, Weidner W, Wu FC. Investigation, treatment and monitoring of lateonset hypogonadism in males: ISA, ISSAM, EAU, EAA and ASA recommendations. Eur J Endocrinol 2008;159:507–14. 2 Bhasin S, Cunningham GR, Hayes FJ, Matsumoto AM, Snyder PJ, Swerdloff RS, Montori VM. Testosterone therapy in adult men with androgen deficiency syndromes: An Endocrine Society Clinical Practice Guideline. J Clin Endocrinol Metab 2006;91:1995–2010. 3 Shabsigh R, Arver S, Channer KS, Eardley I, Fabbri A, Gooren L, Heufelder A, Jones H, Meryn S, Zitzmann M. The triad of erectile dysfunction, hypogonadism and the metabolic syndrome. Int J Clin Pract 2008;62:791–8. 4 Guay A, Jacobson J. The relationship between testosterone levels, the metabolic syndrome (by two criteria), and insulin resistance in a population of men with organic erectile dysfunction. J Sex Med 2007;4:1046–55. 5 Dobs A. Testosterone deficiency and supplementation in chronic disease states. Med Clin N Am 2006;90(1 suppl):14– 8. 6 Rhoden EL, Morgentaler A. Risks of testosteronereplacement therapy and recommendations for monitoring. N Engl J Med 2004;350:482–92. 7 McCann DS, Kirkish LS. Evaluation of free testosterone in serum. J Clin Immunoassay 1985;8:234–6. 8 De Ronde W, van der Schouw YT, Pols HA, Gooren LJ, Muller M, Grobbee DE, de Jong FH. Calculation of bioavailable and free testosterone in men: A comparison of 5 published algorithms. Clin Chem 2006;52:1777–84. 9 Vermeulen A, Verdonck L, Kaufman JM. A critical evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab 1999;84:3666–72. 10 Lepage R. Measurement of testosterone and its sub-fractions in Canada. Clin Biochem 2006;39:97–108. 11 Lazarou S, Reyes-Vallejo L, Morgentaler A. Wide variability in laboratory reference values for serum testosterone. J Sex Med 2006;3:1085–9. 12 Rosner W, Auchus RJ, Azziz R, Sluss PM, Raff H. Position statement: Utility, limitations, and pitfalls in measuring testosterone: An Endocrine Society position statement. J Clin Endocrinol Metab 2007;92:405–13. 13 Morgentaler A. Commentary: Guideline for male testosterone therapy: A clinician’s perspective. J Clin Endocrinol Metab 2007;92:416–7.
1953 14 Winters SJ, Kelley DE, Goodpaster B. The analog free testosterone assay: Are the results in men clinically useful? Clin Chem 1998;44:2178–82. 15 Fritz KS, McKean AJ, Nelson JC, Wilcox RB. Analog-based free testosterone test results linked to total testosterone concentrations, not free testosterone concentrations. Clin Chem 2008;54:512–6. 16 Okamura K, Ando F, Shimokata H. Serum total and free testosterone level of Japanese men: A population-based study. Int J Urol 2005;12:810–4. 17 Carruthers M. The paradox dividing testosterone deficiency symptoms and androgen assays: A closer look at the cellular and molecular mechanisms of androgen action. J Sex Med 2008;5:998–1012. 18 Wu FC, Tajar A, Pye SR, Silman AJ, Finn JD, O’Neill TW, Bartfai G, Casanueva F, Forti G, Giwercman A, Huhtaniemi IT, Kula K, Punab M, Boonen S, Vanderschueren D, European Male Aging Study Group. Hypothalamic-pituitarytesticular axis disruptions in older men are differentially linked to age and modifiable risk factors: The European Male Aging Study. J Clin Endocrinol Metab 2008;93:2737–45. 19 Dawson B, Trapp RG. Basic and clinical biostatistics. 4th edition. Columbus, OH: McGraw-Hill Professional; 2004:48. 20 Nestler JE. Errors in the measurement of plasma free testosterone–Author’s response. J Clin Endocrinol Metab 1997;82:2015. 21 Bremner WJ, Vitiello MV, Prinz PN. Loss of circadian rhythmicity in blood testosterone levels with aging in normal men. J Clin Endocrinol Metab 1983;56:1278–81. 22 Guay A, Miller MG, McWhirter CL. Does early morning versus late morning draw time influence apparent testosterone concentration in men aged > or = 45 years? Data from the Hypogonadism In Males study. Int J Impot Res 2008;20: 162–7. 23 Crawford ED, Barqawi AB, O’Donnell C, Morgentaler A. The association of time of day and serum testosterone concentration in a large screening population. BJU Int. 2007;100:509–13. 24 Reyes-Vallejo L, Lazarou S, Morgentaler A. Subjective sexual response to testosterone replacement therapy based on initial serum levels of total testosterone. J Sex Med 2007;4:1757–62. 25 Morgentaler A, Rhoden EL. Prevalence of prostate cancer among hypogonadal men with prostate-specific antigen of 4.0 ng/mL or less. Urology 2006;68:1263–7. 26 Moreno SA, Morgentaler A. Testosterone deficiency and Peyronie’s disease: Pilot data suggesting a significant relationship. J Sex Med. 2009;6:1729–35. 27 Guay A. Screening for androgen deficiency in women: Methodological and interpretive issues. Fertil Steril 2002;77:S83–8.
J Sex Med 2010;7:1948–1953
ORIGINAL RESEARCH—ENDOCRINOLOGY Comparison of Free Testosterone Results by Analog Radioimmunoassay and Calculated Free Testosterone in an Ambulatory Clinical Population jsm_1473
1948..1953
Sergio A. Moreno, MD, Anita Shyam, MD, and Abraham Morgentaler, MD Men’s Health Boston, Harvard Medical School, Boston, MA, USA DOI: 10.1111/j.1743-6109.2009.01473.x
ABSTRACT
Introduction. The most widely used method for measuring free testosterone (FT) is by analog immunoassay (aFT); however, this assay has been criticized as unreliable based on laboratory studies in small groups of men. Calculated FT (cFT), derived from total testosterone (TT) and sex-hormone binding globulin (SHBG) values has been recommended in its place. There are limited data comparing aFT and cFT in clinical populations. Aim. The purpose of this study was to compare aFT with cFT in a population of ambulatory men in a clinical setting. Methods. Medical records were reviewed for 100 randomly selected men in a urology practice, yielding 140 test results complete for TT, aFT, and SHBG. Calculated FT was determined via an online calculator. Comparisons were made with Pearson rank coefficients. Main Outcome Measures. Pearson rank correlation between aFT and cFT. Results. Mean patient age was 52.3 ⫾ 14.3 years (range 24–80). Mean TT was 443.0 ⫾ 208.3 ng/dL (range 110– 1276). Mean aFT was 1.22 ⫾ 0.54 ng/dL (range 0.24–3.8) and mean cFT 9.4 ⫾ 4.5 ng/dL (range 1.8–27.8). Mean SHBG was 34.2 ⫾ 19.5 nmol/L (range 9–127). A strong correlation was observed for aFT and cFT (r = 0.88, P < 0.0001), particularly at low concentrations. Significant correlations were also noted between aFT and TT (r = 0.73, P < 0.0001), and between cFT and TT (r = 0.82, P < 0.0001). Numerical values for aFT were approximately one-eighth of the values obtained for cFT. Neither aFT nor cFT correlated with SHBG. Conclusions. A strong correlation was observed between aFT and cFT in this clinical population of ambulatory men. Different sets of reference values must be applied for each of these tests. Moreno SA, Shyam A, and Morgentaler A. Comparison of free testosterone results by analog radioimmunoassay and calculated free testosterone in an ambulatory clinical population. J Sex Med 2010;7:1948–1953. Key Words. Testosterone; Hypogonadism; Testosterone Deficiency; Analog Free Testosterone; Calculated Free Testosterone; Testosterone Blood Testing
Introduction
T
here is a growing appreciation of the importance of testosterone deficiency (TD) in men [1–6]. Traditionally, the diagnosis has been made on the basis of serum total testosterone (TT); however, there is a general consensus that serum free testosterone (FT) may more accurately reflect bioavailable T. Bioavailable T consists of the free and albumin-bound fractions of testosterone, and although bioavailable T can also be measured or J Sex Med 2010;7:1948–1953
calculated, much of the focus in recent years has been on FT. Yet, there is considerable uncertainty regarding the reliability and utility of currently available FT assays, particularly the “analog” radioimmunoassay (aFT) available in most laboratories. Equilibrium dialysis (EqD) has been considered the “gold standard” technique for determining FT concentrations [7]; however, this technique is time-consuming, non-automated, labor-intensive, and expensive [8,9]. For this reason, use of EqD © 2009 International Society for Sexual Medicine
1949
Analog and Calculated Free Testosterone has been largely restricted to research laboratories. Only 7 out of 122 laboratories reported using EqD in the September 2004 Ligand Special Survey of the College of American Pathology [10]. A survey of academic and community hospital laboratories in six U.S. states revealed that none of 24 offered EqD, and it was only made available from a national reference laboratory upon special request [11]. The most widely used method for measuring FT in the United States is by aFT, with 70% of U.S. laboratories reporting use of this assay [10]. However, the Endocrine Society and a group of international andrology societies have discouraged use of aFT, asserting it is inaccurate as values obtained by this assay are considerably lower than those obtained via EqD [1,2]. Instead, these groups have recommended use of calculated FT (cFT) for clinical practice, based on its strong correlation with EqD [1,2,12]. Results for cFT are obtained by a complex equation involving concentrations for TT, sex hormone binding globulin (SHBG), and albumin, but can more easily be obtained via an online calculator. The calculation is simplified by the use of a constant value for albumin (4.3 g/dL) in place of measured concentrations, without negatively affecting the correlation of cFT results with those obtained by EqD [9]. The criticism of aFT’s reliability despite its widespread use has created confusion for clinicians, some of whom have argued that aFT provides valuable clinical information regarding testosterone status in the adult man [13]. Moreover, in the landmark study by Vermeulen et al. in which various FT assays were directly compared, the correlation of aFT with EqD approached unity (r = 0.937), and was exceeded by only a minor degree by cFT (r = 0.987) [9]. However, numerical values for aFT were a fraction of those obtained by cFT and EqD, leading to the conclusion that this assay provided inaccurate results. Additional studies performed in small numbers of men in research laboratories have suggested that aFT provided no independent information from TT [14], and that it may not even detect dialyzable T [15]. These studies would appear to support the assertion that aFT may not be clinically useful. On the other hand, a literature search identified only a single publication that directly compared aFT and cFT results, demonstrating a strong correlation (r > 0.80) in 471 Japanese men aged 40–79 years [16].
The goal of this study was to examine the relationship of aFT, cFT, and TT results in a population of ambulatory men seen in a urological clinical practice. Methods
We retrospectively reviewed the medical records of 100 randomly selected men seen in a urology practice specializing in andrology, providing a total of 140 sets of test results that were complete for TT, aFT, and SHBG. The maximum number of samples from any one individual was three. Socio-demographic data and medical history were obtained. Blood samples were obtained during clinic hours of 8:00 am to 5:00 pm. Commercial immunoassays were used for measurement of TT (Tosoh Bioscience, San Francisco, CA, USA) and FT (Diagnostic Products Corp., Los Angeles, CA, USA). SHBG was measured by Immulite/ Chemiluminescence (Diagnostic Products Corp., Los Angeles, CA, USA). Results for cFT were obtained by using the online FT calculator at http://www.issam.ch/freetesto.htm. A standard value of 4.3 g/dL was used for albumin. Tests for TT were performed in-house, while aFT and SHBG were performed by a national reference laboratory (Quest Diagnostics). Statistical analyses were performed with Minitab statistical program version 15 (Minitab Inc., State College, PA, USA). Pearson’s rank test coefficients were used to evaluate relationships between continuous variables. All results are expressed as means ⫾ standard deviation. A P value of <0.05 was considered significant. Results
Mean patient age was 52.3 ⫾ 14.3 years (range 24–80). Erectile dysfunction was reported by 51 patients (51%). The most common additional medical conditions were hypertension (32%), infertility (21%), and dyslipidemia (19%). Fortythree patients (43%) were receiving testosterone therapy when blood tests were obtained. No patient in this study had significant renal or hepatic disease. The characteristics of the study population are presented in Table 1. Mean TT was 443.0 ⫾ 208.3 ng/dL (range 110–1276). Mean aFT was 1.22 ⫾ 0.54 ng/dL (range 0.24–3.8), and mean cFT 9.4 ⫾ 4.5 ng/dL (range 1.8–27.8; multiply by 34.7 to convert to pmol/L). The mean J Sex Med 2010;7:1948–1953
1950 Table 1
Moreno et al. General characteristics of population sample
Characteristics
Value*
Subjects Age (years) Erectile dysfunction Comorbidities Hypertension Infertility Dyslipidemia Diabetes mellitus Coronary artery disease Hormonal profile TT (ng/dL) aFT (ng/dL) cFT (ng/dL) SHBG (nmol/L)
100 52.3 ⫾ 14.3 (24–80) 51 (51) 32 21 19 7 4
(32) (21) (19) (7) (4)
Discussion
There is a great need to have a simple, readily available blood test to identify men with symptomatic TD [17]. Although TT has been used for this purpose, it is now recognized that TT may inaccurately reflect bioavailable serum T in some men
443.0 ⫾ 208.3 (110–1276) 1.22 ⫾ 0.54 (0.24–3.8) 9.4 ⫾ 4.5 (1.8–27.8) 34.2 ⫾ 19.5 (9–127)
*Values expressed as means ⫾ standard deviation (range), frequency (%). TT = total testosterone; aFT = analog free testosterone; cFT = calculated free testosterone; SHBG = sex hormone binding globulin.
SHBG was 34.2 ⫾ 19.5 nmol/L (range 9–127) and the median was 30 nmol/L. Figure 1 shows a strong correlation between aFT and cFT (r = 0.88, P < 0.0001). Examination of the scatter plot reveals a particularly tight correlation at lower values, with greater dispersion at higher concentrations. For aFT values above the median, the correlation with cFT was r = 0.81 (P < 0.0001), and for values below the median, it was r = 0.73 (P < 0.0001). Numerical values for aFT were approximately one-eighth of the values obtained for cFT. Significant correlations were also noted between aFT and TT (r = 0.73, P < 0.0001; Figure 2), and between cFT and TT (r = 0.82, P < 0.0001; Figure 3). No significant correlation was noted between aFT and SHBG (r = -0.13, P = 0.141), or cFT and SHBG (r = -0.12, P = 0.168). However, a significant correlation was observed between TT and SHBG (r = 0.41, P < 0.0001). A sub-analysis was performed by dividing the population into two groups based on SHBG values below or above the median to investigate how these tests correlated in men with relatively high or low SHBG concentrations. Numerically stronger correlations were noted in men in the lower SHBG sub-population. In men with SHBG below the median, the correlation between aFT and cFT was r = 0.90 (P < 0.0001); aFT and TT, r = 0.87 (P < 0.0001); and cFT and TT, r = 0.97 (P < 0.0001). For men with SHBG over the median, the corresponding r values were: aFT and cFT, r = 0.85 (P < 0.0001); aFT and TT, r = 0.76 (P < 0.0001); and cFT and TT, r = 0.85 (P < 0.0001). J Sex Med 2010;7:1948–1953
Figure 1 Analog free testosterone (afT) versus calculated free testosterone (cFT).
Figure 2 Analog free testosterone (aFT) versus total testosterone (TT).
Figure 3 Calculated free testosterone (cFT) versus total testosterone (TT).
Analog and Calculated Free Testosterone because of its tight binding to SHBG, which can vary considerably in concentration from one individual to another. A number of factors may increase SHBG, most notably age, and elevated levels of SHBG may produce normal-appearing TT results even when bioavailable T is depressed. This effect is illustrated by the minor decrease in TT seen with age, in contrast to a substantial decline in FT [18]. This study compared results from the most widely used FT assay, aFT, with results from the FT assay recommended by the Endocrine Society and others, cFT. The recommendation to use cFT clinically is based on studies showing a strong correlation with the “gold standard” technique of EqD, which is impractical for clinical practice and not widely available. The goal of this study was straightforward: How well do results by aFT compare with results from an established, recommended test, i.e. cFT? The primary results of this study revealed a very strong correlation of aFT with cFT (r = 0.88, P < 0.0001). Statistical texts indicate that correlations greater than 0.75 represent “very good to excellent relationships.” [19] Higher or lower values noted by cFT were associated with proportionately higher or lower values by aFT. Visual investigation of the scatterplot comparing aFT and cFT revealed particularly tight correlations at low concentrations, where the diagnosis of TD is made, and slightly greater dispersion at higher concentrations. Weaker but significant correlations were noted between aFT and TT, and cFT and TT. The strong relationship between aFT and cFT appears to be robust, as the study population included men with a wide range of total testosterone values (110–1276 ng/dL), and a broad age range (24–80 years). Moreover, this study included men who were deficient in T at the time of blood testing as well as men with high T values while receiving T therapy. Further investigation revealed that the strong correlation between aFT and cFT held true for men with relatively high or relatively low SHBG, defined as values above or below the median. No significant relationship was noted for either aFT or cFT results with SHBG concentrations, which is reassuring for a clinical test intended to avoid the confounding influence of SHBG on the interpretation of TT results. Numerical results for aFT were approximately one-eighth the values for cFT. This observation has been made previously, and this unfortunate aspect of the aFT assay has resulted in the criticism
1951 that the aFT assay is inaccurate. However, the strong correlation indicates this discrepancy is a calibration issue, one that has already been corrected in clinical practice by the use of a different set of reference values. Because the aFT assay is proprietary (originally Diagnostic Products Corp, now Siemens), it is unknown how numerical values were originally assigned when the test was introduced several decades ago. One may speculate that if the test were introduced today, with FT norms already established by EqD, numerical aFT results would have been calibrated differently to provide values more consistent with those obtained by cFT or EqD. The data in this study were obtained during routine clinical practice, with blood tests for aFT and SHBG performed by a national laboratory that performs many thousands of these tests annually. These results should thus have wide clinical applicability, as similar results would presumably be obtained by a clinician practicing anywhere in the United States. This differs from research studies of aFT that have been performed in small numbers of men under laboratory conditions that may differ from the automated, controlled environment of a large commercial laboratory. The results of this study are consistent with a study of 471 Japanese men that revealed a strong correlation between aFT and cFT with an r value of 0.803 [16]. In that study, the two tests had a 98.7% correspondence for results more than two standard deviations below the mean, representing the standard range categorized as abnormally low [19]. In the current study, there was 100% concordance between aFT and cFT values that fell in this range. Others have also noted a strong and highly significant correlation between aFT and cFT [20]. It is important to note that cFT values in this study were obtained using a standard value of 4.3 g/dL for albumin, which has been shown in healthy men to provide accurate results [9]. It is unknown how the correlation between aFT and cFT would have been influenced by the use of actual albumin values, which may vary considerably in clinical populations. A technical consideration is that blood tests were obtained throughout the day and were not restricted to the early morning, as recommended by some authors [2]. Because this study compared results performed on the same blood samples, it is unlikely that time of day would have influenced the correlation between aFT and cFT. Moreover, substantial blunting of diurnal T variation has been shown in men over 40 years [21–23], which should J Sex Med 2010;7:1948–1953
1952 minimize any impact of time of day on results presented in this study. A number of publications support the clinical relevance of aFT in men. Its use in clinical practice has been found helpful with the diagnosis and monitoring men with TD [13]. Men with characteristic symptoms of TD and aFT results <1.5 ng/dL had largely similar rates of symptomatic response to testosterone therapy regardless of whether TT was severely reduced (<200 ng/dL), more moderately reduced (200–300 ng/dL), or “normal” (>300 ng/dL) [24], suggesting that aFT results below this threshold may have greater clinical utility in the diagnosis of TD than TT, a concept that is consistent with consensus opinions regarding FT in general [1,2]. Other studies have found that men with aFT concentrations less than 1.0 ng/dL have a significantly increased risk of prostate cancer compared with men with milder FT reductions [25], and that aFT concentrations are significantly associated with severity of Peyronie’s disease [26]. These studies argue that aFT appears to have clinical and biological relevance. A number of criticisms of aFT appear in the literature. The Endocrine Society Guidelines assert that aFT results “are affected by alterations in SHBG and are inaccurate.” The single citation for this statement is the classic paper comparing various FT assays by Vermeulen et al. [9], who found a remarkably strong correlation between aFT and EqD of 0.937, albeit with numerically lower aFT values, as noted above. The impact of SHBG concentrations on aFT values noted by Vermeulen in a population of 28 men was weak (r = 0.348). A significant correlation between SHBG and aFT was also noted by Winters et al. [14] in 29 men and by Guay in 76 women [27]. However, no significant impact of SHBG on aFT results was noted by Okamura et al. [20] in a much larger population of 471 men, nor in the current study involving 140 samples. The explanation for these discrepant results is uncertain, but may possibly be influenced by gender and technical issues. In this regard, it is worth noting that aFT results in the study by Okamura et al. and in the present study were obtained from high-volume clinical laboratories. Moreover, the near-unity correlation of 0.97 for aFT and TT noted by Winters et al. [14] has not been reproduced in this or other studies [9,16], and suggests methodological differences in performance of the aFT assay. Despite recommendations against its use, the strong correlation between aFT and cFT suggests J Sex Med 2010;7:1948–1953
Moreno et al. that aFT may serve as a clinically useful test for FT. A lower set of reference values is required for aFT compared with cFT or EqD, and has already been in use by clinical laboratories for many years. Strengths of aFT include automation, speed, convenience, and reasonable cost. Further investigation of the clinical utility of aFT is warranted, particularly in prospective studies of defined groups such as hypogonadal men prior to and following treatment. In addition, it would be valuable to determine whether aFT can reliably identify men with symptoms of testosterone deficiency and predict clinical response to treatment, and to compare such results to prospectively obtained EqD values. Conclusion
In this clinical population of ambulatory men seen in a urological practice, aFT results correlated strongly with cFT results, and were independent of SHBG. Different sets of reference values are required for each test due to differences in numerical values. Corresponding Author: Abraham Morgentaler, MD, One Brookline Place, #624, Brookline, MA 02445, USA. Tel: 617-277-5000; Fax: 617-277-5444; E-mail: [email protected] Conflict of Interest: None.
Statement of Authorship
Category 1 (a) Conception and Design Abraham Morgentaler; Anita Shyam; Sergio A. Moreno (b) Acquisition of Data Abraham Morgentaler; Anita Shyam; Sergio A. Moreno (c) Analysis and Interpretation of Data Abraham Morgentaler; Anita Shyam; Sergio A. Moreno
Category 2 (a) Drafting the Article Abraham Morgentaler; Sergio A. Moreno (b) Revising It for Intellectual Content Abraham Morgentaler; Sergio A. Moreno
Category 3 (a) Final Approval of the Completed Article Abraham Morgentaler; Anita Shyam; Sergio A. Moreno
Analog and Calculated Free Testosterone References 1 Wang C, Nieschlag E, Swerdloff R, Behre HM, Hellstrom WJ, Gooren LJ, Kaufman JM, Legros JJ, Lunenfeld B, Morales A, Morley JE, Schulman C, Thompson IM, Weidner W, Wu FC. Investigation, treatment and monitoring of lateonset hypogonadism in males: ISA, ISSAM, EAU, EAA and ASA recommendations. Eur J Endocrinol 2008;159:507–14. 2 Bhasin S, Cunningham GR, Hayes FJ, Matsumoto AM, Snyder PJ, Swerdloff RS, Montori VM. Testosterone therapy in adult men with androgen deficiency syndromes: An Endocrine Society Clinical Practice Guideline. J Clin Endocrinol Metab 2006;91:1995–2010. 3 Shabsigh R, Arver S, Channer KS, Eardley I, Fabbri A, Gooren L, Heufelder A, Jones H, Meryn S, Zitzmann M. The triad of erectile dysfunction, hypogonadism and the metabolic syndrome. Int J Clin Pract 2008;62:791–8. 4 Guay A, Jacobson J. The relationship between testosterone levels, the metabolic syndrome (by two criteria), and insulin resistance in a population of men with organic erectile dysfunction. J Sex Med 2007;4:1046–55. 5 Dobs A. Testosterone deficiency and supplementation in chronic disease states. Med Clin N Am 2006;90(1 suppl):14– 8. 6 Rhoden EL, Morgentaler A. Risks of testosteronereplacement therapy and recommendations for monitoring. N Engl J Med 2004;350:482–92. 7 McCann DS, Kirkish LS. Evaluation of free testosterone in serum. J Clin Immunoassay 1985;8:234–6. 8 De Ronde W, van der Schouw YT, Pols HA, Gooren LJ, Muller M, Grobbee DE, de Jong FH. Calculation of bioavailable and free testosterone in men: A comparison of 5 published algorithms. Clin Chem 2006;52:1777–84. 9 Vermeulen A, Verdonck L, Kaufman JM. A critical evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab 1999;84:3666–72. 10 Lepage R. Measurement of testosterone and its sub-fractions in Canada. Clin Biochem 2006;39:97–108. 11 Lazarou S, Reyes-Vallejo L, Morgentaler A. Wide variability in laboratory reference values for serum testosterone. J Sex Med 2006;3:1085–9. 12 Rosner W, Auchus RJ, Azziz R, Sluss PM, Raff H. Position statement: Utility, limitations, and pitfalls in measuring testosterone: An Endocrine Society position statement. J Clin Endocrinol Metab 2007;92:405–13. 13 Morgentaler A. Commentary: Guideline for male testosterone therapy: A clinician’s perspective. J Clin Endocrinol Metab 2007;92:416–7.
1953 14 Winters SJ, Kelley DE, Goodpaster B. The analog free testosterone assay: Are the results in men clinically useful? Clin Chem 1998;44:2178–82. 15 Fritz KS, McKean AJ, Nelson JC, Wilcox RB. Analog-based free testosterone test results linked to total testosterone concentrations, not free testosterone concentrations. Clin Chem 2008;54:512–6. 16 Okamura K, Ando F, Shimokata H. Serum total and free testosterone level of Japanese men: A population-based study. Int J Urol 2005;12:810–4. 17 Carruthers M. The paradox dividing testosterone deficiency symptoms and androgen assays: A closer look at the cellular and molecular mechanisms of androgen action. J Sex Med 2008;5:998–1012. 18 Wu FC, Tajar A, Pye SR, Silman AJ, Finn JD, O’Neill TW, Bartfai G, Casanueva F, Forti G, Giwercman A, Huhtaniemi IT, Kula K, Punab M, Boonen S, Vanderschueren D, European Male Aging Study Group. Hypothalamic-pituitarytesticular axis disruptions in older men are differentially linked to age and modifiable risk factors: The European Male Aging Study. J Clin Endocrinol Metab 2008;93:2737–45. 19 Dawson B, Trapp RG. Basic and clinical biostatistics. 4th edition. Columbus, OH: McGraw-Hill Professional; 2004:48. 20 Nestler JE. Errors in the measurement of plasma free testosterone–Author’s response. J Clin Endocrinol Metab 1997;82:2015. 21 Bremner WJ, Vitiello MV, Prinz PN. Loss of circadian rhythmicity in blood testosterone levels with aging in normal men. J Clin Endocrinol Metab 1983;56:1278–81. 22 Guay A, Miller MG, McWhirter CL. Does early morning versus late morning draw time influence apparent testosterone concentration in men aged > or = 45 years? Data from the Hypogonadism In Males study. Int J Impot Res 2008;20: 162–7. 23 Crawford ED, Barqawi AB, O’Donnell C, Morgentaler A. The association of time of day and serum testosterone concentration in a large screening population. BJU Int. 2007;100:509–13. 24 Reyes-Vallejo L, Lazarou S, Morgentaler A. Subjective sexual response to testosterone replacement therapy based on initial serum levels of total testosterone. J Sex Med 2007;4:1757–62. 25 Morgentaler A, Rhoden EL. Prevalence of prostate cancer among hypogonadal men with prostate-specific antigen of 4.0 ng/mL or less. Urology 2006;68:1263–7. 26 Moreno SA, Morgentaler A. Testosterone deficiency and Peyronie’s disease: Pilot data suggesting a significant relationship. J Sex Med. 2009;6:1729–35. 27 Guay A. Screening for androgen deficiency in women: Methodological and interpretive issues. Fertil Steril 2002;77:S83–8.
J Sex Med 2010;7:1948–1953